Using SQLite Using SQLite Jay A. Kreibich Beijing • Cambridge • Farnham • Köln • Sebastopol • Taipei • Tokyo Using SQLite by Jay A. Kreibich Copyright © 2010 Jay A. Kreibich. All rights reserved. Printed in the United States of America. Published by O’Reilly Media, Inc., 1005 Gravenstein Highway North, Sebastopol, CA 95472. O’Reilly books may be purchased for educational, business, or sales promotional use. Online editions are also available for most titles (http://my.safaribooksonline.com). For more information, contact our corporate/institutional sales department: (800) 998-9938 or corporate@oreilly.com. Editor: Mike Loukides Production Editor: Kristen Borg Proofreader: Kiel Van Horn Indexer: Lucie Haskins Cover Designer: Karen Montgomery Interior Designer: David Futato Illustrator: Robert Romano Printing History: August 2010: First Edition. Nutshell Handbook, the Nutshell Handbook logo, and the O’Reilly logo are registered trademarks of O’Reilly Media, Inc. Using SQLite, the image of a great white heron, and related trade dress are trade- marks of O’Reilly Media, Inc. Many of the designations used by manufacturers and sellers to distinguish their products are claimed as trademarks. Where those designations appear in this book, and O’Reilly Media, Inc., was aware of a trademark claim, the designations have been printed in caps or initial caps. While every precaution has been taken in the preparation of this book, the publisher and author assume no responsibility for errors or omissions, or for damages resulting from the use of the information con- tained herein. ISBN: 978-0-596-52118-9 [M] 1281104401 To my Great-Uncle Albert “Unken Al” Kreibich. 1918–1994 He took a young boy whose favorite question was “why?” and taught him to ask the question “how?” (Who also—much to the dismay of his parents and the kitchen telephone—taught him the joy of answering that question, especially if it involved pliers or screwdrivers.) —jk Table of Contents Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv 1. What Is SQLite? ......................................................... 1 Self-Contained, No Server Required 2 Single File Database 4 Zero Configuration 4 Embedded Device Support 5 Unique Features 5 Compatible License 6 Highly Reliable 6 2. Uses of SQLite . . . ....................................................... 9 Database Junior 9 Application Files 10 Application Cache 11 Archives and Data Stores 11 Client/Server Stand-in 11 Teaching Tool 12 Generic SQL Engine 13 Not the Best Choice 13 Big Name Users 15 3. Building and Installing SQLite . . . ......................................... 17 SQLite Products 17 Precompiled Distributions 18 Documentation Distribution 18 Source Distributions 19 The Amalgamation 19 Source Files 19 Source Downloads 20 Building 21 vii Configure 21 Manually 22 Build Customization 23 Build and Installation Options 23 An sqlite3 Primer 24 Summary 26 4. The SQL Language ..................................................... 27 Learning SQL 27 Brief Background 28 Declarative 28 Portability 29 General Syntax 30 Basic Syntax 30 Three-Valued Logic 31 Simple Operators 33 SQL Data Languages 34 Data Definition Language 34 Tables 35 Views 43 Indexes 44 Data Manipulation Language 45 Row Modification Commands 46 The Query Command 49 Transaction Control Language 51 ACID Transactions 51 SQL Transactions 53 Save-Points 55 System Catalogs 57 Wrap-up 58 5. The SELECT Command . . . ............................................... 61 SQL Tables 61 The SELECT Pipeline 62 FROM Clause 63 WHERE Clause 68 GROUP BY Clause 69 SELECT Header 70 HAVING Clause 73 DISTINCT Keyword 74 ORDER BY Clause 74 LIMIT and OFFSET Clauses 75 Advanced Techniques 76 viii | Table of Contents Subqueries 76 Compound SELECT Statements 77 Alternate JOIN Notation 78 SELECT Examples 79 Simple SELECTs 80 Simple JOINs 80 JOIN...ON 81 JOIN...USING, NATURAL JOIN 82 OUTER JOIN 82 Compound JOIN 82 Self JOIN 83 WHERE Examples 83 GROUP BY Examples 84 ORDER BY Examples 85 What’s Next 85 6. Database Design ... .................................................... 87 Tables and Keys 87 Keys Define the Table 87 Foreign Keys 89 Foreign Key Constraints 90 Generic ID Keys 91 Keep It Specific 92 Common Structures and Relationships 93 One-to-One Relationships 93 One-to-Many Relationships 95 Many-to-Many Relationships 97 Hierarchies and Trees 99 Normal Form 102 Normalization 103 Denormalization 103 The First Normal Form 104 The Second Normal Form 104 The Third Normal Form 105 Higher Normal Forms 106 Indexes 107 How They Work 107 Must Be Diverse 108 INTEGER PRIMARY KEYs 109 Order Matters 109 One at a Time 110 Index Summary 111 Transferring Design Experience 112 Table of Contents | ix Tables Are Types 112 Keys Are Backwards Pointers 113 Do One Thing 113 Closing 114 7. C Programming Interface .............................................. 115 API Overview 115 Structure 116 Strings and Unicode 117 Error Codes 118 Structures and Allocations 118 More Info 119 Library Initialization 119 Database Connections 120 Opening 120 Special Cases 121 Closing 122 Example 122 Prepared Statements 123 Statement Life Cycle 123 Prepare 124 Step 126 Result Columns 127 Reset and Finalize 130 Statement Transitions 131 Examples 132 Bound Parameters 133 Parameter Tokens 133 Binding Values 135 Security and Performance 138 Example 140 Potential Pitfalls 141 Convenience Functions 142 Result Codes and Error Codes 146 Standard Codes 146 Extended Codes 148 Error Functions 148 Prepare v2 149 Transactions and Errors 150 Database Locking 151 Utility Functions 156 Version Management 156 Memory Management 157 x | Table of Contents Summary 158 8. Additional Features and APIs . . . ........................................ 159 Date and Time Features 159 Application Requirements 160 Representations 160 Time and Date Functions 162 ICU Internationalization Extension 167 Full-Text Search Module 169 Creating and Populating FTS Tables 169 Searching FTS Tables 170 More Details 171 R*Trees and Spatial Indexing Module 171 Scripting Languages and Other Interfaces 172 Perl 172 PHP 173 Python 173 Java 174 Tcl 174 ODBC 175 .NET 175 C++ 175 Other Languages 176 Mobile and Embedded Development 176 Memory 176 Storage 177 Other Resources 178 iPhone Support 178 Other Environments 179 Additional Extensions 180 9. SQL Functions and Extensions . . . ........................................ 181 Scalar Functions 182 Registering Functions 182 Extracting Parameters 184 Returning Results and Errors 186 Example 189 Aggregate Functions 194 Defining Aggregates 194 Aggregate Context 195 Example 197 Collation Functions 200 Registering a Collation 201 Table of Contents | xi Collation Example 202 SQLite Extensions 204 Extension Architecture 205 Extension Design 206 Example Extension: sql_trig 207 Building and Integrating Static Extensions 209 Using Loadable Extensions 211 Building Loadable Extensions 212 Loadable Extension Security 213 Loading Loadable Extensions 213 Multiple Entry Points 215 Chapter Summary 215 10. Virtual Tables and Modules ............................................. 217 Introduction to Modules 218 Internal Modules 218 External Modules 218 Example Modules 219 SQL for Anything 219 Module API 220 Simple Example: dblist Module 224 Create and Connect 224 Disconnect and Destroy 229 Query Optimization 230 Custom Functions 231 Table Rename 232 Opening and Closing Table Cursors 233 Filtering Rows 235 Extracting and Returning Data 237 Virtual Table Modifications 239 Cursor Sequence 240 Transaction Control 241 Register the Module 243 Example Usage 245 Advanced Example: weblog Module 246 Create and Connect 248 Disconnect and Destroy 249 Other Table Functions 250 Open and Close 250 Filter 252 Rows and Columns 254 Register the Module 259 Example Usage 259 xii | Table of Contents Best Index and Filter 262 Purpose and Need 262 xBestIndex() 263 xFilter() 266 Typical Usage 267 Wrap-Up 268 A. SQLite Build Options ................................................... 269 B. sqlite3 Command Reference . . . ......................................... 287 C. SQLite SQL Command Reference . . . ...................................... 299 D. SQLite SQL Expression Reference . . . ..................................... 341 E. SQLite SQL Function Reference . . . ....................................... 361 F. SQLite SQL PRAGMA Reference . . . ....................................... 381 G. SQLite C API Reference . . . .............................................. 409 Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 491 Table of Contents | xiii Preface This book provides an introduction to the SQLite database product. SQLite is a zero- configuration, standalone, relational database engine that is designed to be embedded directly into an application. Database instances are self-contained within a single file, allowing easy transport and simple setup. Using SQLite is primarily written for experienced software developers that have never had a particular need to learn about relational databases. For one reason or another, you now find yourself with a large data management task, and are hoping a product like SQLite may provide the answer. To help you out, the various chapters cover the SQL language, the SQLite C programming API, and the basics of relational database design, giving you everything you need to successfully integrate SQLite into your ap- plications and development work. The book is divided into two major sections. The first part is a traditional set of chapters that are primarily designed to be read in order. The first two chapters provide an in- depth look at exactly what SQLite provides and how it can be used. The third chapter covers downloading and building the library. Chapters Four and Five provide an in- troduction to the SQL language, while Chapter Six covers database design concepts. Chapter Seven covers the basics of the C API. Chapter Eight builds on that to cover more advanced topics, such as storing times and dates, using SQLite from scripting languages, and utilizing some of the more advanced extensions. Chapters Nine and Ten cover writing your own custom SQL functions, extensions, and modules. To complete the picture, the ten chapters are followed by several reference appendixes. These references cover all of the SQL commands, expressions, and built-in functions supported by SQLite, as well as documentation for the complete SQLite API. SQLite Versions The first edition of this book coves SQLite version 3.6.23.1. As this goes to press, work on SQLite version 3.7 is being finalized. SQLite 3.7 introduces a new transaction journal mode known as Write Ahead Logging, or WAL. In some environments, WAL can pro- vide better concurrent transaction performance than the current rollback journal. This xv performance comes at a cost, however. WAL has more restrictive operational require- ments and requires more advanced support from the operating system. Once WAL has been fully tested and released, look for an article on the O’Reilly website that covers this new feature and how to get the most out of it. Email Lists The SQLite project maintains three mailing lists. If you’re trying to learn more about SQLite, or have any questions that are not addressed in this book or in the project documentation, these are often a good place to start. sqlite-announce@sqlite.org This list is limited to announcements of new releases, critical bug alerts, and other significant events in the SQLite community. Traffic is extremely low, and most messages are posted by the SQLite development team. sqlite-users@sqlite.org This is the main support list for SQLite. It covers a broad range of topics, including SQL questions, programming questions, and questions about how the library works. This list is moderately busy. sqlite-dev@sqlite.org This list is for people working on the internal code of the SQLite library itself. If you have questions about how to use the published SQLite API, those questions belong on the sqlite-users list. Traffic on this list is fairly low. You can find instructions on how to join these mailing lists on the SQLite website. Visit http://www.sqlite.org/support.html for more details. The sqlite-users@sqlite.org email list can be quite helpful, but it is a moderately busy list. If you’re only a casual user and don’t wish to receive that much email, you can also access and search list messages through a web archive. Links to several different archives are available on the SQLite support page. Example Code Download The code examples found in this book are available for download from the O’Reilly website. You can find a link to the examples on the book’s catalog page at http://oreilly .com/catalog/9780596521196/. The files include both the SQL examples and the C examples found in later chapters. xvi | Preface How We Got Here Taking a book from an idea to a finished product involves a great many people. Al- though my name is on the cover, this could not have been possible without all of their help. First, I would like to acknowledge the friendship and support of my primary editor, Mike Loukides. Thanks to some mutual friends, I first started doing technical reviews for Mike over eight years ago. Through the years, Mike gently encouraged me to take on my own project. The first step on that path came nearly three years ago. I had downloaded a set of database exports from the Wikipedia project and was trying to devise a minimal data- base configuration that would (hopefully) cram nearly all the current data onto a small flash storage card. The end goal was to provide a local copy of the Wikipedia articles on an ebook reader I had. SQLite was a natural choice. At some point, frustrated with trying to understand the correct call sequence, I threw my hands up and exclaimed, “Someone should write a book about this!”—Ding!—The proverbial light bulb went off, and many, many (many…) late nights later, here we are. Behind Mike stands the whole staff of O’Reilly Media. Everyone I interacted with did their best to help me out, calm me down, and fix my problems—sometimes all at once. The production staff understands how to make life easy for the author, so that we can focus on writing and leave the details to someone else. I would like to thank D. Richard Hipp, the creator and lead maintainer of SQLite. In addition to coordinating the continued development of SQLite and providing us all with a high-quality software product, he was also gracious enough to answer numerous questions, as well as review a final draft of the manuscript. Some tricky spots went through several revisions, and he was always quick to review things and get back to me with additional comments. A technical review was also done by Jon W. Marks. Jon is an old personal and profes- sional friend with enterprise-class database experience. He has had the opportunity to mentor several experienced developers as they made their first journey into the rela- tional database world. Jon provided very insightful feedback, and was able to pinpoint areas that are often difficult for beginners to grasp. My final two technical reviewers were Jordan Hawker and Erin Moy. Although they are knowledgeable developers, they were relatively new to relational databases. As they went through the learning process, they kept me honest when I started to make too many assumptions, and kept me on track when I started to skip ahead too quickly. Preface | xvii I also owe a thank-you to Mike Kulas and all my coworkers at Volition, Inc. In addition to helping me find the right balance between my professional work and the book work, Mike helped me navigate our company’s intellectual property policies, making sure everything was on the straight and narrow. Numerous coworkers also deserve a thank- you for reviewing small sections, looking at code, asking lots of good questions, and otherwise putting up with me venting about not having enough time in the day. A tip of the hat goes out to the crew at the Aroma Café in downtown Champaign, Illinois. They’re just a few blocks down from my workplace, and a significant portion of this book was written at their coffee shop. Many thanks to Michael and his staff, including Kim, Sara, Nichole, and Jerry, for always having a hot and creamy mocha ready. Finally, I owe a tremendous debt to my wife, Debbie Fligor, and our two sons. They were always willing to make time for me to write and showed enormous amounts of patience and understanding. They all gave more than I had any right to ask, and this accomplishment is as much theirs as it is mine. Conventions Used in This Book The following typographical conventions are used in this book: Italic Indicates new terms, URLs, email addresses, filenames, and file extensions. Constant width Used for program listings, as well as within paragraphs to refer to program elements such as variable or function names, databases, data types, environment variables, statements, and keywords. Constant width bold Shows commands or other text that should be typed literally by the user. Constant width italic Shows text that should be replaced with user-supplied values or by values deter- mined by context. This icon signifies a tip, suggestion, or general note. This icon indicates a warning or caution. xviii | Preface Using Code Examples This book is here to help you get your job done. In general, you may use the code in this book in your programs and documentation. You do not need to contact us for permission unless you’re reproducing a significant portion of the code. For example, writing a program that uses several chunks of code from this book does not require permission. Selling or distributing a CD-ROM of examples from O’Reilly books does require permission. Answering a question by citing this book and quoting example code does not require permission. Incorporating a significant amount of example code from this book into your product’s documentation does require permission. We appreciate, but do not require, attribution. An attribution usually includes the title, author, publisher, and ISBN. For example: “Using SQLite by Jay A. Kreibich. Copyright 2010 O’Reilly Media, Inc., 978-0-596-52118-9.” If you feel your use of code examples falls outside fair use or the permission given above, feel free to contact us at permissions@oreilly.com. Safari® Books Online Safari Books Online is an on-demand digital library that lets you easily search over 7,500 technology and creative reference books and videos to find the answers you need quickly. With a subscription, you can read any page and watch any video from our library online. Read books on your cell phone and mobile devices. Access new titles before they are available for print, and get exclusive access to manuscripts in development and post feedback for the authors. Copy and paste code samples, organize your favorites, down- load chapters, bookmark key sections, create notes, print out pages, and benefit from tons of other time-saving features. O’Reilly Media has uploaded this book to the Safari Books Online service. To have full digital access to this book and others on similar topics from O’Reilly and other pub- lishers, sign up for free at http://my.safaribooksonline.com. How to Contact Us Please address comments and questions concerning this book to the publisher: O’Reilly Media, Inc. 1005 Gravenstein Highway North Sebastopol, CA 95472 800-998-9938 (in the United States or Canada) 707-829-0515 (international or local) 707 829-0104 (fax) Preface | xix We have a web page for this book, where we list errata, examples, and any additional information. You can access this page at: http://oreilly.com/catalog/9780596521196/ To comment or ask technical questions about this book, send email to: bookquestions@oreilly.com For more information about our books, conferences, Resource Centers, and the O’Reilly Network, see our web site at: http://www.oreilly.com xx | Preface CHAPTER 1 What Is SQLite? In the simplest terms, SQLite is a public-domain software package that provides a relational database management system, or RDBMS. Relational database systems are used to store user-defined records in large tables. In addition to data storage and man- agement, a database engine can process complex query commands that combine data from multiple tables to generate reports and data summaries. Other popular RDBMS products include Oracle Database, IBM’s DB2, and Microsoft’s SQL Server on the commercial side, with MySQL and PostgreSQL being popular open source products. The “Lite” in SQLite does not refer to its capabilities. Rather, SQLite is lightweight when it comes to setup complexity, administrative overhead, and resource usage. SQLite is defined by the following features: Serverless SQLite does not require a separate server process or system to operate. The SQLite library accesses its storage files directly. Zero Configuration No server means no setup. Creating an SQLite database instance is as easy as opening a file. Cross-Platform The entire database instance resides in a single cross-platform file, requiring no administration. Self-Contained A single library contains the entire database system, which integrates directly into a host application. Small Runtime Footprint The default build is less than a megabyte of code and requires only a few megabytes of memory. With some adjustments, both the library size and memory use can be significantly reduced. 1 Transactional SQLite transactions are fully ACID-compliant, allowing safe access from multiple processes or threads. Full-Featured SQLite supports most of the query language features found in the SQL92 (SQL2) standard. Highly Reliable The SQLite development team takes code testing and verification very seriously. Overall, SQLite provides a very functional and flexible relational database environment that consumes minimal resources and creates minimal hassle for developers and users. Self-Contained, No Server Required Unlike most RDBMS products, SQLite does not have a client/server architecture. Most large-scale database systems have a large server package that makes up the database engine. The database server often consists of multiple processes that work in concert to manage client connections, file I/O, caches, query optimization, and query process- ing. A database instance typically consists of a large number of files organized into one or more directory trees on the server filesystem. In order to access the database, all of the files must be present and correct. This can make it somewhat difficult to move or reliably back up a database instance. All of these components require resources and support from the host computer. Best practices also dictate that the host system be configured with dedicated service-user accounts, startup scripts, and dedicated storage, making the database server a very intrusive piece of software. For this reason, and for performance concerns, it is cus- tomary to dedicate a host computer system solely for the database server software. To access the database, client software libraries are typically provided by the database vendor. These libraries must be integrated into any client application that wishes to access the database server. These client libraries provide APIs to find and connect to the database server, as well as set up and execute database queries and commands. Figure 1-1 shows how everything fits together in a typical client/server RDBMS. In contrast, SQLite has no separate server. The entire database engine is integrated into whatever application needs to access a database. The only shared resource among ap- plications is the single database file as it sits on disk. If you need to move or back up the database, you can simply copy the file. Figure 1-2 shows the SQLite infrastructure. By eliminating the server, a significant amount of complexity is removed. This simplifies the software components and nearly eliminates the need for advanced operating system support. Unlike a traditional RDBMS server that requires advanced multitasking and high-performance inter-process communication, SQLite requires little more than the ability to read and write to some type of storage. 2 | Chapter 1: What Is SQLite? Figure 1-1. Traditional RDBMS client/server architecture that utilizes a client library. Figure 1-2. The SQLite server-less architecture. This simplicity makes it fairly straightforward to port SQLite to just about any envi- ronment, including mobile phones, handheld media players, game consoles, and other devices, where no traditional database system could ever dare to venture. Self-Contained, No Server Required | 3 Although SQLite does not use a traditional client/server architecture, it is common to speak of applications being “SQLite clients.” This termi- nology is often used to describe independent applications that simulta- neously access a shared SQLite database file, and is not meant to imply that there is a separate server. SQLite is designed to be integrated directly into an executable. This eliminates the need for an external library and simplifies distribution and installation. Removing external dependencies also removes most versioning issues. If the SQLite code is built right into your application, you never have to worry about linking to the correct version of a client library, or that the client library is version-compatible with the database server. Eliminating the server imposes some restrictions, however. SQLite is designed to ad- dress localized storage needs, such as a web server accessing a local database. This means it isn’t well suited for situations where multiple client machines need to access a centralized database. That situation is more representative of a client/server archi- tecture, and is better serviced by a database system that uses the same architecture. Single File Database SQLite packages the entire database into a single file. That single file contains the database layout as well as the actual data held in all the different tables and indexes. The file format is cross-platform and can be accessed on any machine, regardless of native byte order or word size. Having the whole database in a single file makes it trivial to create, copy, or back up the on-disk database image. Whole databases can be emailed to colleagues, posted to a web forum, or checked into a revision control system. Entire databases can be moved, modified, and shared with the same ease as a word-processing document or spread- sheet file. There is no chance of a database becoming corrupt or unavailable because one of a dozen files was accidentally moved or renamed. Perhaps most importantly, computer users have grown to expect that a document, project, or other “unit of application data” is stored as a single file. Having the whole database in a single file allows applications to use database instances as documents, data stores, or preference data, without contradicting customer expectations. Zero Configuration From an end-user standpoint, SQLite requires nothing to install, nothing to configure, and nothing to worry about. While there are a fair number of tuning parameters avail- able to developers, these are normally hidden from the end-user. By eliminating the server and merging the database engine directly into your application, your customers never need to know they’re using a database. It is quite practical to design an application 4 | Chapter 1: What Is SQLite? so that selecting a file is the only customer interaction—an action they are already comfortable doing. Embedded Device Support SQLite’s small code size and conservative resource use makes it well suited for embedded systems running limited operating systems. The ANSI C source code tends toward an older, more conservative style that should be accepted by even the most eccentric embedded processor compiler. Using the default configuration, the compiled SQLite library is less than 700 KB on most platforms, and requires less than 4 MB of memory to operate. By omitting the more advanced features, the library can often be trimmed to 300 KB or less. With minor configuration changes, the library can be made to function on less than 256 KB of memory, making its total footprint not much more than half a megabyte, plus data storage. SQLite expects only minimal support from its host environment and is written in a very modular way. The internal memory allocator can be easily modified or swapped out, while all file and storage access is done through a Virtual File System (VFS) interface that can be modified to meet the needs and requirements of different platforms. In general, SQLite can be made to run on almost anything with a 32-bit processor. Unique Features SQLite offers several features not found in many other database systems. The most notable difference is that SQLite uses a dynamic-type system for tables. The SQLite engine will allow you to put any value into nearly any column, regardless of type. This is a major departure from traditional database systems, which tend to be statically typed. In many ways, the dynamic-type system in SQLite is similar to those found in popular scripting languages, which often have a single scalar type that can accept any- thing from integers to strings. In my own experience, the dynamic-type system has solved many more problems than it has caused. Another useful feature is the ability to manipulate more than one database at a time. SQLite allows a single database connection to associate itself with multiple database files simultaneously. This allows SQLite to process SQL statements that bridge across multiple databases. This makes it trivial to join tables from different databases with a single query, or bulk copy data with a single command. SQLite also has the ability to create fully in-memory databases. These are essentially database “files” that have no backing store and spend their entire lifetime within the file cache. While in-memory databases lack durability and do not provide full transac- tion support, they are very fast (assuming you have enough RAM), and are a great place to store temporary tables and other transient data. Unique Features | 5 There are a number of other features that make SQLite extremely flexible. Many of these, like virtual tables, are based off similar features found in other products, but with their own unique twist. These features and extensions provide a number of powerful tools to adapt SQLite to your own particular problem or situation. Compatible License SQLite, and the SQLite code, have no user license. It is not covered by the GNU General Public License (GPL) or any of the similar open source/free-source licenses. Rather, the SQLite development team has chosen to place the SQLite source code in the public domain. This means that they have explicitly and deliberately relinquished any claim they have to copyright or ownership of the code or derived products. In short, this basically means you can do whatever you want with the SQLite source code, short of claiming to own it. The code and compiled libraries can be used in any way, modified in any way, redistributed in any way, and sold in any way. There are no restrictions, and no requirements or obligations to make third-party changes or mod- ifications available to the project or the public. The SQLite team takes this decision very seriously. Great care is taken to avoid any potential software patents or patented algorithms. All contributions to the SQLite source require a formal copyright release. Commercial contributions also require signed affidavits stating that the authors (and, if applicable, their employers) release their work into the public domain. All of this effort is taken to ensure that integrating SQLite into your product carries along minimal legal baggage or liability, making it a viable option for almost any de- velopment effort. Highly Reliable The purpose of a database is to keep your data safe and organized. The SQLite devel- opment team is aware that nobody will use a database product that has a reputation for being buggy or unreliable. To maintain a high level of reliability, the core SQLite library is aggressively tested before each release. In full, the standard SQLite test suites consists of over 10 million unit tests and query tests. The “soak test,” done prior to each release, consists of over 2.5 billion tests. The suite provides 100% statement coverage and 100% branch coverage, including edge- case errors, such as out-of-memory and out-of-storage conditions. The test suite is designed to push the system to its specified limits and beyond, providing extensive coverage of both code and operating parameters. 6 | Chapter 1: What Is SQLite? This high level of testing keeps the SQLite bug count relatively low. No software is perfect, but bugs that contribute to actual data-loss or database corruption are fairly rare. Most bugs that escape are performance related, where the database will do the right thing, but in the wrong way, leading to longer run-times. Strong testing also keeps backwards compatibility extremely solid. The SQLite team takes backwards compatibility very seriously. File formats, SQL syntax, and program- ming APIs and behaviors have an extremely strong history of backwards compatibility. Updating to a new version of SQLite rarely causes compatibility problems. In addition to keeping the core library reliable, the extensive testing also frees the SQLite developers to be more experimental. Whole subsystems of the SQLite code can be (and have been) ripped out and updated or replaced, with little concern about compatibility or functional differences—as long as all the tests pass. This allows the team to make significant changes with relatively little risk, constantly pushing the product and per- formance forward. Like so many other aspects of the SQLite design, fewer bugs means fewer problems and less to worry about. As much as any complex piece of software can, it just works. Highly Reliable | 7 CHAPTER 2 Uses of SQLite SQLite is remarkably flexible in both where it can be run and how it can be used. This chapter will take a brief look at some of the roles SQLite is designed to fill. Some of these roles are similar to those taken by traditional client/server RDBMS products. Other roles take advantage of SQLite’s size and ease of use, offering solutions you might not consider with a full client/server database. Database Junior Years of experience has taught developers that large client/server RDBMS platforms are powerful tools for safely storing, organizing, and manipulating data. Unfortunately, most large RDBMS products are resource-intensive and require a lot of upkeep. This boosts their performance and capacity, but it also limits how and where they can be practically deployed. SQLite is designed to fill in those gaps, providing the same powerful and familiar tools for safely storing, organizing, and manipulating data in smaller, more resource con- strained environments. SQLite is designed to complement, rather than replace, larger RDBMS platforms in situations where simplicity and ease of use are more important than capacity and concurrency. This complimentary role enables applications and tools to embrace relational data management (and the years of experience that come with it), even if they’re running on smaller platforms without administrative oversight. Developers may laugh at the idea of installing MySQL on a desktop system (or mobile phone!) just to support an address book application, but with SQLite this not only becomes possible, it becomes entirely practical. 9 Application Files Modern desktop applications typically deal with a significant number of files. Most applications and utilities have one or more preference files. There may also be system- wide and per-user configuration files, caches, and other data that must be tracked and stored. Document-based applications also need to store and access the actual document files. Using the SQLite library as an abstract storage layer has many advantages. A fair amount of application metadata, such as caches, state data, and configuration data, fit well with the relational data model. This makes it relatively easy to create an appropriate database design that maps cleanly and easily into an application’s internal data structures. In many cases, SQLite can also work well as a document file format. Rather than cre- ating a custom document format, an application can simply use individual database instances to represent working documents. SQLite supports many standard datatypes, including Unicode text, as well as arbitrary binary data fields that can store images or other raw data. Even if an application does not have particularly strong relational requirements, there are still significant advantages to using the SQLite library as a storage container. The SQLite library provides incremental updates that make it quick and easy to save small changes. The transaction system protects all file I/O against process termination and power disruption, nearly eliminating the possibility of file corruption. SQLite even provides its own file caching layer, so that very large files can be opened and processed in a limited memory footprint, without any additional work on the part of the application. SQLite database files are cross-platform, allowing easy migration. File contents can be easily and safely shared with other applications without worrying about detailed file format specifications. The common file format also makes it easy for automated scripts or troubleshooting utilities to access the files. Multiple applications can even access the same file simultaneously, and the library will transparently take care of all required file locking and cache synchronization. The use of SQLite can also make debugging and troubleshooting much easier, as files can be inspected and manipulated with standard database tools. You can even use standard tools to inspect and modify a database file as your application is using it. Similarly, test files can be programmatically generated outside the application, which is useful for automatic testing suites. Using an entire database instance as a document container may sound a bit unusual, but it is worth considering. The advantages are significant and should help a developer stay focused on the core of their application, rather than worrying about file formats, caching, or data synchronization. 10 | Chapter 2: Uses of SQLite Application Cache SQLite is capable of creating databases that are held entirely in memory. This is ex- tremely useful for creating small, temporary databases that require no permanent storage. In-memory databases are often used to cache results pulled from a more traditional RDBMS server. An application may pull a subset of data from the remote database, place it into a temporary database, and then process multiple detailed searches and refinements against the local copy. This is particularly useful when processing type- ahead suggestions, or any other interactive element that requires very quick response times. Temporary databases can also be used to index and store nearly any type of inter-linked, cross-referenced data. Rather than designing a set of complex runtime data structures which might include multiple hash tables, trees, and cross-referenced pointers, the developer can simply design an appropriate database schema and load the data into the database. While it might seem odd to execute SQL statements in order to extract data from an in-memory data structure, it is surprisingly efficient and can reduce development time and complexity. A database also provides an upgrade path, making it trivial to grow the data beyond the available memory or persist the data across application runs, simply by migrating to an on-disk database. Archives and Data Stores SQLite makes it very easy to package complex data sets into a single, easy-to-access, fully cross-platform file. Having all the data in a single file makes it much easier to distribute or download large, multi-table data stores, such as large dictionaries or geo- location references. Unlike many RDBMS products, the SQLite library is able to access read-only database files. This allows data stores to be read directly from an optical disc or other read-only filesystem. This is especially useful for systems with limited hard drive space, such as video game consoles. Client/Server Stand-in SQLite works well as a “stand-in” database for those situations when a more robust RDBMS would normally be the right choice, were it available. SQLite can be especially useful for the demonstration and evaluation of applications and tools that normally depend on a database. Consider a data analysis product that is designed to pull data from a relational database to generate reports and graphs. It can be difficult to offer downloads and evaluation Client/Server Stand-in | 11 copies of such software. Even if a download is available, the software must be config- ured and authorized to connect to a database that contains applicable data. This presents a significant barrier for a potential customer. Now consider an evaluation download that includes support for a bundled SQLite demonstration database. By simply downloading and running the software, customers can interact and experiment with the sample database. This makes the barrier of entry significantly lower, allowing a customer to go from downloading to running data in just a few seconds. Similar concerns apply to traditional sales and marketing demonstrations. Reliable network connectivity is often unavailable when doing on-site demonstrations to po- tential clients, so it is standard practice to run a local database server for demonstration purposes. Running a local database server consumes significant resources and adds administrative overhead. Database licenses may also be a concern. The use of SQLite eliminates these issues. The database becomes a background piece, allowing the demonstration to focus on the product. There are no database adminis- tration concerns. The simple file format also makes it easy to prepare customer-specific data sets or show off product features that significantly modify the database. All this can be done by simply making a copy of the database file before proceeding. Beyond evaluations and demonstrations, SQLite support can be used to promote a “lite” or “personal edition” of a larger product. Adding an entry-level product that is more suitable and cost-effective for smaller installations can open up a significant number of new customers by providing a low-cost, no-fuss introduction to the product line. SQLite support can even help with development and testing. SQLite databases are small and compact, allowing them to be attached to bug reports. They also provide an easy way to test a wide variety of situations, allowing a product to be tested against hundreds, if not thousands, of unique database instances. Even if a customer never sees an SQLite database, the integration time may easily pay for itself with improved testing and de- bugging capabilities. Teaching Tool For the student looking to learn SQL and the basics of the relational model, SQLite provides an extremely accessible environment that is easy to set up, easy to use, and easy to share. SQLite offers a full-fledged relational system that supports nearly all of the core SQL language, yet requires no server setup, no administration, and no over- head. This allows students to focus on learning SQL and data manipulation without getting bogged down by server configuration and database maintenance. 12 | Chapter 2: Uses of SQLite Given its compact size, it is simple to place a Windows, Mac OS X, and Linux version of the command-line tools, along with several databases, onto a small flash drive. With no installation process and fully cross-platform database files, this provides an “on the go” teaching environment that will work with nearly any computer. The “database in a file” architecture makes it easy for students to share their work. Whole database instances can be attached to an email or posted to a discussion forum. The single-file format also makes it trivial to back up work in progress, allowing stu- dents to experiment and explore different solutions without concern over losing data. Generic SQL Engine SQLite virtual tables allow a developer to define the contents of a table through code. By defining a set of callback functions that fetch and return rows and columns, a de- veloper can create a link between the SQLite data processing engine and any data source. This allows SQLite to run queries against the data source without importing the data into a standard table. Virtual tables are an extremely useful way to generate reports or allow ad hoc queries against logs or any tabular data set. Rather than writing a set of custom search or reporting tools, the data can simply be exposed to the SQLite engine. This allows re- ports and queries to be expressed in SQL, a language that many developers are already familiar with using. It also enables the use of generic database visualization tools and report generators. Chapter 10 shows how to build a virtual table module that provides direct access to live web server logs. Not the Best Choice Although SQLite has proven itself extremely flexible, there are some roles that are outside of its design goals. While SQLite may be able to perform in these areas, it might not be the best fit. If you find yourself with any of these requirements, it may be more practical to consider a more traditional client/server RDBMS product. High Transaction Rates SQLite is able to support moderate transaction rates, but it is not designed to sup- port the level of concurrent access provided by many client/server RDBMS prod- ucts. Many server systems are able to provide table-level or row-level locking, allowing multiple transactions to be processed in parallel without the risk of data loss. Not the Best Choice | 13 The concurrency protection offered by SQLite depends on file locks to protect against data loss. This model allows multiple database connections to access a database at the same time, but the whole database file must be locked in an ex- clusive mode to make any changes. As a result, write transactions are serialized across all database connections, limiting the overall transaction rate. Depending on the size and complexity of your updates, SQLite might be able to handle a few hundred transactions per minute from different processes or threads. If, however, you start to see performance problems, or expect higher transaction rates, a client/server system is likely to provide better transaction performance. Extremely Large Datasets It is not unusual to find SQLite databases that approach a dozen gigabytes or more, but there are some practical limits to the amount of data that can (or should) be stuffed into an SQLite database. Because SQLite puts everything into a single file (and thus, a single filesystem), very large data sets can stress the capability of the operating system or filesystem design. Although most modern filesystems are ca- pable of handling files that are a terabyte or larger, that doesn’t always mean they’re very good at it. Many filesystems see a significant drop in performance for random access patterns if the file starts to get into multiple gigabyte ranges. If you need to store and process several gigabytes or more of data, it might be wise to consider a more performance-oriented product. Access Control An SQLite database has no authentication or authorization data. Instead, SQLite depends on filesystem permissions to control access to the raw database file. This essentially limits access to one of three states: complete read/write access, read- only access, or no access at all. Write access is absolute, and allows both data modification and the ability to alter the structure of the database itself. While the SQLite API provides a basic application-layer authorization mechanism, it is trivial to circumvent if the user has direct access to the database file. Overall, this makes SQLite unsuitable for sensitive data stores that require per-user access control. Client/Server SQLite is specifically designed without a network component, and is best used as a local resource. There is no native support for providing access to multiple com- puters over a network, making it a poor choice as a client/server database system. Having multiple computers access an SQLite file through a shared directory is also problematic. Most networked filesystems have poor file-locking facilities. Without the ability to properly lock the file and keep updates synchronized, the database file can easily become corrupt. 14 | Chapter 2: Uses of SQLite This isn’t to say that client/server systems can’t utilize SQLite. For example, many web servers utilize SQLite. This works because all of the web server processes are running on the same machine and are all accessing the database file from local storage. Replication SQLite has no internal support for database replication or redundancy. Simple replication can be achieved by copying the database file, but this must be done when nothing is attempting to modify the database. Replication systems can be built on top of the basic database API, but such systems tend to be somewhat fragile. Overall, if you’re looking for real-time replication— especially at a transaction-safe level—you’ll need to look at a more complex RDBMS platform. Most of these requirements get into a realm where complexity and administrative over- head is traded for capacity and performance. This makes sense for a large client/server RDBMS platform, but it is somewhat at odds with the SQLite design goals of staying simple and maintenance free. To keep frustration to a minimum, use the right tool for the job. Big Name Users The SQLite website states that, “SQLite is the most widely deployed SQL database engine in the world.” This is a pretty bold claim, especially considering that when most people think of relational database platforms, they usually think of names like Oracle, SQL Server, and MySQL. It is also a claim that is difficult to support with exact numbers. Because there are no license agreements or disclosure requirements, it is hard to guess just how many SQLite databases are out there. Nobody, including the SQLite development team, is fully aware of who is using SQLite, and for what purposes. Regardless, the list of known SQLite users adds up to an impressive list. The Firefox web browser and the Thunderbird email client both use several SQLite databases to store cookies, history, preferences, and other account data. Many products from Skype, Adobe, and McAfee also utilize the SQLite engine. The SQLite library is also integrated into a number of popular scripting languages, including PHP and Python. Apple, Inc., has heavily embraced SQLite, meaning that every iPhone, iPod touch, and iPad, plus every copy of iTunes, and many other Macintosh applications, all ship with several SQLite databases. The Symbian, Android, BlackBerry, and Palm webOS envi- ronments all provide native SQLite support, while WinCE has third-party support. Chances are, if you have a smartphone, it has a number of SQLite databases stored on it. Big Name Users | 15 All of this adds up to millions, if not billions, of SQLite databases in the wild. No doubt that most of these databases only contain a few hundred kilobytes of data, but these low-profile environments are exactly where SQLite is designed to thrive. Large client/server RDBMS platforms have shown thousands of developers the power of relational data management systems. SQLite has brought that power out of the server room to the desktops and mobile devices of the world. 16 | Chapter 2: Uses of SQLite CHAPTER 3 Building and Installing SQLite This chapter is about building SQLite. We’ll cover how to build and install the SQLite distribution on Linux, Mac OS X, and Windows. The SQLite code base supports all of these operating systems natively, and precompiled libraries and executables for all three environments are available from the SQLite website. All downloads, including source and precompiled binaries, can be found on the SQLite download webpage (http://www .sqlite.org/download.html). SQLite Products The SQLite project consists of four major products: SQLite core The SQLite core contains the actual database engine and public API. The core can be built into a static or dynamic library, or it can be built in directly to an application. sqlite3 command-line tool The sqlite3 application is a command-line tool that is built on top of the SQLite core. It allows a developer to issue interactive SQL commands to the SQLite core. It is extremely useful for developing and debugging queries. Tcl extension SQLite has a strong history with the Tcl language. This library is essentially a copy of the SQLite core with the Tcl bindings tacked on. When compiled into a library, this code exposes the SQLite interfaces to the Tcl language through the Tcl Exten- sion Architecture (TEA). Outside of the native C API, these Tcl bindings are the only official programming interface supported directly by the SQLite team. 17 SQLite analyzer tool The SQLite analyzer is used to analyze database files. It displays statistics about the database file size, fragmentation, available free space, and other data points. It is most useful for debugging performance issues related to the physical layout of the database file. It can also be used to determine if it is appropriate to VACUUM (repack and defragment) the database or not. The SQLite website provides pre- compiled sqlite3_analyzer executables for most desktop platforms. The source for the analyzer is only available through the development source distribution. Most developers will be primarily interested in the first two products: the SQLite core and the sqlite3 command-line tool. The rest of the chapter will focus on these two products. The build process for the Tcl extension is identical to building the SQLite core as a dynamic library. The analyzer tool is normally not built, but simply down- loaded. If you want to build your own copy from scratch, you need a full development tree to do so. Precompiled Distributions The SQLite download page includes precompiled, standalone versions of the sqlite3 command-line tool for Linux, Mac OS X, and Windows. If you want to get started experimenting with SQLite, you can simply download the command-line tool, unpack it, run it, and start issuing SQL commands. You may not even have to download it first —Mac OS X and most Linux distributions include a copy of the sqlite3 utility as part of the operating system. The SQLite download page also includes precompiled, stand- alone versions of the sqlite3_analyzer for all three operating systems. Precompiled dynamic libraries of the SQLite core and the Tcl extension are also available for Linux and Windows. The Linux files are distributed as shared objects (.so files), while the Windows downloads contain DLL files. No precompiled libraries are available for Mac OS X. The libraries are only required if you are writing your own application, but do not wish to compile the SQLite core directly into your application. Documentation Distribution The SQLite download page includes a documentation distribution. The sqlite_docs_3 _x_x.zip file contains most of the static content from the SQLite website. The docu- mentation online at the SQLite website is not versioned and always reflects the API and SQL syntax for the most recent version of SQLite. If you don’t plan on continuously upgrading your SQLite distribution, it is useful to grab a copy of the documentation that goes with the version of SQLite you are using. 18 | Chapter 3: Building and Installing SQLite Source Distributions Most open source projects provide a single download that allows you to configure, build, and install the software with just a handful of commands. SQLite works a bit differently. Because the most common way to use SQLite is to integrate the core source directly into a host application, the source distributions are designed to make integra- tion as simple as possible. Most of the source distributions contain only source code and provide minimal (if any) configuration or build support files. This makes it simpler to integrate SQLite into a host application, but if you want to build a library or sqlite3 application, you will often need to do that by hand. As we’ll see, that’s fairly easy. The Amalgamation The official code distribution is known as the amalgamation. The amalgamation is a single C source file that contains the entire SQLite core. It is created by assembling the individual development files into a single C source file that is almost 4 megabytes in size and over 100,000 lines long. The amalgamation, along with its corresponding header file, is all that is needed to integrate SQLite into your application. The amalgamation has two main advantages. First, with everything in one file, it is extremely easy to integrate SQLite into a host application. Many projects simply copy the amalgamation files into their own source directories. It is also possible to compile the SQLite core into a library and simply link the library into your application. Second, the amalgamation also helps improve performance. Many compiler optimiza- tions are limited to a single translation unit. In C, that’s a single source file. By putting the whole library into a single file, a good optimizer can process the whole package at once. Compared to compiling the individual source files, some platforms see a 5% or better performance boost just by using the amalgamation. The only disadvantage of using the amalgamation is size. Some debuggers have issues with files more than 65,535 lines long. Things typically run correctly, but it can be difficult to set breakpoints or look at stack traces. Compiling a source file over 100,000 lines long also takes a fair number of resources. While this is no problem for most desktop systems, it may push the limits of any compilers running on limited platforms. Source Files When working with the amalgamation, there are four important source files: sqlite3.c The amalgamation source file, which includes the entire SQLite core, plus common extensions. Source Distributions | 19 sqlite3.h The amalgamation header file, which exposes the core API. sqlite3ext.h The extension header file, which is used to build SQLite extensions. shell.c The sqlite3 application source, which provides an interactive command-line shell. The first two, sqlite3.c and sqlite3.h, are all that is needed to integrate SQLite into most applications. The sqlite3ext.h file is used to build extensions and modules. Building extensions is covered in “SQLite Extensions” on page 204. The shell.c file contains the source code for the sqlite3 command-line shell. All of these files can be built on Linux, Mac OS X, or Windows, without any additional configuration files. Source Downloads The SQLite website offers five source distribution packages. Most people will be in- terested in one of the first two files. sqlite-amalgamation-3_x_x.zip The Windows amalgamation distribution. sqlite-amalgamation-3.x.x.tar.gz The Unix amalgamation distribution. sqlite-3_x_x-tea.tar.gz The Tcl extension distribution. sqlite-3.x.x.tar.gz The Unix source tree distribution. This is unsupported and the build files are unmaintained. sqlite-source-3_x_x.zip The Windows source distribution. This is unsupported. The Windows amalgamation file consists of the four main files, plus a .def file to build a DLL. No makefile, project, or solution files are included. The Unix amalgamation file, which works on Linux, Mac OS X, and many other flavors of Unix, contains the four main files plus an sqlite3 manual page. The Unix distribu- tion also contains a basic configure script, along with other autoconf files, scripts, and makefiles. The autoconf files should also work under the Minimalist GNU for Windows (MinGW) environment (http://www.mingw.org/). The Tcl extension distribution is a specialized version of the amalgamation. It is only of interest to those working in the Tcl language. See the included documentation for more details. 20 | Chapter 3: Building and Installing SQLite The Unix source tree is an unsupported legacy distribution. This is what the standard distribution looked like before the amalgamation became the officially supported dis- tribution. It is made available for those that have older build environments or devel- opment branches that utilize the old distribution format. This distribution also includes a number of README files that are unavailable elsewhere. Although the source files are kept up to date, the configuration scripts and makefiles included in the Unix source tree distribution are no longer maintained and do not work properly on most platforms. Unless you have some significant need to use the source tree distribution, you should use one of the amalgamation distributions instead. The Windows source distribution is essentially a .zip file of the source directory from the source tree distribution, minus some test files. It is strictly source files and header files, and contains no build scripts, makefiles, or project files. Building There are a number of different ways to build SQLite, depending on what you’re trying to build and where you would like it installed. If you are trying to integrate the SQLite core into a host application, the easiest way to do that is to simply copy sqlite3.c and sqlite3.h into your application’s source directory. If you’re using an IDE, the sqlite3.c file can simply be added to your application’s project file and configured with the proper search paths and build directives. If you want to build a custom version of the SQLite library or sqlite3 utility, it is also easy to do that by hand. All of the SQLite source is written in C. It cannot be compiled by a C++ compiler. If you’re getting errors related to structure definitions, chances are you’re using a C++ compiler. Make sure you use a vanilla C compiler. Configure If you’re using the Unix amalgamation distribution, you can build and install SQLite using the standard configure script. After downloading the distribution, it is fairly easy to unpack, configure, and build the source: $ tar xzf sqlite-amalgamation-3.x.x.tar.gz $ cd sqlite-3.x.x $ ./configure [...] $ make By default, this will build the SQLite core into both static and dynamic libraries. It will also build the sqlite3 utility. These will be built with many of the extra features (such as full text search and R*Tree support) enabled. Once this finishes, the command make install will install these files, along with the header files and sqlite3 manual page. By Building | 21 default, everything is installed into /usr/local, although this can be changed by giving a --prefix=/path/to/install option to configure. Issue the command configure --help for information on other build options. Manually Because the main SQLite amalgamation consists of only two source files and two header files, it is extremely simple to build by hand. For example, to build the sqlite3 shell on Linux, or most other Unix systems: $ cc -o sqlite3 shell.c sqlite3.c -ldl -lpthread The additional libraries are needed to support dynamic linking and threads. Mac OS X includes those libraries in the standard system group, so no additional libraries are required when building for Mac OS X: $ cc -o sqlite3 shell.c sqlite3.c The commands are very similar on Windows, using the Visual Studio C compiler from the command-line: > cl /Fesqlite3 shell.c sqlite3.c This will build both the SQLite core and the shell into one application. That means the resulting sqlite3 executable will not require an installed library in order to operate. If you want to build things with one of the optional modules installed, you need to define the appropriate compiler directives. This shows how to build things on Unix with the FTS3 (full text search) extension enabled: $ cc -DSQLITE_ENABLE_FTS3 -o sqlite3 shell.c sqlite3.c -ldl -lpthread Or, on Windows: > cl /Fesqlite3 /DSQLITE_ENABLE_FTS3 shell.c sqlite3.c Building the SQLite core into a dynamic library is a bit more complex. We need to build the object file, then build the library using that object file. If you’ve already built the sqlite3 utility, and have an sqlite3.o (or .obj) file, you can skip the first step. First, in Linux and most Unix systems: $ cc -c sqlite3.c $ ld -shared -o libsqlite3.so sqlite3.o Some versions of Linux may also require the -fPIC option when compiling. Mac OS X uses a slightly different dynamic library format, so the command to build it is slightly different. It also needs the standard C library to be explicitly linked: $ cc -c sqlite3.c $ ld -dylib -o libsqlite3.dylib sqlite3.o -lc 22 | Chapter 3: Building and Installing SQLite And finally, building a Windows DLL (which requires the sqlite3.def file): > cl /c sqlite3.c > link /dll /out:sqlite3.dll /def:sqlite3.def sqlite3.obj You may need to edit the sqlite3.def file to add or remove functions, depending on which compiler directives are used. Build Customization The SQLite core is aware of a great number of compiler directives. Appendix A covers all of these in detail. Many are used to alter the standard default values, or to adjust some of the maximum sizes and limits. Compiler directives are also used to enable or disable specific features and extensions. There are several dozen directives in all. The default build, without any specific directives, will work well enough for a wide variety of applications. However, if your application requires one of the extensions, or has specific performance concerns, there may be some ways to tune the build. Many of the parameters can also be adjusted at runtime, so a recompile may not always be necessary, but it can make development more convenient. Build and Installation Options There are several different ways to build, integrate, and install SQLite. The design of the SQLite core lends itself to being compiled as a dynamic library. A single library can then be utilized by whatever application requires SQLite. Building a shared library this way is one of the more straightforward ways to integrate and install SQLite, but it is often not the best approach. The SQLite project releases new versions rather frequently. While they take backward compatibility seriously, there are sometimes changes to the default configuration. There are also cases of applications becoming dependent on version-specific bugs or undefined (or undocumented) be- haviors. There are also a large number of custom build options that SQLite supports. All these concerns can be difficult to create a system-wide build that is suitable for every application that uses SQLite. This problem becomes worse as the number of applications utilizing SQLite continues to increase, making for more and more application-specific copies of SQLite. Even if an application (or suite of applications) has its own private copy of an SQLite library, there is still the possibility of incorrect linking and version incompatibilities. To avoid these problems, the recommended way of using SQLite is to integrate the whole database engine directly into your application. This can be done by building a static library and then linking it in, or by simply building the amalgamation source directly into your application code. This method provides a truly custom build that is tightly bound to the application that uses it, eliminating any possibility of version or build incompatibilities. Build and Installation Options | 23 About the only time it may be appropriate to use a dynamic library is when you’re building against an existing system-installed (and system-maintained) library. This in- cludes Mac OS X, many Linux distributions, as well as the majority of phone environ- ments. In that case, you’re depending on the operating system to keep a consistent build. This normally works for reasonably simple needs, but your application needs to be somewhat flexible. System libraries are often frozen with each major release, but chances are that sooner or later the system software (including the SQLite system libraries) will be upgraded. Your application may have to work across different versions of the system library if you need to support different versions of the operating system. For all these same reasons, it is ill-advised to manually replace or upgrade the system copy of SQLite. If you do decide to use your own private library, take great care when linking. It is all too easy to accidentally link against a system library, rather than your private copy, if both are available. Versioning problems, along with many other issues, can be completely avoided if the application simply contains its own copy of the SQLite core. The SQLite source dis- tributions and the amalgamation make direct integration an easy path to take. Libraries have their place, but makes sure you understand the possible implications of having an external library. In specific, unless you control an entire device, never assume you’re the only SQLite user. Try to keep your builds and installs clear of any system-wide library locations. An sqlite3 Primer Once you have some form of SQLite installed, the first step is normally to run sqlite3 and play around. The sqlite3 tool accepts SQL commands from an interactive prompt and passes those commands to the SQLite core for processing. Even if you have no intention of distributing a copy of sqlite3 with your application, it is extremely useful to have a copy around for testing and debugging queries. If your application uses a customized build of the SQLite core, you will likely want to build a copy of sqlite3 using the same build parameters. To get started, just run the SQLite command. If you provide a filename (such as test.db), sqlite3 will open (or create) that file. If no filename is given, sqlite3 will automatically open an unnamed temporary database: $ sqlite3 test.db SQLite version 3.6.23.1 Enter ".help" for instructions Enter SQL statements terminated with a ";" sqlite> 24 | Chapter 3: Building and Installing SQLite The sqlite> prompt means sqlite3 is ready to accept commands. We can start with some basic expressions: sqlite> SELECT 3 * 5, 10; 15|10 sqlite> SQL commands can also be entered across multiple lines. If no terminating semicolon is found, the statement is assumed to continue. In that case, the prompt will change to ...> to indicate sqlite3 is waiting for more input: sqlite> SELECT 1 + 2, ...> 6 + 3; 3|9 sqlite> If you ever find yourself at the ...> prompt unexpectedly, make sure you finished up the previous line with a semicolon. In addition to processing SQL statements, there is a series of shell-specific commands. These are sometimes referred to as “dot-commands” because they start with a period. Dot-commands control the shell’s output formatting, and also provide a number of utility features. For example, the .read command can be used to execute a file full of SQL commands. Dot-commands must be given at the sqlite> prompt. They must be given on one line, and should not end in a semicolon. You cannot mix SQL statements and dot- commands. Two of the more useful dot-commands (besides .help) are .headers and .mode. Both of these control some aspect of the database output. Turning headers on and setting the output mode to column will produce a table that most people find easier to read: sqlite> SELECT 'abc' AS start, 'xyz' AS end; abc|xyz sqlite> .headers on sqlite> .mode column sqlite> SELECT 'abc' AS start, 'xyz' AS end; start end ---------- ---------- abc xyz sqlite> Also helpful is the .schema command. This will list all of the DDL commands (CREATE TABLE, CREATE INDEX, etc.) used to define the database. For a more complete list of all the sqlite3 command-line options and dot-commands, see Appendix A. An sqlite3 Primer | 25 Summary SQLite is designed to integrate into a wide variety of code bases on a broad range of platforms. This flexibility provides a great number of options, even for the most basic situations. While flexibility is usually a good thing, it can make for a lot of confusion when you’re first trying to figure things out. If you’re just starting out, and all you need is a copy of the sqlite3 shell, don’t get too caught up in all the advanced build techniques. You can download one of the precompiled executables or build your own with the one-line commands provided in this chapter. That will get you started. As your needs evolve, you may need a more specific build of sqlite3, or you may start to look at integrating the SQLite library into your own application. At that point you can try out different build techniques and see what best matches your needs and build environment. While the amalgamation is a somewhat unusual form for source distribution, it has proven itself to be quite useful and well suited for integrating SQLite into larger projects with the minimal amount of fuss. It is also the only officially supported source distri- bution format. It works well for the majority of projects. 26 | Chapter 3: Building and Installing SQLite CHAPTER 4 The SQL Language This chapter provides an overview of the Structured Query Language, or SQL. Although sometimes pronounced “sequel,” the official pronunciation is to name each letter as “ess-cue-ell.” The SQL language is the main means of interacting with nearly all modern relational database systems. SQL provides commands to configure the tables, indexes, and other data structures within the database. SQL commands are also used to insert, update, and delete data records, as well as query those records to look up specific data values. All interaction with a relational database is done through the SQL language. This is true when interactively typing commands or when using the programming API. In all cases, data is stored, modified, and retrieved through SQL commands. Many times, people look through the list of API calls, looking for functions that provide direct pro- gram access to the table or index data structures. Functions of this sort do not exist. The API is structured around preparing and issuing SQL commands to the database engine. If you want to query a table or insert a value using the API, you must create and execute the proper SQL command. If you want to do relational database programming, you must know SQL. Learning SQL The goal of this chapter is to introduce you to all the major SQL commands and show some of the basic usage patterns. The first time you read through this chapter, don’t feel you need to absorb everything at once. Get an idea of what structures the database supports, and how they might be used, but don’t feel that you need to memorize the details of every last command. 27 For people just getting started, the most important commands are CREATE TABLE, INSERT, and SELECT. These will let you create a table, insert some data into the table, and then query the data and display it. Once you get comfortable with those commands, you can start to look at the others in more depth. Feel free to refer back to this chapter, or the command reference in Appendix C. The command reference provides detailed descriptions of each command, including some of the more advanced syntax that isn’t covered in this chapter. Always remember that SQL is a command language. It assumes you know what you’re doing. If you’re directly entering SQL commands through the sqlite3 application, the program will not stop and ask for confirmation before processing dangerous or de- structive commands. When entering commands by hand, it is always worth pausing and looking back at what you’ve typed before you hit return. If you are already reasonably familiar with the SQL language, it should be safe to skim this chapter. Much of the information here is on the SQL language in general, but there is some information about the specific dialect of SQL that SQLite recognizes. Again, Appendix C provides a reference to the specific SQL syntax used by SQLite. Brief Background Although the first official SQL specification was published in 1986 by the American National Standards Institute (ANSI), the language traces its roots back to the early 1970s and the pioneering relational database work that was being done at IBM. Current SQL standards are ratified and published by the International Standards Organization (ISO). Although a new standard is published every few years, the last significant set of changes to the core language can be traced to the SQL:1999 standard (also known as “SQL3”). Subsequent standards have mainly been concerned with storing and pro- cessing XML-based data. Overall, the evolution of SQL is firmly rooted in the practical aspects of database development, and in many cases new standards only serve to ratify and standardize syntax or features that have been present in commercial database products for some time. Declarative The core of SQL is a declarative language. In a declarative language, you state what you want the results to be and allow the language processor to figure out how to deliver the desired results. Compare this to imperative languages, such as C, Java, Perl, or Python, where each step of a calculation or operation must be explicitly written out, leaving it up to the programmer to lead the program, step by step, to the correct conclusion. The first SQL standards were specifically designed to make the language approachable and usable by “non-computer people”—at least by the 1980s definition of that term. This is one of the reasons why SQL commands tend to have a somewhat English-like 28 | Chapter 4: The SQL Language syntax. Most SQL commands take the form verb-subject. For example, CREATE (verb) TABLE (subject), DROP INDEX, UPDATE table_name. The almost-English, declarative nature of SQL has both advantages and disadvantages. Declarative languages tend to be simpler (especially for nonprogrammers) to under- stand. Once you get used to the general structure of the commands, the use of English keywords can make the syntax easier to remember. The fixed command structure also makes it much easier for the database engine to optimize queries and return data more efficiently. The predefined nature of declarative statements can sometimes feel a bit limited, how- ever—especially in a command-based language like SQL, where individual, isolated commands are constructed and issued one at a time. If you require a query that doesn’t fit into the processing order defined by the SELECT command, you may find yourself having to patch together nested queries or temporary tables. This is especially true when the problem you’re trying to solve is inherently nonrelational, and you’re forced to jump through extra hoops to account for that. Despite its oddities and somewhat organic evolution, SQL is a powerful language with a surprisingly broad ability to express complex operations. Once you wrap your head around what makes SQL tick, you can often find moderately simple solutions to even the most off-the-beaten-path problems. It can take some adjustment, however, espe- cially if your primary experience is with imperative languages. Portability SQL’s biggest flaw is that formal standardization has almost always followed common implementations. Almost every database product (including SQLite) has custom ex- tensions and enhancements to the core language that help differentiate it from other products, or expose features or control systems that are not covered by the core SQL standard. Often these enhancements are related to performance enhancements, and can be difficult to ignore. While this less-strict approach to language purity has allowed SQL to grow and evolve in very practical ways, it means that “real world” SQL portability is not all that practical. If you strictly limit yourself to standardized SQL syntax, you can achieve a moderate degree of portability, but normally this comes at the cost of lower performance and less data integrity. Generally, applications will write to the specific SQL dialect they’re using and not worry about cross-database compatibility. If cross-database compatibility is important to a specific application, the normal approach is to develop a core list of SQL commands required by the application, with minor tweaks for each specific database product. SQLite makes an effort to follow the SQL standards as much as possible. SQLite will also recognize and correctly parse a number of nonstandard syntax conventions used by other popular databases. This can help with the portability issues. Brief Background | 29 SQL is not without other issues, but considering its lineage, it is surprisingly well suited for the task at hand. Love it or hate it, it is the relational database language of choice, and it is likely to be with us for a long time. General Syntax Before getting into specific commands in SQL, it is worth looking at the general lan- guage structure. Like most languages, SQL has a fairly complete expression syntax that can be used to define command parameters. A more detailed description of the ex- pression support can be found in Appendix D. Basic Syntax SQL consists of a number of different commands, such as CREATE TABLE or INSERT. These commands are issued and processed one at a time. Each command implements a dif- ferent action or feature of the database system. Although it is customary to use all capital letters for SQL commands and keywords, SQL is a case-insensitive* language. All commands and keywords are case insensitive, as are identifiers (such as table names and column names). Identifiers must be given as literals. If necessary, identifiers can be enclosed in the standards compliant double-quotes (" ") to allow the inclusion of spaces or other non- standard characters in an identifier. SQLite also allows identifiers to be enclosed in square brackets ([ ]) or back ticks (` `) for compatibility with other popular database products. SQLite reserves the use of any identifier that uses sqlite_ as a prefix. SQL is whitespace insensitive, including line breaks. Individual statements are separa- ted by a semicolon. If you’re using an interactive application, such as the sqlite3 command-line tool, then you’ll need to use a semicolon to indicate the end of a state- ment. The semicolon is not strictly required for single statements, however, as it is properly a statement separator and not a statement terminator. When passing SQL commands into the programming API, the semicolon is not required unless you are passing more than one command statement within a single string. Single-line comments start with a double dash (--) and go to the end of the line. SQL also supports multi-line comments using the C comment syntax (/* */). As with most languages, numeric literals are represented bare. Both integer (453) and real (rational) numbers (43.23) are recognized, as is exponent-style scientific notation (9.745e-6). In order to avoid ambiguities in the parser, SQLite requires that the decimal point is always represented as a period (.), regardless of the current internationalization setting. * Unless otherwise specified, case insensitivity only applies to ASCII characters. That is, characters represented by values less than 128. 30 | Chapter 4: The SQL Language Text literals are enclosed in single quotes (' '). To represent a string literal that includes a single quote character, use two single quotes in a row (publisher = 'O''Reilly'). C-style backslash escapes ( \' ) are not part of the SQL standard and are not supported by SQLite. BLOB literals (binary data) can be represented as an x (or X) followed by a string literal of hexadecimal characters (x'A554E59C'). Text literals use single quotes. Double quotes are reserved for identifiers (table names, columns, etc.). C-style backslash escapes are not part of the SQL standard. SQL statements and expressions frequently contain lists. A comma is used as the list separator. SQL does not allow for a trailing comma following the last item of a list. In general, expressions can be used any place a literal data value is allowed. Expressions can include both mathematical statements, as well as functions. Function-calling syn- tax is similar to most other computer languages, utilizing the name of the function, followed by a list of parameters enclosed in parentheses. Expressions can be grouped into subexpressions using parentheses. If an expression is evaluated in the context of a row (such as a filtering expression), the value of a row element can be extracted by naming the column. You may have to qualify the column name with a table name or alias. If you’re using cross-database queries, you may also have to specify which database you’re referring to. The syntax is: [[database_name.]table_name.]column_name If no database name is given, it is assumed you’re referring to the main database on the default connection. If the table name/alias is also omitted, the system will make a best- guess using just the column name, but will return an error if the name is ambiguous. Three-Valued Logic SQL allows any value to be assigned a NULL. NULL is not a value in itself (SQLite actually implements it as a unique valueless type), but is used as a marker or flag to represent unknown or missing data. The thought is that there are times when values for a specific row element may not be available or may not be applicable. A NULL may not be a value, but it can be assigned to data elements that normally have values, and can therefore show up in expressions. The problem is that NULLs don’t interact well with other values. If a NULL represents an unknown that might be any possible value, how can we know if the expression NULL > 3 is true or false? To deal with this problem, SQL must employ a concept called three-valued logic. Three- valued logic is often abbreviated TVL or 3VL, and is more formally known as ternary logic. 3VL essentially adds an “unknown” state to the familiar true/false Boolean logic system. General Syntax | 31 Here are the truth tables for the 3VL operators NOT, AND, and OR: Value NOT Value True False False True NULL NULL 3VL AND TRUE FALSE NULL TRUE TRUE FALSE NULL FALSE FALSE FALSE FALSE NULL NULL FALSE NULL 3VL OR TRUE FALSE NULL TRUE TRUE TRUE TRUE FALSE TRUE FALSE NULL NULL TRUE NULL NULL 3VL also dictates how comparisons work. For example, any equality check (=) involving a NULL will evaluate to NULL, including NULL = NULL. Remember that NULL is not a value, it is a flag for the unknown, so the expression NULL = NULL is asking, “Does this unknown equal that unknown?” The only practical answer is, “That is unknown.” It might, but it might not. Similar rules apply to greater-than, less-than, and other comparisons. You cannot use the equality operator (=) to test for NULLs. You must use the IS NULL operator. If you’re having trouble resolving an expression, just remember that a NULL marks an unknown or unresolved value. This is why the expression False AND NULL is false, but True AND NULL is NULL. In the case of the first expression, the NULL can be replaced by either true or false without altering the expression result. That isn’t true of the second expression, where the outcome is unknown (in other words, NULL) because the output depends on the unknown input. 32 | Chapter 4: The SQL Language Simple Operators SQLite supports the following unary prefix operators: - + These adjust the sign of a value. The “-” operator flips the sign of the value, effec- tively multiplying it by -1.0. The “+” operator is essentially a no-op, leaving a value with the same sign it previously had. It does not make negative values positive. ~ As in the C language, the “~” operator performs a bit-wise inversion. This operator is not part of the SQL standard. NOT The NOT operator reverses a Boolean expression using 3VL. There are also a number of binary operators. They are listed here in descending precedence. || String concatenation. This is the only string concatenation operator recognized by the SQL standard. Many other database products allow “+” to be used for con- catenation, but SQLite does not. + - * / % Standard arithmetic operators for addition, subtraction, multiplication, division, and modulus (remainder). | & << >> The bitwise operators or, and, and shift-high/shift-low, as found in the C language. These operators are not part of the SQL standard. < <= => > Comparison test operators. Again, just as in the C language we have less-than, less- than or equal, greater-than or equal, and greater than. These operators are subject to SQL’s 3VL regarding NULLs. = == != <> Equality test operators. Both “=” and “==” will test for equality, while both “!=” and “<>” test for inequality. Being logic operators, these tests are subject to SQL’s 3VL regarding NULLs. Specifically, value = NULL will always return NULL. IN LIKE GLOB MATCH REGEXP These five keywords are logic operators, returning, true, false, or NULL state. See Appendix D for more specifics on these operators. AND OR Logical operators. Again, they are subject to SQL’s 3VL. In addition to these basics, SQL supports a number of specific expression operations. For more information on these and any SQL-specific expressions, see Appendix D. General Syntax | 33 SQL Data Languages SQL commands are divided into four major categories, or languages. Each language defines a subset of commands that serve a related purpose. The first language is the Data Definition Language, or DDL, which refers to commands that define the structure of tables, views, indexes, and other data containers and objects within the database. CREATE TABLE (used to define a new table) and DROP VIEW (used to delete a view) are examples of DDL commands. The second category of commands is known as Data Manipulation Language, or DML. These are all of the commands that insert, update, delete, and query actual data values from the data structures defined by the DDL. INSERT (used to insert new values into a table) and SELECT (used to query or look up data from tables) are examples of DML commands. Related to the DML and DDL is the Transaction Control Language, or TCL. TCL com- mands can be used to control transactions of DML and DDL commands. BEGIN (used to begin a multistatement transaction) and COMMIT (used to end and accept a transac- tion) are examples of TCL commands. The last category is the Data Control Language, or DCL. The main purpose of the DCL is to grant or revoke access control. Much like file permissions, DCL commands are used to allow (or deny) specific database users (or groups of users) permission to utilize or access specific resources within a database. These permissions can apply to both the DDL and the DML. DDL permissions might include the ability to create a real or tem- porary table, while DML permissions might include the ability to read, update, or delete the records of a specific table. GRANT (used to assign a permission) and REVOKE (used to delete an existing permission) are the primary DCL commands. SQLite supports the majority of standardized DDL, DML, and TCL commands but lacks any DCL commands. Because SQLite does not have user names or logins, it does not have any concept of assigned permissions. Rather, SQLite depends on datatype permissions to define who can open and access a database. Data Definition Language The DDL is used to define the structure of data containers and objects within the da- tabase. The most common of these containers and objects are tables, indexes, and views. As you’ll see, most objects are defined with a variation of the CREATE command, such as CREATE TABLE or CREATE VIEW. The DROP command is used to delete an existing object (and all of the data it might contain). Examples include DROP TABLE or DROP INDEX. Because the command syntax is so different, statements like CREATE TABLE or CREATE INDEX are usually considered to be separate commands, and not variations of a single CREATE command. 34 | Chapter 4: The SQL Language In a sense, the DDL commands are similar to C/C++ header files. DDL commands are used to define the structure, names, and types of the data containers within a database, just as a header file typically defines type definitions, structures, classes, and other data structures. DDL commands are typically used to set up and configure a brand new database before data is entered. DDL commands define the basic structure of the database and are typ- ically run when a new database is created. DDL commands are often held in a script file, so that the structure of the database can be easily recreated. Sometimes, especially during development, you may need to re- create only part of the database. To help support this, most CREATE commands in SQLite have an optional IF NOT EXISTS clause. Normally, a CREATE statement will return an error if an object with the requested name already exists. If the optional IF NOT EXISTS clause is present, then this error is sup- pressed and nothing is done, even if the structure of the existing object and the new object are not compatible. Similarly, most DROP statements allow an optional IF EXISTS clause that silently ignores any request to delete an object that isn’t there. In the examples that follow, the IF EXISTS and IF NOT EXISTS command variations are not explicitly called out. Please see the SQLite command reference in Appendix C for the full details on the complete syntax supported by SQLite. Tables The most common DDL command is CREATE TABLE. No data values can be stored in a database until a table is defined to hold that data. At the bare minimum, the CREATE TABLE command defines the table name, plus the name of each column. Most of the time you’ll want to define a type for each column as well, although types are optional when using SQLite. Optional constraints, conditions, and default values can also be assigned to each column. Table-level constraints can also be assigned, if they involve multiple columns. In some large RDBMS systems, CREATE TABLE can be quite complex, defining all kinds of storage options and configurations. SQLite’s version of CREATE TABLE is somewhat simpler, but there are still a great many options available. For full explanation of the command, see CREATE TABLE in Appendix C. Data Definition Language | 35 The basics The most basic syntax for CREATE TABLE looks something like this: CREATE TABLE table_name ( column_name column_type, [...] ); A table name must be provided to identify the new table. After that, there is just a simple list of column names and their types. Table names come from a global namespace of all identifiers, including the names of tables, views, and indexes. Clear and concise identifier names are important. Like the database design itself, some careful thought should be put into the table name, trying to pin down the precise meaning of the data it contains. Much the same could be said of column names. Table names and column names tend to be referenced frequently in queries, so there is some desire to keep them as brief as possible while still keeping their purpose clear. Column types Most databases use strong, static column typing. This means that the elements of a column can only hold values compatible with the column’s defined type. SQLite utilizes a dynamic typing technique known as manifest typing. For each row value, manifest typing records the value’s type along with the value data. This allows nearly any element of any row to hold almost any type of value. In the strictest sense, SQLite supports only five concrete datatypes. These are known as storage classes, and represent the different ways SQLite might choose to store data on disk. Every value has one of these five native storage classes: NULL A NULL is considered its own distinct type. A NULL type does not hold a value. Literal NULLs are represented by the keyword NULL. Integer A signed integer number. Integer values are variable length, being 1, 2, 3, 4, 6, or 8 bytes in length, depending on the minimum size required to hold the specific value. Integer have a range of −9,223,372,036,854,775,808 to +9,223,372, 036,854,775,807, or roughly 19 digits. Literal integers are represented by any bare series of numeric digits (without commas) that does not include a decimal point or exponent. Float A floating-point number, stored as an 8-byte value in the processor’s native format. For nearly all modern processors, this is an IEEE 754 double-precision number. Literal floating-point numbers are represented by any bare series of numeric digits that include a decimal point or exponent. 36 | Chapter 4: The SQL Language Text A variable-length string, stored using the database encoding (UTF-8, UTF-16BE, or UTF-16LE). Literal text values are represented using character strings in single quotes. BLOB A length of raw bytes, copied exactly as provided. Literal BLOBs are represented as hexadecimal text strings preceded by an x. For example, the notation x'1234ABCD' represents a 4-byte BLOB. BLOB stands for Binary Large OBject. SQLite text and BLOB values are always variable length. The maximum size of a text or BLOB value is limited by a compile-time directive. The default limit is exactly one billion bytes, or slightly less than a full gigabyte. The maximum value for this directive is two gigabytes. Since the elements of most columns can hold any value type, the “type” of a column is a somewhat misleading concept. Rather than being an absolute type, as in most data- bases, an SQLite column type (as defined in CREATE TABLE) becomes more of a sugges- tion than a hard and fast rule. This is known as a type affinity, and essentially represents a desired category of type. Each type affinity has specific rules about what types of values it can store, and how different values will be converted when stored in that column. Generally, a type affinity will cause a conversion or migration of types only if it can be done without losing data or precision. Each table column must have one of five type affinities: Text A column with a text affinity will only store values of type NULL, text, or BLOB. If you attempt to store a value with a numeric type (float or integer) it will be converted into a text representation before being stored as a text value type. Numeric A column with a numeric affinity will store any of the five types. Values with integer and float types, along with NULL and BLOB types, are stored without conversion. Any time a value with a text type is stored, an attempt is made to convert the value to a numeric type (integer or float). Assuming the conversion works, the value is stored in an appropriate numeric type. If the conversion fails, the text value is stored without any type of conversion. Integer A column with an integer affinity works essentially the same as a numeric affinity. The only difference is that any value with a float type that lacks a fractional com- ponent will be converted into an integer type. Float A column with a floating-point affinity also works essentially the same as a numeric affinity. The only difference is that most values with integer types are converted into floating-point values and stored as a float type. Data Definition Language | 37 None A column with a none affinity has no preference over storage class. Each value is stored as the type provided, with no attempt to convert anything. Since type affinities are not part of the SQL standard, SQLite has a series of rules that attempt to map traditional column types to the most logical type affinity. The type affinity of a column is determined by the declared type of the column, according to the following rules (substring matches are case-insensitive): 1. If no column type was given, then the column is given the none affinity. 2. If the column type contains the substring “INT,” then the column is given the integer affinity. 3. If the column type contains any of the substrings “CHAR,” “CLOB,” or “TEXT,” then the column is given the text affinity. 4. If the column type contains the substring “BLOB,” then the column is given the none affinity. 5. If the column type contains any of the substrings “REAL,” “FLOA,” or “DOUB,” then it is given the float affinity. 6. If no match is found, the column is assigned the numeric affinity. As implied by the first rule, the column type is completely optional. SQLite will allow you to create a table by simply naming the columns, such as CREATE TABLE t ( i, j, k);. You’ll also notice that there isn’t any specific list of column types that are recog- nized. You can use any column type you want, even making up your own names. This might sound a bit fast and loose for a typing system, but it works out quite well. By keying off specific substrings, rather than trying to define specific types, SQLite is able to handle SQL statements (and their database-specific types) from just about any database, all while doing a pretty good job of mapping the types to an appropriate affinity. About the only type you need to be careful of is “floating point.” The “int” in “point” will trigger rule 2 before the “floa” in “floating” will get to rule 5, and the column affinity will end up being integer. Column constraints In addition to column names and types, a table definition can also impose constraints on specific columns or sets of columns. A more complete view of the CREATE TABLE command looks something like this: CREATE TABLE table_name ( column_name column_type column_constraints..., [... ,] table_constraints, [...] ); 38 | Chapter 4: The SQL Language Here we see that each individual column can have additional, optional constraints and modifiers placed on it. Column constraints only affect the column for which they are defined, while table constraints can be used to define a constraint across one or more columns. Constraints that involve two or more columns must be defined as table constraints. Column constraints can be used to define a custom sort order (COLLATE colla tion_name). Collations determine how text values are sorted. In addition to user-defined collations, SQLite includes a NOCASE collation that ignores case when sorting. A default value (DEFAULT value) can also be assigned. Nearly all columns have a default of NULL. You can use the DEFAULT column constraint to set a different default value. The default can either be a literal or an expression. Expressions must be placed in parentheses. To help with date and time defaults, SQLite also includes three special keywords that may be used as a default value: CURRENT_TIME, CURRENT_DATE, and CURRENT_TIMESTAMP. These will record the UTC time, date, or date and time, respectively, when a new row is inserted. See “Date and Time Features” on page 159 for more information on date and time functions. Column constraints can also impose limits on a column, like denying NULL (NOT NULL) or requiring a unique value for each row (UNIQUE). Remember that a NULL is not considered a value, so UNIQUE does not imply NOT NULL, nor does UNIQUE imply only a single NULL is allowed. If you want to keep NULL assignments out of a UNIQUE column, you need to explicitly mark the column NOT NULL. Column values can also be subjected to arbitrary user-defined constraints before they are assigned (CHECK ( expression )). Some types of constraints also allow you to specify the action to be taken in situations when the constraint would be violated. See CREATE TABLE and UPDATE in Appendix C for more specific details. When multiple column constraints are used on a single column, the constraints are listed one after another without commas. Some examples: CREATE TABLE parts ( part_id INTEGER PRIMARY KEY, stock INTEGER DEFAULT 0 NOT NULL, desc TEXT CHECK( desc != '' ) -- empty strings not allowed ); In order to enforce a UNIQUE column constraint, a unique index will be automatically created over that column. A different index will be created for each column (or set of columns) marked UNIQUE. There is some expense in maintaining an index, so be aware that enforcing a UNIQUE column constraint can have performance considerations. Data Definition Language | 39 Primary keys In addition to these other constraints, a single column (or set of columns) can be des- ignated as the PRIMARY KEY. Each table can have only one primary key. Primary keys must be unique, so designating a column as PRIMARY KEY implies the UNIQUE constraint as well, and will result in an automatic unique index being created. If a column is marked both UNIQUE and PRIMARY KEY, only one index will be created. In SQLite, PRIMARY KEY does not imply NOT NULL. This is in contradiction to the SQL standard and is considered a bug, but the behavior is so long-standing that there are concerns about fixing it and breaking existing applications. As a result, it is always a good idea to explicitly mark at least one column from each PRIMARY KEY as NOT NULL. There are also some good design reasons for defining a primary key, which will be discussed in “Tables and Keys” on page 87, but the only significant, concrete thing that comes out of defining a PRIMARY KEY is the automatic unique index. There are also some minor syntax shortcuts. If, however, the primary key column has a type that is designated as INTEGER (and very specifically INTEGER), then that column becomes the table’s “root” column. SQLite must have some column that can be used to index the base storage for the table. In a sense, that column acts as the master index that is used to store the table itself. Like many other database systems, SQLite will silently create a hidden ROWID column for this purpose. Different database systems use different names so, in an effort to maintain compatibility, SQLite will recognize the names ROWID, OID, or _ROWID_ to ref- erence the root column. Normally ROWID columns are not returned (even for column wildcards), nor are their values included in dump files. If a table includes an INTEGER PRIMARY KEY column, then that column becomes an alias for the automatic ROWID column. You can still reference the column by any of the ROWID names, but you can also reference the column by its “real” user-defined name. Unlike PRIMARY KEY by itself, INTEGER PRIMARY KEY columns do have an automatic NOT NULL constraint associated with them. They are also strictly typed to only accept integer values. There are two significant advantages of INTEGER PRIMARY KEY columns. First, because the column aliases the table’s root ROWID column, there is no need for a secondary index. The table itself acts as the index, providing efficient lookups without the maintenance costs of an external index. Second, INTEGER PRIMARY KEY columns can automatically provide unique default values. When you insert a row without an explicit value for the ROWID (or ROWID alias) column, SQLite will automatically choose a value that is one greater than the largest existing value in the column. This provides an easy means to automatically generate unique keys. If the maximum value is reached, the database will randomly try other values, looking for an unused key. 40 | Chapter 4: The SQL Language INTEGER PRIMARY KEY columns can optionally be marked as AUTOINCREMENT. In that case, the automatically generated ID values will constantly increase, preventing the reuse of an ID value from a previously deleted row. If the maximum value is reached, insertions with automatic INTEGER PRIMARY KEY values are no longer possible. This is unlikely, however, as the INTEGER PRIMARY KEY type domain is large enough to allow 1,000 inserts per second for almost 300 million years. When using either automatic or AUTOINCREMENT values, it is always possible to insert an explicit ROWID (or ROWID alias) value. Other than the INTEGER PRIMARY KEY designation, SQLite offers no other type of automatic sequence feature. In addition to a PRIMARY KEY, columns can also be marked as a FOREIGN KEY. These columns reference rows in another (foreign) table. Foreign keys can be used to create links between rows in different tables. See “Tables and Keys” on page 87 for details. Table constraints Table definitions can also include table-level constraints. In general, table constraints and column constraints work the same way. Table-level constraints still operate on individual rows. The main difference is that using the table constraint syntax, you can apply the constraint to a group of columns rather than just a single column. It is perfectly legal to define a table constraint with only one column, effectively defining a column constraint. Multicolumn constraints are sometimes known as compound constraints. At the table level, SQLite supports the UNIQUE, CHECK, and PRIMARY KEY constraints. The check constraint is very similar, requiring only an expression (CHECK (expression)). Both the UNIQUE and PRIMARY KEY constraints, when given as a table constraint, require a list of columns (e.g., UNIQUE (column_name, [...]), PRIMARY KEY (column_name, [...])). As with column constraints, any table-level UNIQUE or PRIMARY KEY (which im- plies UNIQUE) constraint will automatically create a unique index over the appropriate columns. Table constraints that are applied to multiple columns use the set of columns as a group. For example, when UNIQUE or PRIMARY KEY is applied across more than one column, each individual column is allowed to have duplicate values. The constraint only prevents the set of values across the designated columns from being replicated. If you wanted each individual column to also be UNIQUE, you’d need to add the appropriate constraints to the individual columns. Consider a table that contains records of all the rooms in a multibuilding campus: CREATE TABLE rooms ( room_number INTEGER NOT NULL, building_number INTEGER NOT NULL, [...,] PRIMARY KEY( room_number, building_number ) ); Data Definition Language | 41 Clearly we need to allow for more than one room with the number 101. We also need to allow for more than one room in building 103. But there should only be one room 101 in building 103, so we apply the constraint across both columns. In this example, we’ve chosen to make these columns into a compound primary key, since the building number and room number combine to quintessentially define a specific room. De- pending on the design of the rest of the database, it might have been equally valid to define a simple UNIQUE constraint across these two columns, and designated an arbitrary room_id column as the primary key. Tables from queries You can also create a table from the output of a query. This is a slightly different CREATE TABLE syntax that creates a new table and preloads it with data, all in one command: CREATE [TEMP] TABLE table_name AS SELECT query_statement; Using this form, you do not designate the number of columns or their names or types. Rather, the query statement is run and the output is used to define the column names and preload the new table with data. With this syntax, there is no way to designate column constraints or modifiers. Any result column that is a direct column reference will inherit the column’s affinity, but and all columns are given a NONE affinity. The query statement consists of a SELECT command. More information on SELECT can be found in Chapter 5. Tables created in this manner are not dynamically updated—the query command is run only once when the table is created. Once the data is entered into the new table, it remains unaltered until you change it. If you need a table-like object that can dynam- ically update itself, use a VIEW (“Views” on page 43). This example shows the optional TEMP keyword (the full word TEMPORARY can also be used) in CREATE TEMP TABLE. This modifier can be used on any variation of CREATE TABLE, but is frequently used in conjunction with the ...AS SELECT... variation shown here. Temporary tables have two specific features. First, temporary tables can only be seen by the database connection that created them. This allows the simultaneous re- use of table names without any worry of conflict between different clients. Second, all associated temporary tables are automatically dropped and deleted whenever a data- base connection is closed. Generally speaking, CREATE TABLE...AS SELECT is not the best choice for creating standard tables. If you need to copy data from an old table into a new table, a better choice is to use CREATE TABLE to define an empty table with all of the appropriate column modifiers and table constraints. You can then bulk copy all the data into the new table using a variation of the INSERT command that allows for query statements. See “IN- SERT” on page 46 for details. 42 | Chapter 4: The SQL Language Altering tables SQLite supports a limited version of the ALTER TABLE command. Currently, there are only two operations supported by ALTER TABLE: add column and rename. The add col- umn variant allows you to add new columns to an existing table. It cannot remove them. New columns are always added to the end of the column list. Several other restrictions apply. If you need to make a more significant change while preserving as much data as possible, you can use the rename variant to rename the existing table, create a new table under the original name, and then copy the data from the old table to the new table. The old table can then be safely dropped. For full details, see ALTER TABLE in Appendix C. Dropping tables The CREATE TABLE command is used to create tables and DROP TABLE is used to delete them. The DROP TABLE command deletes a table and all of the data it contains. The table definition is also removed from the database system catalogs. The DROP TABLE command is very simple. The only argument is the name of the table you wish to drop: DROP TABLE table_name; In addition to deleting the table, DROP TABLE will also drop any indexes associated with the table. Both automatically created indexes (such as those used to enforce a UNIQUE constraint) as well as manually created indexes will be dropped. Virtual tables Virtual tables can be used to connect any data source to SQLite, including other data- bases. A virtual table is created with the CREATE VIRTUAL TABLE command. Although very similar to CREATE TABLE, there are important differences. For example, virtual tables cannot be made temporary, nor do they allow for an IF NOT EXISTS clause. To drop a virtual table, you use the normal DROP TABLE command. For more information on virtual tables, including the full syntax for CREATE VIRTUAL TABLE, see Chapter 10. Views Views provide a way to package queries into a predefined object. Once created, views act more or less like read-only tables. Just like tables, new views can be marked as TEMP, with the same result. The basic syntax of the CREATE VIEW command is: CREATE [TEMP] VIEW view_name AS SELECT query_statement Data Definition Language | 43 The CREATE VIEW syntax is almost identical to the CREATE TABLE...AS SELECT command. This is because both commands serve a similar purpose, with one important difference. The result of a CREATE TABLE command is a new table that contains a full copy of the data. The SELECT statement is run exactly once and the output of the query is stored in the newly defined table. Once created, the table will hold its own, independent copy of the data. A view, on the other hand, is fully dynamic. Every time the view is referenced or queried, the underlying SELECT statement is run to regenerate the view. This means the data seen in a view automatically updates as the data changes in the underlying tables. In a sense, views are almost like named queries. Views are commonly used in one of two ways. First, they can be used to package up commonly used queries into a more convenient form. This is especially true if the query is complex and prone to error. By creating a view, you can be sure to get the same query each time. Views are also commonly used to create user-friendly versions of standard tables. A common example are tables with date and time records. Normally, any time or date value is recorded in Coordinated Universal Time, or UTC. UTC is a more proper format for dates and times because it is unambiguous and time-zone independent. Unfortu- nately, it can also be a bit confusing if you’re several time zones away. It is often useful to create a view that mimics the base table, but converts all the times and dates from UTC into the local time zone. This way the data in the original tables remains un- changed, but the presentation is in units that are more user-friendly. Views are dropped with the DROP VIEW command: DROP VIEW view_name; Dropping a view will not have any effect on the tables it references. Indexes Indexes (or indices) are a means to optimize database lookups by pre-sorting and in- dexing one or more columns of a table. Ideally, this allows specific rows in a table to be found without having to scan every row in the table. In this fashion, indexes can provide a large performance boost to some types of queries. Indexes are not free, how- ever, requiring updates with each modification to a table as well as additional storage space. There are even some situations when an index will cause a drop in performance. See “Indexes” on page 107 for more information on when it makes sense to use an index. The basic syntax for creating an index specifies the name of the new index, as well as the table and column names that are indexed. Indexes are always associated with one (and only one) table, but they can include one or more columns from that table: CREATE [UNIQUE] INDEX index_name ON table_name ( column_name [, ...] ); 44 | Chapter 4: The SQL Language Normally indexes allow duplicate values. The optional UNIQUE keyword indicates that duplicate entries are not allowed, and any attempt to insert or update a table with a nonunique value will cause an error. For unique indexes that reference more than one column, all the columns must match for an entry to be considered duplicate. As dis- cussed with CREATE TABLE, NULL isn’t considered a value, so a UNIQUE index will not prevent one or more NULLs. If you want to prevent NULLs, you must indicate NOT NULL in the original table definition. Each index is tied to a specific table, but they all share a common namespace. Although you can name an index anything you like, it is standard practice to name an index with a standard prefix (such as idx_), and then include the table name and possibly the names of the columns included in the index. For example: CREATE INDEX idx_employees_name ON employees ( name ); This makes for long index names but, unlike table or view names, you typically only reference an index’s name when you create it and when you drop it. As with all the DROP commands, the DROP INDEX command is very simple, and requires only the name of the index that is being dropped: DROP INDEX index_name; Dropping an index will remove the index from the database, but will leave the associ- ated table intact. As described earlier, the CREATE TABLE command will automatically create unique in- dexes to enforce a UNIQUE or PRIMARY KEY constraint. All automatic indexes will start with an sqlite_ prefix. Because these indexes are required to enforce the table defini- tion, they cannot be manually dropped with the DROP INDEX command. Dropping the automatic indexes would alter the table behavior as defined by the original CREATE TABLE command. Conversely, if you have manually defined a UNIQUE index, dropping that index will allow the database to insert or update redundant data. Be careful when auditing indexes and remember that not all indexes are created for performance reasons. Data Manipulation Language The Data Manipulation Language is all about getting user data in and out of the data- base. After all the data structures and other database objects have been created with DDL commands, DML commands can be used to load those data structures full of useful data. The DML supported by SQLite falls into two basic categories. The first category con- sists of the “update” commands, which includes the actual UPDATE command, as well as the INSERT and DELETE commands. As you might guess, these commands are used to update (or modify), insert, and delete the rows of a table. All of these commands alter Data Manipulation Language | 45 the stored data in some way. The update commands are the primary means of managing all the data within a database. The second category consists of the “query” commands, which are used to extract data from the database. Actually, there is only one query command: SELECT. The SELECT command not only prints returned values, but provides a great number of options to combine different tables and rows and otherwise manipulate data before returning the final result. SELECT is, unquestionably, the most complex SQL command. It is also, arguably, the most important SQL command. This chapter will only cover the very basics of SELECT, and then we will spend the next chapter going through all of its parts, bit by bit. To address the full command syntax in detail, SELECT gets a whole chapter to itself (Chapter 5). Row Modification Commands There are three commands used for adding, modifying, and removing data from the database. INSERT adds new rows, UPDATE modifies existing rows, and DELETE removes rows. These three commands are used to maintain all of the actual data values within the database. All three update commands operate at a row level, adding, altering, or removing the specified rows. Although all three commands are capable of acting on multiple rows, each command can only directly act upon rows contained within a single table. INSERT The INSERT command is used to create new rows in the specified table. There are two meaningful versions of the command. The first version uses a VALUES clause to specify a list of values to insert: INSERT INTO table_name (column_name [, ...]) VALUES (new_value [, ...]); A table name is provided, along with a list of columns and a list of values. Both lists must have the same number of items. A single new row is created and each value is recorded into its respective column. The columns can be listed in any order, just as long as the list of columns and the list of values line up correctly. Any columns that are not listed will receive their default values: INSERT INTO parts ( name, stock, status ) VALUES ( 'Widget', 17, 'IN STOCK' ); In this example, we attempt to insert a new row into a “parts” table. Note the use of single quotes for text literals. Technically, the list of column names is optional. If no explicit list of columns is pro- vided, the INSERT command will attempt to pair up values with the table’s full list of columns: INSERT INTO table_name VALUES (new_value [, ...]); 46 | Chapter 4: The SQL Language The trick with this format is that the number and order of values must exactly match the number and order of columns in the table definition. That means it is impossible to use default values, even on INTEGER PRIMARY KEY columns. More often than not, this is not actually desirable. This format is also harder to maintain within application source code, since it must be meticulously updated if the table format changes. In general, it is recommended that you always explicitly list out the columns in an INSERT statement. When bulk importing data, it is common to loop over data sets, calling INSERT over and over. Processing these statements one at a time can be fairly slow, since each command will update both the table and any relevant indexes, and then make sure the data is fully written out to physical disk before (finally!) starting the next INSERT. This is a fairly lengthly process, since it requires physical I/O. To speed up bulk inserts, it is common to wrap groups of 1,000 to 10,000 INSERT statements into a single transaction. Grouping the statement together will substantially increase the overall speed of the inserts by delaying the physical I/O until the end of the transaction. See “Transaction Control Language” on page 51 for more informa- tion on transactions. Bulk inserts can be sped up by wrapping large groups of INSERT com- mands inside a transaction. The second version of INSERT allows you to define values by using a query statement. This is very similar to the CREATE TABLE...AS SELECT command, although the table must already exist. This is the only version of INSERT that can insert more than one row with a single command: INSERT INTO table_name (column_name, [...]) SELECT query_statement; This type of INSERT is most commonly used to bulk copy data from one table to another. This is a common operation when you need to update the definition of a table, but you don’t want to lose all the data that already exists in the database. The old table is renamed, the new table is defined, and the data is copied from the old table into the new table using an INSERT INTO...SELECT command. This form can also be used to populate temporary tables or copy data from one attached database to another. As with the VALUES version of INSERT, the column list is technically optional but, for all the same reasons, it is still recommended that you provide an explicit column list. All versions of the INSERT command also support an optional conflict resolution clause. This conflict clause determines what should be done if the results of the INSERT would violate a database constraint. The most common example is INSERT OR REPLACE, which comes into play when the INSERT would, as executed, cause a UNIQUE constraint viola- tion. If the REPLACE conflict resolution is present, any existing row that would cause a Data Manipulation Language | 47 UNIQUE constraint violation is first deleted, and then the INSERT is allowed to continue. This specific usage pattern is so common that the whole INSERT OR REPLACE phrase can be replaced by just REPLACE. For example, REPLACE INTO table_name.... See INSERT and UPDATE in Appendix C for more information on the details of conflict resolution. UPDATE The UPDATE command is used to assign new values to one or more columns of existing rows in a table. The command can update more than one row, but all of the rows must be part of the same table. The basic syntax is: UPDATE table_name SET column_name=new_value [, ...] WHERE expression The command requires a table name followed by a list of column name/value pairs that should be assigned. Which rows are updated is determined by a conditional expression that is tested against each row of the table. The most common usage pattern uses the expression to check for equality on some unique column, such as a PRIMARY KEY column. If no WHERE condition is given, the UPDATE command will attempt to up- date the designated columns in every row of a table. It is not considered an error if the WHERE expression evaluates to false for every row in the table, resulting in no actual updates. Here is a more specific example: -- Update the price and stock of part_id 454: UPDATE parts SET price = 4.25, stock = 75 WHERE part_id = 454; This example assumes that the table parts has at least three columns: price, stock, and part_id. The database will find each row with a part_id of 454. In this case, it can be assumed that part_id is a PRIMARY KEY column, so only one row will be updated. The price and stock columns of that row are then assigned new values. The full syntax for UPDATE can be found at UPDATE in Appendix C. DELETE As you might guess, the DELETE command is used to delete or remove one or more rows from a single table. The rows are completely deleted from the table: DELETE FROM table_name WHERE expression; The command requires only a table name and a conditional expression to pick out rows. The WHERE expression is used to select specific rows to delete, just as it is used in the UPDATE command. 48 | Chapter 4: The SQL Language If no WHERE condition is given, the DELETE command will attempt to delete every row of a table. As with UPDATE, it is not considered an error if the WHERE expression evaluates to false for every row in the table, resulting in no actual deletions. Some specific examples: -- Delete the row with rowid 385: DELETE FROM parts WHERE part_id = 385; -- Delete all rows with a rowid greater than or equal to 43 -- and less than or equal to 246: DELETE FROM parts WHERE part_id >= 43 AND part_id <= 246; These examples assume we have a table named parts that contains at least one unique column named part_id. As noted, if no WHERE clause is given, the DELETE command will attempt to delete every row in a table. SQLite optimizes this specific case, truncating the full table, rather than processing each individual row. Truncating the table is much faster than deleting each individual row, but truncation bypasses the individual row processing. If you wish to process each row as it is deleted, provide a WHERE clause that always evaluates to true: DELETE FROM parts WHERE 1; -- delete all rows, force per-row processing The existence of the WHERE clause will prevent the truncation, allowing each row to be processed in turn. The Query Command The final DML command to cover is the SELECT command. SELECT is used to extract or return values from the database. Almost any time you want to extract or return some kind of value, you’ll need to use the SELECT command. Generally, the returned values are derived from the contents of the database, but SELECT can also be used to return the value of simple expressions. This is a great way to test out expressions, for example: sqlite> SELECT 1+1, 5*32, 'abc'||'def', 1>2; 1+1 5*32 'abc' || 'def' 1>2 ---------- ---------- -------------- ---------- 2 160 abcdef 0 SELECT is a read-only command, and will not modify the database (unless the SELECT is embedded in a different command, such as a CREATE TABLE...AS SELECT or an INSERT INTO...SELECT). Without question, SELECT is the most complex SQL command, both in terms of syntax as well as function. The SELECT syntax tries to represent a generic framework that is capable of expressing a great many different types of queries. While it is somewhat successful at this, there are areas where SELECT has traded away simplicity for more Data Manipulation Language | 49 flexibility. As a result, SELECT has a large number of optional clauses, each with its own set of options and formats. Understanding how to mix and match these optional clauses to get the result you’re looking for can take some time. While the most basic syntax can be shown with a good set of examples, to really wrap your head around SELECT, it is best to understand how it actually works and what it is trying to accomplish. Because SELECT can be so complex, and because SELECT is an extremely important com- mand, we will spend the whole next chapter looking very closely at SELECT and each of its clauses. There will be some discussion about what is going on behind the scenes, to provide more insight into how to read and write complex queries. For now, we’ll just give you a taste. That should provide enough information to play around with the other commands in this chapter. The most basic form of SELECT is: SELECT output_list FROM input_table WHERE row_filter; The output list is a list of expressions that should be evaluated and returned for each resulting row. Most commonly, this is simply a list of columns. The output list can also include a wildcard (*) that indicates all known columns should be returned. The FROM clause defines the source of the table data. The next chapter will show how tables can be linked and joined, but for now we’ll stick with querying one table at a time. The WHERE clause is a conditional filtering expression that is applied to each row. It is essentially the same as the WHERE clause in the UPDATE and DELETE commands. Those rows that evaluate to true will be part of the result, while the other rows will be filtered out. Consider this table: sqlite> CREATE TABLE tbl ( a, b, c, id INTEGER PRIMARY KEY ); sqlite> INSERT INTO tbl ( a, b, c ) VALUES ( 10, 10, 10 ); sqlite> INSERT INTO tbl ( a, b, c ) VALUES ( 11, 15, 20 ); sqlite> INSERT INTO tbl ( a, b, c ) VALUES ( 12, 20, 30 ); We can return the whole table like this: sqlite> SELECT * FROM tbl; a b c id ---------- ---------- ---------- ---------- 10 10 10 1 11 15 20 2 12 20 30 3 We can also just return specific columns: sqlite> SELECT a, c FROM tbl; a c ---------- ---------- 10 10 11 20 12 30 50 | Chapter 4: The SQL Language Or specific rows: sqlite> SELECT * FROM tbl WHERE id = 2; a b c id ---------- ---------- ---------- ---------- 11 15 20 2 For more specifics, see Chapter 5 and SELECT in Appendix C. Transaction Control Language The Transaction Control Language is used in conjunction with the Data Manipulation Language to control the processing and exposure of changes. Transactions are a fundamental part of how relational databases protect the integrity and reliability of the data they hold. Transactions are automatically used on all DDL and DML commands. ACID Transactions A transaction is used to group together a series of low-level changes into a single, logical update. A transaction can be anything from updating a single value to a complex, mul- tistep procedure that might end up inserting several rows into a number of different tables. The classic transaction example is a database that holds account numbers and balances. If you want to transfer a balance from one account to another, that is a simple two-step process: subtract an amount from one account balance and then add the same amount to the other account balance. That process needs to be done as a single logical unit of change, and should not be broken apart. Both steps should either succeed completely, resulting in the balance being correctly transferred, or both steps should fail completely, resulting in both accounts being left unchanged. Any other outcome, where one step succeeds and the other fails, is not acceptable. Typically a transaction is opened, or started. As individual data manipulation com- mands are issued, they become part of the transaction. When the logical procedure has finished, the transaction can be committed, which applies all of the changes to the permanent database record. If, for any reason, the commit fails, the transaction is rolled back, removing all traces of the changes. A transaction can also be manually rolled back. The standard for reliable, robust transactions is the ACID test. ACID stands for Atomic, Consistent, Isolated, and Durable. Any transaction system worth using must possess these qualities. Atomic A transaction should be atomic, in the sense that the change cannot be broken down into smaller pieces. When a transaction is committed to the database, the entire transaction must be applied or the entire transaction must not be applied. It should be impossible for only part of a transaction to be applied. Transaction Control Language | 51 Consistent A transaction should also keep the database consistent. A typical database has a number of rules and limits that help ensure the stored data is correct and consistent with the design of the database. Assuming a database starts in a consistent state, applying a transaction must keep the database consistent. This is important, be- cause the database is allowed to (and is often required to) become inconsistent while the transaction is open. For example, while transferring funds, there is a moment between the subtraction from one account and the addition to another account that the total amount of funds represented in the database is altered and may become inconsistent with a recorded total. This is acceptable, as long as the transaction, as a whole, is consistent when it is committed. Isolated An open transaction must also be isolated from other clients. When a client opens a transaction and starts to issue individual change commands, the results of those commands are visible to the client. Those changes should not, however, be visible to any other system accessing the database, nor should they be integrated into the permanent database record until the entire transaction is committed. Conversely, changes committed by other clients after the transaction was started should not be visible to this transaction. Isolation is required for transactions to be atomic and consistent. If other clients could see half-applied transactions, the transactions could not claim to be atomic in nature, nor would they preserve the consistency of the database, as seen by other clients. Durable Last of all, a transaction must be durable. If the transaction is successfully com- mitted, it must have become a permanent and irreversible part of the database record. Once a success status is returned, it should not matter if the process is killed, the system loses power, or the database filesystem disappears—upon restart, the committed changes should be present in the database. Conversely, if the system loses power before a transaction is committed, then upon restart the changes made within the transaction should not be present. Most people think that the atomic nature of transactions is their most important quality, but all four aspects must work together to ensure the overall integrity of the database. Durability, especially, is often overlooked. SQLite tries extremely hard to guarantee that if a transaction is successfully committed, those changes are actually physically written to permanent storage and are there to stay. Compare this to tradi- tional filesystem operations, where writes might go into an operating system file cache. Updates may sit in the cache anywhere from a few seconds to a few minutes before finally being spooled off to storage. Even then, it is possible for the data to wait around in device buffers before finally being committed to physical storage. While this type of buffering can increase efficiency, it means that a normal application really has no idea when its data is safely committed to permanent storage. 52 | Chapter 4: The SQL Language Power failures and disappearing filesystems may seem like rare occurrences, but that’s not really the point. Databases are designed to deal with absolutes, especially when it comes to reliability. Besides, having a filesystem disappear is not that radical of an idea when you consider the prevalence of flash drives and USB thumb drives. Disappearing media and power failures are even more commonplace when you consider the number of SQLite databases that are found on battery-operated, handheld devices such as mobile phones and media players. The use of transactions is even more important on devices like this, since it is nearly impossible to run data recovery tools in that type of environment. These types of devices must be extremely robust and, no matter what the user does (including yanking out flash drives at inconvenient times), the system must stay consistent and reliable. Use of a transactional system can provide that kind of reliability. Transactions are not just for writing data. Opening a transaction for an extended read- only operation is sometimes useful if you need to gather data with multiple queries. Having the transaction open keeps your view of the database consistent, ensuring that the data doesn’t change between queries. That is useful if, for example, you use one query to gather a bunch of record IDs, and then issue a series of queries against each ID value. Wrapping all the queries in a transaction guarantees all of the queries see the same set of data. SQL Transactions Normally, SQLite is in autocommit mode. This means that SQLite will automatically start a transaction for each command, process the command, and (assuming no errors were generated) automatically commit the transaction. This process is transparent to the user, but it is important to realize that even individually entered commands are processed within a transaction, even if no TCL commands are used. The autocommit mode can be disabled by explicitly opening a transaction. The BEGIN command is used to start or open a transaction. Once an explicit transaction has been opened, it will remain open until it is committed or rolled back. The keyword TRANSAC TION is optional: BEGIN [ DEFERRED | IMMEDIATE | EXCLUSIVE ] [TRANSACTION] The optional keywords DEFERRED, IMMEDIATE, or EXCLUSIVE are specific to SQLite and control how the required read/write locks are acquired. If only one client is accessing the database at a time, the locking mode is largely irrelevant. When more than one client may be accessing the database, the locking mode defines how to balance peer access with ensured success of the transaction. By default, all transactions (including autocommit transactions) use the DEFERRED mode. Under this mode, none of the database locks are acquired until they are required. This is the most “neighborly” mode and allows other clients to continue accessing and using the database until the transaction has no other choice but to lock them out. This allows Transaction Control Language | 53 other clients to continue using the database, but if the locks are not available when the transaction requires them, the transaction will fail and may need to be rolled back and restarted. BEGIN IMMEDIATE attempts to acquire a reserved lock immediately. If it succeeds, it guarantees the write locks will be available to the transaction when they are needed, but still allows other clients to continue to access the database for read-only operations. The EXCLUSIVE mode attempts to lock out all other clients, including read-only clients. Although the IMMEDIATE and EXCLUSIVE modes are more restrictive to other clients, the advantage is that they will fail immediately if the required locks are not available, rather than after you’ve issued your DDL or DML commands. Once a transaction is open, you can continue to issue other SQL commands, including both DML and DDL commands. You can think of the changes resulting from these commands as “proposed” changes. The changes are only visible to the local client and have not been fully and permanently applied to the database. If the client process is killed or the server loses power in the middle of an open transaction, the transaction and any proposed changes it has will be lost, but the rest of the database will remain intact and consistent. It is not until the transaction is closed that the proposed changes are committed to the database and made “real.” The COMMIT command is used to close out a transaction and commit the changes to the database. You can also use the alias END. As with BEGIN, the TRANSACTION keyword is optional. COMMIT [TRANSACTION] END [TRANSACTION] Once a COMMIT has successfully returned, all the proposed changes are fully committed to the database and become visible to other clients. At that point, if the system loses power or the client process is killed, the changes will remain safely in the database. Things don’t always go right, however. Rather than committing the proposed changes, the transaction can be manually rolled back, effectively canceling the transaction and all of the changes it contains. Rolling back a set of proposed changes is useful if an error is encountered. This might be a database error, such as running out of disk space half- way through inserting a series of related records, or it might be an application logic error, such as trying to assign an invoice to an order that doesn’t exist. In such cases, it usually doesn’t make sense to continue with the transaction, nor does it make sense to leave inconsistent data in the database. Pretty much the only choice is to back out and try again. To cancel the transaction and roll back all the proposed changes, you can use the ROLLBACK command. Again, the keyword TRANSACTION is optional: ROLLBACK [TRANSACTION] 54 | Chapter 4: The SQL Language ROLLBACK will undo and revert all the proposed changes made by the current transaction and then close the transaction. It does not necessarily return the database to its prior state, as other clients may have been making changes in parallel. A ROLLBACK only cancels the proposed changes made by this client within the current transaction. Both COMMIT and ROLLBACK will end the current transaction, putting SQLite back into autocommit mode. Save-Points In addition to ACID-compliant transactions, SQLite also supports save-points. Save- points allow you to mark specific points in the transaction. You can then accept or rollback to individual save-points without having to commit or rollback an entire transaction. Unlike transactions, you can have more than one save-point active at the same time. Save-points are sometimes called nested transactions. Save-points are generally used in conjunction with large, multistep transactions, where some of the steps or sub-procedures require rollback ability. Save-points allow a trans- action to proceed and (if required) roll back one step at a time. They also allow an application to explore different avenues, attempting one procedure, and if that doesn’t work, trying another, without having to roll back the entire transaction to start over. In a sense, save-points can be thought of as “undo” markers in SQL command stream. You can create a save-point with the SAVEPOINT command. Since multiple save-points can be defined, you must provide a name to identify the save-point: SAVEPOINT savepoint_name Save-points act as a stack. Whenever you create a new one, it is put at the top of the stack. Save-point identifiers do not need to be unique. If the same save-point identifier is used more than once, the one nearest to the top of the stack is used. To release a save-point and accept all of the proposed changes made since the save- point was set, use the RELEASE command: RELEASE [SAVEPOINT] savepoint_name The RELEASE command does not commit any changes to disk. Rather, it flattens all of the changes in the save-point stack into the layer below the named save-point. The save-point is then removed. Any save-points contained by the named save-point are automatically released. To cancel a set of commands and undo everything back to where a save-point was set, use the ROLLBACK TO command: ROLLBACK [TRANSACTION] TO [SAVEPOINT] savepoint_name Transaction Control Language | 55 Unlike a transaction ROLLBACK, a save-point ROLLBACK TO does not close out and eliminate the save-point. ROLLBACK TO rolls back and cancels any changes issued since the save- point was established, but leaves the transaction state exactly as it was after the SAVE POINT command was issued. Consider the following series of SQL statements. The indentation is used to show the save-point stack: CREATE TABLE t (i); BEGIN; INSERT INTO t (i) VALUES 1; SAVEPOINT aaa; INSERT INTO t (i) VALUES 2; SAVEPOINT bbb; INSERT INTO t (i) VALUES 3; At this point, if the command ROLLBACK TO bbb is issued, the state of the database would be as if the following commands were entered: CREATE TABLE t (i); BEGIN; INSERT INTO t (i) VALUES 1; SAVEPOINT aaa; INSERT INTO t (i) VALUES 2; SAVEPOINT bbb; Again, notice that rolling back to save-point bbb still leaves the save-point in place. Any new commands will be associated with SAVEPOINT bbb. For example: CREATE TABLE t (i); BEGIN; INSERT INTO t (i) VALUES 1; SAVEPOINT aaa; INSERT INTO t (i) VALUES 2; SAVEPOINT bbb; DELETE FROM t WHERE i=1; Continuing, if the command RELEASE aaa was issued, we would get the equivalent of: CREATE TABLE t (i); BEGIN; INSERT INTO t (i) VALUES 1; INSERT INTO t (i) VALUES 2; DELETE FROM t WHERE i=1; In this case, the proposed changes from both the aaa and the enclosed bbb save-points were released and merged outward. The transaction is still open, however, and a COMMIT would still be required to make the proposed changes permanent. Even if you have open save-points, you can still issue transaction commands. If the enclosing transaction is committed, all outstanding save-points will automatically be released and then committed. If the transaction is rolled back, all the save-points are rolled back. 56 | Chapter 4: The SQL Language If the SAVEPOINT command is issued when SQLite is in autocommit mode—that is, outside of a transaction—then a standard autocommit BEGIN DEFERRED TRANSACTION will be started. However, unlike with most commands, the autocommit transaction will not automatically commit after the SAVEPOINT command returns, leaving the system inside an open transaction. The automatic transaction will remain active until the orig- inal save-point is released, or the outer transaction is either explicitly committed or rolled back. This is the only situation when a save-point RELEASE will have a direct effect on the enclosing transaction. As with other save-points, if an autocommit save-point is rolled back, the transaction will remain open and the original save-point will be open, but empty. System Catalogs Many relational database systems, including SQLite, keep system state data in a series of data structures known as system catalogs. All of the SQLite system catalogs start with the prefix sqlite_. Although many of these catalogs contain internal data, they can be queried, using SELECT, just as if they were standard tables. Most system catalogs are read-only. If you encounter an unknown database and you’re not sure what’s in it, examining the system catalogs is a good place to start. All nontemporary SQLite databases have an sqlite_master catalog. This is the master record of all database objects. If any of the tables has a populated AUTOINCREMENT col- umn, the database will also have an sqlite_sequence catalog. This catalog is used to keep track of the next valid sequence value (for more information on AUTOINCREMENT, see “Primary keys” on page 40). If the SQL command ANALYZE has been used, it will also generate one or more sqlite_stat# tables, such as sqlite_stat1 and sqlite_stat2. These tables hold various statistics about the values and distributions in various indexes, and are used to help the query optimizer pick the more efficient query solution. For more information, see ANALYZE in Appendix C. The most important of these system catalogs is the sqlite_master table. This catalog contains information on all the objects within a database, including the SQL used to define them. The sqlite_master table has five columns: Column name Column type Meaning type Text Type of database object name Text Identifier name of object tbl_name Text Name of associated table rootpage Integer Internal use only sql Text SQL used to define object System Catalogs | 57 The type column can be table (including virtual tables), index, view, or trigger. The name column gives the name of the object itself, while the tbl_name column gives the name of the table or view the object is associated with. For tables and views, the tbl_name is just a copy of the name column. The final sql column holds a full copy of the original SQL command used to define the object, such as a CREATE TABLE or CREATE TRIGGER command. Temporary databases do not have an sqlite_master system catalog. Rather, they have an sqlite_temp_master table instead. Wrap-up This is a long chapter packed with a huge amount of information. Even if you’re familiar with a number of traditional programming languages, the declarative nature of SQL often takes time to wrap your head around. One of the best ways to learn SQL is to simply experiment. SQLite makes it easy to open up a test database, create some tables, and try things out. If you’re having problems understanding the details of a command, be sure to look it up in Appendix C. In addition to more detailed descriptions, Appen- dix C contains detailed syntax diagrams . If you wish to make a deeper study into SQL, there are literally hundreds of books to choose from. O’Reilly alone publishes a dozen or so titles just on the SQL language. While there are some differences between the SQL supported by SQLite and other major database systems, SQLite follows the standard fairly closely. Most of the time, SQLite deviates from the standard, it does so in an attempt to support common nota- tions or usage in other popular database products. If you’re working on wrapping you’re head around some of the higher level concepts, or basic query structures, a tutorial or book written for just about any database product is likely to help. There might be a small bit of tweaking to get the queries to run under SQLite, but the changes are usually minimal. Popular O’Reilly books covering the SQL language include Learning SQL (Beaulieu), SQL in a Nutshell (Kline, Kline, Hunt), and the SQL Cookbook (Molinaro). More ad- vanced discussions can be found in The Art of SQL (Faroult, Robson). Popular reference books also include SQL For Smarties (Celko, Morgan Kaufmann) and Introduction to SQL (van der Lans, Addison Wesley). These two are large, but very complete. There is also The SQL Guide to SQLite (van der Lans, lulu.com), which takes a much deeper look at the SQL dialect specifically used by SQLite. There are also thousands of websites and online tutorials, communities, and forums, including the SQLite mailing lists, where you can often get insightful answers to intel- ligent questions. 58 | Chapter 4: The SQL Language Before trying too much, be sure to read the next chapter. The next chapter is devoted to the SELECT command. In addition to covering the syntax of the command, it dives a bit deeper into what is going on behind the scenes. That foundation knowledge should make it much easier to break down and understand complex queries. It should also make it much easier to write them. Wrap-up | 59 CHAPTER 5 The SELECT Command The SELECT command is used to extract data from the database. Any time you want to query or return user data stored in a database table, you’ll need to use the SELECT command. In terms of both syntax and functionality, SELECT is the most complex SQL command. In most applications, it is also one of the most frequently used commands. If you want to get the most out of your database (and your database designs), you’ll need a solid understanding of how to properly use SELECT. Generally, the returned values are derived from the contents of the database, but SELECT can also be used to return the value of simple expressions. SELECT is a read-only command, and will not modify the database (unless the SELECT is embedded in a dif- ferent command, such as INSERT INTO...SELECT). SQL Tables The main SQL data structure is the table. Tables are used for both storage and for data manipulation. We’ve seen how to define tables with the CREATE TABLE command, but let’s look at some of the details. A table consists of a heading and a body. The heading defines the name and type (in SQLite, the affinity) of each column. Column names must be unique within the table. The heading also defines the order of the columns, which is fixed as part of the table definition. The table body contains all of the rows. Each row consists of one data element for each column. All of the rows in a table must have the same number of data elements, one for each column. Each element can hold exactly one data value (or a NULL). SQL tables are allowed to hold duplicate rows. Tables can contain multiple rows where every user-defined column has an equivalent corresponding value. Duplicate rows are normally undesirable in practice, but they are allowed. 61 The rows in an SQL table are unordered. When a table is displayed or written down, it will have some inherent ordering, but conceptually tables have no ordering. The order of insertion has no meaning to the database. The rows of an SQL table have no defined order. A very common mistake is to assume a given query will always return rows in the same order. Unless you’ve specifically asked a query to sort the returned rows in a specific order, there is no guarantee the rows will continue to come back in the same order. Don’t let your application code become dependent on the natural ordering of an un- ordered query. A different version of SQLite may optimize the query differently, re- sulting in a different row ordering. Even something as simple as adding or dropping an index can alter the row ordering of an unsorted result. To verify your code is making no assumptions about row order, you can turn on PRAGMA reverse_unordered_selects. This will cause SQLite to reverse the natural row ordering of any SELECT statement that does not have an explicit order (an ORDER BY clause). See reverse_unordered_selects in Appendix F for more details. The SELECT Pipeline The SELECT syntax tries to represent a generic framework that is capable of expressing many different types of queries. To achieve this, SELECT has a large number of optional clauses, each with its own set of options and formats. The most general format of a standalone SQLite SELECT statement looks like this: SELECT [DISTINCT] select_heading FROM source_tables WHERE filter_expression GROUP BY grouping_expressions HAVING filter_expression ORDER BY ordering_expressions LIMIT count OFFSET count Every SELECT command must have a select heading, which defines the returned values. Each additional line (FROM, WHERE, GROUP BY, etc.) represents an optional clause. Each clause represents a step in the SELECT pipeline. Conceptually, the result of a SELECT statement is calculated by generating a working table, and then passing that table through the pipeline. Each step takes the working table as input, performs a spe- cific operation or manipulation, and passes the modified table on to the next step. Manipulations operate the whole working table, similar to vector or matrix operations. 62 | Chapter 5: The SELECT Command Practically, the database engine takes a few shortcuts and makes plenty of optimizations when processing a query, but the end result should always match what you would get from independently going through each step, one at a time. The clauses in a SELECT statement are not evaluated in the same order they are written. Rather, their evaluation order looks something like this: 1. FROM source_tables Designates one or more source tables and combines them together into one large working table. 2. WHERE filter_expression Filters specific rows out of the working table. 3. GROUP BY grouping_expressions Groups sets of rows in the working table based off similar values. 4. SELECT select_heading Defines the result set columns and (if applicable) grouping aggregates. 5. HAVING filter_expression Filters specific rows out of the grouped table. Requires a GROUP BY. 6. DISTINCT Eliminates duplicate rows. 7. ORDER BY ordering_expressions Sorts the rows of the result set. 8. OFFSET count Skips over rows at the beginning of the result set. Requires a LIMIT. 9. LIMIT count Limits the result set output to a specific number of rows. No matter how large or complex a SELECT statement may be, they all follow this basic pattern. To understand how any query works, break it down and look at each individual step. Make sure you understand what the working table looks like before each step, how that step manipulates and modifies the table, and what the working table looks like when it is passed to the next step. FROM Clause The FROM clause takes one or more source tables from the database and combines them into one large table. Source tables are usually named tables from the database, but they can also be views or subqueries (see “Subqueries” on page 76 for more details on subqueries). The SELECT Pipeline | 63 Tables are combined using the JOIN operator. Each JOIN combines two tables into a larger table. Three or more tables can be joined together by stringing a series of JOIN operators together. JOIN operators are evaluated left-to-right, but there are several dif- ferent types of joins, and not all of them are commutative or associative. This makes the ordering and grouping very important. If necessary, parentheses can be used to group the joins correctly. Joins are the most important and most powerful database operator. Joins are the only way to bring together information stored in different tables. As we’ll see in the next chapter, nearly all of database design theory assumes the user is comfortable with joins. If you can master joins, you’ll be well on your way to mastering relational databases. SQL defines three major types of joins: the CROSS JOIN, the INNER JOIN, and the OUTER JOIN. CROSS JOIN A CROSS JOIN matches every row of the first table with every row of the second table. If the input tables have x and y columns, respectively, the resulting table will have x+y columns. If the input tables have n and m rows, respectively, the resulting table will have n·m rows. In mathematics, a CROSS JOIN is known as a Cartesian product. The syntax for a CROSS JOIN is quite simple: SELECT ... FROM t1 CROSS JOIN t2 ... Figure 5-1 shows how a CROSS JOIN is calculated. Because CROSS JOINs have the potential to generate extremely large tables, care must be taken to only use them when appropriate. Figure 5-1. In a CROSS JOIN, each row from the first table is matched to each row in the second table. 64 | Chapter 5: The SELECT Command INNER JOIN An INNER JOIN is very similar to a CROSS JOIN, but it has a built-in condition that is used to limit the number of rows in the resulting table. The conditional is normally used to pair up or match rows from the two source tables. An INNER JOIN without any type of conditional expression (or one that always evaluates to true) will result in a CROSS JOIN. If the input tables have x and y columns, respectively, the resulting table will have no more than x+y columns (in some cases, it can have fewer). If the input tables have n and m rows, respectively, the resulting table can have anywhere from zero to n·m rows, depending on the condition. An INNER JOIN is the most common type of join, and is the default type of join. This makes the INNER keyword optional. There are three primary ways to specify the conditional. The first is with an ON expres- sion. This provides a simple expression that is evaluated for each potential row. Only those rows that evaluate to true are actually joined. A JOIN...ON looks like this: SELECT ... FROM t1 JOIN t2 ON conditional_expression ... An example of this is shown in Figure 5-2. Figure 5-2. In an INNER JOIN, the rows are matched based off a condition. If the input tables have C and D columns, respectively, a JOIN...ON will always result in C+D columns. The conditional expression can be used to test for anything, but the most common type of expression tests for equality between similar columns in both tables. For example, in a business employee database, there is likely to be an employee table that contains (among other things) a name column and an eid column (employee ID number). Any other table that needs to associate rows to a specific employee will also have an eid column that acts as a pointer or reference to the correct employee. This relationship makes it very common to have queries with ON expressions similar to: SELECT ... FROM employee JOIN resource ON employee.eid = resource.eid ... The SELECT Pipeline | 65 This query will result in an output table where the rows from the resource table are correctly matched to their corresponding rows in the employee table. This JOIN has two issues. First, that ON condition is a lot to type out for something so common. Second, the resulting table will have two eid columns, but for any given row, the values of those two columns will always be identical. To avoid redundancy and keep the phrasing shorter, inner join conditions can be declared with a USING expres- sion. This expression specifies a list of one or more columns: SELECT ... FROM t1 JOIN t2 USING ( col1 ,... ) ... Queries from the employee database would now look something like this: SELECT ... FROM employee JOIN resource USING ( eid ) ... To appear in a USING condition, the column name must exist in both tables. For each listed column name, the USING condition will test for equality between the pairs of columns. The resulting table will have only one instance of each listed column. If this wasn’t concise enough, SQL provides an additional shortcut. A NATURAL JOIN is similar to a JOIN...USING, only it automatically tests for equality between the values of every column that exists in both tables: SELECT ... FROM t1 NATURAL JOIN t2 ... If the input tables have x and y columns, respectively, a JOIN...USING or a NATURAL JOIN will result in anywhere from max(x,y) to x+y columns. Assuming eid is the only column identifier to appear in both the employee and resource table, our business query becomes extremely simple: SELECT ... FROM employee NATURAL JOIN resource ... NATURAL JOINs are convenient, as they are very concise, and allow changes to the key structure of various tables without having to update all of the corresponding queries. They can also be a tad dangerous unless you follow some discipline in naming your columns. Because none of the columns are explicitly named, there is no error checking in the sanity of the join. For example, if no matching columns are found, the JOIN will automatically (and without warning) degrade to a CROSS JOIN, just like any other INNER JOIN. Similarly, if two columns accidentally end up with the same name, a NATURAL JOIN will automatically include them in the join condition, if you wanted it or not. OUTER JOIN The OUTER JOIN is an extension of the INNER JOIN. The SQL standard defines three types of OUTER JOINs: LEFT, RIGHT, and FULL. Currently, SQLite only supports the LEFT OUTER JOIN. OUTER JOINs have a conditional that is identical to INNER JOINs, expressed using an ON, USING, or NATURAL keyword. The initial results table is calculated the same way. Once the primary JOIN is calculated, an OUTER join will take any unjoined rows from one or 66 | Chapter 5: The SELECT Command both tables, pad them out with NULLs, and append them to the resulting table. In the case of a LEFT OUTER JOIN, this is done with any unmatched rows from the first table (the table that appears to the left of the word JOIN). Figure 5-3 shows an example of a LEFT OUTER JOIN. Figure 5-3. An OUTER JOIN is just like an INNER JOIN, only unmatched rows are included in the results table. This shows a LEFT OUTER JOIN, where unmatched rows from the left (t1) table are added to the results. The result of a LEFT OUTER JOIN will contain at least one instance of every row from the lefthand table. If the input tables have x and y columns, respectively, the resulting table will have no more than x+y columns (the exact number depends on which conditional is used). If the input tables have n and m rows, respectively, the resulting table can have anywhere from n to n·m rows. Because they include unmatched rows, OUTER JOINs are often specifically used to search for unresolved or “dangling” rows. Table aliases Because the JOIN operator combines the columns of different tables into one, larger table, there may be cases when the resulting working table has multiple columns with the same name. To avoid ambiguity, any part of the SELECT statement can qualify any column reference with a source-table name. However, there are some cases when this is still not enough. For example, there are some situations when you need to join a table to itself, resulting in the working table having two instances of the same source-table. Not only does this make every column name ambiguous, it makes it impossible to distinguish them using the source-table name. Another problem is with subqueries, as they don’t have concrete source-table names. The SELECT Pipeline | 67 To avoid ambiguity within the SELECT statement, any instance of a source-table, view, or subquery can be assigned an alias. This is done with the AS keyword. For example, in the cause of a self-join, we can assign a unique alias for each instance of the same table: SELECT ... FROM x AS x1 JOIN x AS x2 ON x1.col1 = x2.col2 ... Or, in the case of a subquery: SELECT ... FROM ( SELECT ... ) AS sub ... Technically, the AS keyword is optional, and each source-table name can simply be followed with an alias name. This can be quite confusing, however, so it is re- commended you use the AS keyword. If any of the subquery columns conflict with a column from a standard source table, you can now use the sub qualifier as a table name. For example, sub.col1. Once a table alias has been assigned, the original source-table name becomes invalid and cannot be used as a column qualifier. You must use the alias instead. WHERE Clause The WHERE clause is used to filter rows from the working table generated by the FROM clause. It is very similar to the WHERE clause found in the UPDATE and DELETE commands. An expression is provided that is evaluated for each row. Any row that causes the expression to evaluate to false or NULL is discarded. The resulting table will have the same number of columns as the original table, but may have fewer rows. It is not con- sidered an error if the WHERE clause eliminates every row in the working table. Fig- ure 5-4 shows how the WHERE clause works. Figure 5-4. The WHERE clause filters rows based off a filter expression. Some WHERE clauses can get quite complex, resulting in a long series of AND operators used to join sub-expressions together. Most filter for a specific row, however, searching for a specific key value. 68 | Chapter 5: The SELECT Command GROUP BY Clause The GROUP BY clause is used to collapse, or “flatten,” groups of rows. Groups can be counted, averaged, or otherwise aggregated together. If you need to perform any kind of inter-row operation that requires data from more than one row, chances are you’ll need a GROUP BY. The GROUP BY clause provides a list of grouping expressions and optional collations. Very often the expressions are simple column references, but they can be arbitrary expressions. The syntax looks like this: GROUP BY grouping_expression [COLLATE collation_name] [,...] The grouping process has two steps. First, the GROUP BY expression list is used to arrange table rows into different groups. Once the groups are defined, the SELECT header (dis- cussed in the next section) defines how those groups are flattened down into a single row. The resulting table will have one row for each group. To split up the working table into groups, the list of expressions is evaluated across each row of the table. All of the rows that produce equivalent values are grouped together. An optional collation can be given with each expression. If the grouping ex- pression involves text values, the collation is used to determine which values are equiv- alent. For more information on collations, see “ORDER BY Clause” on page 74. Figure 5-5 shows how the rows are grouped together with the GROUP BY clause. Figure 5-5. The GROUP BY clause groups rows based off a list of grouping expressions. Once grouped together, each collection of rows is collapsed into a single row. This is typically done using aggregate functions that are defined in the SELECT heading, which is described in the next section, on page 70. Because it is common to GROUP BY using expressions that are defined in the SELECT header, it is possible to simply reference SELECT heading expressions in the GROUP BY expression list. If a GROUP BY expression is given as a literal integer, that number is used as a column index in the result table defined by the SELECT header. Indexes start at one The SELECT Pipeline | 69 with the leftmost column. A GROUP BY expression can also reference a result column alias. Result column aliases are explained in the next section. SELECT Header The SELECT header is used to define the format and content of the final result table. Any column you want to appear in the final results table must be defined by an expression in the SELECT header. The SELECT heading is the only required step in the SELECT com- mand pipeline. The format of the header is fairly simple, consisting of a list of expressions. Each ex- pression is evaluated in the context of each row, producing the final results table. Very often the expressions are simple column references, but they can be any arbitrary expression involving column references, literal values, or SQL functions. To generate the final query result, the list of expressions is evaluated once for each row in the work- ing table. Additionally, you can provide a column name using the AS keyword: SELECT expression [AS column_name] [,...] Don’t confuse the AS keyword used in the SELECT header with the one used in the FROM clause. The SELECT header uses the AS keyword to assign a column name to one of the output columns, while the FROM clauses uses the AS keyword to assign a source-table alias. Providing an output column name is optional, but recommended. The column name assigned to a results table is not strictly defined unless the user provides an AS column alias. If your application searches for a specific column name in the query results, be sure to assign a known name using AS. Assigning a column name will also allow other parts of the SELECT statement to reference an output column by name. Steps in the SELECT pipeline that are processed before the SELECT header, such as the WHERE and GROUP BY clause, can also reference output columns by name, just as long as the column ex- pression does not contain an aggregate function. If there is no working table (no FROM clause), the expression list is evaluated a single time, producing a single row. This row is then used as the working table. This is useful to test and evaluate standalone expressions. Although the SELECT header appears to filter columns from the working table, much like the WHERE clause filters rows, this isn’t exactly correct. All of the columns from the original working table are still available to clauses that are processed after the SELECT header. For example, it is possible to sort the results (via ORDER BY, which is processed after the SELECT header) using a column that doesn’t appear in the query output. It would be more accurate to say that the SELECT header tags specific columns for output. Not until the whole SELECT pipeline has been processed and the results are ready to be returned, are the unused columns stripped out. Figure 5-6 illustrates this point. 70 | Chapter 5: The SELECT Command In addition to the standard expressions, SELECT supports two wildcards. A simple as- terisk (*) will cause every user-defined column from every source table in the FROM clause to be output. You can also target a specific table (or table alias) using the format table_name.*. Although both of these wildcards are capable of returning more than one column, they can be mixed along with other expressions in the expression list. Wild- cards cannot use a column alias, since they often return more than one column. Be aware that the SELECT wildcards will not return any automatically generated ROWID columns. To return both the ROWID and the user-defined columns, simply ask for them both: SELECT ROWID, * FROM table; Wildcards do include any user-defined INTEGER PRIMARY KEY column that have replaced the standard ROWID column. See “Primary keys” on page 40 for more information about how ROWID and INTEGER PRIMARY KEY columns interact. In addition to determining the columns of the query result, the SELECT header deter- mines how row-groups (produced by the GROUP BY clause) are flattened into a single row. This is done using aggregate functions. An aggregate function takes a column expression as input and aggregates, or combines, all of the column values from the rows of a group and produces a single output value. Common aggregate functions include count(), min(), max(), and avg(). Appendix E provides a full list of all the built-in aggregate functions. Any column or expression that is not passed through an aggregate function will assume whatever value was contained in the last row of the group. However, because SQL tables are unordered, and because the SELECT header is processed before the ORDER BY Figure 5-6. The SELECT heading tags specific columns for output. The unused columns are not removed until the query result is actually returned. Later SELECT clauses (such as ORDER BY) still have access to columns that are not part of the query result. The SELECT Pipeline | 71 clause, we don’t really know which row is “last.” This means the values for any unag- gregated output will be taken from some essentially random row in the group. Fig- ure 5-7 shows how this works. Figure 5-7. The SELECT header will flatten any row groups created by GROUP BY. This figure shows how different columns from one row-group are flattened into an output row. Any value not computed by an aggregate function comes from the last row. Because column A was used as a GROUP BY expression, all the rows are known to have the same value, and it is safe to return. Column B is run through an aggregate function, and is also safe to return. Column C is not safe to return, as the order of the rows within a group is undefined. In some cases, picking the value from a random row is not a bad thing. For example, if a SELECT header expression is also used as a GROUP BY expression, then we know that column has an equivalent value in every row of a group. No matter which row you choose, you will always get the same value. Where you can run into trouble is when the SELECT header uses column references that were not part of the GROUP BY clause, nor were they passed through aggregate functions. In those cases there is no deterministic way to figure out what the output value will be. To avoid this, when using a GROUP BY clause, SELECT header expressions should only use column references as aggregate function inputs, or the header expressions should match those used in the GROUP BY clause. Here are some examples. In this case all of the expressions are bare column references to help make things clear: SELECT col1, sum( col2 ) FROM tbl GROUP BY col1; -- well formed This is a well formed statement. The GROUP BY clause shows that the rows are being grouped based off the values in col1. That makes it safe for col1 to appear in the SELECT header, since every row in a particular group will have an equivalent value in 72 | Chapter 5: The SELECT Command col1. The SELECT header also references col2, but it is fed into an aggregate function. The aggregate function will take all of the col2 values from different rows in the group and produce a logical answer—in this case, a numerical summation. The result of this statement will be two columns. The first column will have one row for each unique value from col1. Each row of the second column will have the sum of all the values in col2 that are associated with the col1 value listed in the first result column. More detailed examples can be found at the end of the chapter. This next statement is not well formed: SELECT col1, col2 FROM tbl GROUP BY col1; -- NOT well formed As before, the rows are grouped based off the value in col1, which makes it safe for col1 to appear in the SELECT header. The column col2 appears bare, however, and not as an aggregate parameter. When this statement is run, the second return column will contain random values from the original col2 column. Although every row within a group should have an equivalent value in a column or expression that was used as a grouping key, that doesn’t always mean the values are the exact same. If a collation such as NOCASE was used, different values (such as 'ABC' and 'abc') are considered equivalent. In these cases, there is no way to know the specific value that will be returned from a SELECT header. For example: CREATE TABLE tbl ( t ); INSERT INTO tbl VALUES ( 'ABC' ); INSERT INTO tbl VALUES ( 'abc' ); SELECT t FROM tbl GROUP BY t COLLATE NOCASE; This query will only return one row, but there is no way to know which specific value will be returned. Finally, if the SELECT header contains an aggregate function, but the SELECT statement has no GROUP BY clause, the entire working table is treated as a single group. Since flattened groups always return one row, this will cause the query to return only one row—even if the working table contained no rows. HAVING Clause Functionally, the HAVING clause is identical to the WHERE clause. The HAVING clause con- sists of a filter expression that is evaluated for each row of the working table. Any row that evaluates to false or NULL is filtered out and removed. The resulting table will have the same number of columns, but may have fewer rows. The main difference between the WHERE clause and the HAVING clause is where they appear in the SELECT pipeline. The HAVING clause is processed after the GROUP BY and SELECT clauses, allowing HAVING to filter rows based off the results of any GROUP BY aggregate. HAVING clauses can even have their own aggregates, allowing them to filter on aggregate results that are not part of the SELECT header. The SELECT Pipeline | 73 HAVING clauses should only contain filter expressions that depend on the GROUP BY out- put. All other filtering should be done in the WHERE clause. Both the HAVING and WHERE clauses can reference result column names defined in the SELECT header with the AS keyword. The main difference is that the WHERE clause can only reference expressions that do not contain aggregate functions, while the HAVING clause can reference any result column. DISTINCT Keyword The DISTINCT keyword will scan the result set and eliminate any duplicate rows. This ensures the returned rows constitute a proper set. Only the columns and values speci- fied in the SELECT header are considered when determining if a row is a duplicate or not. This is one of the few cases when NULLs are considered to have “equality,” and will be eliminated. Because SELECT DISTINCT must compare every row against every other row, it is an expensive operation. In a well-designed database, it is also rarely required. Therefore, its usage is somewhat unusual. ORDER BY Clause The ORDER BY clause is used to sort, or order, the rows of the results table. A list of one or more sort expressions is provided. The first expression is used to sort the table. The second expression is used to sort any equivalent rows from the first sort, and so on. Each expression can be sorted in ascending or descending order. The basic format of the ORDER BY clause looks like this: ORDER BY expression [COLLATE collation_name] [ASC|DESC] [,...] The expression is evaluated for each row. Very often the expression is a simple column reference, but it can be any expression. The resulting value is then compared against those values generated by other rows. If given, the named collation is used to sort the values. A collation defines a specific sorting order for text values. The ASC or DESC key- words can be used to force the sort in an ascending or descending order. By default, values are sorted in an ascending order using the default collation. An ORDER BY expression can utilize any source column, including those that do not appear in the query result. Like GROUP BY, if an ORDER BY expression consists of a literal integer, it is assumed to be a column index. Column indexes start on the left with 1, so the phrase ORDER BY 2 will sort the results table by its second column. Because SQLite allows different datatypes to be stored in the same column, sorting can get a bit more interesting. When a mixed-type column is sorted, NULLs will be sorted to the top. Next, integer and real values will be mixed together in proper numeric order. The numbers will be followed by text values, with BLOB values at the end. There will 74 | Chapter 5: The SELECT Command be no attempt to convert types. For example, a text value holding a string representation of a number will be sorted in with the other text values, and not with the numeric values. In the case of numeric values, the natural sort order is well defined. Text values are sorted by the active collation, while BLOB values are always sorted using the BINARY collation. SQLite comes with three built-in collation functions. You can also use the API to define your own collation functions. The three built-in collations are: BINARY Text values are sorted according to the semantics of the POSIX memcmp() call. The encoding of a text value is not taken into account, essentially treating it as a large binary string. BLOB values are always sorted with this collation. This is the default collation. NOCASE Same as BINARY, only ASCII uppercase characters are converted to lowercase before the comparison is done. The case-conversion is strictly done on 7-bit ASCII values. The normal SQLite distribution does not support UTF-aware collations. RTRIM Same as BINARY, only trailing (righthand) whitespace is ignored. While ORDER BY is extremely useful, it should only be used when it is actually needed— especially with very large result tables. Although SQLite can sometimes make use of an index to calculate the query results in order, in many cases SQLite must first calculate the entire result set and then sort it, before rows are returned. In that case, the inter- mediate results table must be held in memory or on disk until it is fully computed and can then be sorted. Overall, there are plenty of situations where ORDER BY is justified, if not required. Just be aware there can be some significant costs involved in its use, and you shouldn’t get in the habit of tacking it on to every query “just because.” LIMIT and OFFSET Clauses The LIMIT and OFFSET clauses allow you to extract a specific subset of rows from the final results table. LIMIT defines the maximum number of rows that will be returned, while OFFSET defines the number of rows to skip before returning the first row. If no OFFSET is provided, the LIMIT is applied to the top of the table. If a negative LIMIT is provided, the LIMIT is removed and will return the whole table. There are three ways to define a LIMIT and OFFSET: LIMIT limit_count LIMIT limit_count OFFSET offset_count LIMIT offset_count, limit_count The SELECT Pipeline | 75 Note that if both a limit and offset are given using the third format, the order of the numbers is reversed. Here are some examples. Notice that the OFFSET value defines how many rows are skipped, not the position of the first row: LIMIT 10 -- returns the first 10 rows (rows 1 - 10) LIMIT 10 OFFSET 3 -- returns rows 4 - 13 LIMIT 3 OFFSET 20 -- returns rows 21 - 23 LIMIT 3, 20 -- returns rows 4 - 23 (different from above!) Although it is not strictly required, you usually want to define an ORDER BY if you’re using a LIMIT. Without an ORDER BY, there is no well-defined order to the result, making the limit and offset somewhat meaningless. Advanced Techniques Beyond the basic SELECT syntax, there are a few advanced techniques for expressing more complex queries. Subqueries The SELECT command provides a great deal of flexibility, but there are times when a single SELECT command cannot fully express a query. To help with these situations, SQL supports subqueries. A subquery is nothing more than a SELECT statement that is embedded in another SELECT statement. Subqueries are also known as sub-selects. Subqueries are most commonly found in the FROM clause, where they act as a computed source table. This type of subquery can return any number of rows or columns, and is similar to creating a view or running the query, recording the results into a temporary table, and then referencing that table in the main query. The main advantage of using an in-line subquery is that the query optimizer is able to merge the subquery into the main SELECT statement and look at the whole problem, often leading to a more efficient query plan. To use a subquery in the FROM clause, simply enclose it in parentheses. The following two statements will produce the same output: SELECT * FROM TblA AS a JOIN TblB AS b; SELECT * FROM TblA AS a JOIN (SELECT * FROM TblB) AS b; Subqueries can show up in other places, including general expressions used in any SQL command. The EXISTS and IN operators both utilize subqueries. In fact, you can use a subquery any place an expression expects a list of literal values (a subquery cannot be used to generate a list of identifiers, however). See Appendix D for more details on SQL expressions. 76 | Chapter 5: The SELECT Command Compound SELECT Statements In addition to subqueries, multiple SELECT statements can be combined together to form a compound SELECT. Compound SELECTs use set operators on the rows generated by a series of SELECT statements. In order to combine correctly, each SELECT statement must generate the same number of columns. The column names from the first SELECT statement will be used for the overall result. Only the last SELECT statement can have an ORDER BY, LIMIT or OFFSET clause, which get applied to the full compound result table. The syntax for a compound SELECT looks like this: SELECT ... FROM ... WHERE ... GROUP BY ... HAVING ... compound_operator SELECT ... FROM ... WHERE ... GROUP BY ... HAVING ... [...] ORDER BY ... LIMIT ... OFFSET ... Multiple compound operators can be used to include additional SELECT statements. UNION ALL The UNION ALL operator concatenates all of the rows returned by each SELECT into one large table. If the two SELECT blocks generate N and M rows, respectively, the resulting table will have N+M rows. UNION The UNION operator is very similar to the UNION ALL operator, but it will eliminate any duplicate rows, including duplicates that came from the same SELECT block. If the two SELECT blocks generate N and M rows, respectively, the resulting table can have anywhere from 1 to N+M rows. INTERSECT The INTERSECT operator will return one instance of any row that is found (one or more times) in both SELECT blocks. If the two SELECT blocks generate N and M rows, respectively, the resulting table can have anywhere from 0 to MIN(N,M) rows. EXCEPT The EXCEPT operator will return all of the rows in the first SELECT block that are not found in the second SELECT block. It is essentially a subtraction operator. If there are duplicate rows in the first block, they will all be eliminated by a single, matching row in the second block. If the two SELECT blocks generate N and M rows, respectively, the resulting table can have anywhere from 0 to N rows. SQLite supports the UNION, UNION ALL, INTERSECT, and EXCEPT compound operators. Figure 5-8 shows the result of each operator. Advanced Techniques | 77 Once all the compound operators have been combined, any trailing ORDER BY, LIMIT, and OFFSET is applied to the final result table. In the case of compound SELECT state- ments, the expressions present in any ORDER BY clause must be exactly match one of the result columns, or use a column index. Alternate JOIN Notation There are two styles of join notation. The style shown earlier in this chapter is known as explicit join notation. It is named such because it uses the keyword JOIN to explicitly describe how each table is joined to the next. The explicit join notation is also known as ANSI join notation, as it was introduced when SQL went through the standardization process. The older, original join notation is known as implicit join notation. Using this notation, the FROM clause is simply a comma-separated list of tables. The tables in the list are combined using a Cartesian product and the relevant rows are extracted with additional WHERE conditions. In effect, it degrades every join to a CROSS JOIN and then moves the join conditions out of the FROM clause and into the WHERE clause. This first statement uses the explicit join notation we learned earlier in the chapter: SELECT ... FROM employee JOIN resource ON ( employee.eid = resource.eid ) WHERE ... Figure 5-8. The compound operators UNION ALL, UNION, INTERSECT and EXCEPT. 78 | Chapter 5: The SELECT Command This is the same statement written with the implicit join notation: SELECT ... FROM employee, resource WHERE employee.eid = resource.eid AND ... There is no performance difference between the two notations: it is purely a matter of syntax. In general, the explicit notion (the first one) has become the standard way of doing things. Most people find the explicit notation easier to read, making the intent of the query more transparent and easier to understand. I’ve always felt the explicit notation is a bit cleaner, as it puts the complete join specification into the FROM clause, leaving the WHERE clause free for query-specific filters. Using the explicit notation, the FROM clause (and the FROM clause alone) fully and independently specifies what you’re se- lecting “from.” The explicit notation also lets you be much more specific about the type and order of each JOIN. In SQLite, you must use the explicit notation if you want an OUTER JOIN— the implicit notation can only be used to indicate a CROSS JOIN or INNER JOIN. If you’re learning SQL for the first time, I would strongly suggest you become com- fortable with the explicit notation. Just be aware that there is a great deal of SQL code out there (including older books and tutorials) using the older, implicit notation. SELECT Examples The SELECT command is very complex, and it can be difficult to see how these different clauses can be fit together into something useful. Some of this will become more obvious in the next chapter, when we look at standard database design practices, but to get you started, we’re going to look at several examples. All of these examples will use this data: CREATE TABLE x ( a, b ); INSERT INTO x VALUES ( 1, 'Alice' ); INSERT INTO x VALUES ( 2, 'Bob' ); INSERT INTO x VALUES ( 3, 'Charlie' ); CREATE TABLE y ( c, d ); INSERT INTO y VALUES ( 1, 3.14159 ); INSERT INTO y VALUES ( 1, 2.71828 ); INSERT INTO y VALUES ( 2, 1.61803 ); CREATE TABLE z ( a, e ); INSERT INTO z VALUES ( 1, 100 ); INSERT INTO z VALUES ( 1, 150 ); INSERT INTO z VALUES ( 3, 300 ); INSERT INTO z VALUES ( 9, 900 ); SELECT Examples | 79 These examples show the sqlite3 command-line tool. The following dot-commands were issued to make the output easier to understand. The last command will cause sqlite3 to print the string [NULL] whenever a NULL is encountered. Normally, a NULL will produce a blank output that is indistinguishable from an empty string: .headers on .mode column .nullvalue [NULL] This dataset is available on the book’s download page on the O’Reilly website, as both an SQL file and an SQLite database. I suggest you sit down with a copy of sqlite3 and try these commands out. Try experimenting with different variations. If one of these examples doesn’t quite make sense, just break the SELECT statement down into its individual parts and step through them bit by bit. Simple SELECTs Let’s start with a simple select that returns all of the columns and rows in table x. The SELECT * syntax returns all columns by default: sqlite> SELECT * FROM x; a b ---------- ---------- 1 Alice 2 Bob 3 Charlie We can also return expressions, rather than just columns: sqlite> SELECT d, d*d AS dSquared FROM y; d dSquared ---------- ------------ 3.14159 9.8695877281 2.71828 7.3890461584 1.61803 2.6180210809 Simple JOINs Now some joins. By default, the bare keyword JOIN indicates an INNER JOIN. However, when no additional condition is put on the JOIN, it reverts to a CROSS JOIN. As a result, all three of these queries produce the same results. The last line uses the implicit join notation. 80 | Chapter 5: The SELECT Command sqlite> SELECT * FROM x JOIN y; sqlite> SELECT * FROM x CROSS JOIN y; sqlite> SELECT * FROM x, y; a b c d ---------- ---------- ---------- ---------- 1 Alice 1 3.14159 1 Alice 1 2.71828 1 Alice 2 1.61803 2 Bob 1 3.14159 2 Bob 1 2.71828 2 Bob 2 1.61803 3 Charlie 1 3.14159 3 Charlie 1 2.71828 3 Charlie 2 1.61803 In the case of a cross join, every row in table a is matched to every row in table y. Since both tables had three rows and two columns, the result set has nine rows (3·3) and four columns (2+2). JOIN...ON Next, a fairly simple inner join using a basic ON join condition: sqlite> SELECT * FROM x JOIN y ON a = c; a b c d ---------- ---------- ---------- ---------- 1 Alice 1 3.14159 1 Alice 1 2.71828 2 Bob 2 1.61803 This query still generates four columns, but only those rows that fulfill the join condi- tion are included in the result set. The following statement requires the columns to be qualified, since both table x and table z have an a column. Notice that two different a columns are returned, one from each source table: sqlite> SELECT * FROM x JOIN z ON x.a = z.a; a b a e ---------- ---------- ---------- ---------- 1 Alice 1 100 1 Alice 1 150 3 Charlie 3 300 SELECT Examples | 81 JOIN...USING, NATURAL JOIN If we use a NATURAL JOIN or the USING syntax, the duplicate column will be eliminated. Since both table x and table z have only column a in common, both of these statements produce the same output: sqlite> SELECT * FROM x JOIN z USING ( a ); sqlite> SELECT * FROM x NATURAL JOIN z; a b e ---------- ---------- ---------- 1 Alice 100 1 Alice 150 3 Charlie 300 OUTER JOIN A LEFT OUTER JOIN will return the same results as an INNER JOIN, but will also include rows from table x (the left/first table) that were not matched: sqlite> SELECT * FROM x LEFT OUTER JOIN z USING ( a ); a b e ---------- ---------- ---------- 1 Alice 100 1 Alice 150 2 Bob [NULL] 3 Charlie 300 In this case, the Bob row from table x has no matching row in table z. Those column values normally provided by table z are padded out with NULL, and the row is then included in the result set. Compound JOIN It is also possible to JOIN multiple tables together. In this case we join table x to table y, and then join the result to table z: sqlite> SELECT * FROM x JOIN y ON x.a = y.c LEFT OUTER JOIN z ON y.c = z.a; a b c d a e ---------- ---------- ---------- ---------- ---------- ---------- 1 Alice 1 3.14159 1 100 1 Alice 1 3.14159 1 150 1 Alice 1 2.71828 1 100 1 Alice 1 2.71828 1 150 2 Bob 2 1.61803 [NULL] [NULL] 82 | Chapter 5: The SELECT Command If you don’t see what is going on here, work through the joins one at a time. First look at what FROM x JOIN y ON x.a = y.c will produce (shown in one of the previous ex- amples). Then look at how this result set would combine with table z using a LEFT OUTER JOIN. Self JOIN Our last join example shows a self-join, where a table is joined against itself. This creates two unique instances of the same table and necessitates the use of table aliases: sqlite> SELECT * FROM x AS x1 JOIN x AS x2 ON x1.a + 1 = x2.a; a b a b ---------- ---------- ---------- ---------- 1 Alice 2 Bob 2 Bob 3 Charlie Also, notice that the join condition is a more arbitrary expression, rather than being a simple test of column references. WHERE Examples Moving on, the WHERE clause is used to filter rows. We can pick out a specific row: sqlite> SELECT * FROM x WHERE b = 'Alice'; a b ---------- ---------- 1 Alice Or a range of values: sqlite> SELECT * FROM y WHERE d BETWEEN 1.0 AND 3.0; c d ---------- ---------- 1 2.71828 2 1.61803 In this case, the WHERE expression references the output column by its assigned name: sqlite> SELECT c, d, c+d AS sum FROM y WHERE sum < 4.0; c d sum ---------- ---------- ---------- 1 2.71828 3.71828 2 1.61803 3.61803 SELECT Examples | 83 GROUP BY Examples Now let’s look at a few GROUP BY statements. Here we group table z by the a column. Since there are three unique values in z.a, the output has three rows. Only the grouping a=1 has more than one row, however. We can see this in the count() values returned by the second column: sqlite> SELECT a, count(a) AS count FROM z GROUP BY a; a count ---------- ---------- 1 2 3 1 9 1 This is a similar query, only now the second output column represents the sum of all the z.e values in each group: sqlite> SELECT a, sum(e) AS total FROM z GROUP BY a; a total ---------- ---------- 1 250 3 300 9 900 We can even compute our own average and compare that to the avg() aggregate: sqlite> SELECT a, sum(e), count(e), ...> sum(e)/count(e) AS expr, avg(e) AS agg ...> FROM z GROUP BY a; a sum(e) count(e) expr agg ---------- ---------- ---------- ---------- ---------- 1 250 2 125 125.0 3 300 1 300 300.0 9 900 1 900 900.0 A HAVING clause can be used to filter rows based off the results of the sum() aggregation: sqlite> SELECT a, sum(e) AS total FROM z GROUP BY a HAVING total > 500; a total ---------- ---------- 9 900 84 | Chapter 5: The SELECT Command ORDER BY Examples The output can also be sorted. Most of these examples already look sorted, but that’s mostly by chance. The ORDER BY clause enforced a specific order: sqlite> SELECT * FROM y ORDER BY d; c d ---------- ---------- 2 1.61803 1 2.71828 1 3.14159 Limits and offsets can also be applied to pick out specific rows from an ordered result. Conceptually, these are fairly simple, however. These tables and queries are available as part of the book download. Feel free to load the data into sqlite3 and try out different queries. Don’t worry about creating SELECT statements that use every available clause. Start with simple queries where you can understand all the steps, then start to combine clauses to build larger and more complex queries. What’s Next Now that we’ve had a much closer look at the SELECT command, you’ve seen most of the core SQL language. The next chapter will take a more detailed look at database design. This will help you lay out your tables and data in a way that reinforces the strengths of the database system. Understanding some of the core design ideas should also make it more clear why SELECT works the way it does. What’s Next | 85 CHAPTER 6 Database Design Relational databases have only one type of data container: the table. When designing any database, the main concern is defining the tables that make up the database and defining how those tables reference and interact with each other. Designing tables for a database is a lot like designing classes or data structures for an application. Simple designs are largely driven by common sense, but as things get larger and more complex, it takes some experience and insight to keep the design clean and on target. Understanding basic design principles and standard approaches can be a big help. Tables and Keys Tables may look like simple two-dimensional grids of simple values, but a well-defined table has a fair amount of structure. Different columns can play different roles. Some columns act as unique identifiers that define the intent and purpose of each row. Other columns hold supporting data. Still other columns act as external references that link rows in one table to rows in another table. When designing a table, it is important to understand why each column is there, and what role each column is playing. Keys Define the Table When designing a table, you usually start by specifying the primary key. The primary key consists of one or more columns that uniquely identify each row of a table. In a sense, the primary key values represent the fundamental identity of each row in the table. The primary key columns identify what the table is all about. All the other col- umns should provide supporting data that is directly relevant to the primary key. 87 Sometimes the primary key is an actual unique data value, such as a room number or a hostname. Very often the primary key is simply an arbitrary identification number, such as an employee or student ID number. The important point is that primary keys must be unique over every possible entry in the table, not just the rows that happen to be in there right now. This is why names are normally not used as primary keys—in a large group of people, it is too easy to end up with duplicate (or very similar) names. While there is nothing particularly special about the columns that make up a primary key, the keys themselves play a very important role in the design and use of a database. Their role as a unique identifier for each row makes them analogous to the key of a hash table, or the key to a dictionary class of data structure. They are essentially “lookup” values for the rows of a table. A primary key can be identified in the CREATE TABLE command. Explicitly identifying the primary key will cause the database system to automatically create a UNIQUE index across all of the primary key columns. Declaring the primary key also allows for some syntax shortcuts when establishing relationships between tables. For example, this table definition defines the employee_id field to be a primary key: CREATE TABLE employee ( employee_id INTEGER PRIMARY KEY NOT NULL, name TEXT NOT NULL /* ...etc... */ ); For more information on the syntax used to define a primary key, see the section “Primary keys” on page 40. In schema documentation, primary keys are often indicated with the abbreviation “PK.” It is also common to double underline primary keys when drawing out tables, as shown in Figure 6-1. Figure 6-1. Primary keys are sometimes identified with the abbreviation PK. It is also common to use a double underline when diagramming the table. 88 | Chapter 6: Database Design Many database queries use a table’s primary key as either the input or output. The database might be asked to return the row with a given key, such as, “return the record for employee #953.” It is also common to ask for collections of keys, such as, “gather the ID values for all employees hired more than two years ago.” This set of keys might then be joined to another table as part of a report. Foreign Keys In addition to identifying each row in a table, primary keys are also central to joining tables together. Since the primary key acts as a unique row identifier, you can create a reference to a specific row by recording the row’s primary key. This is known as a foreign key. A foreign key is a copy or recording of a primary key, and is used as a reference or pointer to a different (or “foreign”) row, most often in a different table. Like the primary key, columns that make up a foreign key can be identified within the CREATE TABLE command. In this example, we define the format of a task assignment. Each task gets assigned to a specific employee by referencing the employee_id field from the employee table: CREATE TABLE task_assignment ( task_assign_id INTEGER PRIMARY KEY, task_name TEXT NOT NULL, employee_id INTEGER NOT NULL REFERENCES employee( employee_id ) /* ...etc... */ ); The REFERENCES constraint indicates that this column is a foreign key. The constraint indicates which table is referenced and, optionally, which column. If no column is indicated, the foreign key will reference the primary key (meaning the column reference used in the prior example is not required, since employee_id is the primary key of the employee table). The vast majority of foreign keys will reference a primary key, but if a column other than the primary key is used, that column must have a UNIQUE constraint, or it must have a single-column UNIQUE index. A foreign key can also be defined as a table constraint. In that case, there may be multiple local columns that refer to multiple columns in the referenced table. The referenced columns must be a multicolumn primary key, or they must otherwise have a multicol- umn UNIQUE index. A foreign key definition can include several other optional parts. For the full syntax, see CREATE TABLE in Appendix C. Unlike a table’s own primary key, foreign keys are not required to be unique. This is because multiple foreign key values (multiple rows) in one table may refer to the same row in another table. This is known as a one-to-many relationship. Please see the section “One-to-Many Relationships” on page 95. Foreign keys are often marked with the abbreviation “FK,” as shown in Figure 6-2. Tables and Keys | 89 Figure 6-2. Foreign keys are copies of the primary key from another row. Foreign keys act as references or pointers to other rows. They are often identified with the abbreviation FK. Foreign Key Constraints Declaring foreign keys in the table definition allows the database to enforce foreign key constraints. Foreign key constraints are used to keep foreign key references in sync. Among other things, foreign key constraints can prevent “dangling references” by re- quiring that all foreign key values correctly match a row value from the columns of the referenced table. Foreign keys can also be set to NULL. A NULL clearly marks the foreign key as unassigned, which is a bit different than having an invalid value. In many cases, unassigned foreign keys don’t fit the design of the database. In that case, the foreign key columns should be declared with a NOT NULL constraint. Using our previous example, foreign key constraints would require that every task_assignment.employee_id element needs to contain the value of a valid employee.employee_id. By default, a NULL foreign key would also be allowed, but we’ve defined the task_assignment.employee_id column with a NOT NULL constraint. This de- mands that every task reference a valid employee. Native foreign key support was added in SQLite 3.6.19, but is turned off by default. You must use the PRAGMA foreign_keys command to turn on foreign key constraints. This was done to avoid problems with existing applications and database files. A future version of SQLite may have foreign key constraints enabled by default. If your appli- cation is dependent on this setting, it should explicitly turn it on or off. Modifications to either the foreign key table or the referenced table can potentially cause violations of the foreign key constraint. For example, if a statement attempted to update a task_assignment.employee_id value to an invalid employee_id, the foreign key constraint would be violated. Similarly, if an employee row was assigned a new employee_id value, any existing task_assignment references that point to the old value would become invalid. This would also violate the foreign key constraint. 90 | Chapter 6: Database Design If the database detects a change that would cause a foreign key violation, there are several actions that can be taken. The default action is to prohibit the change. For example, if you attempted to delete an employee that still had tasks assigned to them, the delete would fail. You would need to delete the tasks or transfer them to a different employee before you could delete the original employee. Other conflict resolutions are also available. For example, using an ON DELETE CASCADE foreign key definition, deleting an employee would cause the database to automatically delete any tasks assigned to that employee. For more information on conflict resolutions and other advanced foreign key options, please see the SQLite website. Up-to-date documentation on SQLite’s support for foreign keys can be found at http://www.sqlite.org/foreignkeys.html. Correctly defining foreign keys is one of the most critical aspects of data integrity and security. Once defined, foreign key constraints make sure that data relationships remain valid and consistent. Generic ID Keys If you look at most database designs, you’ll notice that a large number of the tables have a generic ID column that is used as the primary key. The ID is typically an arbitrary integer value that is automatically assigned as new rows are inserted. The number may be incrementally assigned, or it might be assigned from a sequence mechanism that is guaranteed to never assign the same ID number twice. When an SQLite column is defined as an INTEGER PRIMARY KEY, that column will replace the hidden ROWID column that acts as the root column of every table. Using an INTEGER PRIMARY KEY allows for some significant performance enhancements. It also allows SQLite to automatically assign sequenced ID values. For more details, see “Primary keys” on page 40. At first, a generic ID field might seem like a design cheat. If each table should have a specific and well-defined role, then the primary key should reflect what makes any given row unique—reflecting, in part, the essential definition of the items in a table. Using a generic and arbitrary ID to define that uniqueness seems to be missing the point. From a theoretical standpoint, that may be correct, but this is one of those areas where theory bumps into reality, and reality usually wins. Practically speaking, many datasets don’t have a truly unique representation. For ex- ample, the names of people are not sufficiently unique to be used as database keys. Names are reasonably unique, and they do a fair job at identifying individuals in person, but they lack the inherent and complete uniqueness that good database design demands. Names also change from time to time, such as when people get married. The more you dig around, the more you’ll find that the world is full of data like this. Data that is sufficiently unique and stable for casual use, but not truly, absolutely Tables and Keys | 91 unique or fixed enough to use as a smart database key. In these situations, the best solution is to simply use an arbitrary ID that the database designer has total control over, even if it is meaningless outside of the database. This type of key is known as a surrogate key. There are also situations when the primary key consists of three or four (or more!) columns. This is somewhat rare, but there are situations when it does come up. If you’ve got a large multicolumn primary key in the middle of a complex set of relationships, it can be a big pain to create and match all those multicolumn foreign keys. To simplify such situations, it is often easier to simply create an arbitrary ID and use that as the primary key. Using an arbitrary ID is also useful if the customary primary key is physically large. Because each foreign key is a full copy of the primary key, it is unwise to use a lengthy text value or BLOB as the primary key. Rather, it would be better to use an arbitrary identifier and simply reference to the identifier. One final comment on key names. There is often a temptation to name a generic ID field something simple, such as id. After all, if you’ve got an employee table, it might seem somewhat redundant to name the primary key employee_id; you end up with a lot of column references that read employee.employee_id, when it seems that employee.id is clear enough. Well, by itself, it is clear enough, but primary keys tend to show up in other tables as foreign keys. While employee.employee_id might be slightly redundant, the name task_assignment.employee_id is not. That name also gives you significant clues about the column’s function (a foreign key) and what table and column it references (the employee_id column, which is the PK column of the employee table). Using the same name for primary keys and foreign keys makes the inherent meaning and linkage a lot more obvious. It also allows shortcuts, such as the NATURAL JOIN or JOIN...USING( ) syntax. Both of these forms require that matching columns have the exact same name. Using a more explicit name also avoids the problem of having multiple tables, each with a column named id. Such a common name can make things somewhat confusing if you join together three or four tables. While I wouldn’t necessarily prefix every col- umn name, keeping primary key names unique within a database (and using the exact same name for foreign keys) can make the intent of the database design a lot more clear. Keep It Specific The biggest stumbling block for beginning database developers is that they don’t create enough tables. Less experienced developers tend to view tables as large and monu- mental structures. There tends to be a feeling that tables are important, and that each table needs a fair number of columns containing significant amounts of data to justify its existence. If a design change calls for a new column or set of data points, there tends 92 | Chapter 6: Database Design to be a strong desire to lump everything together into large centralized tables rather than breaking things apart into logical groups. The “tables must be big” mentality will quickly lead to poor designs. While you will have a few large, significant tables at the core of most designs, a typical design will also have a fair number of smaller auxiliary tables that may only hold two or three columns of data. In fact, in some cases you may find yourself building tables that consist of nothing but external references to other tables. Creating a new table is the only means you have to define a new data structure or data container, so any time you find yourself needing a new container, that’s going to indicate a new table. Every table should have a well-defined and clear role, but that doesn’t always mean it will be big or stuffed with data. Common Structures and Relationships Database design has a small number of design structures and relationships that act as basic building blocks. Once you master how to use them and how to manipulate them, you can use these building blocks to build much larger and more complex data representations. One-to-One Relationships The most basic kind of inter-table relationship is the one-to-one relationship. As you can guess, this type of relationship establishes a reference from a single row in one table to a single row in another table. Most commonly, one-to-one relationships are repre- sented by having a foreign key in one table reference a primary key in another table. If the foreign key column is made unique, only one reference will be allowed. As Fig- ure 6-3 illustrates, a unique foreign key creates a one-to-one relationship between the rows of the two tables. Figure 6-3. In a one-to-one relationship, table B has a foreign key that references the primary key of table A. This associates every non-NULL foreign key in table B with some row in table A. Common Structures and Relationships | 93 In the strictest sense, a foreign key relationship is a one-to-(zero or one) relationship. When two tables are involved in a one-to-one relationship, there is nothing to enforce that every primary key has an incoming reference from a foreign key. For that matter, if the foreign key allows NULLs, there may be unassigned foreign keys as well. One-to-one relationships are commonly used to create detail tables. As the name implies, a detail table typically holds details that are linked to the records in a more prominent table. Detail tables can be used to hold data that is only relevant to a small subsection of the database application. Breaking the detail data apart from the main tables allows different sections of the database design to evolve and grow much more independently. Detail tables can also be helpful when extended data only applies to a limited number of records. For example, a website might have a sales_items table that lists common information (price, inventory, weight, etc.) for all available items. Type-specific data can then be held in detail tables, such as cd_info for CDs (artist name, album name, etc.) or dvd_info (directors, studio, etc.) for DVDs. Although the sales_items table would have a unique one-to-one relationship with every type-specific info table, each individual row in the sales_item table would be referenced by only one type of detail table. One-to-one relationships can also be used to isolate very large data elements, such as BLOBs. Consider an employee database that contains an image of each employee. Due to the data storage and I/O overhead, it might be unwise to include a photo column directly in the employee table, but it is easy to create a photo table that references the employee table. Consider these two tables: CREATE TABLE employee ( employee_id INTEGER NOT NULL PRIMARY KEY, name TEXT NOT NULL /* ...etc... */ ); CREATE TABLE employee_photo ( employee_id INTEGER NOT NULL PRIMARY KEY REFERENCES employee, photo_data BLOB /* ...etc... */ ); This example is a bit unique, because the employee_photo.employee_id column is both the primary key for the employee_photo table, as well as a foreign key to the employee table. Since we want a one-to-one relationship, it makes sense to just pair up primary keys. Because this foreign key does not allow NULL keys, every employee_photo row must be matched to a specific employee row. The database does not guarantee that every employee will have a matching employee_photo, however. 94 | Chapter 6: Database Design One-to-Many Relationships One-to-many relationships establish a link between a single row in one table to multiple rows in another table. This is done by expanding a one-to-one relationship, allowing one side to contain duplicate keys. One-to-many relationships are often used to asso- ciate lists or arrays of items back to a single row. For example, multiple shipping addresses can be linked to a single account. Unless otherwise specified, “many” usually means “zero or more.” The only difference between a one-to-one relationship and a one-to-many relationship is that the one-to-many relationship allows for duplicate foreign key values. This allows multiple rows in one table (the many table) to refer back to a single row in another table (the One table). In building a one-to-many relationship, the foreign key must always be on the many side. Figure 6-4 illustrates the relationship in more detail. Figure 6-4. When building a one-to-many relationship, the primary keys must be unique, but foreign key columns may contain duplicate values. This means a one-to-many relationship must have the foreign key on the many side. Here we see a foreign key value in Table B refer to the primary key in Table A. Any time you find yourself wondering how to stuff an array or list into the column of a table, the solution is to separate the array elements out into their own table and establish a one-to-many relationship. If you need to represent a list or array, try using a one-to-many relationship. Common Structures and Relationships | 95 The same is true any time you start to contemplate sets of columns, such as item0, item1, item2, etc., that are all designed to hold instances of the same type of value. Such designs have inherent limits, and the insert, update, and removal process becomes quite complex. It is much easier to just break the data out into its own table and establish a proper relationship. One example of a one-to-many relationship is music albums, and the songs they con- tain. Each album has a list of songs associated with that album. For example: CREATE TABLE albums ( album_id INTEGER NOT NULL PRIMARY KEY, album_name TEXT ); CREATE TABLE tracks ( track_id INTEGER NOT NULL PRIMARY KEY, track_name TEXT, track_number INTEGER, track_length INTEGER, -- in seconds album_id INTEGER NOT NULL REFERENCES albums ); Each album and track has a unique ID. Each track also has a foreign key reference back to its album. Consider: INSERT INTO albums VALUES ( 1, "The Indigo Album" ); INSERT INTO tracks VALUES ( 1, "Metal Onion", 1, 137, 1 ); INSERT INTO tracks VALUES ( 2, "Smooth Snake", 2, 212, 1 ); INSERT INTO tracks VALUES ( 3, "Turn A", 3, 255, 1 ); INSERT INTO albums VALUES ( 2, "Morning Jazz" ); INSERT INTO tracks VALUES ( 4, "In the Bed", 1, 214, 2 ); INSERT INTO tracks VALUES ( 5, "Water All Around", 2, 194, 2 ); INSERT INTO tracks VALUES ( 6, "Time Soars", 3, 265, 2 ); INSERT INTO tracks VALUES ( 7, "Liquid Awareness", 4, 175, 2 ); To get a simple list of tracks and their associated album, we just join the tables back together. We can also sort by both album name and track number: sqlite> SELECT album_name, track_name, track_number ...> FROM albums JOIN tracks USING ( album_id ) ...> ORDER BY album_name, track_number; album_name track_name track_number ------------ ---------- ------------ Morning Jazz In the Bed 1 Morning Jazz Water All 2 Morning Jazz Time Soars 3 Morning Jazz Liquid Awa 4 The Indigo A Metal Onio 1 The Indigo A Smooth Sna 2 The Indigo A Turn A 3 We can also manipulate the track groupings: sqlite> SELECT album_name, sum( track_length ) AS runtime, count(*) AS tracks ...> FROM albums JOIN tracks USING ( album_id ) ...> GROUP BY album_id; 96 | Chapter 6: Database Design album_name runtime tracks ---------------- ---------- ---------- The Indigo Album 604 3 Morning Jazz 848 4 This query groups the tracks based off their album, and then aggregates the track data together. Many-to-Many Relationships The next step is the many-to-many relationship. A many-to-many relationship asso- ciates one row in the first table to many rows in the second table while simultaneously allowing individual rows in the second table to be linked to multiple rows in the first table. In a sense, a many-to-many relationship is really two one-to-many relationships built across each other. Figure 6-5 shows the classic many-to-many example of people and groups. One person can belong to many different groups, while each group is made up of many different people. Common operations are to find all the groups a person belongs to, or to find all the people in a group. Figure 6-5. A many-to-many relationship is like two one-to-many relationships built across each other. In this example, each individual person can be a member of one or more groups, while each group can contain one or more people. Many-to-many relationships are a bit more complex than other relationships. Although the tables have a many-to-many relationship with each other, the entries in both tables must remain unique. We cannot duplicate either person rows or group rows for the purpose of matching keys. This is a problem, since each foreign key can only reference one row. This makes it impossible for a foreign key of one table (such as a group) to refer to multiple rows of another table (such as people). To solve this, we go back to the advice from before: if you need to add a list to a row, break out that list into its own table and establish a one-to-many relationship with the new table. You cannot directly represent a many-to-many relationship with only two tables, but you can take a pair of one-to-many relationships and link them together. The link requires a small table, known as a link table, or bridge table, that sits between the two many tables. Each many-to-many relationship requires a unique bridge table. Common Structures and Relationships | 97 In its most basic form, the bridge table consists of nothing but two foreign keys—one for each of the tables it is linking together. Each row of the bridge table links one row in the first many table to one row in the second many table. In our People-to-Groups example, the link table defines a membership of one person in one group. This is illustrated in Figure 6-6. Figure 6-6. Implementing a many-to-many relationship requires a bridge table. In this example, each row of the bridge table represents a membership of one person in one group. Note that the primary key of the bridge table is a multicolumn key over (p_id, g_id). This keeps memberships unique. The logical representation of a many-to-many relationship is actually a one-to-many- to-one relationship, with the bridge table (acting as a record of membership) in the middle. Each person has a one-to-many relationship with memberships, just as groups have a one-to-many relationship with memberships. In addition to binding the two tables together, the bridge table can also hold additional information about the mem- bership itself, such as a first-joined date or an expiration date. Here are three tables. The people and groups table are obvious enough. The p_g_bridge table acts as the bridge table between the people table and groups table. The two columns are both foreign key references, one to people and one to groups. Estab- lishing a primary key across both foreign key columns ensures the memberships remain unique: CREATE TABLE people ( pid INTEGER PRIMARY KEY, name TEXT, ... ); CREATE TABLE groups ( gid INTEGER PRIMARY KEY, name TEXT, ... ); CREATE TABLE p_g_bridge( pid INTEGER NOT NULL REFERENCES people, gid INTEGER NOT NULL REFERENCES groups, PRIMARY KEY ( pid, gid ) ); 98 | Chapter 6: Database Design This query will list all the groups that a person belongs to: SELECT groups.name AS group_name FROM people JOIN p_g_bridge USING ( pid ) JOIN groups USING ( gid ) WHERE people.name = search_person_name; The query simply links people to groups using the bridge table, and then filters out the appropriate rows. We don’t always need all three tables. This query counts all the members of a group without utilizing the people table: SELECT name AS group_name, count(*) AS members FROM groups JOIN p_g_bridge USING ( gid ) GROUP BY gid; There are many other queries we can do, like finding all the groups that have no members: SELECT name AS group_name FROM groups LEFT OUTER JOIN p_g_bridge USING ( gid ) WHERE pid IS NULL; This query performs an outer join from the groups table to the p_g_bridge table. Any unmatched group rows will be padded out with a NULL in the p_g_bridge.pid column. Since this column is marked NOT NULL, we know the only possible way for a row to be NULL in that column is from the outer join, meaning the row is unmatched to any memberships. A very similar query could be used to find any people that have no memberships. Hierarchies and Trees Hierarchies and other tree-style data relationships are common and frequently show up in database design. Modeling one in a database can be a challenge because you tend to ask different types of questions when dealing with hierarchies. Common tree operations include finding all the sub-nodes under a given node, or querying the depth of a given node. These operations have an inherent recursion in them—a concept that SQL doesn’t support. This can lead to clumsy queries or complex representations. There are two common methods for representing a tree relation using database tables. The first is the adjacency model, which uses a simple representation that is easy to modify but complex to query. The other common representation is the nested set, which allows relatively simple queries, but at the cost of a more complex representation that can be expensive to modify. Common Structures and Relationships | 99 Adjacency Model A tree is basically a series of nodes that have a one-to-many relationship between the parents and the children. We already know how to define a one-to-many relationship: give each child a foreign key that points to the primary key of the parent. The only trick is that the same table is sitting on both sides of the relationship. For example, here is a basic adjacency model table: CREATE TABLE tree ( node INTEGER NOT NULL PRIMARY KEY, name TEXT, parent INTEGER REFERENCES tree ); Each node in the tree will have a unique node identifier. Each node will also have a reference to its parent node. The root of the tree can simply have a NULL parent ref- erence. This allows multiple trees to be stored in the same table, as multiple nodes can be defined as a root node of different trees. If we want to represent this tree: A A.1 A.1.a A.2 A.2.a A.2.b A.3 We would use the following data: INSERT INTO tree VALUES ( 1, 'A', NULL ); INSERT INTO tree VALUES ( 2, 'A.1', 1 ); INSERT INTO tree VALUES ( 3, 'A.1.a', 2 ); INSERT INTO tree VALUES ( 4, 'A.2', 1 ); INSERT INTO tree VALUES ( 5, 'A.2.a', 4 ); INSERT INTO tree VALUES ( 6, 'A.2.b', 4 ); INSERT INTO tree VALUES ( 7, 'A.3', 1 ); The following query will give a list of nodes and parents by joining the tree table to itself: sqlite> SELECT n.name AS node, p.name AS parent ...> FROM tree AS n JOIN tree AS p ON n.parent = p.node; node parent ---------- ---------- A.1 A A.1.a A.1 A.2 A A.2.a A.2 A.2.b A.2 A.3 A Inserting or removing nodes is fairly straightforward, as is moving subtrees around to different parents. 100 | Chapter 6: Database Design What isn’t easy are tree-centric operations, like counting all of the nodes in a subtree, or computing the depth of a specific node (often used for output formatting). You can do limited traversals (like finding a grand-parent) by joining the tree table to itself multiple times, but you cannot write a single query that can compute an answer to these types of questions on a tree of arbitrary size. The only choice is to write application routines that loop over different levels in the tree, computing the answers you seek. Overall, the adjacency model is easy to understand, and the trees are easy to modify. The model is based on foreign keys, and can take full advantage of the database’s built- in referential integrity. The major disadvantage is that many types of common data queries require the application code to loop over several individual database queries and assist in calculating answers. Nested set As the name implies, the nested set representation depends on nesting groups of nodes inside other groups. Rather than representing some type of parent-child relationship, each node holds bounding data about the full subtree underneath it. With some clever math, this allows us to query all kinds of information about a node. A nested set table might look like this: CREATE TABLE nest ( name TEXT, lower INTEGER NOT NULL UNIQUE, upper INTEGER NOT NULL UNIQUE, CHECK ( lower < upper ) ); The nested set can be visualized by converting the tree structure into a parenthetical representation. We then count the parentheses and record the upper and lower bound of each nested set. The index numbers can also be calculated with a depth-first tree traversal, where the lower bound is a pre-visit count and the upper bound is a post- visit count. A( A.1( A.1.a( ) ), A.2( A.2.a( ), A.2.b( ) ), A.3( ) ) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Or, in SQL: INSERT INTO nest VALUES ( 'A', 1, 14 ); INSERT INTO nest VALUES ( 'A.1', 2, 5 ); INSERT INTO nest VALUES ( 'A.1.a', 3, 4 ); INSERT INTO nest VALUES ( 'A.2', 6, 11 ); INSERT INTO nest VALUES ( 'A.2.a', 7, 8 ); INSERT INTO nest VALUES ( 'A.2.b', 9, 10 ); INSERT INTO nest VALUES ( 'A.3', 12, 13 ); This might not look that useful, but it allows a number of different queries. For example, if you want to find all the leaf nodes (nodes without children) just look for nodes that have an upper and lower value that are next to each other: SELECT name FROM nest WHERE lower + 1 = upper; Common Structures and Relationships | 101 You can find the depth of a node by counting all of its ancestors: sqlite> SELECT n.name AS name, count(*) AS depth ...> FROM nest AS n JOIN nest AS p ...> ON p.lower <= n.lower AND p.upper >= n.upper ...> GROUP BY n.name; name depth ---------- ---------- A 1 A.1 2 A.1.a 3 A.2 2 A.2.a 3 A.2.b 3 A.3 2 There are many other queries that match patterns or differences in the upper and lower bounds. Nested sets are very efficient at calculating many types of queries, but they are expensive to change. For the math to work correctly, there can’t be any gaps in the numbering sequence. This means that any insert or delete requires renumbering a significant num- ber of entries in the table. This isn’t bad for something with a few dozen nodes, but it can quickly prove impractical for a tree with hundreds of nodes. Additionally, because nested sets aren’t based off any kind of key reference, the data- base can’t help enforce the correctness of the tree structure. This leaves database integrity and correctness in the hands of the application—something that is normally avoided. More information This is just a brief overview of how to represent tree relationships. If you need to im- plement a tree, I suggest you do a few web searches on adjacency model or nested set. Many of the larger SQL books mentioned in “Wrap-up” on page 58 also have sections on tree and hierarchies. Normal Form You won’t get too far into most database design books without reading a discussion on normalization and the Normal Forms. The Normal Forms are a series of forms, or table design specifications, that describe the best way to lay out data in a database. The higher the normal form, the more normalized the database is. Each form builds on the previous one, adding additional rules or conditions that must be met. Normalization is the process of removing data duplication, more clearly defining key relationships, and generally moving towards a more idealized database form. It is possible for different tables in the same database to be at different levels. 102 | Chapter 6: Database Design Most people recognize five normal forms simply referred to as the First Normal Form through the Fifth Normal Form. These are often abbreviated 1NF through 5NF. There are also a few named forms, such as the Boyce-Codd Normal Form (BCNF). Most of these other forms are roughly equivalent to one of the numbered forms. For example, BCNF is a slight extension to the Third Normal Form. Some folks also recognize higher levels of normalization, such as a Sixth Normal Form and beyond, but these extreme levels of normalization are well beyond the practical concerns of most database designers. Normalization The normalization process is useful for two reasons. First, normalization specifies de- sign criteria that can act as a guide in the design process. If you have a set of tables that are proving to be difficult to work with, that often points to a deeper design problem or assumption. The normalization process provides a set of rules and conditions that can help identify trouble spots, as well as provide possible solutions to reorganize the data in a more consistent and clean fashion. The other advantage, which shows up more at runtime, is that data integrity is much easier to enforce and maintain in a normalized database. Although the overall database design is often more complex (i.e., more tables), the individual parts are usually much simpler and fill more clearly defined roles. This often translates to better INSERT, UPDATE, and DELETE performance, since changes are often smaller and more localized. Localizing data is a core goal of data normalization. Most of the normal forms deal with eliminating redundant or replicated data so that each unique token of data is stored once—and only once—in the database. Everything else simply references that definitive copy. This makes updates easier, since there is only one place an update needs to be applied, but it also makes the data more consistent, as it is impossible for multiple copies of the data to become out of sync with each other. When working on a schema design, a question you should constantly ask yourself is, “If this piece of data changes, how many different places will I need to make that change?” If the answer is anything other than one, chances are you’re not in Normal Form. Denormalization Normalizing a database and spreading the data out into different tables means that queries usually involve joining several tables back together. This can occasionally lead to performance concerns, especially for complex reports that require data from a large number of tables. These concerns can sometimes lead to the process of denormaliza- tion, where duplicate copies of the same data are intentionally introduced to reduce the number of joins required for common queries. This is typically done on systems that are primarily read-only, such as data-warehouse databases, and is often done by com- puting temporary tables from properly normalized source data. Normal Form | 103 While a large number of joins can lead to performance concerns, database optimization is just like code optimization—don’t start too early and don’t make assumptions. In general, the advantages of normalization far outweigh the costs. A correct database that runs a tad slower is infinitely more useful than a very fast database that returns incorrect or inconsistent answers. The First Normal Form The First Normal Form, or 1NF, is the lowest level of normalization. It is primarily concerned with making sure a table is in the proper format. There are three conditions that must be met for a table to be in 1NF. The first condition relates to ordering. To be in 1NF, the individual rows of a table cannot have any meaningful or inherent order. Each row should be an isolated, stand- alone record. The meaning of a value in one row cannot depend on any of the data values from neighboring rows, either by insertion order, or by some sorted order. This condition is usually easy to meet, as SQL does not guarantee any kind of row ordering. The second condition is uniqueness. Every row within a 1NF table must be unique, and must be unique by those columns that hold meaningful data for the application. For example, if the only difference between two rows is the database-maintained ROWID column, then the rows aren’t really unique. However, it is perfectly fine to con- sider an arbitrary sequence ID (such as an INTEGER PRIMARY KEY) to be part of the ap- plication data. This condition establishes that the table must have some type of PRIMARY KEY, consisting of one or more columns that creates a unique definition of what the table represents. The third and final condition for 1NF requires that every column of every row holds one (and only one) logical value that cannot be broken down any further. The concern is not with compound types, such as dates (which might be broken down into integer day, month, and year values) but with arrays or lists of logical values. For example, you shouldn’t be recording a text value that contains a comma-separated list of logical, independent values. Arrays or lists should be broken out into their own one-to-many relationships. The Second Normal Form The Second Normal Form, or 2NF, deals with compound keys (multicolumn keys) and how other columns relate to such keys. 2NF has only one condition: every column that is not part of the primary key must be relevant to the primary key as a whole, and not just a sub-part of the key. Consider a table that lists all the conference rooms at a large corporate campus. At minimum, the conf_room table has columns for building_numb and room_numb. Taken together, these two columns will uniquely identify any conference room across the whole campus, so that will be our compound primary key. 104 | Chapter 6: Database Design Next, consider a column like seating_capacity. The values in this column are directly dependent on each specific conference room. That, by definition, makes the column dependent on both the building number and the room number. Including the seating_capacity column will not break 2NF. Now consider a column like building_address. This column is dependent on the building_numb column, but it is not dependent on the room_numb column. Since building_address is dependent on only part of the primary key, including this column in the conf_room table would break 2NF. Because 2NF is specifically concerned with multicolumn keys, any table with a single- column primary key that is in 1NF is automatically in 2NF. To recognize a column that might be breaking 2NF, look for columns that have dupli- cate values. If the duplicate values tend to line up with duplicate values in one of the primary key columns, that is a strong indication of a problem. For example, the build ing_address column will have a number of duplicate values (assuming most buildings have more than one conference room). The duplicate address values can be matched to duplicate values in the building_numb column. This alignment shows how the address column is tied to only the building_numb column specifically, and not the whole primary key. The Third Normal Form The Third Normal Form, or 3NF, extends the 2NF to eliminate transitive key depend- encies. A transitive dependency is when A depends on B, and B depends on C, and therefore A depends on C. 3NF requires that each nonprimary key column has a direct (nontransitive) dependency on the primary key. For example, consider an inventory database that is used to track laptops at a small business. The laptop table will have a primary key that uniquely identifies each laptop, such as an inventory control number. It is likely the table would have other columns that include the make and model of the machine, the serial number, and perhaps a purchase date. For our example, the laptop table will also include a responsible_ person_id column. When an employee is assigned a laptop, their employee ID number is put in this column. Within a row, the value of the responsible_person_id column is directly dependent on the primary key. In other words, each individual laptop is assigned a specific responsible person, making the values in the responsible_person_id column directly dependent on the primary key of the laptop table. Now consider what happens when we add a column like responsible_person_email. This is a column that holds the email address of the responsible person. The value of this column is still dependent on the primary key of the laptop table. Each individual laptop has a specific responsible_person_email field that is just as unique as the respon sible_person_id field. Normal Form | 105 The problem is that the values in the responsible_person_email column are not directly dependent on an individual laptop. Rather, the email column is tied to the responsible_person_id, and the responsible_person_id is, in turn, dependent on the individual laptop. This transitive dependency breaks 3NF, indicating that the respon sible_person_email column doesn’t belong there. In the employee table, we will also find both a person_id column and an email column. This is perfectly acceptable if the person_id is the primary key (likely). That would make the email column directly dependent on the primary key, keeping the table in 3NF. A good way to recognize columns that may break 3NF is to look for pairs or sets of unrelated columns that need to be kept in sync with each other. Consider the laptop table. If a system was reassigned to a new person, you would always update both the responsible_person_id column and the responsible_person_email column. The need to keep columns in sync with each other is a strong indication of a dependency to each other, rather than to the primary key. Higher Normal Forms We’re not going to get into the details of BCNF, or the Fourth or Fifth (or beyond) Normal Forms, other than to mention that the Fourth and Fifth Normal Forms start to deal with inter-table relationships and how different tables interact with each other. Most database designers make a solid effort to get everything into 3NF and then stop worrying about it. It turns out that if you get the hang of things and tend to turn out table designs that are in 3NF, chances are pretty good that your tables will also meet the conditions for 4NF and 5NF, if not higher. To a large extent, the higher Normal Forms are formal ways of addressing some edge cases that are somewhat unusual, especially in simpler designs. Although the conditions of the Normal Forms build on each other, the typical design process doesn’t actually iterate over the individual Forms. You don’t sit down with a new design and alter it until everything is 1NF, just to turn around and muck with the design until everything is 2NF, and so on, in a isolated step-by-step manner. Once you understand the ideas and concepts behind the First, Second, and Third Normal Forms, it becomes second nature to design directly to 3NF. Stepping over the conditions one at a time can help you weed out especially difficult trouble spots, but it doesn’t take long to gain a sense of when a design looks clean and when something “just ain’t right.” The core concept to remember is that each table should try to represent one and only one thing. The primary key(s) for that table should uniquely and inherently identify the concept behind the table. All other columns should provide supporting data specific to that one concept. When speaking of the first three Normal Forms in a 1982 CACM article, William Kent wrote that each non-key column “ . . . must provide a fact about the key, the whole key, and nothing but the key.” If you incorporate only one formal aspect of database theory into your designs, that would be a great place to start. 106 | Chapter 6: Database Design Indexes Indexes (or indices) are auxiliary data structures used by the database system to enforce unique constraints, speed up sort operations, and provide faster access to specific re- cords. They are often created in hopes of making queries run faster by avoiding table scans and providing a more direct lookup method. Understanding where and when to create indexes can be an important factor in ach- ieving good performance, especially as datasets grow. Without proper indexes in place, the database system has no option but to do full table scans for every lookup. Table scans can be especially expensive when joining tables together. Indexes are not without cost, however. Each index must maintain a one-to-one corre- spondence between index entries and table rows. If a new row is inserted, updated, or deleted from a table, that same modification must be made to all associated indexes. Each new index will add additional overhead to the INSERT, UPDATE, and DELETE com- mands. Indexes also consume space, both on disk as well as in the SQLite page cache. A proper, well-placed index is worth the cost, but it isn’t wise to just create random indexes, in hopes that one of them will prove useful. A poorly placed index still has all the costs, and can actually slow queries down. You can have too much of a good thing. How They Work Each index is associated with a specific table. Indexes can be either single column or multicolumn, but all the columns of a given index must belong to the same table. There is no limit to the number of indexes a table can have, nor is there a limit on the number of indexes a column can belong to. You cannot create an index on a view or on a virtual table. Internally, the rows of a normal table are stored in an indexed structure. SQLite uses a B-Tree for this purpose, which is a specific type of multi-child, balanced tree. The details are unimportant, other than understanding that as rows are inserted into the tree, the rows are sorted, organized, and optimized, so that a row with a specific, known ROWID can be retrieved relatively directly and quickly. When you create an index, the database system creates another tree structure to hold the index data. Rather than using the ROWID column as the sort key, the tree is sorted and organized using the column or columns you’ve specified in the index definition. The index entry consists of a copy of the values from each of the indexed columns, as well as a copy of the corresponding ROWID value. This allows an indexed entry to be found very quickly using the values from the indexed columns. If we have a table like this: CREATE TABLE tbl ( a, b, c, d ); Indexes | 107 And then create an index that looks like this: CREATE INDEX idx_tbl_a_b ON tbl ( a, b ); The database generates an internal data structure that is conceptually similar to: SELECT a, b, ROWID FROM tbl ORDER BY a, b; If SQLite needs to quickly find all the rows where, for example, a = 45, it can use the sorted index to quickly jump to that range of values and extract the relevant index entries. If it’s looking for the value of b, it can simply extract that from the index and be done. If we need any other value in the row, it needs to fetch the full row. This is done by looking up the ROWID. The last value of any index entry is the ROWID of its corresponding table row. Once SQLite has a list of ROWID values for all rows where a = 45, it can efficiently look up those rows in the original table and retrieve the full row. If everything works correctly, the process of looking up a small set of index entries, and then using those to look up a small set of table rows, will be much faster and more efficient than doing a full table scan. Must Be Diverse To have a positive impact on query performance, indexes must be diverse. The cost of fetching a single row through an index is significantly higher than fetching a single row through a table scan. To reduce the overall cost, an index must overcome this overhead by eliminating the vast majority of row fetches. By significantly reducing the number of row lookups, the total query cost can be reduced, even if the individual row lookups are more expensive. The break-even point for index performance is somewhere in the 10% to 20% range. Any query that fetches more rows from a table will do better with a table scan, while any query that fetches fewer rows will see an improvement by using an index. If an index cannot isolate rows to this level, there is no reason for it to be there. In fact, if the query optimizer mistakenly uses an index that returns a large percentage of rows, the index can reduce performance and slow things down. This creates two concerns. First, if you’re trying to improve the performance of a query that fetches a moderate percentage of rows, adding an index is unlikely to help. An index can only improve performance if it is used to target a focused number of rows. Second, even if the query is asking for a specific value, this can still lead to a higher percentage of fetched rows if the indexed columns are not reasonably unique. For ex- ample, if a column only has four unique values, a successful query will never fetch fewer than approximately 25% of the rows (assuming a reasonable distribution of values or queries). Adding an index to this type of column will not improve performance. Cre- ating an index on a true/false column would be even worse. When the query optimizer is trying to figure out if it should use an index or not, it typically has very little information to go on. For the most part, it assumes that if an 108 | Chapter 6: Database Design index exists, it must be a good index, and it will tend to use it. The ANALYZE command can help mitigate this by providing statistical information to the query optimizer (see ANALYZE for more details), but keeping that updated and well maintained can be difficult in many environments. To avoid problems, you should avoid creating indexes on columns that are not rea- sonably unique. Indexing such columns rarely provides any benefit, and it can confuse the query optimizer. INTEGER PRIMARY KEYs When you declare one or more columns to be a PRIMARY KEY, the database system automatically creates a unique index over those columns. The fundamental purpose of this index is to enforce the UNIQUE constraint that is implied with every PRIMARY KEY. It also happens that many database operations typically involve the primary key, such as natural joins, or conditional lookups in UPDATE or DELETE commands. Even if the index wasn’t required to enforce the UNIQUE constraint, chances are good you would want an index over those columns anyway. There are further advantages to using an INTEGER PRIMARY KEY. As we’ve discussed, when an INTEGER PRIMARY KEY is declared, that column replaces the automatic ROWID column, and becomes the root column for the tree. In essence, the column becomes the index used to store the table itself, eliminating the need for a second, external index. INTEGER PRIMARY KEY columns also provide better performance than standard indexes. INTEGER PRIMARY KEYs have direct access to the full set of row data. A normal index simply references the ROWID value, which is then looked up in the table root. INTEGER PRIMARY KEYs can provide indexed lookups, but skip this second lookup step, and di- rectly access the table data. If you’re using generic row ID values, it is worth the effort to define them as INTEGER PRIMARY KEY columns. Not only will this reduce the size of the database, it can make your queries faster and more efficient. Order Matters When creating a multicolumn index for performance purposes, the column order is very important. If an index has only one column, that column is used as the sort key. If additional columns are specified in the index definition, the additional columns will be used as secondary, tertiary, etc., sort keys. All of the data is sorted by the first column, then any groups of duplicate values are further sorted by the second column, and so on. This is very similar to a typical phonebook. Most phone books sort by last name. Any group of common last names is then sorted by first name, and then by middle name or initial. Indexes | 109 In order to utilize a multicolumn index, a query must contain conditions that are able to utilize the sort keys in the same order they appear in the index definition. If the query does not contain a condition that keys off the first column, the index cannot be used. Consider looking up a name in the phonebook. You can use a phonebook to quickly find the phone number for “Jennifer T. Smith.” First, you look up the last name “Smith.” Then you refine the search, looking for the first name, “Jennifer,” and finally the middle initial “T” (if required). This sequence should allow you to focus in on a specific entry very quickly. A phonebook isn’t much use to quickly find all the people with the first name “Jenni- fer.” Even though the first name is part of the phonebook index, it isn’t the first column of the index. If our query cannot provide a specific condition for the first column of a multicolumn index, the index cannot be used, and our only option is to do a full scan. At that point it is usually more efficient to do a full scan on the table itself, rather than the index. It is not necessary to utilize every column, only that you utilize the columns in order. If you’re looking up a very unique name in the phone book, you may not need to search off the first name or middle initial. Or, perhaps your query is to find every “Smith.” In that case, the conditions of the query are satisfied with just the first column. In the end, you need to be very careful when considering the order of a multicolumn index. The additional columns should be viewed as a refinement on the first column, not a replacement. A single multicolumn index is very different from multiple single- column indexes. If you have different queries that use different columns as the main lookup condition, you may be better off with multiple small indexes, rather than one large multicolumn index. One at a Time Multicolumn indexes are very useful in some situations, but there are limitations on when they can be effectively used. In some cases, it is more appropriate to create mul- tiple single-column indexes on the same table. This too has limitations. You cannot create a single-column index on every column of a table and expect the query system to mix and match whatever indexes it needs. Once a subset of rows is extracted from a table (via index or full table scan), that set of rows conceptually becomes an independent temporary table that is no longer part of the original table. At that point, any index associated with the source table becomes unavailable. In general, the query optimizer can only utilize one index per table-instance per query. Multiple indexes on a single table only make sense if you have a different queries that extract rows using different columns (or have multiple UNIQUE constraints). Different queries that key off different columns can pick the index that is most appropriate, but 110 | Chapter 6: Database Design each individual query will only use one index per table-instance. If you want a single query to utilize multiple indexed columns, you need a multicolumn index. The major exception is a series of OR conditions. If a WHERE clause has a chain of OR conditions on one or more columns of the same table, then there are cases when SQLite may go ahead and use multiple indexes for the same table in a single query. Index Summary For all their power, indexes are often a source of great frustration. The right index will result in a huge performance boost, while the wrong index can significantly slow things down. All indexes add cost to table modifications and add bulk to the database size. A well-placed index is often worth the overhead costs, but it can be difficult to understand what makes a well-placed index. To complicate matters, you don’t get to tell the query optimizer when or how to use your indexes—the query planner makes its own decisions. Having everything be au- tomatic means that queries will work no matter what, but it can also make it difficult to determine if an index is being used. It can be even more difficult to figure out why an index is being ignored. This essentially leaves the database designer in the position of second-guessing the query optimizer. Changes and optimizations must be searched for on something of a trial-and-error basis. To make things worse, as your application changes and evolves, changes in the query structure or patterns can cause a change in index use. All in all, it can be a difficult situation to manage. Traditionally, this is where the role of the DBA, or Database Administrator, comes in. This person is responsible for the care and feeding of a large database server. They do things like monitor query efficiency, adjust tuning parameters and indexes, and make sure all the statistical data is kept up to date. Unfortunately, that whole idea is somewhat contrary to what SQLite is all about. So how do you maintain performance? For starters, have a solid design. A well-crafted and normalized database is going to go a long way toward defining your access patterns. Taking the time to correctly identify and define your primary keys will allow the data- base system to create appropriate indexes and give the query optimizer (and future developers) some insight into your design and intentions. Also remember that (with the exception of UNIQUE constraints) a database will operate correctly without any indexes. It might be slow, but all the answers will be correct. This lets you worry about design first, and performance later. Once you have a working design, it is always possible to add indexes after the fact, without needing to alter your table structure or your query commands. The queries of a fully normalized database will typically have quite a few JOIN opera- tions. Nearly all of these joins are across primary and foreign keys. Primary keys are Indexes | 111 automatically indexed, but in most cases you need to manually create indexes on your foreign keys. With those in place, your database should already have the most critical indexes defined. Start from here. Measure the performance of the different types of queries your appli- cation uses and find the problem areas. Looking at these queries, try to find columns where an index might boost performance. Look for any constraints and conditions that may benefit from an index, especially in join conditions that reference nonkey columns. Use EXPLAIN and EXPLAIN QUERY PLAN to understand how SQLite is accessing the data. These commands can also be used to verify if a query is using an index or not. For more information, see EXPLAIN in Appendix C. You can also use the sqlite3_stmt_status() function to get a more measured understanding of statement efficiency. See sqlite3_stmt_status() in Appendix G for more details. All in all, you shouldn’t get too hung up on precise index placement. You may need a few well-placed indexes to deal with larger tables, but the stock PRIMARY KEY indexes do surprisingly well in most cases. Further index placement should be considered per- formance tuning and shouldn’t be worried about until an actual need is identified. Never forget that indexes have cost, so you shouldn’t create them unless you can iden- tify a need. Transferring Design Experience If you have some experience designing application data structures or class hierarchies, you may have noticed some similarities between designing runtime structures and database tables. Many of the organization principles are the same and, thankfully, much of the design knowledge and experience gained in the application development world will transfer to the database world. There are differences, however. Although these differences are minor, they often prove to be significant stumbling blocks for experienced developers that are new to database design. With a little insight, we can hopefully avoid the more common misconceptions, opening up the world of database design to those with existing experience in data structure and class design. Tables Are Types The most common misconception is to think of tables as instances of a compound data structure. A table looks a whole lot like an array or a dynamic list, so it is easy to make this mistake. Tables should be thought of as type definitions. You should never use a named table as a data organizer or record grouping. Rather, each table definition should be treated like a data structure definition or a class definition. SQL DDL commands such as CREATE TABLE are conceptually similar to those C/C++ header files that define an application’s 112 | Chapter 6: Database Design data structures and classes. The table itself should be considered a global management pool for all instances of that type. If you need a new instance of that type, you simply insert a new row. If you need to group or catalog sets of instances, do that with key associations, not by creating new tables. If you ever find yourself creating a series of tables with identical definitions, that’s usually a big warning. Any time your application uses string manipulation to program- matically build table names, that’s also a big warning—especially if the table names are derived from values stored elsewhere in the database. Keys Are Backwards Pointers Another stumbling block is the proper use of keys. Keys are very similar to pointers. A primary key is used to identify a unique instance of a data structure. This is similar to the address of a record. Anything that wants to reference that record needs to record its address as a pointer or, in the cases of databases, as a foreign key. Foreign keys are essentially database pointers. The trick is that database references are backwards. Rather than pointers that indicate ownership (“I manage that”), foreign keys indicate a type of possession (“I am managed by that”). In C or C++, if a main data record manages a list of sub-records, the main data record would have some kind of pointer list. Each pointer would reference a specific sub- record associated with this main record. If you are dealing with the main data record and need the list of sub-records, you simply look at the pointer list. Databases do it the other way around. In a one-to-many relationship, the main record (the “one” side row) would simply have a primary key. All the sub-records (the “many” side rows) would have foreign keys that point back at the main record. If you are dealing with the main record and need the list of sub-records, you ask the database system to look at the global pool of all subrecord instances and return just those subrecords that are managed by this main record. This tends to make application developers uncomfortable. This is not the way tradi- tional programming language data structures tend to be organized, and the idea of searching a large global record pool just to retrieve a small number of records tends to raise all kinds of performance concerns. Thankfully, this is exactly the kind of thing that databases are very good at doing. Do One Thing My final design advice is more general. As with data structures or classes, the funda- mental idea behind a table is that it should represent instances of one single idea or “thing.” It might represent a set of nouns or things, such as people, or it might represent verbs or actions, such as a transaction log. Tables can even represent less concrete Transferring Design Experience | 113 things, such as a many-to-many bridge table that records membership. But no matter what it is, each table should have one, and only one, clearly defined role in the database. Normally the problem isn’t too many tables, it is too few. If the meaning of one field is ever dependent on the value of another field, the design is heading in a bad direction. Two different meanings should have two different tables (or two different columns). Closing As with application development, database design is part science and part art. It may seem quite complex, with keys to set up, different relationships to define, Normal Forms to follow, and indexes to create. Thankfully, the basics usually fall into place fairly quickly. If you start to get into larger or more complex designs, some reading on more formal methods of data modeling and database design might be in order, but most developers can get pretty far by just lev- eraging their knowledge and experience in designing application data structures. In the end, you’re just defining data structures and hooking them together. Finally, don’t expect to get it right the first time. Very often, when going through the database design process, you realize that your understanding of the thing you’re trying to store is incorrect or incomplete. Just as you might refactor a class hierarchy, don’t be afraid to refactor a database design. It should be an evolving thing, just like your application code and data structures. 114 | Chapter 6: Database Design CHAPTER 7 C Programming Interface The sqlite3 command-line utility is designed to provide an interactive interface for end users. It is extremely useful to design and test out SQL queries, debug database files, and experiment with new features of SQLite, but it was never meant to interface with other applications. While the command-line utility can be used for very basic scripting and automated tasks, if you want to write an application that utilizes the SQLite library in any serious manner, it is expected that you’ll use a programming interface. The native SQLite programming interface is in C, and that is the interface this chapter will cover. If you’re working in something else, there are wrappers and extensions available for many other languages, including most popular scripting languages. With the exception of the Tcl interface, all of these wrappers are provided by third parties and are not part of the SQLite product. For more information on language wrappers, see “Scripting Languages and Other Interfaces” on page 172. Using the C API allows your application to interface directly with the SQLite library and the database engine. You can link a static or dynamic build of the SQLite library into your application, or simply include the amalgamation source file in your applica- tion’s build process. The best choice depends on your specific situation. See “Build and Installation Options” on page 23 for more details. The C API is fairly extensive, and provides full access to all of SQLite’s features. In fact, the sqlite3 command-line utility is written using the public C API. This chapter covers the core features of the API, while the following chapters cover more advanced features. API Overview Even when using the programming interface, the primary way of interacting with your data is to issue SQL commands to the database engine. This chapter focuses on the core of the API that is used to convey SQL command strings to the database engine. It is important to understand that there are no public functions to walk the internal structure of a table or, for example, access the tree structure of an index. You must use SQL to query data from the database. In order to be successful with the SQLite API, 115 you not only need to understand the C API, but you also need to know enough SQL to form meaningful and efficient queries. Structure The C API for SQLite 3 includes a dozen-plus data structures, a fair number of con- stants, and well over one hundred different function calls. While the API is somewhat large, using it doesn’t have to be complex. A fair number of the functions are highly specialized and infrequently used by most developers. Many of the remaining functions are simple variations of the same basic operation. For example, there are a dozen variations on the sqlite3_value_xxx() function, such as sqlite3_value_int(), sqlite3_value_double(), and sqlite3_value_text(). All of these functions perform the same basic operation and can be considered simple type variations of the same basic interface. When referring to a whole category of functions, either in text or in pseudo code, I’ll simply refer to them as the sqlite3_value_xxx() func- tions. Much of the SQLite documentation refers to them as sqlite3_value_*(), but I prefer to use the xxx notation to avoid any confusion with pointers. There are no actual SQLite3 functions with the letter sequence xxx in the name. All public API function calls and datatypes have the prefix sqlite3_, indicating they are part of version 3.x of the SQLite product. Most of the constants, such as error codes, use the prefix SQLITE_. The design and API differences between SQLite 2.x and 3.x were significant enough to warrant a complete change of all API names and structures. The depth of these changes required anyone upgrading from SQLite 2 to SQLite 3 to modify their application, so changing the names of the API functions only helped keep the names distinct and keep any version questions clear. The distinct names also allowed applications that were in transition to link to both libraries at the same time. In addition to the sqlite3_ prefix, public function calls can be identified by the use of lowercase letters and underscores in their names. Private functions use run-together capitalized words (also known as CamelCase). For example, sqlite3_create_function() is a public API function (used to register a user-defined SQL function), while sqlite3CreateFunc() is an internal function that should never be called directly. Internal functions are not in the public header file, are not documented, and are subject to change at any time. The stability of the public interface is extremely important to the SQLite development team. An existing API function call will not be altered once it has been made public. The only possible exceptions are brand new interfaces that are marked experimental, and even experimental interfaces tend to become fairly solid after a few releases. 116 | Chapter 7: C Programming Interface If a revised version of a function call is needed, the newer function will generally be introduced with the suffix _v2. For example, when a more flexible version of the existing sqlite3_open() function was introduced, the old version of the function was retained as is and the new, improved sqlite3_open_v2() was introduced. Although no _v3 (or higher) functions currently exist, it is possible they may be introduced in the future. By adding a new function, rather than modifying the parameters of an existing function, new code could take advantage of the newer features, while existing, unmodified code could continue to link against updated versions of the SQLite library. The use of a different function name also means that if a newer application is ever accidentally linked against an older version of the library, the result will be a link error rather than a program crash, making it much easier to track down and resolve the problem. Strings and Unicode There are a number of API functions that have a 16 variant. For instance, both an sqlite3_column_text() function and an sqlite3_column_text16() function are availa- ble. The first requests a text value in UTF-8 format, while the second will request a text value in UTF-16. All of the strings in an SQLite database file are stored using the same encoding. SQLite database files support the UTF-8, UTF-16LE, and UTF-16BE encodings. A database’s encoding is determined when the database is created. Regardless of the database, you can insert or request text values in either UTF-8 or UTF-16. SQLite will automatically convert text values between the database encoding and the API encoding. The UTF-16 encoding passed by the 16 APIs will always be in the machine’s native byte order. UTF-16 buffers use a void* C data type. The wchar_t data type is not used, as its size is not fixed, and not all platforms define a 16-bit type. Most of the string- and text-related functions have some type of length parameter. SQLite does not assume input text values are null-terminated, so explicit lengths are often required. These lengths are always given in bytes, not characters, regardless of the string encoding. All string lengths are given in bytes, not characters, even if the string uses a multi-byte encoding such as UTF-16. This difference is important to keep in mind when using international strings. API Overview | 117 Error Codes SQLite follows the convention of returning integer error codes in any situation when there is a chance of failure. If data needs to be passed back to the function caller, it is returned through a reference parameter. In all cases, if a function succeeds, it will return the constant SQLITE_OK, which happens to have the value zero. If something went wrong, API functions will return one of the standard error codes to indicate the nature of the error. More recently, a set of extended error codes were introduced. These provide a more specific indication of what went wrong. However, to keep things backwards compat- ible, these extended codes are only available when you activate them. The situation is complex enough to warrant its own discussion later in the chapter. It will be much easier to explain the different error codes once you’ve had a chance to see how the API works. See “Result Codes and Error Codes” on page 146 for more details. I also have to give the standard “do as I say, not as I do” caveat about properly checking error codes and return results. The example code in this chapter and elsewhere in this book tends to have extremely terse (as in, almost none at all) error checking. This is done to keep the examples short and clear. Needless to say, this isn’t the best approach for production code. When working with your own code, do the right thing and check your error codes. Structures and Allocations Although the native SQLite API is often referred to as a C/C++ API, technically the interface is only available in C. As mentioned in “Building” on page 21, the SQLite source code is strictly C based, and as such can only be compiled with a C compiler. Once compiled, the library can be easily linked to, and called from, both C and C++ code, as well as any other language that follows the C linking conventions for your platform. Although the API is written in C, it has a distinct object-like flavor. Most of the program state is held in a series of opaque data structures that act like objects. The most common data structures are database connections and prepared statements. You should never directly access the fields of these data structures. Instead, functions are provided to create, destroy, and manipulate these structures in much the same way that object methods are used to manipulate object instances. This results in an API design that has similar feelings to an object-oriented design. In fact, if you download one of the third- party C++ wrappers, you’ll notice that the wrappers tend to be rather thin, owing most of their structure to the underlying C functions and the data structures. It is important that you allow SQLite to allocate and manage its own data structures. The design of the API means that you should never manually allocate one of these structures, nor should you put these structures on the stack. The API provides calls to 118 | Chapter 7: C Programming Interface internally allocate the proper data structures, initialize them, and return them. Simi- larly, for every function that allocates a data structure, there is some function that is used to clean it up and release it. As with memory management, you need to be sure these calls are balanced, and that every data structure that is created is eventually released. More Info The core of the SQLite API focuses on opening database connections, preparing SQL statements, binding parameter values, executing statements, and finally stepping through the results. These procedures are the focus of this chapter. There are also interfaces to create your own SQL functions, load dynamic modules, and create code-driven virtual tables. We’ll be covering some of these more advanced interfaces in other chapters. Beyond that, there are a fair number of management and customization functions. Not all of these are covered in the main part of the book, but a reference for the full API can be found in Appendix G. If the description of a function leaves you with additional questions, be sure to check that appendix for more specific details. Library Initialization Before an application can use the SQLite library to do anything, the library must first be initialized. This process allocates some standard resources and sets up any OS- specific data structures. By default, most major API function calls will automatically initialize the library, if it has not already been initialized. It is considered a good practice to manually initialize the library, however. int sqlite3_initialize( ) Initializes the SQLite library. This function should be called prior to any other function in the SQLite API. Calling this function after the library has already been initialized is harmless. This function can be called after a shutdown to reinitialize the library. A return value of SQLITE_OK indicates success. When an application is finished using the SQLite library, the library should be shut down. int sqlite3_shutdown( ) Releases any resources allocated by sqlite3_initialize(). Calling this function before the library has been initialized or after the library has already been shut down is harmless. A return value of SQLITE_OK indicates success. Because of the automatic initialization features, many applications never call either of these functions. Rather, they call sqlite3_open(), or one of the other primary functions, and depend on the library to automatically initialize itself. In most cases this is safe enough, but for maximum compatibility it is best to call these functions explicitly. Library Initialization | 119 Database Connections Before we can prepare or execute SQL statements, we must first establish a database connection. Most often this is done by opening or creating an SQLite3 database file. When you are done with the database connection, it must be closed. This verifies that there are no outstanding statements or allocated resources before closing the database file. Opening Database connections are allocated and established with one of the sqlite3_ open_xxx() commands. These pass back a database connection in the form of an sqlite3 data structure. There are three variants: int sqlite3_open( const char *filename, sqlite3 **db_ptr ) int sqlite3_open16( const void *filename, sqlite3 **db_ptr ) Opens a database file and allocates an sqlite3 data structure. The first parameter is the filename of the database file you wish to open, given as a null-terminated string. The second parameter is a reference to an sqlite3 pointer, and is used to pass back the new connection. If possible, the database will be opened read/write. If not, it will be opened read-only. If the given database file does not exist, it will be created. The first variant assumes that the database filename is encoded in UTF-8, while the second assumes that the database filename is encoded in UTF-16. int sqlite3_open_v2( const char *filename, sqlite3 **db_ptr, int flags, const char *vfs_name ) The _v2 variant offers more control over how the database file is created and opened. The first two parameters are the same. The filename is assumed to be in UTF-8. There is no UTF-16 variant of this function. A third parameter is a set of bit-field flags. These flags allow you to specify if SQLite should attempt to open the database read/write (SQLITE_OPEN_READWRITE), or read- only (SQLITE_OPEN_READONLY). If you ask for read/write access but only read-only access is available, the database will be opened in read-only mode. This variant of open will not create a new file for an unknown filename unless you explicitly allow it using the SQLITE_OPEN_CREATE flag. This only works if the database is being opened in read/write mode. There are also a number of other flags dealing with thread and cache management. See sqlite3_open() in Appendix G for more details. The standard version of open is equivalent to the flag values of (SQLITE_READWRITE | SQLITE_CREATE). 120 | Chapter 7: C Programming Interface The final parameter allows you to name a VFS (Virtual File System) module to use with this database connection. The VFS system acts as an abstraction layer between the SQLite library and any underlying storage system (such as a filesystem). In nearly all cases, you will want to use the default VFS module and can simply pass in a NULL pointer. For new code, it is recommended that you use the call sqlite3_open_v2(). The newer call allows more control over how the database is opened and processed. The use of the double pointer may be a bit confusing at first, but the idea behind it is simple enough. The pointer-to-a-pointer is really nothing more than a pointer that is passed by reference. This allows the function call to modify the pointer that is passed in. For example: sqlite3 *db = NULL; rc = sqlite3_open_v2( "database.sqlite3", &db, SQLITE_OPEN_READWRITE, NULL ); /* hopefully, db now points to a valid sqlite3 structure */ Note that db is an sqlite3 pointer (sqlite3*), not an actual sqlite3 structure. When we call sqlite3_open_xxx() and pass in the pointer reference, the open function will allocate a new sqlite3 data structure, initialize it, and set our pointer to point to it. This approach, including the use of a pointer reference, is a common theme in the SQLite APIs that are used to create or initialize something. They all basically work the same way and, once you get the hang of them, they are pretty straightforward and easy to use. There is no standard file extension for an SQLite3 database file, although .sqlite3, .db, and .db3 are popular choices. The extension .sdb should be avoided, as this extension has special meaning on some Microsoft Windows platforms, and may suffer from sig- nificantly slower I/O performance. The string encoding used by a database file is determined by the function that is used to create the file. Using sqlite3_open() or sqlite3_open_v2() will result in a database with the default UTF-8 encoding. If sqlite3_open16() is used to create a database, the default string encoding will be UTF-16 in the native byte order of the machine. You can override the default string encoding with the SQL command PRAGMA encoding. See encoding in Appendix F for more details. Special Cases In addition to recognizing standard filenames, SQLite recognizes a few specialized filename strings. If the given filename is a NULL pointer or an empty string (""), then an anonymous, temporary, on-disk database is created. An anonymous database can only be accessed through the database connection that created it. Each call will create a new, unique database instance. Like all temporary items, this database will be destroyed when the connection is closed. Database Connections | 121 If the filename :memory: is used, then a temporary, in-memory database is created. In- memory databases live in the database cache and have no backing store. This style of database is extremely fast, but requires sufficient memory to hold the entire database image in memory. As with anonymous databases, each open call will create a new, unique, in-memory database, making it impossible for more than one database con- nection to access a given in-memory database. In-memory databases make great structured caches. It is not possible to directly image an in-memory database to disk, but you can copy the contents of an in-memory data- base to disk (or disk to memory) using the database backup API. See the section sqlite3_backup_init() in Appendix G for more details. Closing To close and release a database connection, call sqlite3_close(). int sqlite3_close( sqlite3 *db ) Closes a database connection and releases any associated data structures. All tem- porary items associated with this connection will be deleted. In order to succeed, all prepared statements associated with this database connection must be finalized. See “Reset and Finalize” on page 130 for more details. Any pointer returned by a call to sqlite3_open_xxx(), including a NULL pointer, can be passed to sqlite3_close(). This function verifies there are no outstanding changes to the database, then closes the file and frees the sqlite3 data structure. If the database still has nonfinalized statements, the SQLITE_BUSY error will be returned. In that case, you need to correct the problem and call sqlite3_close() again. In most cases, sqlite3_open_xxx() will return a pointer to an sqlite3 structure, even when the return code indicates a problem. This allows the caller to retrieve an error message with sqlite3_errmsg(). (See “Result Codes and Error Codes” on page 146.) In these situations, you must still call sqlite3_close() to free the sqlite3 structure. Example Here is the outline of a program that opens a database, performs some operations, and then closes it. Most of the other examples in this chapter will build from this example by inserting code into the middle: #include "sqlite3.h" #include int main( int argc, char **argv ) { char *file = ""; /* default to temp db */ sqlite3 *db = NULL; int rc = 0; 122 | Chapter 7: C Programming Interface if ( argc > 1 ) file = argv[1]; sqlite3_initialize( ); rc = sqlite3_open_v2( file, &db, SQLITE_OPEN_READWRITE | SQLITE_OPEN_CREATE, NULL ); if ( rc != SQLITE_OK) { sqlite3_close( db ); exit( -1 ); } /* perform database operations */ sqlite3_close( db ); sqlite3_shutdown( ); } The default filename is an empty string. If passed to sqlite3_open_xxx(), this will result in a temporary, on-disk database that will be deleted as soon as the database connection is closed. If at least one argument is given, the first argument will be used as a filename. If the database does not exist, it will be created and then opened for read/write access. It is then immediately closed. If this example is run using a new filename, it will not create a valid database file. The SQLite library delays writing the database header until some actual data operation is performed. This “lazy” initialization gives an application the chance to adjust any relevant pragmas, such as the text encoding, page size, and database file format, before the database file is fully created. Prepared Statements Once a database connection is established, we can start to execute SQL commands. This is normally done by preparing and stepping through statements. Statements are held in sqlite3_stmt data structures. Statement Life Cycle The life cycle of a prepared statement is a bit complex. Unlike database connections, which are typically opened, used for some period of time, and then closed, a statement can be in a number of different states. A statement might be prepared, but not run, or it might be in the middle of processing. Once a statement has run to completion, it can be reset and re-executed multiple times before eventually being finalized and released. Prepared Statements | 123 The life cycle of a typical sqlite3_stmt looks something like this (in pseudo-code): /* create a statement from an SQL string */ sqlite3_stmt *stmt = NULL; sqlite3_prepare_v2( db, sql_str, sql_str_len, &stmt, NULL ); /* use the statement as many times as required */ while( ... ) { /* bind any parameter values */ sqlite3_bind_xxx( stmt, param_idx, param_value... ); ... /* execute statement and step over each row of the result set */ while ( sqlite3_step( stmt ) == SQLITE_ROW ) { /* extract column values from the current result row */ col_val = sqlite3_column_xxx( stmt, col_index ); ... } /* reset the statement so it may be used again */ sqlite3_reset( stmt ); sqlite3_clear_bindings( stmt ); /* optional */ } /* destroy and release the statement */ sqlite3_finalize( stmt ); stmt = NULL; The prepare process converts an SQL command string into a prepared statement. That statement can then have values bound to any statement parameters. The statement is then executed, or “stepped through.” In the case of a query, each step will make a new results row available for processing. The column values of the current row can then be extracted and processed. The statement is stepped through, row by row, until no more rows are available. The statement can then be reset, allowing it to be re-executed with a new set of bindings. Preparing a statement can be somewhat costly, so it is a common practice to reuse statements as much as possible. Finally, when the statement is no longer in use, the sqlite3_stmt data structure can be finalized. This releases any internal resources and frees the sqlite3_stmt data structure, effectively deleting the statement. Prepare To convert an SQL command string into a prepared statement, use one of the sqlite3_prepare_xxx() functions: 124 | Chapter 7: C Programming Interface int sqlite3_prepare( sqlite3 *db, const char *sql_str, int sql_str_len, sqlite3_stmt **stmt, const char **tail ) int sqlite3_prepare16( sqlite3 *db, const void *sql_str, int sql_str_len, sqlite3_stmt **stmt, const void **tail ) It is strongly recommended that all new developments use the _v2 version of these functions. int sqlite3_prepare_v2( sqlite3 *db, const char *sql_str, int sql_str_len, sqlite3_stmt **stmt, const char **tail ) int sqlite3_prepare16_v2( sqlite3 *db, const void *sql_str, int sql_str_len, sqlite3_stmt **stmt, const void **tail ) Converts an SQL command string into a prepared statement. The first parameter is a database connection. The second parameter is an SQL command encoded in a UTF-8 or UTF-16 string. The third parameter indicates the length of the com- mand string in bytes. The fourth parameter is a reference to a statement pointer. This is used to pass back a pointer to the new sqlite3_stmt structure. The fifth parameter is a reference to a string (char pointer). If the command string contains multiple SQL statements and this parameter is non-NULL, the pointer will be set to the start of the next statement in the command string. These _v2 calls take the exact same parameters as the original versions, but the internal representation of the sqlite3_stmt structure that is created is somewhat different. This enables some extended and automatic error handling. These dif- ferences are discussed later in “Result Codes and Error Codes” on page 146. If the length parameter is negative, the length will be automatically computed by the prepare call. This requires that the command string be properly null-terminated. If the length is positive, it represents the maximum number of bytes that will be parsed. For optimal performance, provide a null-terminated string and pass a valid length value that includes the null-termination character. If the SQL command string passed to sqlite3_prepare_xxx() consists of only a single SQL statement, there is no need to terminate it with a semicolon. Once a statement has been prepared, but before it is executed, you can bind parameter values to the statement. Statement parameters allow you to insert a special token into the SQL command string that represents an unspecified literal value. You can then bind specific values to the parameter tokens before the statement is executed. After execu- tion, the statement can be reset and new parameter values can be assigned. This allows you to prepare a statement once and then re-execute it multiple times with different parameter values. This is commonly used with commands, such as INSERT, that have a common structure but are repeatedly executed with different values. Parameter binding is a somewhat in-depth topic, so we’ll get back to that in the next section. See “Bound Parameters” on page 133 for more details. Prepared Statements | 125 Step Preparing an SQL statement causes the command string to be parsed and converted into a set of byte-code commands. This byte-code is fed into SQLite’s Virtual Database Engine (VDBE) for execution. The translation is not a consistent one-to-one affair. Depending on the database structure (such as indexes), the query optimizer may gen- erate very different VDBE command sequences for similar SQL commands. The size and flexibility of the SQLite library can be largely attributed to the VDBE architecture. To execute the VDBE code, the function sqlite3_step() is called. This function steps through the current VDBE command sequence until some type of program break is encountered. This can happen when a new row becomes available, or when the VDBE program reaches its end, indicating that no more data is available. In the case of a SELECT query, sqlite3_step() will return once for each row in the result set. Each subsequent call to sqlite3_step() will continue execution of the statement until the next row is available or the statement reaches its end. The function definition is quite simple: int sqlite3_step( sqlite3_stmt *stmt ) Attempts to execute the provided prepared statement. If a result set row becomes available, the function will return with a value of SQLITE_ROW. In that case, individ- ual column values can be extracted with the sqlite3_column_xxx() functions. Ad- ditional rows can be returned by making further calls to sqlite3_step(). If the statement execution reaches its end, the code SQLITE_DONE will be returned. Once this happens, sqlite3_step() cannot be called again with this prepared statement until the statement is first reset using sqlite3_reset(). If the first call to sqlite3_step() returns SQLITE_DONE, it means that the statement was successfully run, but there was no result data to make available. This is the typical case for most commands, other than SELECT. If sqlite3_step() is called repeatedly, a SELECT command will return SQLITE_ROW for each row of the result set before finally returning SQLITE_DONE. If a SELECT command returns no rows, it will return SQLITE_DONE on the first call to sqlite3_step(). There are also some PRAGMA commands that will return a value. Even if the return value is a simple scalar value, that value will be returned as a one-row, one-column result set. This means that the first call to sqlite3_step() will return SQLITE_ROW, indicating result data is available. Additionally, if PRAGMA count_changes is set to true, the INSERT, UPDATE, and DELETE commands will return the number of rows they modified as a one- row, one-column integer value. Any time sqlite3_step() returns SQLITE_ROW, new row data is available for processing. Row values can be inspected and extracted from the statement using the sqlite3_ column_xxx() functions, which we will look at next. To resume execution of the state- ment, simply call sqlite3_step() again. It is common to call sqlite3_step() in a loop, processing each row until SQLITE_DONE is returned. 126 | Chapter 7: C Programming Interface Rows are returned as soon as they are computed. In many cases, this spreads the pro- cessing costs out across all of the calls to sqlite3_step(), and allows the first row to be returned reasonably quickly. However, if the query has a GROUP BY or ORDER BY clause, the statement may be forced to first gather all of the rows within the result set before it is able to complete the final processing. In these cases, it may take a considerable time for the first row to become available, but subsequent rows should be returned very very quickly. Result Columns Any time sqlite3_step() returns the code SQLITE_ROW, a new result set row is available within the statement. You can use the sqlite3_column_xxx() functions to inspect and extract the column values from this row. Many of these functions require a column index parameter (cidx). Like C arrays, the first column in a result set always has an index of zero, starting from the left. int sqlite3_column_count( sqlite3_stmt *stmt ) Returns the number of columns in the statement result. If the statement does not return values, a count of zero will be returned. Valid column indexes are zero through the count minus one. (N columns have the indexes 0 through N-1). const char* sqlite3_column_name( sqlite3_stmt *stmt, int cidx ) const void* sqlite3_column_name16( sqlite3_stmt *stmt, int cidx ) Returns the name of the specified column as a UTF-8 or UTF-16 encoded string. The returned string is the name provided by the AS clause within the SELECT header. For example, this function would return person_id for column zero of the SQL statement SELECT pid AS person_id,.... If no AS expression was given, the name is technically undefined and may change from one version of SQLite to another. This is especially true of columns that consist of an expression. The returned pointers will remain valid until one of these functions is called again on the same column index, or until the statement is destroyed with sqlite3_final ize(). The pointers will remain valid (and unmodified) across calls to sqlite3_step() and sqlite3_reset(), as column names do not change from one execution to the next. These pointers should not be passed to sqlite3_free(). int sqlite3_column_type( sqlite3_stmt *stmt, int cidx ) Returns the native type (storage class) of the value found in the specified column. Valid return codes can be SQLITE_INTEGER, SQLITE_FLOAT, SQLITE_TEXT, SQLITE_BLOB, or SQLITE_NULL. To get the correct native datatype, this function should be called before any attempt is made to extract the data. This function returns the type of the actual value found in the current row. Because SQLite allows different types to be stored in the same column, the type returned for a specific column index may vary from row to row. This is also how you detect the presence of a NULL. Prepared Statements | 127 These sqlite3_column_xxx() functions allow your code to get an idea of what the avail- able row looks like. Once you’ve figured out the correct value type, you can extract the value with one of these typed sqlite3_column_xxx() functions. All of these functions take the same parameters: a statement pointer and a column index. const void* sqlite3_column_blob( sqlite_stmt *stmt, int cidx ) Returns a pointer to the BLOB value from the given column. The pointer may be invalid if the BLOB has a length of zero bytes. The pointer may also be NULL if a type conversion was required. double sqlite3_column_double( sqlite_stmt *stmt, int cidx ) Returns a 64-bit floating-point value from the given column. int sqlite3_column_int( sqlite_stmt *stmt, int cidx ) Returns a 32-bit signed integer from the given column. The value will be truncated (without warning) if the column contains an integer value that cannot be repre- sented in 32 bits. sqlite3_int64 sqlite3_column_int64( sqlite_stmt *stmt, int cidx ) Returns a 64-bit signed integer from the given column. const unsigned char* sqlite3_column_text( sqlite_stmt *stmt, int cidx ) const void* sqlite3_column_text16( sqlite_stmt *stmt, int cidx ) Returns a pointer to a UTF-8 or UTF-16 encoded string from the given column. The string will always be null-terminated, even if it is an empty string. Note that the returned char pointer is unsigned and will likely require a cast. The pointer may also be NULL if a type conversion was required. sqlite3_value* sqlite3_column_value( sqlite_stmt *stmt, int cidx ) Returns a pointer to an unprotected sqlite3_value structure. Unprotected sqlite3_value structures cannot safely undergo type conversion, so you should not attempt to extract a primitive value from this structure using the sqlite3_value _xxx() functions. If you want a primitive value, you should use one of the other sqlite3_column_xxx() functions. The only safe use for the returned pointer is to call sqlite3_bind_value() or sqlite3_result_value(). The first is used to bind the value to another prepared statement, while the second is used to return a value in a user-defined SQL function (see “Binding Values” on page 135, or “Returning Results and Errors” on page 186). There is no sqlite3_column_null() function. There is no need for one. If the native datatype is NULL, there is no additional value or state information to extract. Any pointers returned by these functions become invalid if another call to any sqlite3_column_xxx() function is made using the same column index, or when sqlite3_step() is next called. Pointers will also become invalid if the statement is reset or finalized. SQLite will take care of all the memory management associated with these pointers. 128 | Chapter 7: C Programming Interface If you request a datatype that is different from the native value, SQLite will attempt to convert the value. Table 7-1 describes the conversion rules used by SQLite. Table 7-1. SQLite type conversion rules. Original type Requested type Converted value NULL Integer 0 NULL Float 0.0 NULL Text NULL pointer NULL BLOB NULL pointer Integer Float Converted float Integer Text ASCII number Integer BLOB Same as text Float Integer Rounds towards zero Float Text ASCII number Float BLOB Same as text Text Integer Internal atoi() Text Float Internal atof() Text BLOB No change BLOB Integer Converts to text, atoi() BLOB Float Converts to text, atof() BLOB Text Adds terminator Some conversions are done in place, which can cause subsequent calls to sqlite3_ column_type() to return undefined results. That’s why it is important to call sqlite3_column_type() before trying to extract a value, unless you already know exactly what datatype you want. Although numeric values are returned directly, text and BLOB values are returned in a buffer. To determine how large that buffer is, you need to ask for the byte count. That can be done with one of these two functions. int sqlite3_column_bytes( sqlite3_stmt *stmt, int cidx ) Returns the number of bytes in a BLOB or in a UTF-8 encoded text value. If re- turning the size of a text value, the size will include the terminator. int sqlite3_column_bytes16( sqlite3_stmt *stmt, int cidx ) Returns the number of bytes in a UTF-16 encoded text value, including the terminator. Prepared Statements | 129 Be aware that these functions can cause a data conversion in text values. That conver- sion can invalidate any previously returned pointer. For example, if you call sqlite3_column_text() to get a pointer to a UTF-8 encoded string, and then call sqlite3_column_bytes16() on the same column, the internal column value will be con- verted from a UTF-8 encoded string to a UTF-16 encoded string. This will invalidate the character pointer that was originally returned by sqlite3_column_text(). Similarly, if you first call sqlite3_column_bytes16() to get the size of UTF-16 encoded string, and then call sqlite3_column_text(), the internal value will be converted to a UTF-8 string before a string pointer is returned. That will invalidate the length value that was originally returned. The easiest way to avoid problems is to extract the datatype you want and then call the matching bytes function to find out how large the buffer is. Here are examples of safe call sequences: /* correctly extract a blob */ buf_ptr = sqlite3_column_blob( stmt, n ); buf_len = sqlite3_column_bytes( stmt, n ); /* correctly extract a UTF-8 encoded string */ buf_ptr = sqlite3_column_text( stmt, n ); buf_len = sqlite3_column_bytes( stmt, n ); /* correctly extract a UTF-16 encoded string */ buf_ptr = sqlite3_column_text16( stmt, n ); buf_len = sqlite3_column_bytes16( stmt, n ); By matching the correct bytes function for your desired datatype, you can avoid any type conversions keeping both the pointer and length valid and correct. You should always use sqlite3_column_bytes() to determine the size of a BLOB. Reset and Finalize When a call to sqlite3_step() returns SQLITE_DONE, the statement has successfully fin- ished execution. At that point, there is nothing further you can do with the statement. If you want to use the statement again, it must first be reset. int sqlite3_reset( sqlite3_stmt *stmt ) Resets a prepared statement so that it is ready for another execution. A statement should be reset as soon as you’re done using it. This will ensure any locks are released. The function sqlite3_reset() can be called any time after sqlite3_step() is called. It is valid to call sqlite3_reset() before a statement is finished executing (that is, before sqlite3_step() returns SQLITE_DONE or an error indicator). You can’t cancel a running sqlite3_step() call this way, but you can short-circuit the return of additional SQLITE_ROW values. 130 | Chapter 7: C Programming Interface For example, if you only want the first six rows of a result set, it is perfectly valid to call sqlite3_step() only six times and then reset the statement, even if sqlite3_step() would continue to return SQLITE_ROW. The function sqlite3_reset() simply resets a statement, it does not release it. To de- stroy a prepared statement and release its memory, the statement must be finalized. int sqlite3_finalize( sqlite3_stmt *stmt ) Destroys a prepared statement and releases any associated resources. The function sqlite3_finalize() can be called at any time on any statement that was successfully prepared. All of the prepared statements associated with a database con- nection must be finalized before the database connection can be closed. Although both of these functions can return errors, they always perform their function. Any error that is returned was generated by the last call to sqlite3_step(). See “Result Codes and Error Codes” on page 146 for more details. It is a good idea to reset or finalize a statement as soon as you are done using it. A call to sqlite3_reset() or sqlite3_finalize() ensures the statement will release any locks it might be holding, and frees any resources associated with the prior statement exe- cution. If an application keeps statements around for an extended period of time, they should be kept in a reset state, ready to be bound and executed. Statement Transitions Prepared statements have a significant amount of state. In addition to the currently bound parameter values and other details, every prepared statement is always in one of three major states. The first is the “ready” state. Any freshly prepared or reset state- ment will be “ready.” This indicates that the statement is ready to execute, but hasn’t been started. The second state is “running,” indicating that a statement has started to execute, but hasn’t yet finished. The final state is “done,” which indicates the statement has completed executing. Knowing the current state of a statement is important. Although some API functions can be called at any time (like sqlite3_reset()), other API functions can only be called when a statement is in a specific state. For example, the sqlite3_bind_xxx() functions can only be called when a statement is in its “ready” state. Figure 7-1 shows the different states and how a statement transitions from one state to another. There is no way to query the current state of a statement. Transitions between states are normally controlled by the design and flow of the application. Prepared Statements | 131 Figure 7-1. Prepared statement transitions. A statement can be in one of three states. Depending on the current state, only some API functions are valid. Calling a function in an inappropriate state will result in an SQLITE_MISUSE error. Examples Here are two examples of using prepared statements. The first example executes a CREATE TABLE statement by first preparing the SQL string and then calling sqlite3_step() to execute the statement: sqlite3_stmt *stmt = NULL; /* ... open database ... */ rc = sqlite3_prepare_v2( db, "CREATE TABLE tbl ( str TEXT )", -1, &stmt, NULL ); if ( rc != SQLITE_OK) exit( -1 ); rc = sqlite3_step( stmt ); if ( rc != SQLITE_DONE ) exit ( -1 ); sqlite3_finalize( stmt ); /* ... close database ... */ The CREATE TABLE statement is a DDL command that does not return any type of value and only needs to be “stepped” once to fully execute the command. Remember to reset or finalize statements as soon as they’re finished executing. Also remember that all statements associated with a database connection must be fully finalized before the connection can be closed. 132 | Chapter 7: C Programming Interface This second example is a bit more complex. This code performs a SELECT and loops over sqlite3_step() extracting all of the rows in the table. Each value is displayed as it is extracted: const char *data = NULL; sqlite3_stmt *stmt = NULL; /* ... open database ... */ rc = sqlite3_prepare_v2( db, "SELECT str FROM tbl ORDER BY 1", -1, &stmt, NULL ); if ( rc != SQLITE_OK) exit( -1 ); while( sqlite3_step( stmt ) == SQLITE_ROW ) { data = (const char*)sqlite3_column_text( stmt, 0 ); printf( "%s\n", data ? data : "[NULL]" ); } sqlite3_finalize( stmt ); /* ... close database ... */ This example does not check the type of the column value. Since the value will be displayed as a string, the code depends on SQLite’s internal conversion process and always requests a text value. The only tricky bit is that the string pointer may be NULL, so we need to be prepared to deal with that in the printf() statement. Bound Parameters Statement parameters are special tokens that are inserted into the SQL command string before it is passed to one of the sqlite3_prepare_xxx() functions. They act as a place- holder for any literal value, such as a bare number or a single quote string. After the statement is prepared, but before it is executed, you can bind specific values to each statement parameter. Once you’re done executing a statement, you can reset the state- ment, bind new values to the parameters, and execute the statement again—only this time with the new values. Parameter Tokens SQLite supports five different styles of statement parameters. These short string tokens are placed directly into the SQL command string, which can then be passed to one of the sqlite3_prepare_xxx() functions. Once the statement is prepared, the individual parameters are referenced by index. ? An anonymous parameter with automatic index. As the statement is processed, each anonymous parameter is assigned a unique, sequential index value, starting with one. Bound Parameters | 133 ? Parameter with explicit numeric index. Duplicate indexes allow the same value to be bound multiple places in the same statement. : A named parameter with an automatic index. Duplicate names allow the same value to be bound multiple places in the same statement. @ A named parameter with an automatic index. Duplicate names allow the same value to be bound multiple places in the same statement. Works exactly like the colon parameter. $ A named parameter with an automatic index. Duplicate names allow the same value to be bound multiple places in the same statement. This is an extended syntax to support Tcl variables. Unless you’re doing Tcl programming, I suggest you use the colon format. To get an idea of how these work, consider this INSERT statement: INSERT INTO people (id, name) VALUES ( ?, ? ); The two statement parameters represent the id and name values being inserted. Param- eter indexes start at one, so the first parameter that represents the id value has an index of one, and the parameter used to reference the name value has an index of two. Notice that the second parameter, which is likely a text value, does not have single quotes around it. The single quotes are part of the string-literal representation, and are not required for a parameter value. Statement parameters should not be put in quotes. The notation '?' designates a one-character text value, not a parameter. Once this statement has been prepared, it can be used to insert multiple rows. For each row, simply bind the appropriate values, step through the statement, and then reset the statement. After the statement has been reset, new values can be bound to the parameters and the statement can be stepped again. You can also use explicit index values: INSERT INTO people (id, name) VALUES ( ?1, ?2 ); Using explicit parameter indexes has two major advantages. First, you can have mul- tiple instances of the same index value, allowing the same value to be bound to more than one place in the same statement. 134 | Chapter 7: C Programming Interface Second, explicit indexes allow the parameters to appear out of order. There can even be gaps in the index sequence. This can help simplify application code maintenance if the query is modified and parameters are added or removed. This level of abstraction can be taken even further by using named parameters. In this case, you allow SQLite to assign parameter index values as it sees fit, in a similar fashion to anonymous parameters. The difference is that you can ask SQLite to tell you the index value of a specific parameter based off the name you’ve given it. Consider this statement: INSERT INTO people (id, name) VALUES ( :id, :name ); In this case, the parameter values are quite explicit. As we will see in the next section, the code that binds values to these parameters is also quite explicit, making it very clear what is going on. Best of all, it doesn’t matter if new parameters are added. As long as the existing names remain unchanged, the code will properly find and bind the named parameters. Note, however, that parameters can only be used to replace literal values, such as quo- ted strings or numeric values. Parameters cannot be used in place of identifiers, such as table names or column names. The following bit of SQL is invalid: SELECT * FROM ?; -- INCORRECT: Cannot use a parameter as an identifier If you attempt to prepare this statement, it will fail. This is because the parameter (which acts as an unknown literal value) is being used where an identifier is required. This is invalid, and the statement will not prepare correctly. Within a statement, it is best to choose a specific parameter style and stick with it. Mixing anonymous parameters with explicit indexes or named parameters is likely to cause confusion about what index belongs to which parameter. Personally, I prefer to used the colon-name-style parameters. Using named parameters eliminates the need to know any specific index values, allowing you to just reference the name at runtime. The use of short, significant names can also make the intent of both your SQL statements and your bind code easier to understand. Binding Values When you first prepare a statement with parameters, all of the parameters start out with a NULL assigned to them. Before you execute the statement, you can bind specific values to each parameter using the sqlite3_bind_xxx() family of functions. There are nine sqlite3_bind_xxx() functions available, plus a number of utility func- tions. These functions can be called any time after the statement is prepared, but before sqlite3_step() is called for the first time. Once sqlite3_step() has been called, these functions cannot be called again until the statement is reset. Bound Parameters | 135 All the sqlite3_bind_xxx() functions have the same first and second parameters and return the same result. The first parameter is always a pointer to an sqlite3_stmt, and the second is the index of the parameter to bind. Remember that for anonymous pa- rameters, the first index value starts with one. For the most part, the third parameter is the value to bind. The fourth parameter, if present, indicates the length of the data value in bytes. The fifth parameter, if present, is a function pointer to a memory man- agement callback. Remember that bind index values start with one (1), unlike result column indexes, which start with zero (0). All the bind functions return an integer error code, which is equal to SQLITE_OK upon success. The bind functions are: int sqlite3_bind_blob( sqlite3_stmt *stmt, int pidx, const void *data, int data_len, mem_callback ) Binds an arbitrary length binary data BLOB. int sqlite3_bind_double( sqlite3_stmt *stmt, int pidx, double data ) Binds a 64-bit floating point value. int sqlite3_bind_int( sqlite3_stmt *stmt, int pidx, int data ) Binds a 32-bit signed integer value. int sqlite3_bind_int64( sqlite3_stmt *stmt, int pidx, sqlite3_int64 ) Binds a 64-bit signed integer value. int sqlite3_bind_null( sqlite3_stmt *stmt, int pidx ) Binds a NULL datatype. int sqlite3_bind_text( sqlite3_stmt *stmt, int pidx, const char *data, int data_len, mem_callback ) Binds an arbitrary length UTF-8 encoded text value. The length is in bytes, not characters. If the length parameter is negative, SQLite will compute the length of the string up to, but not including, the null terminator. It is recommended that the manually computed lengths do not include the terminator (the terminator will be included when the value is returned). int sqlite3_bind_text16( sqlite3_stmt *stmt, int pidx, const void *data, int data_len, mem_callback ) Binds an arbitrary length UTF-16 encoded text value. The length is in bytes, not characters. If the length parameter is negative, SQLite will compute the length of the string up to, but not including, the null terminator. It is recommended that the manually computed lengths do not include the terminator (the terminator will be included when the value is returned). 136 | Chapter 7: C Programming Interface int sqlite3_bind_zeroblob( sqlite3_stmt *stmt, int pidx, int len ) Binds an arbitrary length binary data BLOB, where each byte is set to zero (0x00). The only additional parameter is a length value, in bytes. This function is partic- ularly useful for creating large BLOBs that can then be updated with the incre- mental BLOB interface. See sqlite3_blob_open() in Appendix G for more details. In addition to these type-specific bind functions, there is also a specialized function: int sqlite3_bind_value( sqlite3_stmt *stmt, int pidx, const sqlite3_value *data_value ) Binds the type and value of an sqlite3_value structure. An sqlite3_value structure can hold any data format. The text and BLOB variants of sqlite3_bind_xxx() require you to pass a buffer pointer for the data value. Normally this buffer and its contents must remain valid until a new value is bound to that parameter, or the statement is finalized. Since that might be some time later in the code, these bind functions have a fifth parameter that controls how the buffer memory is handled and possibly released. If the fifth parameter is either NULL or the constant SQLITE_STATIC, SQLite will take a hands-off approach and assume the buffer memory is either static or that your appli- cation code is taking care of maintaining and releasing any memory. If the fifth parameter is the constant SQLITE_TRANSIENT, SQLite will make an internal copy of the buffer. This allows you to release your buffer immediately (or allow it to go out of scope, if it happens to be on the stack). SQLite will automatically release the internal buffer at an appropriate time. The final option is to pass a valid void mem_callback( void* ptr ) function pointer. This callback will be called when SQLite is done with the buffer and wants to release it. If the buffer was allocated with sqlite3_malloc() or sqlite3_realloc(), you can pass a reference to sqlite3_free() directly. If you allocated the buffer with a different set of memory management calls, you’ll need to pass a reference to a wrapper function that calls the appropriate memory release function. Once a value has been bound to a parameter, there is no way to extract that value back out of the statement. If you need to reference a value after it has been bound, you must keep track of it yourself. To help you figure out what parameter index to use, there are three utility functions: int sqlite3_bind_parameter_count( sqlite3_stmt *stmt ) Returns an integer indicating the largest parameter index. If no explicit numeric indexes are used ( ? ), this will the be the number of unique parameters that appear in a statement. If explicit numeric indexes are used, there may be gaps in the number sequence. Bound Parameters | 137 int sqlite3_bind_parameter_index( sqlite3_stmt *stmt, const char *name ) Returns the index of a named parameter. The name must include any leading character (such as “:”) and must be given in UTF-8, even if the statement was prepared from UTF-16. A zero is returned if a parameter with a matching name cannot be found. const char* sqlite3_bind_parameter_name( sqlite3_stmt *stmt, int pidx ) Returns the full text representation of a specific parameter. The text is always UTF-8 encoded and includes the leading character. Using the sqlite3_bind_parameter_index() function, you can easily find and bind named parameters. The sqlite3_bind_xxx() functions will properly detect an invalid index range, allowing you to look up the index and bind a value in one line: sqlite3_bind_int(stmt, sqlite3_bind_parameter_index(stmt, ":pid"), pid); If you want to clear all of the bindings back to their initial NULL defaults, you can use the function sqlite3_clear_bindings(): int sqlite3_clear_bindings( sqlite3_stmt *stmt ) Clears all parameter bindings in a statement. After calling, all parameters will have a NULL bound to them. This will cause the memory management callback to be called on any text or BLOB values that were bound with a valid function pointer. Currently, this function always returns SQLITE_OK. If you want to be absolutely sure bound values won’t leak from one statement execution to the next, it is best to clear the bindings any time you reset the statement. If you’re doing manual memory management on data buffers, you can free any memory used by bound values after this function is called. Security and Performance There are significant security advantages to using bound parameters. Many times peo- ple will manipulate SQL strings to substitute the values they want to use. For example, consider building an SQL statement in C using the string function snprintf(): snprintf(buf, buf_size, "INSERT INTO people( id, name ) VALUES ( %d, '%s' );", id_val, name_val); In this case we do need single quotes around the string value, as we’re trying to form a literal representation. If we pass in these C values: id_val = 23; name_val = "Fred"; Then we get the following SQL statement in our buffer: INSERT INTO people( id, name ) VALUES ( 23, 'Fred'); 138 | Chapter 7: C Programming Interface This seems simple enough, but the danger with a statement like this is that the variables need to be sanitized before they’re passed into the SQL statement. For example, con- sider these values: id_val = 23; name_val = "Fred' ); DROP TABLE people;"; This would cause our snprintf() to create the following SQL command sequence, with the individual commands split out onto their own lines for clarity: INSERT INTO people( id, name ) VALUES ( 23, 'Fred' ); DROP TABLE people; ' ); While that last statement is nonsense, the second statement is cause for concern. Thankfully, things are not quite as bad as they seem. The sqlite3_prepare_xxx() func- tions will only prepare a single statement (up to the first semicolon), unless you explicitly pass the remainder of the SQL command string to another sqlite3_pre pare_xxx() call. That limits what can be done in a case like this, unless your code automatically prepares and executes multiple statements from a single command buffer. Be warned, however, that the interfaces provided by many scripting languages will do exactly that, and will automatically process multiple SQL statements passed in with a single call. The SQLite convenience functions, including sqlite3_exec(), will also au- tomatically process multiple SQL commands passed in through a single string. What makes sqlite3_exec() particularly dangerous is that the convenience functions don’t allow the use of bound values, forcing you to programmatically build SQL command statements and opening you up to problems. Later in the chapter, we’ll take a closer look at sqlite3_exec() and why it usually isn’t the best choice. Even if SQLite will only process the first command, damage can still be done with subqueries and other commands. Bad input can also force a statement to fail. Consider the result if the name value is: Fred', 'extra junk If you’re updating a series of records based off this id value, you had better wrap all the commands up in a transaction and be prepared to roll it back if you encounter an error. If you just assume the commands will work, you’ll end up with an inconsistent database. This type of attack is known as an SQL injection attack. An SQL injection attack inserts SQL command fragments into data values, causing the database to execute arbitrary SQL commands. Unfortunately, it is extremely common for websites to be susceptible to this kind of attack. It also borders on inexcusable, because it is typically very easy to avoid. Bound Parameters | 139 One defense against SQL injections is to try to sanitize any string values received from an untrusted source. For example, you might try to substitute all single quote characters with two single quote characters (the standard SQL escape mechanism). This can get quite complex, however, and you’re putting utter faith in the code’s ability to correctly sanitize untrusted strings. A much easier way to defend yourself against SQL injections is to use SQL statement parameters. Injection attacks depend on a data value being represented as a literal value in an SQL command statement. The attack only works if the attack value is passed through the SQL parser, where it alters the meaning of the surrounding SQL commands. In the case of SQL parameters, the bound values are never passed through the SQL parser. An SQL statement is only parsed when the command is prepared. If you’re using parameters, the SQL engine parses only the parameter tokens. Later, when you bind a value to a specific parameter, that value is bound directly in its native format (i.e. string, integer, etc.) and is not passed through the SQL parser. As long as you’re careful about how you extract and display the string, it is perfectly safe to directly bind an untrusted string value to a parameter value without fear of an SQL injection. Besides avoiding injection attacks, parameters can also be faster and use less memory than string manipulations. Using a function such as snprintf() requires an SQL com- mand template, and a sufficiently large output buffer. The string manipulation func- tions also need working memory, plus you may need additional buffers to copy and sanitize values. Additionally, a number of datatypes, such as integers and floating-point numbers (and especially BLOBs), often take up significantly more space in their string representation. This further increases memory usage. Finally, once the final command buffer has been created, all the data needs to be passed through the SQL parser, where the literal data values are converted back into their native format and stored in addi- tional buffers. Compare that to the resource usage of preparing and binding a statement. When using parameters, the SQL command statement is essentially static, and can be used as is, without modification or additional buffers. The parser doesn’t need to deal with con- verting and storing literal values. In fact, the data values normally never leave their native format, further saving time and memory by avoiding conversion in and out of a string representation. Using parameters is the safe and wise choice, even for situations when a statement is only used once. It may take a few extra lines of code, but the process will be safer, faster, and more memory efficient. Example This example executes an INSERT statement. Although this statement is only executed once, it still uses bind parameters to protect against possible injection attacks. This eliminates the need to sanitize the input value. 140 | Chapter 7: C Programming Interface The statement is first prepared with a statement parameter. The data value is then bound to the statement parameter before we execute the prepared statement: char *data = ""; /* default to empty string */ sqlite3_stmt *stmt = NULL; int idx = -1; /* ... set "data" pointer ... */ /* ... open database ... */ rc = sqlite3_prepare_v2( db, "INSERT INTO tbl VALUES ( :str )", -1, &stmt, NULL ); if ( rc != SQLITE_OK) exit( -1 ); idx = sqlite3_bind_parameter_index( stmt, ":str" ); sqlite3_bind_text( stmt, idx, data, -1, SQLITE_STATIC ); rc = sqlite3_step( stmt ); if (( rc != SQLITE_DONE )&&( rc != SQLITE_ROW )) exit ( -1 ); sqlite3_finalize( stmt ); /* ... close database ... */ In this case we look for either an SQLITE_DONE or an SQLITE_ROW return value. Both are possible. Although the INSERT itself will be fully executed on the first call to sqlite3_step(), if PRAGMA count_changes is enabled, then the statement may return a value. In this case, we want to ignore any potential return value without triggering an error, so we must check for both possible return codes. For more details, see count_changes in Appendix F. Potential Pitfalls It is important to understand that all parameters must have some literal associated with them. As soon as you prepare a statement, all the parameters are set to NULL. If you fail to bind an alternate value, the parameter still has a literal NULL associated with it. This has a few ramifications that are not always obvious. The general rule of thumb is that bound parameters act as literal string substitutions. Although they offer additional features and protections, if you’re trying to figure out the expected behavior of a parameter substitution, it is safe to assume you’ll get the exact same behavior as if the parameter was a literal string substitution. In the case of an INSERT statement, there is no way to force a default value to be used. For example, if you have the following statement: INSERT INTO membership ( pid, gid, type ) VALUES ( :pid, :gid, :type ); Bound Parameters | 141 Even if the type column has a default value available, there is no way this statement can use it. If you fail to bind a value to the :type parameter, a NULL will be inserted, rather than the default value. The only way to insert a default value into the type column is to use a statement that doesn’t reference it, such as: INSERT INTO membership ( pid, gid ) VALUES ( :pid, :gid ); This means that if you’re heavily dependent on database-defined default values, you may need to prepare several variations of an INSERT statement to cover the different cases when different data values are available. Of course, if your application code is aware of the proper default values, it can simply bind that value to the proper parameter. The other area where parameters can cause surprises is in NULL comparisons. For example, consider the statement: SELECT * FROM employee WHERE manager = :manager; This works for normal values, but if a NULL is bound to the :manager parameter, no rows will ever be returned. If you need the ability to test for a NULL in the manager column, make sure you use the IS operator: SELECT * FROM employee WHERE manager IS :manager; For more details, see IS in Appendix D. This behavior also makes it tricky to “stack” conditionals. For example, if you have the statement: SELECT * FROM employee WHERE manager = :manager AND project = :project; you must provide a meaningful value for both :manager and :project. If you want the ability to search on a manager, on a project, or on a manager and a project, you need to prepare multiple statements, or you need to add a bit more logic: ...WHERE ( manager = :manager OR :manager IS NULL ) AND ( project = :project OR :project IS NULL ); This query will ignore one (or both) of the value conditions if you assign a NULL to the appropriate parameters. This expression won’t let you explicitly search for NULLs, but that can be done with additional parameters and logic. Preparing more flexible statements reduces the number of unique statements you need to manage, but it also tends to make them more complex and can make them run slower. If they get too complex, it might make more sense to simply define a new set of statements, rather than adding more and more parameter logic to the same statement. Convenience Functions SQLite includes a number of convenience functions that can be used to prepare, step, and finalize an SQL statement in one call. Most of these functions exist for historical reasons and, as the name says, convenience. 142 | Chapter 7: C Programming Interface While they’re not fully deprecated, there are a number of reasons why their use is not exactly encouraged. First off, understand that there is nothing special under the hood. Both of these functions eventually call the same sqlite3_prepare_xxx(), sqlite3_ step(), and sqlite3_finalize() calls that are available in the public API. These func- tions are not faster, nor are they more efficient. Second, since the API doesn’t support the use of bound parameters, you’re forced to use string manipulations to build your SQL commands. That means these functions are slower to process and much more vulnerable to SQL injection attacks. This is par- ticularly dangerous because all the convenience functions are designed to automatically process multiple SQL statements from a single command string. If input strings are not properly sanitized, this situation effectively gives anyone providing input data full access to the database engine, including the ability to delete data or drop whole tables. These functions also tend to be a bit slower. All results are returned in a string repre- sentation, without any kind of type information. This can make it difficult to determine the type of a return value, and can lead to a lot of extra type conversions. For all their disadvantages, there is still the simple fact that these functions are very convenient. If you’re just trying to throw together a quick and dirty snippet of code, these functions provide an easy means of doing that. They’re also perfectly acceptable for DDL commands, such as CREATE TABLE. For any type of DML command, especially those that involve values from unsanitized sources, I strongly recommend using the normal prepare, step, and finalize routines. You’ll end up with safer code and better performance. The first function allows for fairly generic execution of any SQL command string. int sqlite3_exec( sqlite3 *db, const char *sql, callback_ptr, void *userData, char **errMsg ) Prepares and executes one or more SQL statements, calling the optional callback for each result set row for each statement. The first parameter is a valid database connection. The second parameter is a UTF-8 encoded string that consists of one or more SQL statements. The third parameter is a pointer to a callback function. The prototype of this function is given below. This function pointer can be NULL. The fourth parameter is a user-data pointer that will be passed to the callback. The value can be whatever you want, including NULL. The fifth parameter is a reference to a character pointer. If an error is generated and this parameter is non-NULL, sqlite3_exec() will allocate a string buffer and return it. If the passed-back pointer is non-NULL, you are responsible for releasing the buffer with sqlite3_free() once you are done with it. If the SQL string consists of multiple SQL statements separated by semicolons, each statement will be executed in turn. Convenience Functions | 143 If the call is successful and all statements are processed without errors, SQLITE_OK will be returned. Otherwise, just about any of the other return codes are possible, since this one function runs through the whole statement preparation and execu- tion process. The sqlite3_exec() function is reasonably all encompassing, and can be used to exe- cute any SQL statement. If you’re executing a table query and want to access the result set, you will need to supply a function pointer that references a user-defined callback. This callback will be called once for each row returned. If you’re executing an SQL statement that does not normally return any database value, there is no need to provide a callback function. The success or failure of the SQL command will be indicated in the return value. The sqlite3_exec() function makes any database results available through a user- defined callback function. As each result row is computed, the callback is called to make the row data available to your code. Essentially, each internal call to sqlite3_step() that results in a return value of SQLITE_ROW results in a callback. The format of the callback looks like this: int user_defined_exec_callback( void *userData, int numCol, char **colData, char **colName ) This function is not part of the SQLite API. Rather, this shows the required format for a user-defined sqlite3_exec() callback. The first parameter is the user-data pointer passed in as the fourth parameter to sqlite3_exec(). The second parameter indicates how many columns exist in this row. The third and fourth parameters both return an array of strings (char pointers). The third parameter holds the data values for this row, while the forth parameter holds the column names. All values are returned as strings. There is no type information. Normally, the callback should return a zero value. If a nonzero value is returned, execution is stopped and sqlite3_exec() will return SQLITE_ABORT. The second, third, and fourth parameters act very similar to the traditional C variables argc and argv (and an extra argv) in main( int argc, char **argv ), the traditional start to every C program. The column value and name arrays will always be the same size for any given callback, but the specific size of the arrays and the column names can change over the course of processing a multi-statement SQL string. There is no need to release any of these values. Once your callback function returns, sqlite3_exec() will handle all the memory management. If you’d prefer not to mess with a callback, you can use sqlite3_get_table() to extract a whole table at once. Be warned, however, that this can consume a large amount of memory, and must be used carefully. While you can technically call sqlite3_get_table() with any SQL command string, it is specifically designed to work with SELECT statements. 144 | Chapter 7: C Programming Interface int sqlite3_get_table( sqlite3 *db, const char *sql, char ***result, int *numRow, int *numCol, char **errMsg ); Prepares and executes an SQL command string, consisting of one or more SQL statements. The full contents of the result set(s) is returned in an array of UTF-8 strings. The first parameter is a database connection. The second parameter is a UTF-8 encoded SQL command string that consists of one or more SQL statements. The third parameter is a reference to a one-dimensional array of strings (char pointers). The results of the query are passed back through this reference. The fourth and fifth parameters are integer references that pass back the number of rows and the number of columns, respectively, in the result array. The sixth and final parameter is a reference to a character string, and is used to return any error message. The result array consists of (numCol * (numRow + 1)) entries. Entries zero through numCol - 1 hold the column names. Each additional set of numCol entries holds one row worth of data. If the call is successful and all statements are processed without errors, SQLITE_OK will be returned. Otherwise, just about any of the other return codes are possible, since this one function runs through the whole statement preparation and execu- tion process. void sqlite3_free_table( char **result ) Correctly frees the memory allocated by a successful call to sqlite3_get_table(). Do not attempt to free this memory yourself. As indicated, you must release the result of a call to sqlite3_get_table() with a call to sqlite3_free_table(). This will properly release the individual allocations used to build the result value. As with sqlite3_exec(), you must call sqlite3_free() on any errMsg value that is returned. The result array is a one-dimensional array of character pointers. You must compute your own offsets into the array using the formula: /* offset to access column C of row R of **result */ int offset = ((R + 1) * numCol) + C; char *value = result[offset]; The “+ 1” used to compute the row offset is required to skip over the column names, which are stored in the first row of the result. This assumes that the first row and column would be accessed with an index of zero. As a convenience function, there is nothing special about sqlite3_get_table(). In fact, it is just a wrapper around sqlite3_exec(). It offers no additional performance benefits over the prepare, step, and finalize interfaces. In fact, between all the type conversions inherent in sqlite3_exec(), and all the memory allocations, sqlite3_get_table() has substantial overhead over other methods. Convenience Functions | 145 Since sqlite3_get_table() is a wrapper around sqlite3_exec(), it is possible to pass in an SQL command string that consists of multiple SQL statements. In the case of sqlite3_get_table(), this must be done with care, however. If more than one SELECT statement is passed in, there is no way to determine where one result set ends and the next begins. All the resulting rows are run together as one large result array. All of the statements must return the same number of columns, or the whole sqlite3_get_table() command will fail. Additionally, only the first statement will return any column names. To avoid these issues, it is best to call sqlite3_get_table() with single SQL commands. There are a number of reasons why these convenience functions may not be the best choice. Their use requires building an SQL command statement using string manipu- lation functions, and that process tends to be error prone. However, if you insist, your best bet is to use one of SQLite’s built-in string-building functions: sqlite3_mprintf(), sqlite3_vmprintf(), or sqlite3_snprintf(). See Appendix G for more details. Result Codes and Error Codes You may have noticed that I’ve been fairly quiet about the result codes that can be expected from a number of these API calls. Unfortunately, error handling in SQLite is a bit complex. At some point, it was recognized that the original error reporting mech- anism was a bit too generic and somewhat difficult to use. To address these concerns, a newer “extended” set of error codes was added, but this new system had to be layered on top of the existing system without breaking backward compatibility. As a result, we have both the older and newer error codes, as well as specific API calls that will alter the meaning of some of the codes. This all makes for a somewhat complex situation. Standard Codes Before we get into when things go wrong, let’s take a quick look at when things go right. Generally, any API call that simply needs to indicate, “that worked,” will return the constant SQLITE_OK. Not all non-SQLITE_OK return codes are errors, however. Recall that sqlite3_step() returns SQLITE_ROW or SQLITE_DONE to indicate specific return state. Table 7-2 provides a quick overview of the standard error codes. At this point in the development life cycle, it is unlikely that additional standard error codes will be added. Additional extended error codes may be added at any time, however. Table 7-2. SQLite standard return codes Return code constant Return code meaning SQLITE_OK Operation successful SQLITE_ERROR Generic error SQLITE_INTERNAL Internal SQLite library error 146 | Chapter 7: C Programming Interface Return code constant Return code meaning SQLITE_PERM Access permission denied SQLITE_ABORT User code or SQL requested an abort SQLITE_BUSY A database file is locked (usually recoverable) SQLITE_LOCKED A table is locked SQLITE_NOMEM Memory allocation failed SQLITE_READONLY Attempted to write to a read-only database SQLITE_INTERRUPT sqlite3_interrupt() was called SQLITE_IOERR Some type of I/O error SQLITE_CORRUPT Database file is malformed SQLITE_FULL Database is full SQLITE_CANTOPEN Unable to open requested database file SQLITE_EMPTY Database file is empty SQLITE_SCHEMA Database schema has changed SQLITE_TOOBIG TEXT or BLOB exceeds limit SQLITE_CONSTRAINT Abort due to constraint violation SQLITE_MISMATCH Datatype mismatch SQLITE_MISUSE API used incorrectly SQLITE_NOLFS Host OS cannot provide required functionality SQLITE_AUTH Authorization denied SQLITE_FORMAT Auxiliary database format error SQLITE_RANGE Bad bind parameter index SQLITE_NOTADB File is not a database Many of these errors are fairly specific. For example, SQLITE_RANGE will only be returned by one of the sqlite3_bind_xxx() functions. Other codes, like SQLITE_ERROR, provide almost no information about what went wrong. Of specific interest to the developer is SQLITE_MISUSE. This indicates an attempt to use a data structure or API call in an incorrect or otherwise invalid way. For example, trying to bind a new value to a prepared statement that is in the middle of an sqlite3_step() sequence would result in a misuse error. Occasionally, you’ll get an SQLITE_MISUSE that results from failing to properly deal with a previous error, but many times it is a good indication that there is some more basic conceptual misunderstanding about how the library is designed to work. Result Codes and Error Codes | 147 Extended Codes The extended codes were added later in the SQLite development cycle. They provide more specific details on the cause of an error. However, because they can change the value returned by a specific error condition, they are turned off by default. You need to explicitly enable them for the older API calls, indicating to the SQLite library that you’re aware of the extended error codes and willing to accept them. All of the standard error codes fit into the least-significant byte of the integer value that is returned by most API calls. The extended codes are all based off one of the standard error codes, but provide additional information in the higher-order bytes. In this way, the extended codes can provide more specific details about the cause of the error. Cur- rently, most of the extended error codes provide specific details for the SQLITE_IOERR result. You can find a full list of the extended error codes at http://sqlite.org/c3ref/c_ioerr _access.html. Error Functions The following APIs are used to enable the extended error codes and extract more in- formation about any current error conditions. int sqlite3_extended_result_codes( sqlite3 *db, int onoff ) Turns extended result and error codes on or off for this database connection. Database connections returned by any version of sqlite3_open_xxx() will have extended codes off by default. You can turn them on by passing a nonzero value in the second parameter. This function always returns SQLITE_OK—there is no way to extract the current result code state. int sqlite3_errcode( sqlite3 *db ) If a database operation returns a non-SQLITE_OK status, a subsequent call to this function will return the error code. By default, this will only return a standard error code, but if extended result codes have been enabled, it may also return one of the extended codes. int sqlite3_extended_errcode( sqlite3 *db ) Essentially the same as sqlite3_errcode(), except that extended results are always returned. const char* sqlite3_errmsg( sqlite3 *db ) const void* sqlite3_errmsg16( sqlite3 *db ) Returns a null-terminated, human-readable, English language error string that is encoded in UTF-8 or UTF-16. Any additional calls to the SQLite APIs using this database connection may result in these pointers becoming invalid, so you should either use the string before attempting any other operations, or you should make a private copy. It is also possible for these functions to return a NULL pointer, so check the result value before using it. Extended error codes are not used. 148 | Chapter 7: C Programming Interface It is acceptable to leave extended error codes off and intermix calls to sqlite3_ errcode() and sqlite3_extended_errcode(). Because the error state is stored in the database connection, it is easy to end up with race conditionals in a threaded application. If you’re sharing a database connection across threads, it is best to wrap your core API call and error-checking code in a critical section. You can grab the database connection’s mutex lock with sqlite3_db_ mutex(). See sqlite3_db_mutex() in Appendix G for more details. Similarly, the error handling system can’t deal with multiple errors. If there is an error that goes unchecked, the next call to a core API function is likely to return SQLITE_ MISUSE, indicating the attempt to use an invalid data structure. In this and similar sit- uations where multiple errors have been encountered, the state of the error message can become inconsistent. You need to check and handle any errors after each API call. Prepare v2 In addition to the standard and extended codes, the newer _v2 versions of sqlite3_pre pare_xxx() change the way prepared statement errors are processed. Although the newer and original versions of sqlite3_prepare_xxx() share the same parameters, the sqlite3_stmt returned by the _v2 versions is slightly different. The most noticeable difference is in how errors from sqlite3_step() are handled. For statements prepared with the original version of sqlite3_prepare_xxx(), the majority of errors within sqlite3_step() will return the rather generic SQLITE_ERROR. To find out the specifics of the situation, you had to call sqlite3_reset() or sqlite3_finalize() to extract a more detailed error code. This would, of course, cause the statement to be reset or finalized, which limited your recovery options. Things work a bit differently if the statement was prepared with the _v2 version. In that case, sqlite3_step() will return the specific error directly. The call sqlite3_step() may return a standard code or an extended code, depending if extended codes are enabled or not. This allows the developer to extract the error directly, and provides for more recovery options. The other major difference is how database schema changes are handled. If any Data Definition Language command (such as DROP TABLE) is issued, there is a chance the prepared statement is no longer valid. For example, the prepared statement may refer to a table or index that is no longer there. The only way to resolve any possible problems is to reprepare the statement. The _v2 versions of sqlite3_prepare_xxx() make a copy of the SQL statement used to prepare a statement. (This SQL can be extracted. See sqlite3_sql() in Appendix G for more details.) By keeping an internal copy of the SQL, a statement is able to reprepare itself if the database schema changes. This is done automatically any time SQLite de- tects the need to rebuild the statement. Result Codes and Error Codes | 149 The statements created with the original version of prepare didn’t save a copy of the SQL command, so they were unable to recover themselves. As a result, any time the schema changed, an API call involving any previously prepared statement would return SQLITE_SCHEMA. The program would then have to reprepare the statement using the original SQL and try again. If the schema change was significant enough that the SQL was no longer valid, sqlite3_prepare_xxx() would return an appropriate error when the program attempted to reprepare the SQL command. Statements created with the _v2 version of prepare can still return SQLITE_SCHEMA. If a schema change is detected and the statement is unable to automatically reprepare itself, it will still return SQLITE_SCHEMA. However, under the _v2 prepare, this is now considered a fatal error, as there is no way to recover the statement. Here is a side-by-side comparison of the major differences between the original and _v2 version of prepare: Statement prepared with original version Statement prepared with v2 version Created with sqlite3_prepare() or sqlite3_pre pare16(). Created with sqlite3_prepare_v2() or sqlite3_pre pare16_v2(). Most errors in sqlite3_step() return SQLITE_ERROR. sqlite3_step() returns specific errors directly. sqlite3_reset() or sqlite3_finalize() must be called to get full error. Standard or extended error codes may be returned. No need to call anything additional. sqlite3_step() may return a standard or extended error code. Schema changes will make any statement function return SQLITE_SCHEMA. Application must manually finalize and reprepare statement. Schema changes will make the statement reprepare itself. If application-provided SQL is no longer valid, the prepare will fail. If internal SQL is no longer valid, any statement function will return SQLITE_SCHEMA. This is a statement-fatal error, and the only choice is to finalize the statement. Original SQL is not associated with statement. Statement keeps a copy of SQL used to prepare. SQL can be recovered with sqlite3_sql(). Limited debugging. sqlite3_trace() can be used. Because the _v2 error handling is a lot simpler, and because of the ability to automat- ically recover from schema changes, it is strongly recommended that all new develop- ment use the _v2 versions of sqlite3_prepare_xxx(). Transactions and Errors Transactions and checkpoints add a unique twist to the error recovery process. Nor- mally, SQLite operates in autocommit mode. In this mode, SQLite automatically wraps each SQL command in its own transaction. In terms of the API, that’s the time from when sqlite3_step() is first called until SQLITE_DONE is returned by sqlite3_step() (or when sqlite3_reset() or sqlite3_finalize() is called). 150 | Chapter 7: C Programming Interface If each statement is wrapped up in its own transaction, error recovery is reasonably straightforward. Any time SQLite finds itself in an error state, it can simply roll back the current transaction, effectively canceling the current SQL command and putting the database back into the state it was in prior to the command starting. Once a BEGIN TRANSACTION command is executed, SQLite is no longer in autocommit mode. A transaction is opened and held open until the END TRANSACTION or COMMIT TRANSACTION command is given. This allows multiple commands to be wrapped up in the same transaction. While this is useful to group together a series of discrete com- mands into an atomic change, it also limits the options SQLite has for error recovery. When an error is encountered during an explicit transaction, SQLite attempts to save the work and undo just the current statement. Unfortunately, this is not always possi- ble. If things go seriously wrong, SQLite will sometimes have no choice but to roll back the current transaction. The errors most likely to result in a rollback are SQLITE_FULL (database or disk full), SQLITE_IOERR (disk I/O error or locked file), SQLITE_BUSY (database locked), SQLITE_NOMEM (out of memory), and SQLITE_INTERRUPT (interrupt requested by applica- tion). If you’re processing an explicit transaction and receive one of these errors, you need to deal with the possibility that the transaction was rolled back. To figure out which action was taken by SQLite, you can use the sqlite3_get_autocom mit() function. int sqlite3_get_autocommit( sqlite3 *db ) Returns the current commit state. A nonzero return value indicates the database is in autocommit mode, and not in an explicit transaction. A zero value indicates the database is currently inside an explicit transaction. If SQLite was forced to do a full rollback, the database will once again be in autocommit mode. If the database is not in autocommit mode, it must still be in a transaction, indicating that a rollback was not required. Although there are situations when it is possible to recover and continue a transaction, it is considered a best practice to always issue a ROLLBACK if one of these errors is encountered. In situations when SQLite was already forced to roll back the transaction and has returned to autocommit mode, the ROLLBACK will do nothing but return an error that can be safely ignored. Database Locking SQLite employs a number of different locks to protect the database from race condi- tions. These locks allow multiple database connections (possibly from different pro- cesses) to access the same database file simultaneously without fear of corruption. The locking system is used for both autocommit transactions (single statements) as well as explicit transactions. Result Codes and Error Codes | 151 The locking system involves several different tiers of locks that are used to reduce con- tention and avoid deadlocking. The details are somewhat complex, but the system allows multiple connections to read a database file in parallel, but any write operation requires full, exclusive access to the entire database file. If you want the full details, see http://www.sqlite.org/lockingv3.html. Most of the time the locking system works reasonably well, allowing applications to easily and safely share the database file. If coded properly, most write operations only last a fraction of a second. The library is able to get in, make the required modifications, verify them, and then get out, quickly releasing any locks and making the database available to other connections. However, if more than one connection is trying to access the same database at the same time, sooner or later they’ll bump into each other. Normally, if an operation requires a lock that the database connection is unable to acquire, SQLite will return the error SQLITE_BUSY or, in some more extreme cases, SQLITE_IOERR (extended code SQLITE_IOERR_BLOCKED). The functions sqlite3_prepare_xxx(), sqlite3_step(), sqlite3_reset(), and sqlite3_finalize() can all return SQLITE_BUSY. The functions sqlite3_backup_step() and sqlite3_blob_open() can also return SQLITE_BUSY, as these functions use sqlite3_prepare_xxx() and sqlite3_step() internally. Finally, sqlite3_close() may return SQLITE_BUSY if there are unfinalized statements associated with the database connection, but that’s not related to locking. Gaining access to a needed lock is often a simple matter of waiting until the current holder finishes up and releases the lock. In most cases, this is not a particularly long period of time. The waiting can either be done by the application, which can respond to an SQLITE_BUSY by simply trying to reprocess the statement and trying again, or it can be done with a busy handler. Busy handlers A busy handler is a callback function that is called by the SQLite library any time it is unable to acquire a lock, but has determined it is safe to try and wait for it. The busy handler can instruct SQLite to keep trying to acquire the lock, or to give up and return an SQLITE_BUSY error. SQLite includes an internal busy handler that uses a timer. If you set a timeout period, SQLite will keep trying to acquire the locks it requires until the timer expires. int sqlite3_busy_timeout( sqlite3 *db, int millisec ) Sets the given database connection to use the internal timer-based busy handler. If the second parameter is greater than zero, the handler is set to use a timeout value provided in milliseconds (thousandths of a second). If the second parameter is zero or negative, any busy handler will be cleared. If you want to write your own busy handler, you can set the callback function directly: 152 | Chapter 7: C Programming Interface int sqlite3_busy_handler( sqlite3 *db, callback_func_ptr, void *udp ) Sets a busy handler for the given database. The second parameter is a function pointer to the busy handler, and the third parameter is a user-data pointer that is passed to the callback. Setting a NULL function pointer will remove the busy handler. int user_defined_busy_handler_callback( void *udp, int incr ) This is not an SQLite library call, but the format of a user-defined busy handler. The first parameter is the user-data pointer passed to sqlite3_busy_handler() when the callback was set. The second parameter is a counter that is incremented each time the busy handler is called while waiting for a specific lock. A return value of zero will cause SQLite to give up and return an SQLITE_BUSY error, while a nonzero return value will cause SQLite to keep trying to acquire the lock. If the lock is successfully acquired, command processing will continue. If the lock is not acquired, the busy handler will be called again. Be aware that each database connection has only one busy handler. You cannot set an application busy handler and configure a busy timeout value at the same time. Any call to either of these functions will cancel out the other one. Deadlocks Setting a busy handler will not fix every problem. There are some situations when waiting for a lock will cause the database connection to deadlock. The deadlock hap- pens when a pair of database connections each have some set of locks and both need to acquire additional locks to finish their task. If each connection is attempting to access a lock currently held by the other connection, both connections will stall in a deadlock. This can happen if two database connections both attempt to write to the database at the same time. In this case, there is no point in both database connections waiting for the locks to be released, since the only way to proceed is if one of the connections gives up and releases all of its locks. If SQLite detects a potential deadlock situation, it will skip the busy handler and will have one of the database connections return SQLITE_BUSY immediately. This is done to encourage the applications to release their locks and break the deadlock. Breaking the deadlock is the responsibility of the application(s) involved—SQLite cannot handle this situation for you. Avoiding SQLITE_BUSY When developing code for a system that requires any degree of database concurrency, the easiest approach is to use sqlite3_busy_timeout() to set a timeout value that is reasonable for your application. Start with something between 250 to 2,000 millisec- onds and adjust from there. This will help reduce the number of SQLITE_BUSY response codes, but it will not eliminate them. Result Codes and Error Codes | 153 The only way to completely avoid SQLITE_BUSY is to ensure a database never has more than one database connection. This can be done by setting PRAGMA locking_mode to EXCLUSIVE. If this is unacceptable, an application can use transactions to make an SQLITE_BUSY return code easier to deal with. If an application can successfully start a transaction with BEGIN EXCLUSIVE TRANSACTION, this will eliminate the possibility of getting an SQLITE_BUSY. The BEGIN itself may return an SQLITE_BUSY, but in this case the application can simply reset the BEGIN statement with sqlite3_reset() and try again. The disad- vantage of BEGIN EXCLUSIVE is that it can only be started when no other connection is accessing the database, including any read-only transactions. Once an exclusive trans- action is started, it also locks out all other connections from accessing the database, including read-only transactions. To allow more concurrency, an application can use BEGIN IMMEDIATE TRANSACTION. If an IMMEDIATE transaction is successfully started, the application is very unlikely to re- ceive an SQLITE_BUSY until the COMMIT statement is executed. In all cases (including the COMMIT), if an SQLITE_BUSY is encountered, the application can reset the statement, wait, and try again. As with BEGIN EXCLUSIVE, the BEGIN IMMEDIATE statement itself can return SQLITE_BUSY, but the application can simply reset the BEGIN statement and try again. A BEGIN IMMEDIATE transaction can be started while other connections are reading from the database. Once started, no new writers will be allowed, but read-only connections can continue to access the database up until the point that the immediate transaction is forced to modify the database file. This is normally when the transaction is commit- ted. If all database connections use BEGIN IMMEDIATE for all transactions that modify the database, then a deadlock is not possible and all SQLITE_BUSY errors (for both the IMME DIATE writers and other readers) can be handled with a retry. Finally, if an application is able to successfully begin a transaction of any kind (including the default, DEFERRED), it should never get an SQLITE_BUSY (or risk a deadlock) unless it attempts to modify the database. The BEGIN itself may return an SQLITE_BUSY, but the application can reset the BEGIN statement and try again. Attempts to modify the database within a BEGIN DEFERRED transaction (or within an autocommit) are the only situations when the database may deadlock, and are the only situations when the response to an SQLITE_BUSY needs to go beyond simply waiting and trying again (or beyond letting the busy handler deal with it). If an application performs modifications within a deferred transaction, it needs to be prepared to deal with a possible deadlock situation. Avoiding deadlocks The rules to avoid deadlocks are fairly simple, although their application can cause significant complexity in code. 154 | Chapter 7: C Programming Interface First, the easy ones. The functions sqlite3_prepare_xxx(), sqlite3_backup_step(), and sqlite3_blob_open() cannot cause a deadlock. If an SQLITE_BUSY code is returned from one of these functions at any time, simply wait and call the function again. If the function sqlite3_step(), sqlite3_reset(), or sqlite3_finalize() returns SQLITE_BUSY from within a deferred transaction, the application must back off and try again. For statements that are not part of an explicit transaction, the prepared statement can simply be reset and re-executed. For statements that are inside an explicit deferred transaction, the whole transaction must be rolled back and started over from the beginning. In most cases, this will happen on the first attempt to modify the database. Just remember that the whole reason for the rollback is that some other database con- nection needs to modify the database. If an application has done several read operations to prepare a write operation, it would be best to reread that information in a new transaction to confirm the data is still valid. Whatever you do, don’t ignore SQLITE_BUSY errors. They can be rare, but they can also be a source of great frustration if handled improperly. When BUSY becomes BLOCKED When a database connection needs to modify the database, a lock is placed that makes the database read-only. This allows other connections to continue to read the database, but prevents them from making modifications. The actual changes are held in the database page cache and not yet written to the database file. Writing the changes out would make the changes visible to the other database connections, breaking the isola- tion rule of transactions. Since the changes have not yet been committed, it is perfectly safe to have them cached in memory. When all the necessary modifications have been made and the transaction is ready to commit, the writer further locks the database file so that new read-only transactions cannot get started. This allows the existing readers to finish up and release their own database locks. When all readers are finished, the writer should have exclusive access to the database and may finally flush the changes out of the page cache and into the database file. This process allows read-only transactions to continue running while the write trans- action is in progress. The readers need to be locked out only when the writer actually commits its transaction. However, a key assumption in this process is that the changes fit into the database page cache and do not need to be written until the transaction is committed. If the cache fills up with pages that contain pending changes, a writer has no option but to put an exclusive lock on the database and flush the cache prior to the commit stage. The transaction can still be rolled back at any point, but the writer must be given immediate access to the exclusive write lock in order to perform the cache flush. Result Codes and Error Codes | 155 If this lock is not immediately available, the writer is forced to abort the entire trans- action. The write transaction will be rolled back and the extended result code SQLITE_IOERR_BLOCKED (standard code SQLITE_IOERR) will be returned. Because the transaction is automatically rolled back, there aren’t many options for the application, other than to start the transaction over. To avoid this situation, it is best to start large transactions that modify many rows with an explicit BEGIN EXCLUSIVE. This call may fail with SQLITE_BUSY, but the application can simply retry the command until it succeeds. Once an exclusive transaction has started, the write transaction will have full access to the database, eliminating the chance of an SQLITE_IOERR_BLOCKED, even if the transaction spills out of the cache prior to the commit. Increasing the size of the database cache can also help. Utility Functions The SQLite library contains a number of utility functions that are useful for both ap- plication developers, and those working on SQLite extensions. Most of these are not required for basic database tasks, but if your code is strongly tied to SQLite, you may find these particularly useful. Version Management There are several functions available to query the version of the SQLite library. Each API call has a corresponding #define macro that declares the same value. SQLITE_VERSION const char* sqlite3_libversion( ) Returns the SQLite library version as a UTF-8 string. SQLITE_VERSION_NUMBER int sqlite3_libversion_number( ) Returns the SQLite library version as an integer. The format is MNNNPPP, where M is the major version (3, in this case), N is the minor number, and P is the point release. This format allows for releases up to 3.999.999. If a sub-point release is made, it will not be indicated in this version number. SQLITE_SOURCE_ID const char* sqlite3_sourceid( ) Returns the check-in stamp of the code used in this release. The string consists of a date, time stamp, and an SHA1 hash of the source from the source repository. If you’re building your own application, you can use the #define macros and the func- tion calls to verify that you’re using the correct header for the available library. The #define values come from the header file, and are set when your application is compiled. The function calls return the same values that were baked into the library when it was compiled. 156 | Chapter 7: C Programming Interface If you’re using a dynamic library of some sort, you can use these macros and functions to prevent your application from linking with a library version other than the one it was originally compiled against. This might not be a good thing, however, as it will also prevent upgrades. If you need to lock in a specific version, you should probably be using a static library. If you want to check the validity of a dynamic library, it might be better to do something like this: if ( SQLITE_VERSION_NUMBER > sqlite3_libversion_number( ) ) { /* library too old; report error and exit. */ } Remember that the macro will hold the version used when your code was compiled, while the function call will return the version of the SQLite library. In this case, we report an error if the library is older (smaller version) than the one used to compile and build the application code. Memory Management When SQLite needs to dynamically allocate memory, it normally calls the default memory handler of the underlying operating system. This causes SQLite to allocate its memory from the application heap. However, SQLite can also be configured to do its own internal memory management (see sqlite3_config() in Appendix G). This is especially important on embedded and hand-held devices where memory is limited and overallocation may lead to stability problems. Regardless, you can access whatever memory manager SQLite is using with these SQLite memory management functions: void* sqlite3_malloc( int numBytes ) Allocates and returns a buffer of the size specified. If the memory cannot be allo- cated, a NULL pointer is returned. Memory will always be 8-byte (64-bit) aligned. This is a replacement for the standard C library malloc() function. void* sqlite3_realloc( void *buffer, int numBytes ) Used to resize a memory allocation. Buffers can be made larger or smaller. Given a buffer previously returned by sqlite3_malloc() and a byte count, *_realloc() will allocate a new buffer of the specified size and copy as much of the old buffer as will fit into the new buffer. It will then free the old buffer and return the new buffer. If the new buffer cannot be allocated, a NULL is returned and the original buffer is not freed. If the buffer pointer is NULL, the call is equivalent to a call to sqlite3_malloc(). If the numBytes parameter is zero or negative, the call is equivalent to a call to sqlite3_free(). This is a replacement for the standard C library realloc() function. Utility Functions | 157 void sqlite3_free( void *buffer ) Releases a memory buffer previously allocated by sqlite3_malloc() or sqlite3_realloc(). Also used to free the results or buffers of a number of SQLite API functions that call sqlite3_malloc() internally. This is a replacement for the standard C library free() function. While a number of SQLite calls require the use of sqlite3_free(), application code is free to use whatever memory management is most appropriate. Where these functions become extremely useful is in writing custom functions, virtual tables, or any type of loadable module. Since this type of code is meant to operate in any SQLite environment, you will likely want to use these memory management functions to ensure the maxi- mum portability for your code. Summary With the information in this chapter, you should be able to write code that opens a database file and executes SQL commands against it. With that ability, and the right SQL commands, you should be able to create new databases, specify tables, insert data, and build complex queries. For a large number of applications, this is enough to service most of their database needs. The next chapters look at more advanced features of the SQLite C API. This includes the ability to define your own SQL functions. That will enable you to extend the SQL used by SQLite with simple functions, as well as aggregators (used with GROUP BY) and sorting collations. Additional chapters will look at how to implement virtual tables and other advanced features. 158 | Chapter 7: C Programming Interface CHAPTER 8 Additional Features and APIs This chapter touches on a number of different areas, mostly having to do with features and interfaces that go beyond the basic database engine. The first section covers the SQLite time and date features, which are provided as a small set of scalar functions. We’ll also briefly look at some of the standard extensions that ship with SQLite, such as the ICU internationalization extension, the FTS3 text search module, and the R*Tree module. We’ll also be looking at some of the alternate interfaces available for SQLite in different scripting languages and other environments. Finally, we’ll wrap things up with a quick discussion on some of the things to watch out for when doing development on mobile or embedded systems. Date and Time Features Most relational database products have several native datatypes that deal with record- ing dates, times, timestamps, and durations of all sorts. SQLite does not. Rather than having specific datatypes, SQLite provides a small set of time and date conversion functions. These functions can be used to convert time, date, or duration information to or from one of the more generic datatypes, such as a number or a text value. This approach fits well with the simple and flexible design goals of SQLite. Dates and times can get extremely complicated. Odd time zones and changing daylight saving time (DST) rules can complicate time values, while dates outside of the last few hundred years are subject to calendaring systems that have been changed and modified through- out history. Creating a native type would require picking a specific calendaring system and a specific set of conversion rules that may or may not be suitable for the task at hand. This is one of the reasons a typical database has so many different time and date datatypes. 159 Using external conversion functions is much more flexible. The developer can choose a format and datatype that best fits the needs of the application. Using the simpler underlying datatypes is also a much better fit for SQLite’s dynamic typing system. A more generic approach also keeps the internal code simpler and smaller, which is a plus for most SQLite environments. Application Requirements When creating date and time data values, there are a few basic questions that need to be answered. Many of these seem obvious enough, but skipping over them too quickly can lead to big problems down the road. First, you need to figure out what you’re trying to store. It might be a time of day, such as a standalone hour, minute, second value without any associated date. You may need a date record, that refers to a specific day, month, year, but has no associated time. Many applications require timestamps, which include both a date and time to mark a specific point in time. Some applications need to record a specific day of the year, but not for a specific year (for example, a holiday). Many applications also use durations (time deltas). Even if the application doesn’t store durations or offsets, they are often computed for display purposes, such as the amount of time between “now” and a specific event. It is also worth considering the range and precision required by your application. As already discussed, dates in the far past (or far future) are poorly represented by some calendaring systems. A database that holds reservations for a conference room may require minute precision, while a database that holds network packet dumps may require precision of a microsecond or better. Representations As with many other datatypes, the class of data an application needs to store, along with the required range and precision, often drives the decision on what representation to use. The two most common representations used by SQLite are some type of for- matted text-based value or a floating-point value. Julian Day The simplest and most compact representation is the Julian Day. This is a single floating-point value used to count the number of days since noon, Greenwich time, on 24 November 4714 BCE. SQLite uses the proleptic Gregorian calendar for this repre- sentation. The Julian value for midnight, 1 January 2010 is 2455197.5. When stored as a 64-bit floating-point value, modern age dates have a precision a tad better than one millisecond. 160 | Chapter 8: Additional Features and APIs Many developers have never encountered the Julian Day calendar, but conceptually it is not much different than the more familiar POSIX time() value—it just uses a different value (days, rather than seconds) and a different starting point. Julian Day values have a relatively compact storage format and are fairly easy to work with. Durations and differences are simple and efficient to calculate, and use the same data representation as points in time. Values are automatically normalized and can be utilize simple mathematic operations. Julian values are also able to express a very broad range of dates, making them useful for historic records. The main disadvantage is that they require conversion before being displayed. Text values The other popular representation is a formatted text value. These are typically used to hold a date value, a time value, or a combination of both. Although SQLite recognizes a number of formats, most commonly dates are given in the format YYYY-MM-DD, while times are formatted HH:MM:SS, using an hour value of 00 to 23. If a full timestamp is required, these values can be combined. For example, YYYY-MM-DD HH:MM:SS. Although this style of date may not be the most natural representation, these formats are based off the ISO 8601 international standard for representing dates and times. They also have the advantage of sorting chronologically using a simple string sort. The main advantage of using a text representation is that they are very easy to read. The stored values do not require any kind of translation and can be easily browsed and understood in their native format. You can also pick and choose what parts of the data value are required, storing only a date or only a time of day, making it a bit more clear about the intention of the value. Or, at least, that would be true if it wasn’t for the time zone issue. As we’ll see, times and dates are rarely stored relative to the local time zone, so even text values usually require conversion before being displayed. The major disadvantage of text values is that any operation (other than display) requires a significant amount of data conversion. Time and date conversions require some com- plex math, and can make a noticeable impact in the performance of some applications. For example, moving a date one week into the future requires a conversion of the original date into some generalized format, offsetting the value, and converting it back into an appropriate text value. Calculating durations also requires a significant amount of conversion. The conversion cost may not be significant for a simple update or insert, but it can make a very noticeable difference if found in a search conditional. Text values also require careful normalization of all input values into a standardized format. Many operations, such as sorts and simple comparisons, require that values use the exact same format. Alternate formats can result in equivalent time values being represented by nonequivalent text values. This can lead to inconsistent results from any procedures that operate directly on the text representation. The problem is not just limited to single columns. If a time or date value is used as a key or join column, these operations will only work properly if all of the time and date values use the same format. Date and Time Features | 161 For all these concerns, there is still no denying that text values are the easiest to display and debug. While there is significant value in this, make sure you consider the full range of pros and cons of text values (or any other format) before you make a choice. Time zones You may have noticed that none of these formats support a time zone field. SQLite assumes all time and date information is stored in UTC, or Coordinated Universal Time. UTC is essentially Greenwich Mean Time, although there are some minor technical differences. There are some significant advantages to using UTC time. First and foremost, UTC is unaffected by location. This may seem like a minor thing, but if your database is sitting on a mobile device, it is going to move. Occasionally, it is going to move across time zones. Any displayed time value better shift with the device. If your database is accessible over the Internet, chances are good it will be accessed from more than one time zone. In short, you can’t ignore the time zone issue, and sooner or later you’re going to have to translate between time zones. Having a universal base format makes this process much easier. Similarly, UTC is not affected by Daylight Saving Time . There are no shifts, jumps, or repeats of UTC values. DST rules are extremely complex, and can easily differ by lo- cation, time of year, or even the year itself, as switch-over times are shifted and moved. DST essentially adds a second, calendar-sensitive time zone to any location, com- pounding the problems of location and local time conversions. All of these issues can create a considerable number of headaches. In the end, there are very few justifiable reasons to use anything except UTC. As the name implies, it provides a universal time system that best represents unique moments in time without any context or translation. It might seem silly to convert values to UTC as you input them, and convert them back to local time to display them, but it has the advantage of working correctly, even if the local time zone changes or the DST state changes. Thankfully, SQLite makes all of these conversions simple. Time and Date Functions Nearly all of the time and date functionality within SQLite comes from five SQL func- tions. One of these functions is essentially a universal translator, designed to convert nearly any time or date format into any other format. The other four functions act as convenience wrappers that provide a fixed, predefined output format. In addition to the conversion functions, SQLite also provides a three literal expressions. When an expression is evaluated, these literals will be translated into an appropriate time value that represents “now.” 162 | Chapter 8: Additional Features and APIs Conversion Function The main utility to manipulate time and date values is the strftime() SQL function: strftime( format, time, modifier, modifier... ) The strftime() SQL function is modeled after the POSIX strftime() C function. It uses printf() style formatting markers to specify an output string. The first parameter is the format string, which defines the format of the returned text value. The second param- eter is a source time value that represents the base input time. This is followed by zero or more modifiers that can be used to shift or translate the input value before it is formatted. Typically all of these parameters are text expressions or text literals, al- though the time value may be numeric. In addition to any literal characters, the format string can contain any of the following markers: • %d — day of the month (DD), 01-31 • %f — seconds with fractional part (SS.sss), 00-59 plus decimal portion • %H — hour (HH), 00-23 • %j — day of the year (NNN), 001-366 • %J — Julian day number (DDDDDDD.ddddddd) • %m — month (MM), 01-12 • %M — minute (MM), 00-59 • %s — seconds since 1970-01-01 (POSIX time value) • %S — seconds (SS), 00-59 • %w — day of the week (N), 0-6, starting with Sunday as 0 • %W — week of the year (WW), 00-53 • %Y — year (YYYY) • %% — a literal % For example, the time format HH:MM:SS.sss (including fractional seconds) can be rep- resented by the format string '%H:%M:%f'. SQLite understands a number of input values. If the format of the time string is not recognized and cannot be decoded, strftime() will return NULL. All of the following input formats will be recognized: • YYYY-MM-DD • YYYY-MM-DD HH:MM • YYYY-MM-DD HH:MM:SS • YYYY-MM-DD HH:MM:SS.sss • YYYY-MM-DDTHH:MM • YYYY-MM-DDTHH:MM:SS Date and Time Features | 163 • YYYY-MM-DDTHH:MM:SS.sss • HH:MM • HH:MM:SS • HH:MM:SS.sss • now • DDDDDDD • DDDDDDD.ddddddd In the case of the second, third, and fourth formats, there is a single literal space char- acter between the date portion and the time portion. The fifth, six, and seventh formats have a literal T between the date and time portions. This format is specified by a number of ISO standards, including the standard format for XML timestamps. The last two formats are assumed to be a Julian Day or (with a modifier) a POSIX time value. These last two don’t require a specific number of digits, and can be passed in as numeric values. Internally, strftime() will always compute a full timestamp that contains both a date and time value. Any fields that are not specified by the input time string will assume default values. The default hour, minute, and second values are zero, or midnight, while the default date is 1 January 2000. In addition to doing translations between formats and representations, the strftime() function can also be used to manipulate and modify time values before they are formatted and returned. Zero or more modifiers can be provided: • [+-]NNN day[s] • [+-]NNN hour[s] • [+-]NNN minute[s] • [+-]NNN second[s] • [+-]NNN.nnn second[s] • [+-]NNN month[s] • [+-]NNN year[s] • start of month • start of year • start of day • weekday N • unixepoch • localtime • utc 164 | Chapter 8: Additional Features and APIs The first seven modifiers simply add or subtract the specified amount of time. This is done by translating the time and date into a segregated representation and then adding or subtracting the specified value. This can lead to invalid dates, however. For example, applying the modifier '+1 month' to the date '2010-01-31' would result in the date '2010-02-31', which does not exist. To avoid this problem, after each modifier is applied, the date and time values are normalized back to legitimate dates. For example, the hypothetical date '2010-02-31' would become '2010-03-03', since the unnormalized date was three days past the end of February. The fact that the normalization is done after each modifier is applied means the order of the modifiers can be very important. Careful consideration should be given to how modifiers are applied, or you may encounter some unexpected results. For example, applying the modifier '+1 month' followed by '-1 month' to the date '2010-01-31', will result in the date '2010-02-03', which is three days off from the original value. This is because the first modifier gets normalized to '2010-03-03', which is then moved back to '2010-02-03'. If the modifiers are applied in the opposite order, '-1 month' will convert our starting date to '2009-12-31', and the '+1 month' modifier will then convert the date back to the original starting date of '2010-01-31'. In this instance we end up back at the original date, but that might not always be the case. The three start of... modifiers shift the current date back in time to the specified point, while the weekday modifier will shift the date forward zero to six days, in order to find a date that falls on the specified day of the week. Acceptable weekday values are 0-6, with Sunday being 0. The unixepoch modifier can only be used as an initial modifier to a numeric time value. In that case, the value is assumed to represent a POSIX time, rather than a Julian Day, and is translated appropriately. Although the unixepoch modifier must appear as the first modifier, additional modifiers can still be applied. The last two modifiers are used to translate between UTC and local time representa- tions. The modifier name describes the translation destination, so localtime assumes a UTC input and produces a local output. Conversely, the utc modifier assumes a local time input and produces a UTC output. SQLite is dependent on the local operating system (and its time zone and DST configuration) for these translations. As a result, these modifiers are subject to any errors and bugs that may be present in the time and date libraries of the host operating system. Convenience functions In an effort to help standardized text formats, avoid errors, and provide a more con- venient way to covert dates and times into their most common representations, SQLite has a number convenience functions. Conceptually, these are wrapper functions around strftime() that output the date or time in a fixed format. All four of these functions take the same parameter set, which is essentially the same as the parameters used by strftime(), minus the initial format string. Date and Time Features | 165 date( timestring, modifier, modifier... ) Translates the time string, applies any modifiers, and outputs the date in the format YYYY-MM-DD. Equivalent to the format string '%Y-%m-%d'. time( timestring, modifier, modifier... ) Translates the time string, applies any modifiers, and outputs the date in the format HH:MM:SS. Equivalent to the format string '%H:%M:%S'. datetime( timestring, modifier, modifier... ) Translates the time string, applies any modifiers, and outputs the date in the format YYYY-MM-DD HH:MM:SS. Equivalent to the format string '%Y-%m-%d %H:%M:%S'. julianday( timestring, modifier, modifier... ) Translates the time string, applies any modifiers, and outputs the Julian Day. Equivalent to the format string '%J'. This function differs slightly from the strftime() function, as strftime() will return a Julian Day as a text representation of a floating-point number, while this function will return an actual floating-point number. All four of these functions recognize the same time string and modifier values that strftime() uses. Time literals SQLite recognizes three literal expressions. When an expression that contains one of these literals is evaluated, the literal will take on the appropriate text representation of the current date or time in UTC. CURRENT_TIME Provides the current time in UTC. The format will be HH:MM:SS with an hour value between 00 and 23, inclusive. This is the same as the SQL expression time( 'now' ). CURRENT_DATE Provides the current date in UTC. The format will be YYYY-MM-DD. This is the same as the SQL expression date( 'now' ). CURRENT_TIMESTAMP Provides the current date and time in UTC. The format will be YYYY-MM-DD HH:MM:SS. There is a single space character between the date and time segments. This is the same as the SQL expression datetime( 'now' ). Note that the name of the SQL function is datetime(), while the literal is _TIMESTAMP. Because these literals return the appropriate value in UTC, an expression such as SELECT CURRENT_TIMESTAMP; may not return the expected result. To get date and time in the local representation, you need to use an expression such as: SELECT datetime( CURRENT_TIMESTAMP, 'localtime' ); In this case, the literal CURRENT_TIMESTAMP could also be replaced with 'now'. 166 | Chapter 8: Additional Features and APIs Examples In some ways, the simplicity of the date and time functions can mask their power. The following examples demonstrate how to accomplish basic tasks. Here is an example of how to take a local timestamp and store it as a UTC Julian value: julianday( input_value, 'utc' ) This type of expression might appear in an INSERT statement. To insert the current time, this could be simplified to the 'now' value, which is always given in UTC: julianday( 'now' ) If you want to display a Julian value out of the database, you’ll want to convert the UTC Julian value to the local time zone and format it. This can be done like this: datetime( jul_date, 'localtime' ) It might also be appropriate to put an expression like this into a view. If you wanted to present the date in a format more comfortable for readers from the United States, you might do something like this: strftime( '%m/%d/%Y', '2010-01-31', 'localtime' ); This will display the date as 01/31/2010. The second parameter could also be a Julian value, or any other recognized date format. To get the current POSIX time value (which is always in UTC): strftime( '%s', 'now' ) Or to display the local date and time, given a POSIX time value: datetime( time_value, 'unixepoch', 'localtime' ) Don’t forget that the input value is usually a simple text value. The value can be built up by concatenating together individual values, if required. For example, the following will calculate a Julian value from individual year, month, and day values that are bound to the statement parameters. Just be sure to bind them as text values with the proper number of leading zeros: julianday( :year || '-' || :month || '-' || :day ) As you can see, once you understand how to combine different input formats with the correct modifiers, moving back and forth between time representations is fairly easy. This makes it much simpler to store date and time values using native representations that are otherwise unintelligible to most people. ICU Internationalization Extension SQLite provides full support for Unicode text values. Unicode provides a way to encode many different character representations, allowing a string of bytes to represent written characters, glyphs, and accents from a multitude of languages and writing systems. ICU Internationalization Extension | 167 What Unicode does not provide is any information or understanding of the sorting rules, capitalization rules, or equivalence rules and customs of a given language or location. This is a problem for pattern matching, sorting, or anything that depends on comparing text values. For example, most text-sorting systems will ignore case differences between words. Some languages will also ignore certain accent marks, but often those rules depend on the specific accent mark and character. Occasionally, the rules and con- ventions used within a language change from location to location. By default, the only character system SQLite understands is 7-bit ASCII. Any character encoding of 128 or above will be treated as a binary value with no awareness of capitalization or equiva- lence conventions. While this is often sufficient for English, it is usually insufficient for other languages. For more complete internationalization support, you’ll need to build SQLite with the ICU extension enabled. The International Components for Unicode project is an open- source library that implements a vast number of language-related functions. These functions are customized for different locales. The SQLite ICU extension allows SQLite to utilize different aspects of the ICU library, allowing locale-aware sorts and compar- isons, as well as locale-aware versions of upper() and lower(). To use the ICU extension, you must first download and build the ICU library. The library source code, along with build instructions, can be downloaded from the project website at http://www.icu-project.org/. You must then build SQLite with the ICU ex- tension enabled, and link it against the ICU library. To enable the ICU extension in an amalgamation build, define the SQLITE_ENABLE_ICU compiler directive. You’ll want to take a look at the original README document. It explains how to utilize the extension to create locale-specific collations and operators. You can find a copy of the README file in the full source distribution (in the ext/icu directory) or online at http://www.sqlite.org/src/artifact?ci=trunk&filename=ext/icu/README.txt. The main disadvantage of the ICU library is size. In addition to the library itself, the locale information for all of the languages and locations adds up to a considerable bulk. This extra data may not be significant for a desktop system, but it may prove impractical on a handheld or embedded device. Although the ICU extension can provide location-aware sorting and comparison ca- pabilities, you still need to pick a specific locale to define those sorting and comparison rules. This is simple enough if you’re only working with one language in one location, but it can be quite complex when languages are mixed. If you must deal with cross- locale sorts or other complex internationalization issues, it may be easier to pull that logic up into your application’s code. 168 | Chapter 8: Additional Features and APIs Full-Text Search Module SQLite includes a Full-Text Search (FTS) engine. The current version is known as FTS3. The FTS3 engine is designed to catalog and index large bodies of text. Once indexed, the FTS3 engine can quickly search for documents based off various types of keyword searches. Although the FTS3 source is now maintained by the SQLite team, parts of the engine were originally contributed by members of Google’s engineering team. The FTS3 engine is a virtual table module. Virtual tables are similar to views, in that they wrap a data source to make it look and act like a normal table. Views get their data from a SELECT statement, while virtual tables depend on user-defined C functions. All of the functions required to implement a virtual table are wrapped up in an extension known as a module. For more information on how SQLite modules and virtual tables work, see Chapter 10. Full-Text Search is an important and evolving technology, and is one of the areas that is targeted for improvements and enhancements as this book goes to press. Although this section gives a brief overview of the core FTS3 features, if you find yourself con- sidering the FTS3 module, I would encourage you to review the full documentation on the SQLite website. The FTS3 engine is included in all standard distributions of the SQLite source (includ- ing the amalgamation), but is turned off by default. To enable basic FTS functionality, define the SQLITE_ENABLE_FTS3 compiler directive when building the SQLite library. To enable the more advanced matching syntax, also define SQLITE_ENABLE_FTS3 _PARENTHESIS. Creating and Populating FTS Tables Once SQLite has been compiled with the FTS3 engine enabled, you can create a docu- ment table with an SQL statement similar to this: CREATE VIRTUAL TABLE table_name USING FTS3 ( col1,... ); In addition to providing a table name, you can define zero or more column names. The name of the column is the only information that will actually be used. Any type infor- mation or column constraints will be ignored. If no column names are given, FTS will automatically create one column with the name content. FTS tables are often used to hold whole documents, in which case they only need one column. Other times, they are used to hold different categories of related information, and require multiple columns. For example, if you wanted to store email messages in an FTS table, it might make sense to create separate columns for the "SUBJECT:" line, "FROM:" line, "TO:" line, and message body. This would allow you to limit searches to a specific column (and the data it contains). The column specification for an FTS table is largely determined by how you want to search for the data. FTS also provides an optimized way to look for a search term across all of the indexed columns. Full-Text Search Module | 169 You can use the standard INSERT, UPDATE, and DELETE statements to manipulate data within an FTS table. Like traditional tables, FTS tables have a ROWID column that con- tains a unique integer for each entry in the table. This column can also be referred to using the alias DOCID. Unlike traditional tables, the ROWID of an FTS table is stable through a vacuum (VACUUM in Appendix C), so it can be reliably referenced through a foreign key. Additionally, FTS tables have an internal column with the same name as the table name. This column is used for special operations. You cannot insert or update data in this column. Any virtual table, including FTS tables, can be deleted with the standard DROP TABLE command. Searching FTS Tables FTS tables are designed so that any SELECT statement will work correctly. You can even search for specific text values or patterns directly with the == or LIKE operators. These will work, although they’ll be somewhat slow, since standard operators will require a full table scan. The real power of the FTS system comes from a custom MATCH operator. This operator is able to take advantage of the indexes built around the individual text values, allowing very fast searches over large bodies of text. Generally, searches are done using a query similar to: SELECT * FROM fts_table WHERE fts_column MATCH search_term; The search term used by the MATCH operator has very similar semantics to those used in a web search engine. Search terms are broken down and matched against words and terms found in the text values of the FTS table. Generally, the FTS MATCH operator is case-insensitive and will only match against whole words. For example, the search term 'data' will match 'research data', but not 'database'. The order of the search terms does not matter. The terms 'cat dog' and 'dog cat' will match the same set of rows. By default, MATCH will only match records that contain every word or term found in the search term. If the extended syntax is enabled, more complex logic statements can also be used to define more complex search patterns. Normally, the terms will only be matched against the specified column. However, every FTS table has a special hidden column that has the same name as the table itself. If this column is specified in the match expression, then all of the columns will be searched. This makes it easy to construct “find all” type searches. 170 | Chapter 8: Additional Features and APIs More Details The FTS module is fairly advanced, and offers a large number of search options and index optimizations. If you plan on using the FTS engine in your application, I strongly suggest you spend some time reading the online documentation (http://www.sqlite.org/ fts3.html). The official documentation is quite extensive and covers the more advanced search features in some detail, complete with examples. In addition to explaining the different search patterns, the online documentation also covers index optimization and maintenance. While this level of detail isn’t always re- quired, it can be beneficial for applications that heavily depend on FTS. The documents also explain how to provide a custom tokenizer to adapt the FTS engine to specific applications. For those that want to dig even deeper, later sections of the document explain some of the internal workings of the FTS index. Information is provided on the shadow tables used by the FTS engine, as well as the process used to generate the token index. This level of knowledge isn’t required to use the FTS system, but it is extremely useful if you are looking to modify the code, or if you’re just curious about what is going on under the hood. R*Trees and Spatial Indexing Module The R*Tree module is a standard extension to SQLite that provides an index structure that is optimized for multi-dimensional ranged data. The R*Tree name refers to the internal algorithm used to organize and query the stored data. For example, in a two- dimensional R*Tree, the rows might contain rectangles, in the form of a minimum and maximum longitude value, along with a minimum and maximum latitude. Queries can be made to quickly find all of the rows that contain or overlap a specific geological location or area. Adding more dimensions, such as altitude, allows more complex and specific searches. R*Trees are not limited to just spacial information, but can be used with any type of numeric range data that includes pairs of minimum and maximum values. For example, an R*Tree table might be used to index the start and stop times of events. The index could then quickly return all of the events that were active at a specific point or range of time. The R*Tree implementation included with SQLite can index up to five dimensions of data (five sets of min/max pairs). Tables consist of an integer primary key column, followed by one to five pairs of floating-point columns. This will result in a table with an odd number of 3 to 11 columns. Data values must always be given in pairs. If you wish to store a point, simply use the same value for both the minimum and maximum component. R*Trees and Spatial Indexing Module | 171 Generally, R*Trees act as detail tables for more traditional tables. A traditional table can store whatever data is required to define the object in question, including a key reference to the R*Tree data. The R*Tree table is used to hold just the dimensional data. Multi-dimensional R*Trees, especially those used to store bounding rectangles or bounding volumes, are often approximations of the records they are indexing. In these cases, R*Trees are not always able to provide an exact result set, but are used to effi- ciently provide a first approximation. Essentially, the R*Tree is used as an initial filter to quickly and efficiently screen out all but a small percentage of the total rows. A more specific (and often more expensive) filter expression can be applied to get the final result set. In most cases, the query optimizer understands how to best utilize the R*Tree, so that it is applied before any other conditional expressions. R*Trees are quite powerful, but they serve a very specific need. Because of this, we won’t be spending the time to cover all the details. If an R*Tree index sounds like something your application can take advantage of, I encourage you to check out the online doc- umentation (http://www.sqlite.org/rtree.html). This will provide a full description on how to create, populate, and utilize an R*Tree index. Scripting Languages and Other Interfaces Like most database products, SQLite has bindings that allow the functionality of the C APIs to be accessed from a number of different scripting languages and other envi- ronments. In many cases, these extensions try to follow a standardized database inter- face developed by the language community. In other cases, the driver or scripting extension attempts to faithfully represent the native SQLite API. With the exception of the Tcl extension, all of these packages were developed by the SQLite user community. As such, they are not part of the core SQLite project, nor are they supported by the SQLite development team. If you’re having issues installing or configuring one of these drivers, asking for support on the main SQLite user’s mailing list may produce limited results. You may have more luck on a project mailing list or, failing that, a more general language-specific support forum. As is common with this type of software, support and long-term maintenance can be somewhat hit-or-miss. At the time of publishing, most of the drivers listed here have a good history of keeping current and in sync with new releases of the SQLite library. However, before you build your whole project around a specific wrapper or extension, make sure the project is still active. Perl The preferred Perl module is DBD::SQLite, and is available on CPAN (http://www.cpan .org). This package provides a standardized, DBI-compliant interface to SQLite, as well as a number of custom functions that provide support for SQLite specific features. 172 | Chapter 8: Additional Features and APIs The DBI provides a standard interface for SQL command processing. The custom functions provide some additional coverage of the SQLite API, and provide the ability to define SQL functions, aggregates, and collations using Perl. While the custom func- tions do not provide full coverage of the SQLite API, most of the more common oper- ations are included. PHP As the PHP language has evolved, so have the SQLite access methods. PHP5 includes several different SQLite extensions that provide both vendor-specific interfaces, as well as drivers for the standardized PDO (PHP Data Objects) interface. There are two vendor-specific extensions. The sqlite extension has been included and enabled by default since PHP 5.0, and provides support for the SQLite v2 library. The sqlite3 extension has been included and enabled by default since PHP 5.3.0 and, as you might guess, provides an interface for the current SQLite 3 library. The sqlite3 library provides a pretty basic class interface to the SQL command APIs. It also supports the creation of SQL functions and aggregates using PHP. PHP 5.1 introduced the PDO interfaces. The PDO extension is the latest solution to the problem of providing unified database access mechanisms. PDO acts as a replace- ment for the PEAR-DB and MDB2 interfaces found in other versions of PHP. The PDO_SQLITE extension provides a PDO driver for the current SQLite v3 library. In addition to supporting the standard PDO access methods, this driver also provides custom methods to create SQL functions and aggregates using PHP. Given that there is very little functional difference between the SQLite 3 vendor-specific library and the PDO SQLite 3 library, I suggest that new development utilize the PDO driver. Python There are two popular Python interfaces available. Each wrapper addresses a different set of needs and requirements. At the time of this writing, both modules were under active development. The PySQLite module (http://code.google.com/p/pysqlite/) offers a standardized Python DB-API 2.0 compliant interface to the SQLite engine. PySQLite allows applications to develop against a relatively database-independent interface. This is very useful for sys- tems that need to support more than one database. Using a standardized interface also allows rapid prototyping with SQLite, while leaving a migration path to larger, more complex database systems. As of Python 2.5, PySQLite has become part of the Python Standard Library. The APSW module (Another Python SQLite Wrapper; http://code.google.com/p/apsw/) has a very different design goal. The APSW provides a very minimal abstraction layer Scripting Languages and Other Interfaces | 173 that is designed to mimic the native SQLite C API as much as possible. APSW makes no attempt to provide compatibility with any other database product, but provides very broad coverage of the SQLite library, including many of the low-level features. This allows very fine-grain control, including the ability to create user-defined SQL func- tions, aggregates, and collations in Python. APSW can even be used to write a virtual table implementation in Python. Both modules have their strong points. Which module is right for your application depends on your needs. If your database needs are fairly straightforward and you want a standardized interface that allows future migration, then PySQLite is a better fit. If you don’t care about other database engines, but need very detailed control over SQLite, then APSW is worth a look. Java There are a number of interfaces available to the Java language. Some of these are wrappers around the native C API, while others conform to the standardized Java Database Compatibility (JDBC) API. One of the older wrappers is Java SQLite (http://www.ch-werner.de/javasqlite/), which provides support for both SQLite 2 and SQLite 3. The core of this library uses Java Native Interface (JNI) to produce an interface based off the native C interface. The library also contains a JDBC interface. It is a good choice if you need direct access to the SQLite API. A more modern JDBC-only driver is the SQLiteJDBC package (http://www.xerial.org/ trac/Xerial/wiki/SQLiteJDBC). This is a rather nice distribution, as the JAR file contains both the Java classes, as well as native SQLite libraries for Windows, Mac OS X, and Intel-based Linux. This makes cross-platform distribution quite easy. The driver is also heavily utilized by Xerial, so it tends to be well maintained. Tcl SQLite has a strong history with the Tcl language. In fact, what we now know as SQLite started life as a Tcl extension. Much of the testing and development tools for SQLite are written in Tcl. In addition to the native C API, the Tcl extension is the only API supported by the core SQLite team. To enable the Tcl bindings, download the TEA (Tcl Extension Architecture) distribu- tion of the SQLite source from the SQLite website (http://www.sqlite.org/download .html). This version of the code is essentially the amalgamation distribution with the Tcl bindings appended to the end. This will build into a Tcl extension that can then be imported into any Tcl environment. Documentation on the Tcl interface can be found at http://www.sqlite.org/tclsqlite.html. 174 | Chapter 8: Additional Features and APIs ODBC The ODBC (Open Database Connectivity) specification provides a standardized data- base API for a wide variety of database products. Like many of the language-specific extensions, ODBC drivers act as a bridge between an ODBC library and a specific database API. Using ODBC allows developers to write to a single API, which can then use any number of connectors to speak to a wide variety of database products. Many generic database tools utilize ODBC to support a broad range of database systems. The best known connector for SQLite is SQLiteODBC (http://www.ch-werner.de/sqli teodbc/). SQLiteODBC is tested with a number of ODBC libraries, ensuring it should work with most tools and applications that utilize ODBC support. .NET There are a number of independent SQLite projects that use .NET technologies. Some of these are simple C# wrappers that do little more than provide an object context for the SQLite API. Other projects attempt to integrate SQLite into larger frameworks, such as ADO (ActiveX Data Objects). One of the more established open source projects is the System.Data.SQLite (http:// sqlite.phxsoftware.com/) package. This package provides broad ADO support, as well as LINQ support. There are also commercial ADO and LINQ drivers available. See the SQLite wiki for more information. C++ Although the SQLite C API can be accessed directly by C++ applications, some people prefer a more object-oriented interface. If you would prefer to use an existing library, there are a number of wrappers available. You can check the SQLite website or do some web searches if you’re interested. Be warned that few of these libraries are well maintained. You might be better off just writing and maintaining your own wrapper classes. The SQLite API has a rather object- influenced design, with most functions performing some type of manipulation or action on a specific SQLite data structure. As a result, most C++ wrappers are somewhat thin, providing little more than syntactical translation. Maintaining a private wrapper is normally not a significant burden. Just remember that the core SQLite library is C, not C++, and cannot be compiled with most C++ compilers. Even if you choose to wrap the SQLite API in a C++ class-based interface, you’ll still need to compile the core SQLite library with a C compiler. Scripting Languages and Other Interfaces | 175 Other Languages In addition to those languages listed there, there are wrappers, libraries, and extensions for a great number of other languages and environments. The wiki section of the SQLite website has an extensive list of third-party drivers at http://www.sqlite.org/cvstrac/wiki ?p=SqliteWrappers. Many of the listed drivers are no longer actively maintained, so be sure to research the project websites before investing in a particular driver. Those that are known to be abandoned are marked as such, but it is difficult to keep this kind of information up to date. Mobile and Embedded Development As the power and capacity of smartphones, mobile devices, and other embedded sys- tems continue to increase, these devices are able to deal with larger and more complex data. Many mobile devices are centered around organizing, searching, and displaying large quantities of structured data. This might be something as simple as an address book, or something much more complex, like mapping and route applications. In many cases, application requirements for data storage and management fit very well with the relational model provided by SQLite. SQLite is a fairly small and very resource- aware product, making it run well in restricted environments. The database-in-a-file model also makes it easy to copy or back up datastores easily and quickly. Given all these factors, it should come as no surprise that almost every major smartphone SDK supports SQLite out of the box, or allows it to be easily compiled for their platform. Memory Most mobile devices have limited memory resources. Applications must be conscious of their memory usage, and often need to limit the resources that may be consumed. In most cases, the majority of SQLite memory usage comes from the page cache. By picking a sensible page size and cache size, the majority of memory use can be con- trolled. Remember that each open or attached database normally has its own, inde- pendent cache. The page size can be adjusted at database creation with the PRAGMA page_size command, while the cache size can be adjusted at any time with PRAGMA cache_size. See page_size and cache_size in Appendix F for more details. Be aware that larger cache sizes are not always significantly better. Because some types of flash storage systems have no effective seek time, it is sometimes possible to utilize a relatively small page cache while still maintaining acceptable performance. The faster response time of the storage system reduces the cost of pulling pages into memory, lessening the impact of the cache. Just how fast the storage system can respond has a great deal to do with types of flash chips and how they are configured in the device, but depending on your system, you may find it acceptable to use a relatively small cache. This should help keep your memory footprint under control. 176 | Chapter 8: Additional Features and APIs If you’re working in an extremely constrained environment, you can preallocate buffers and make them available to SQLite through the sqlite3_config() interface. Different buffers can be assigned for the SQLite heap, scratch buffers, and page cache. If buffers are configured before the SQLite library is initialized, all memory allocations will come from these buffers. This allows a host application precise control over the memory resources SQLite may use. Most other issues are fairly self-evident. For example, the use of in-memory databases is generally discouraged on memory-bound devices, unless the database is very small and the performance gains are significant. Similarly, be aware of queries that can gen- erate large intermediate result sets, such as ORDER BY clauses. In some cases it may make more sense to pull some of the processing or business logic out of the database and into your application, where you have better control over resource utilization. Storage Nearly all mobile devices use some type of solid-state storage media. The storage may be on-board, or it may be an expansion card, such as an SD (Secure Digital) card, or even an external thumb drive. Although these storage systems provide the same basic functionality as their “real computer” counterparts, these storage devices often have noticeably different operating characteristics from traditional mass-store devices. If possible, try to match the SQLite page size to the native block size of the storage system. Matching the block sizes will allow the system to write database pages more quickly and more efficiently. You want the database page size to be the same size as one or more whole filesystem blocks. Pages that use partial blocks will be much slower. You don’t want to make the page too large, however, or you’ll be limiting your cache performance. Finding the right balance can take some experimentation. Normally, SQLite depends heavily on filesystem locking to provide proper concurrency support. Unfortunately, this functionality can be limited on mobile and embedded platforms. To avoid problems, it is best to forego multiple database connections to the same database file, even from the same application. If multiple connections are re- quired, make sure the operating system is providing proper locking, or use an alternate locking system. Also consider configuring database connections to acquire and hold any locks (use the PRAGMA locking_mode command; see locking_mode in Appendix F). While this makes access exclusive to one connection, it increases performance while still providing protection. It may be tempting to turn off SQLite’s synchronization and journaling mechanism, but you should consider any possible consequences of disabling these procedures. While there are often significant performance gains to be found in disabling synchro- nizations and journal files, there is also the significant danger of unrecoverable data corruption. Mobile and Embedded Development | 177 For starters, mobile devices run off batteries. As we all know, batteries have a tendency to go dead at very annoying times. Even if the operating system provides low-power warnings and automatic sleep modes, on many models it is all too easy to instantly dislodge the battery if the device is dropped or mishandled. Additionally, many devices utilize removable storage, which has a tendency to be removed and disappear at in- convenient times. In all cases, the main defense against a corrupt database is the storage synchronization and journaling procedure. Storage failures and database corruption can be particularly devastating in a mobile or embedded environment. Because mobile platforms tend to be more closed to the user, it is often difficult for the end-user to back up and recover data from individual appli- cations, even if they are disciplined enough to regularly back up their data. Finally, data entry is often slow and cumbersome on mobile devices, making the prospect of manual recovery a long and undesirable prospect. Mobile applications should be as forgiving and error tolerant as possible. In many cases, losing a customer’s data will result in losing a customer. Other Resources Beyond special memory handlers and storage considerations, most other concerns boil down to being aware of the limitations of your platform and keeping resource use under control. If you’re preparing a custom SQLite build for your application, you should take some time to run through all the available compiler directives and see if there are any defaults you want to alter or any features you might want to disable (see Appen- dix A). Disabling unused features can reduce the code and memory footprints even further. Disabling some features also provides minor performance increases. It is also a good idea to read through the available PRAGMA statements and see if there are any further configuration options to customize SQLite behavior for your specific environment. PRAGMA commands can also be used to dynamically adjust resource use. For example, it might be possible to temporarily boost the cache size for an I/O intensive operation if it is done at a time when you know more memory is available. The cache size could then be reduced, allowing the memory to be used elsewhere in the application. iPhone Support When the iPhone and iPod touch were first released, Apple heavily advocated the use of SQLite. The SQLite library was provided as a system framework and was well docu- mented, complete with code examples, in the SDK. With the release of version 3.0, Apple has made their Core Data system available on the iPhone OS. Core Data has been available on the Macintosh platform for a number of years, and provides a high-level data abstraction framework that offers integrated 178 | Chapter 8: Additional Features and APIs design tools and runtime support to address complex data management needs. Unlike SQLite, the Core Data model is not strictly relational in nature. Now that the higher level library is available on their mobile platform, Apple is en- couraging people to migrate to Core Data. Most of the SQLite documentation and code examples have been removed from the SDK, and the system-level framework is no longer available. However, because Core Data utilizes SQLite in its storage layer, there is still a standard SQLite system library available for use. It is also relatively easy to compile application-specific versions of the SQLite library. This is required if you want to take advantage of some of the more recent features, as the system version of SQLite is often several versions behind. Core Data has some significant advantages. Apple provides development tools that allow a developer to quickly lay out their data requirements and relationships. This can reduce development time and save on code. The Core Data package is also well inte- grated into current Mac OS X systems, allowing data to move back and forth between the platforms quite easily. For all the advantages that Core Data provides, there are still situations where it makes sense to use SQLite directly. The most obvious consideration is if your development needs extend beyond Apple platforms. Unlike Core Data, the SQLite library is available on nearly any platform, allowing data files to be moved and accessed almost anywhere on any platform. Core Data also uses a different storage and retrieval model than SQLite. If your application is particularly well suited to the Relational Model, there may be advantages to having direct SQL query access to the data storage layer. Using the SQLite library directly also eliminates a number of abstraction layers from the ap- plication design. While this may lead to more detailed code, it is also likely to result in better performance, especially with larger datasets. Like many engineering choices, there are benefits and concerns with both approaches. Assuming the platform limitations aren’t a concern, Core Data can provide a very rapid solution for moderately simple systems. On the other hand, SQLite allows better cross- platform compatibility (in both code and data), and allows direct manipulation of complex data models. If you’re not dependent on the latest version of SQLite, you may even be able to reduce the size of your application by using the existing SQLite system library. Which set of factors has higher value to you is likely to be dependent on your platform requirements, and the complexity of your data model. Other Environments A number of smartphone environments require application development to be done in Java or some similar language. These systems often provide no C compilers and limit the ability to deploy anything but byte-code. While most of these platforms provide custom wrappers to system SQLite libraries, these wrappers are often somewhat limited. Typically, the system libraries are several versions behind, and the Java wrap- pers are often limited to the essential core function calls. Mobile and Embedded Development | 179 While this may limit the ability to customize the SQLite build and use advanced features of the SQLite product, these libraries still provide full access to the SQL layer and all the functionality that comes with it, including constraints, triggers, and transactions. To get around some limitations (like the lack of user-defined functions), it may some- times be necessary to pull some of the business logic up into the application. This is best done by designing an access layer into the application that centralizes all of the database functions. Centralizing allows the application code to consistently enforce any database design constraints, even when the database is unable to fully do so. It is also a good idea to include some type of verification function that can scan a database, identifying (and hopefully correcting) any problems. Additional Extensions In addition to those interfaces and modules covered here, there are numerous other extensions and third-party packages available for SQLite. Some are simple but useful extensions, such as a complete set of mathematical functions. The best way to find these are through web searches, or by asking on the SQLite User’s mailing list. You can also start by having a look at http://sqlite.org/contrib/ for a list of some of the older code contributions. In addition to database extensions, there are also several SQLite-specific tools and database managers available. In addition to the command-line shell, several GUI interfaces exist. One of the more popular is SQLite Manager, a Firefox extension avail- able at http://code.google.com/p/sqlite-manager/. A small number of commercial extensions for SQLite also exist. As already discussed, some of the database drivers require a commercial license. Hwaci, Inc., (the company responsible for developing SQLite) offers two commercial extensions. The SQLite En- cryption Extension (SEE) encrypts database pages as they are written to disk, effectively encrypting any database. The Compressed and Encrypted Read-Only Database (CEROD) extension goes further by compressing database pages. The compression reduces the size of the database file, but also makes the database read-only. This extension can be useful for distributing licensed data archives or reference materials. For more informa- tion on these extensions, see http://www.sqlite.org/support.html. 180 | Chapter 8: Additional Features and APIs CHAPTER 9 SQL Functions and Extensions SQLite allows a developer to expand the SQL environment by creating custom SQL functions. Although these functions are used in SQL statements, the code to implement the function is written in C. SQLite supports three kinds of custom functions. Simple scalar functions are the first type. These take some set of parameters and return a single value. An example would be the built-in function abs(), which takes a single numeric parameter and returns the absolute value of that number. The second type of function is an aggregate function, or aggregator. These are SQL functions, such as sum() or avg(), that are used in conjunction with GROUP BY clauses to summarize or otherwise aggregate a series of values together into a final result. The last type of custom function is a collation. Collations are used to define a custom sort orders for an index or an ORDER BY clause. Conceptually, collation functions are quite simple: they take two text values and return a greater than, less than, or equal status. In practice, collations can become quite complex, especially when dealing with natural language strings. This chapter will also take a look at how to package up a set of custom features into an SQLite extension. Extensions are a standard way to package custom functions, aggregations, collations, virtual tables (see Chapter 10), or any other custom feature. Extensions are a handy and standardized way to bundle up sets of related functions or customizations into SQL function libraries. 181 Extensions can be statically linked into an application, or they can be built into loadable extensions. Loadable extensions act as “plug-ins” for the SQLite library. Loadable ex- tensions are a particularly useful way to load your custom functions into sqlite3, pro- viding the ability to test queries or debug problems in the same SQL environment that is found in your application. The source code to the examples found in this chapter can be found in the book down- load. Downloading the source will make it much easier to build the examples and try them out. See “Example Code Download” on page xvi for more information on where to find the source code. Scalar Functions The structure and purpose of SQL scalar functions are similar to C functions or tradi- tional mathematical functions. The caller provides a series of function parameters and the function computes and returns a value. Sometimes these functions are purely func- tional (in the mathematical sense), in that they compute the result based purely off the parameters with no outside influences. In other cases, the functions are more proce- dural in nature, and are called to invoke specific side effects. The body of a function can do pretty much anything you want, including calling out to other libraries. For example, you could write a function that allows SQLite to send email or query the status of a web server all through SQL functions. Your code can also interact with the database and run its own queries. Although scalar functions can take multiple parameters, they can only return a single value, such as an integer or a string. Functions cannot return rows (a series of values), nor can they return a result set, with rows and columns. Scalar functions can still be used to process sets of data, however. Consider this SQL statement: SELECT format( name ) FROM employees; In this query, the scalar function format() is applied to every row in the result set. This is done by calling the scalar function over and over for each row, as each row is com- puted. Even though the format() function is only referenced once in this SQL statement, when the query is executed, it can result in many different invocations of the function, allowing it to process each value from the name column. Registering Functions To create a custom SQL function, you must bind an SQL function name to a C function pointer. The C function acts as a callback. Any time the SQL engine needs to invoke the named SQL function, the registered C function pointer is called. This provides a way for an SQL statement to call a C function you have written. 182 | Chapter 9: SQL Functions and Extensions These functions allow you to create and bind an SQL function name to a C function pointer: int sqlite3_create_function( sqlite3 *db, const char *func_name, int num_param, int text_rep, void *udp, func_ptr, step_func, final_func ) int sqlite3_create_function16( sqlite3 *db, const void *func_name, int num_param, int text_rep, void *udp, func_ptr, step_func, final_func ) Creates a new SQL function within a database connection. The first parameter is the database connection. The second parameter is the name of the function as either a UTF-8 or UTF-16 encoded string. The third parameter is the number of expected parameters to the SQL function. If this value is negative, the number of expected parameters is variable or undefined. Fourth is the expected representation for text values passed into the function, and can be one of SQLITE_UTF8, SQLITE_UTF16, SQLITE_UTF16BE, SQLITE_UTF16LE, or SQLITE_ANY. This is followed by a user-data pointer. The last three parameters are all function pointers. We will look at the specific prototypes for these function pointers later. To register and create a scalar function, only the first function pointer is used. The other two function pointers are used to register aggregate functions and should be set to NULL when defining a scalar function. SQLite allows SQL function names to be overloaded based off both the number of parameters and the text representation. This allows multiple C functions to be associ- ated with the same SQL function name. You can use this overloading capability to register different C implementations of the same SQL function. This might be useful to efficiently handle different text encodings, or to provide different behaviors, de- pending on the number of parameters. You are not required to register multiple text encodings. When the SQLite library needs to make a function call, it will attempt to find a registered function with a matching text representation. If it cannot find an exact match, it will convert any text values and call one of the other available functions. The value SQLITE_ANY indicates that the func- tion is willing to accept text values in any possible encoding. You can update or redefine a function by simply reregistering it with a different function pointer. To delete a function, call sqlite3_create_function_xxx() with the same name, parameter count, and text representation, but pass in NULL for all of the function pointers. Unfortunately, there is no way to find out if a function name is registered or not, outside of keeping track yourself. That means there is no way to tell the difference between a create action and a redefine action. It is permissible to create a new function at any time. There are limits on when you can change or delete a function, however. If the database connection has any prepared statements that are currently being executed (sqlite3_step() has been called at least Scalar Functions | 183 once, but sqlite3_reset() has not), you cannot redefine or delete a custom function, you can only create a new one. Any attempt to redefine or delete a function will return SQLITE_BUSY. If there are no statements currently being executed, you may redefine or delete a custom function, but doing so invalidates all the currently prepared statements (just as any schema change does). If the statements were prepared with sqlite3_prepare_v2(), they will automatically reprepare themselves next time they’re used. If they were prepared with an original version of sqlite3_prepare(), any use of the statement will return an SQLITE_SCHEMA error. The actual C function you need to write looks like this: void custom_scalar_function( sqlite3_context *ctx, int num_values, sqlite3_value **values ) This is the prototype of the C function used to implement a custom scalar SQL function. The first parameter is an sqlite3_context structure, which can be used to access the user-data pointer as well as set the function result. The second parameter is the number of parameter values present in the third parameter. The third parameter is an array of sqlite3_value pointers. The second and third parameters (int num_values, sqlite3_value **values) work together in a very similar fashion to the traditional C main parameters (int argc, char **argv). In a threaded application, it may be possible for different threads to call into your function at the same time. As such, user-defined functions should be thread-safe. Most user-defined functions follow a pretty standard pattern. First, you’ll want to ex- amine the sqlite3_value parameters to verify their types and extract their values. You can also extract the user-data pointer passed into sqlite3_create_function_xxx(). Your code can then perform whatever calculation or procedure is required. Finally, you can set the return value of the function or return an error condition. Extracting Parameters SQL function parameters are passed into your C function as an array of sqlite3_value structures. Each of these structures holds one parameter value. To extract working C values from the sqlite3_value structures, you need to call one of the sqlite3_value_xxx() functions. These functions are extremely similar to the sqlite3_column_xxx() functions in use and design. The only major difference is that these functions take a single sqlite3_value pointer, rather than a prepared statement and a column index. 184 | Chapter 9: SQL Functions and Extensions Like their column counterparts, the value functions will attempt to automatically con- vert the value into whatever datatype is requested. The conversion process and rules are the same as those used by the sqlite3_column_xxx() functions. See Table 7-1 for more details. const void* sqlite3_value_blob( sqlite3_value *value ) Extracts and returns a pointer to a BLOB. double sqlite3_value_double( sqlite3_value *value ) Extracts and returns a double-precision floating point value. int sqlite3_value_int( sqlite3_value *value ) Extracts and returns a 32-bit signed integer value. The returned value will be clip- ped (without warning) if the parameter value contains an integer value that cannot be represented with only 32 bits. sqlite3_int64 sqlite3_value_int64( sqlite3_value *value ) Extracts and returns a 64-bit signed integer value. const unsigned char* sqlite3_value_text( sqlite3_value *value ) Extracts and returns a UTF-8 encoded text value. The value will always be null- terminated. Note that the returned char pointer is unsigned and will likely require a cast. The pointer may also be NULL if a type conversion was required. const void* sqlite3_value_text16( sqlite3_value *value ) const void* sqlite3_value_text16be( sqlite3_value *value ) const void* sqlite3_value_text16le( sqlite3_value *value ) Extracts and returns a UTF-16 encoded string. The first function returns a string in the native byte ordering of the machine. The other two functions will return a string that is always encoded in big-endian or little-endian. The value will always be null-terminated. The pointer may also be NULL if a type conversion was required. There are also a number of helper functions to query the native datatype of a value, as well as query the size of any returned buffers. int sqlite3_value_type( sqlite3_value *value ) Returns the native datatype of the value. The return value can be one of SQLITE_BLOB, SQLITE_INTEGER, SQLITE_FLOAT, SQLITE_TEXT, or SQLITE_NULL. This value can change or become invalid if a type conversion takes place. int sqlite3_value_numeric_type( sqlite3_value *value ) This function attempts to convert a value into a numeric type (either SQLITE_FLOAT or SQLITE_INTEGER). If the conversion can be done without loss of data, then the conversion is made and the datatype of the new value is returned. If a conversion cannot be done, the value will not be converted and the original datatype of the value will be returned. This can be any value that is returned by sqlite3_value_type(). Scalar Functions | 185 The main difference between this function and simply calling sqlite3_value_dou ble() or sqlite3_value_int(), is that the conversion will only take place if it is meaningful and will not result in lost data. For example, sqlite3_value_double() will convert a NULL into the value 0.0, while this function will not. Similarly, sqlite3_value_int() will convert the first part of the string '123xyz' into the integer 123, ignoring the trailing 'xyz'. This function will not, however, because no sense can be made of the trailing 'xyz' in a numeric context. int sqlite3_value_bytes( sqlite3_value *value ) Returns the number of bytes in a BLOB or in a UTF-8 encoded string. If returning the size of a text value, the size will include the null-terminator. int sqlite3_value_bytes16( sqlite3_value *value ) Returns the number of bytes in a UTF-16 encoded string, including the null- terminator. As with the sqlite3_column_xxx() functions, any returned pointers can become invalid if another sqlite3_value_xxx() call is made against the same sqlite3_value structure. Similarly, data conversions can take place on text datatypes when calling sqlite3_ value_bytes() or sqlite3_value_bytes16(). In general, you should follow the same rules and practices as you would with the sqlite3_column_xxx() functions. See “Result Columns” on page 127 for more details. In addition to the SQL function parameters, the sqlite3_context parameter also carries useful information. These functions can be used to extract either the database connec- tion or the user-data pointer that was used to create the function. void* sqlite3_user_data( sqlite3_context *ctx ) Extracts the user-data pointer that was passed into sqlite3_create_function_ xxx() when the function was registered. Be aware that this pointer is shared across all invocations of this function within this database connection. sqlite3* sqlite3_context_db_handle( sqlite3_context *ctx ) Returns the database connection that was used to register this function. The database connection returned by sqlite3_context_db_handle() can be used by the function to run queries or otherwise interact with the database. Returning Results and Errors Once a function has extracted and verified its parameters, it can set about its work. When a result has been computed, that result needs to be passed back to the SQLite engine. This is done by using one of the sqlite3_result_xxx() functions. These func- tions set a result value in the function’s sqlite3_context structure. Setting a result value is the only way your function can communicate back to the SQLite engine about the success or failure of the function call. The C function itself has a void return type, so any result or error has to be passed back through the context structure. Normally, one of the sqlite3_result_xxx() functions is called just prior to 186 | Chapter 9: SQL Functions and Extensions calling return within your C function, but it is permissible to set a new result multiple times throughout the function. Only the last result will be returned, however. The sqlite3_result_xxx() functions are extremely similar to the sqlite3_bind_xxx() functions in use and design. The main difference is that these functions take an sqlite3_context structure, rather than a prepared statement and parameter index. A function can only return one result, so any call to an sqlite3_result_xxx() function will override the value set by a previous call. void sqlite3_result_blob( sqlite3_context* ctx, const void *data, int data_len, mem_callback ) Encodes a data buffer as a BLOB result. void sqlite3_result_double( sqlite3_context *ctx, double data ) Encodes a 64-bit floating-point value as a result. void sqlite3_result_int( sqlite3_context *ctx, int data ) Encodes a 32-bit signed integer as a result. void sqlite3_result_int64( sqlite3_context *ctx, sqlite3_int64 data ) Encodes a 64-bit signed integer as a result. void sqlite3_result_null( sqlite3_context *ctx ) Encodes an SQL NULL as a result. void sqlite3_result_text( sqlite3_context *ctx, const char *data, int data_len, mem_callback ) Encodes a UTF-8 encoded string as a result. void sqlite3_result_text16( sqlite3_context *ctx, const void *data, int data_len, mem_callback ) void sqlite3_result_text16be( sqlite3_context *ctx, const void *data, int data_len, mem_callback ) void sqlite3_result_text16le( sqlite3_context *ctx, const void *data, int data_len, mem_callback ) Encodes a UTF-16 encoded string as a result. The first function is used for a string in the native byte format, while the last two functions are used for strings that are explicitly encoded as big-endian or little-endian, respectively. void sqlite3_result_zeroblob( sqlite3_context *ctx, int length ) Encodes a BLOB as a result. The BLOB will contain the number of bytes specified, and each byte will all be set to zero (0x00). void sqlite3_result_value( sqlite3_context *ctx, sqlite3_value *result_value ) Encodes an sqlite3_value as a result. A copy of the value is made, so there is no need to worry about keeping the sqlite3_value parameter stable between this call and when your function actually exits. This function accepts both protected and unprotected value objects. You can pass one of the sqlite3_value parameters to this function if you wish to return one of the SQL function input parameters. You can also pass a value obtained from a call to sqlite3_column_value(). Scalar Functions | 187 Setting a BLOB or text value requires the same type of memory management as the equivalent sqlite3_bind_xxx() functions. The last parameter of these functions is a callback pointer that will properly free and release the given data buffer. You can pass a reference to sqlite3_free() directly (assuming the data buffers were allocated with sqlite3_malloc()), or you can write your own memory manager (or wrapper). You can also pass in one of the SQLITE_TRANSIENT or SQLITE_STATIC flags. See “Binding Val- ues” on page 135 for specifics on how these flags can be used. In addition to encoding specific datatypes, you can also return an error status. This can be used to indicate a usage problem (such as an incorrect number of parameters) or an environment problem, such as running out of memory. Returning an error code will result in SQLite aborting the current SQL statement and returning the error back to the application via the return code of sqlite3_step() or one of the convenience functions. void sqlite3_result_error( sqlite3_context *ctx, const char *msg, int msg_size ) void sqlite3_result_error16( sqlite3_context *ctx, const void *msg, int msg_size ) Sets the error code to SQLITE_ERROR and sets the error message to the provided UTF-8 or UTF-16 encoded string. An internal copy of the string is made, so the application can free or modify the string as soon as this function returns. The last parameter indicates the size of the message in bytes. If the string is null-terminated and the last parameter is negative, the string size is automatically computed. void sqlite3_result_error_toobig( sqlite3_context *ctx ) Indicates the function could not process a text or BLOB value due to its size. void sqlite3_result_error_nomem( sqlite3_context *ctx ) Indicates the function could not complete because it was unable to allocate re- quired memory. This specialized function is designed to operate without allocating any additional memory. If you encounter a memory allocation error, simply call this function and have your function return. void sqlite3_result_error_code( sqlite3_context *ctx, int code ) Sets a specific SQLite error code. Does not set or modify the error message. It is possible to return both a custom error message and a specific error code. First, call sqlite3_result_error() (or sqlite3_result_error16()) to set the error message. This will also set the error code to SQLITE_ERROR. If you want a different error code, you can call sqlite3_result_error_code() to override the generic error code with something more specific, leaving the error message untouched. Just be aware that sqlite3_ result_error() will always set the error code to SQLITE_ERROR, so you must set the error message before you set a specific error code. 188 | Chapter 9: SQL Functions and Extensions Example Here is a simple example that exposes the SQLite C API function sqlite3_limit() to the SQL environment as the SQL function sql_limit(). This function is used to adjust various limits associated with the database connection, such as the maximum number of columns in a result set or the maximum size of a BLOB value. Here’s a quick introduction to the C function sqlite3_limit(), which can be used to adjust the soft limits of the SQLite environment: int sqlite3_limit( sqlite3 *db, int limit_type, int limit_value ) For the given database connection, this sets the limit referenced by the second parameter to the value provided in the third parameter. The old limit is returned. If the new value is negative, the limit value will remain unchanged. This can be used to probe an existing limit. The soft limit cannot be raised above the hard limit, which is set at compile time. For more specific details on sqlite3_limit(), see sqlite3_limit() in Appendix G. You don’t need a full understanding of how this API call works to understand these examples. Although the sqlite3_limit() function makes a good example, it might not be the kind of thing you’d want to expose to the SQL language in a real-world application. In practice, exposing this C API call to the SQL level brings up some security concerns. Anyone that can issue arbitrary SQL calls would have the capability of altering the SQLite soft limits. This could be used for some types of denial-of-service attacks by raising or lowering the limits to their extremes. sql_set_limit In order to call the sqlite3_limit() function, we need to determine the limit_type and value parameters. This will require an SQL function that takes two parameters. The first parameter will be the limit type, expressed as a text constant. The second parameter will be the new limit. The SQL function can be called like this to set a new expression- depth limit: SELECT sql_limit( 'EXPR_DEPTH', 400 ); The C function that implements the SQL function sql_limit() has four main parts. The first task is to verify that the first SQL function parameter (passed in as val ues[0]) is a text value. If it is, the function extracts the text to the str pointer: static void sql_set_limit( sqlite3_context *ctx, int num_values, sqlite3_value **values ) { sqlite3 *db = sqlite3_context_db_handle( ctx ); const char *str = NULL; int limit = -1, val = -1, result = -1; Scalar Functions | 189 /* verify the first param is a string and extract pointer */ if ( sqlite3_value_type( values[0] ) == SQLITE_TEXT ) { str = (const char*) sqlite3_value_text( values[0] ); } else { sqlite3_result_error( ctx, "sql_limit(): wrong parameter type", -1 ); return; } Next, the function verifies that the second SQL parameter (values[1]) is an integer value, and extracts it into the val variable: /* verify the second parameter is an integer and extract value */ if ( sqlite3_value_type( values[1] ) == SQLITE_INTEGER ) { val = sqlite3_value_int( values[1] ); } else { sqlite3_result_error( ctx, "sql_limit(): wrong parameter type", -1 ); return; } Although our SQL function uses a text value to indicate which limit we would like to modify, the C function sqlite3_limit() requires a predefined integer value. We need to decode the str text value into an integer limit value. I’ll show the code to decode_limit_str() in just a bit: /* translate string into integer limit */ limit = decode_limit_str( str ); if ( limit == -1 ) { sqlite3_result_error( ctx, "sql_limit(): unknown limit type", -1 ); return; } After verifying our two SQL function parameters, extracting their values, and translating the text limit indicator into a proper integer value, we finally call sqlite3_ limit(). The result is set as the result value of the SQL function and the function returns: /* call sqlite3_limit(), return result */ result = sqlite3_limit( db, limit, val ); sqlite3_result_int( ctx, result ); return; } The decode_limit_str() function is very simple, and simply looks for a predefined set of text values: int decode_limit_str( const char *str ) { if ( str == NULL ) return -1; if ( !strcmp( str, "LENGTH" ) ) return SQLITE_LIMIT_LENGTH; if ( !strcmp( str, "SQL_LENGTH" ) ) return SQLITE_LIMIT_SQL_LENGTH; if ( !strcmp( str, "COLUMN" ) ) return SQLITE_LIMIT_COLUMN; if ( !strcmp( str, "EXPR_DEPTH" ) ) return SQLITE_LIMIT_EXPR_DEPTH; if ( !strcmp( str, "COMPOUND_SELECT" ) ) return SQLITE_LIMIT_COMPOUND_SELECT; if ( !strcmp( str, "VDBE_OP" ) ) return SQLITE_LIMIT_VDBE_OP; if ( !strcmp( str, "FUNCTION_ARG" ) ) return SQLITE_LIMIT_FUNCTION_ARG; if ( !strcmp( str, "ATTACHED" ) ) return SQLITE_LIMIT_ATTACHED; if ( !strcmp( str, "LIKE_LENGTH" ) ) return SQLITE_LIMIT_LIKE_PATTERN_LENGTH; 190 | Chapter 9: SQL Functions and Extensions if ( !strcmp( str, "VARIABLE_NUMBER" ) ) return SQLITE_LIMIT_VARIABLE_NUMBER; if ( !strcmp( str, "TRIGGER_DEPTH" ) ) return SQLITE_LIMIT_TRIGGER_DEPTH; return -1; } With these two functions in place, we can create the sql_limit() SQL function by registering the sql_set_limit() C function pointer. sqlite3_create_function( db, "sql_limit", 2, SQLITE_UTF8, NULL, sql_set_limit, NULL, NULL ); The parameters for this function include the database connection (db), the name of the SQL function (sql_limit), the required number of parameters (2), the expected text encoding (UTF-8), the user-data pointer (NULL), and finally the C function pointer that implements this function (sql_set_limit). The last two parameters are only used when creating aggregate functions, and are set to NULL. Once the SQL function has been created, we can now manipulate the limits of our SQLite environment by issuing SQL commands. Here are some examples of what the sql_limit() SQL function might look like if we integrated it into the sqlite3 tool (we’ll see how to do this using a loadable extension later in the chapter). First, we can look up the current COLUMN limit by passing a new limit value of -1: sqlite> SELECT sql_limit( 'COLUMN', -1 ); 2000 We verify the function works correctly by setting the maximum column limit to two, and then generating a result with three columns. The previous limit value is returned when we set the new value: sqlite> SELECT sql_limit( 'COLUMN', 2 ); 2000 sqlite> SELECT 1, 2, 3; Error: too many columns in result set We see from the error that the soft limit is correctly set, meaning our function is working. One thing you might be wondering about is parameter value count. Although the sql_set_limit() function carefully checks the types of the parameters, it doesn’t ac- tually verify that num_values is equal to two. In this case, it doesn’t have to, since it was registered with sqlite3_create_function() with a required parameter count of two. SQLite will not even call our sql_set_limit() function unless we have exactly two parameters: sqlite> SELECT sql_limit( 'COLUMN', 2000, 'extra' ); Error: wrong number of arguments to function sql_limit() SQLite sees the wrong number of parameters and generates an error for us. This means that as long as a function is registered correctly, SQLite will do some of our value checking for us. Scalar Functions | 191 sql_get_limit While having a fixed parameter count simplifies the verification code, it might be useful to provide a single-parameter version that can be used to look up the current value. This can be done a few different ways. First, we can define a second C function called sql_get_limit(). This function would be the same as sql_set_limit(), but with the second block of code removed: /* remove this block of code from a copy of */ /* sql_set_limit() to produce sql_get_limit() */ if ( sqlite3_value_type( values[1] ) == SQLITE_INTEGER ) { val = sqlite3_value_int( values[1] ); } else { sqlite3_result_error( ctx, "sql_limit(): wrong parameter type", -1 ); return; } With this code removed, the function will never decode the second SQL function parameter. Since val is initialized to –1, this effectively makes every call a query call. We register each of these functions separately: sqlite3_create_function( db, "sql_limit", 1, SQLITE_UTF8, NULL, sql_get_limit, NULL, NULL ); sqlite3_create_function( db, "sql_limit", 2, SQLITE_UTF8, NULL, sql_set_limit, NULL, NULL ); This dual registration overloads the SQL function name sql_limit(). Overloading is allowed because the two calls to sqlite3_create_function() have a different number of required parameters. If the SQL function sql_limit() is called with one parameter, then the C function sql_get_limit() is called. If two parameters are provided to the SQL function, then the C function sql_set_limit() is called. sql_getset_limit Although the two C functions sql_get_limit() and sql_set_limit() provide the correct functionality, the majority of their code is the same. Rather than having two functions, it might be simpler to combine these two functions into one function that can deal with either one or two parameters, and is capable of both getting or setting a limit value. This combine sql_getset_limit() function can be created by taking the original sql_set_limit() function and modifying the second section. Rather than eliminating it, as we did to create sql_get_limit(), we’ll simply wrap it in an if statement, so the second section (which extracts the second SQL function parameter) is only run if we have two parameters: /* verify the second parameter is an integer and extract value */ if ( num_values == 2 ) { if ( sqlite3_value_type( values[1] ) == SQLITE_INTEGER ) { val = sqlite3_value_int( values[1] ); 192 | Chapter 9: SQL Functions and Extensions } else { sqlite3_result_error( ctx, "sql_limit(): wrong parameter type", -1 ); return; } } We register the same sql_getset_limit() C function under both parameter counts: sqlite3_create_function( db, "sql_limit", 1, SQLITE_UTF8, NULL, sql_getset_limit, NULL, NULL ); sqlite3_create_function( db, "sql_limit", 2, SQLITE_UTF8, NULL, sql_getset_limit, NULL, NULL ); For this specific task, this is likely the best choice. SQLite will verify the SQL function sql_limit() has exactly one or two parameters before calling our C function, which can easily deal with either one of those two cases. sql_getset_var_limit If for some reason you don’t like the idea of registering the same function twice, we could also have SQLite ignore the parameter count and call our function no matter what. This leaves verification of a valid parameter count up to us. To do that, we’d start with the sql_getset_limit() function and change it to sql_getset_var_limit(), by adding this block at the top of the function: if ( ( num_values < 1 )||( num_values > 2 ) ) { sqlite3_result_error( ctx, "sql_limit(): bad parameter count", -1 ); return; } We register just one version. By passing a required parameter count of -1, we’re telling the SQLite engine that we’re willing to accept any number of parameters: sqlite3_create_function( db, "sql_limit", -1, SQLITE_UTF8, NULL, sql_getset_var_limit, NULL, NULL ); Although this works, the sql_getset_limit() version is still my preferred version. The registration makes it clear which versions of the function are considered valid, and the function code is reasonably clear and compact. Completely free-form parameter counts are usually used by items like the built-in function coalesce(). The coalesce() function will take any number of parameters (greater than one) and return the first non-NULL value in the list. Since you might pass anywhere from two to a dozen or more parameters, it is impractical to register each possible configuration, and is better to just allow the function to do its own parameter management. On the other hand, something like sql_getset_limit() can really only accept two con- figurations: one parameter or two. In that case, I find it easier to explicitly register both parameter counts and allow SQLite to do my parameter verification for me. Scalar Functions | 193 Aggregate Functions Aggregate functions are used to collapse values from a grouping of rows into a single result value. This can be done with a whole table, as is common with the aggregate function count(*), or it can be done with groupings of rows from a GROUP BY clause, as is common with something like avg() or sum(). Aggregate functions are used to sum- marize, or aggregate, all of the individual row values into some single representative value. Defining Aggregates SQL aggregate functions are created using the same sqlite3_create_function_xxx() function that is used to create scalar functions (See “Scalar Functions” on page 182). When defining a scalar function, you pass in a C function pointer in the sixth parameter and set the seventh and eighth parameter to NULL. When defining an aggregate func- tion, the sixth parameter is set to NULL (the scalar function pointer) and the seventh and eighth parameters are used to pass in two C function pointers. The first C function is a “step” function. It is called once for each row in an aggregate group. It acts similarly to an scalar function, except that it does not return a result (it may return an error, however). The second C function is a “finalize” function. Once all the SQL rows have been stepped over, the finalize function is called to compute and set the final result. The finalize function doesn’t take any SQL parameters, but it is responsible for setting the result value. The two C functions work together to implement the SQL aggregate function. Consider the built-in avg() aggregate, which computes the numeric average of all the rows in a column. Each call to the step function extracts an SQL value for that row and updates both a running total and a row count. The finalize function divides the total by the row count and sets the result value of the aggregate function. The C functions used to implement an aggregate are defined like this: void user_aggregate_step( sqlite3_context *ctx, int num_values, sqlite3_value **values ) The prototype of a user-defined aggregate step function. This function is called once for each row of an aggregate calculation. The prototype is the same as a scalar function and all of the parameters have similar meaning. The step function should not set a result value with sqlite3_result_xxx(), but it may set an error. void user_aggregate_finalize( sqlite3_context *ctx ) The prototype of a user-defined aggregate finalize function. This function is called once, at the end of an aggregation, to make the final calculation and set the result. This function should set a result value or error condition. 194 | Chapter 9: SQL Functions and Extensions Most of the rules about SQL function overloading that apply to scalar functions also apply to aggregate functions. More than one set of C functions can be registered under the same SQL function name if different parameter counts or text encodings are used. This is less commonly used with aggregates, however, as most aggregate functions are numeric-based and the majority of aggregates take only one parameter. It is also possible to register both scalar and aggregate functions under the same name, as long as the parameter counts are different. For example, the built-in min() and max() SQL functions are available as both scalar functions (with two parameters) and aggregate functions (with one parameter). Step and finalize functions can be mixed and matched—they don’t always need to be unique pairs. For example, the built-in sum() and avg() aggregates both use the same step function, since both aggregates need to compute a running total. The only differ- ence between these aggregates is the finalize function. The finalize function for sum() simply returns the grand total, while the finalize function for avg() first divides the total by the row count. Aggregate Context Aggregate functions typically need to carry around a lot of state. For example, the built- in avg() aggregate needs to keep track of the running total, as well as the number of rows processed. Each call to the step function, as well as the finalize function, needs access to some shared block of memory that holds all the state values. Although aggregate functions can call sqlite3_user_data() or sqlite3_context_db_han dle(), you can’t use the user-data pointer to store aggregate state data. The user-data pointer is shared by all instances of a given aggregate function. If more than one instance of the aggregate function is active at the same time (for example, an SQL query that averages more than one column), each instance of the aggregate needs a private copy of the aggregate state data, or the different aggregate calculations will get intermixed. Thankfully, there is an easy solution. Because almost every aggregate function requires some kind of state data, SQLite allows you to attach a data-block to each specific ag- gregate instance. void* sqlite3_aggregate_context( sqlite3_context *ctx, int bytes ) This function can be called inside an aggregate step function or finalize function. The first parameter is the sqlite3_context structure passed into the step or finalize function. The second parameter represents a number of bytes. The first time this function is called within a specific aggregate instance, the func- tion will allocate an appropriately sized block of memory, zero it out, and attach it to the aggregate context before returning a pointer. This function will return the same block of memory in subsequent invocations of the step and finalize functions. The memory block is automatically deallocated when the aggregate goes out of scope. Aggregate Functions | 195 Using this API call, you can have the SQLite engine automatically allocate and release your aggregate state data on a per-instance basis. This allows multiple instances of your aggregate function to be active simultaneously without any extra work on your part. Typically, one of the first things a step or finalize function will do is call sqlite3_aggre gate_context(). For example, consider this oversimplified version of sum: void simple_sum_step( sqlite3_context *ctx, int num_values, sqlite3_value **values ) { double *total = (double*)sqlite3_aggregate_context( ctx, sizeof( double ) ); *total += sqlite3_value_double( values[0] ); } void simple_sum_final( sqlite3_context *ctx ) { double *total = (double*)sqlite3_aggregate_context( ctx, sizeof( double ) ); sqlite3_result_double( ctx, *total ); } /* ...inside an initialization function... */ sqlite3_create_function( db, "simple_sum", 1, SQLITE_UTF8, NULL, NULL, simple_sum_step, simple_sum_final ); In this case, we’re only allocating enough memory to hold a double-precision floating- point value. Most aggregate functions will allocate a C struct with whatever fields are required to compute the aggregate, but everything works the same way. The first time simple_sum_step() is called, the call to sqlite3_aggregate_context() will allocate enough memory to hold a double and zero it out. Subsequent calls to simple_sum_ step() that are part of the same aggregation calculation (have the same sqlite3_con text) will have the same block of memory returned, as will simple_sum_final(). Because sqlite3_aggregate_context() may need to allocate memory, it is also a good idea to make sure the returned value is not NULL. The above code, in both the step and finalize functions, should really look something like this: double *total = (double*)sqlite3_aggregate_context( ctx, sizeof( double ) ); if ( total == NULL ) { sqlite3_result_error_nomem( ctx ); return; } The only caution with sqlite3_aggregate_context() is in properly dealing with data structure initialization. Because the context data structure is silently allocated and zeroed out on the first call, there is no obvious way to tell the difference between a newly allocated structure, and one that was allocated in a previous call to your step function. If the default all-zero state of a newly allocated context is not appropriate, and you need to somehow initialize the aggregate context, you’ll need to include some type of initi- alization flag. For example: 196 | Chapter 9: SQL Functions and Extensions typedef struct agg_state_s { int init_flag; /* other fields used by aggregate... */ } agg_state; The aggregate functions can use this flag to determine if it needs to initialize the aggregate context data or not: agg_state *st = (agg_state*)sqlite3_aggregate_context( ctx, sizeof( agg_state ) ); /* ...return nonmem error if st == NULL... */ if ( st->init_flag == 0 ) { st->init_flag = 1; /* ...initialize the rest of agg_state... */ } Since the structure is zeroed out when it is first allocated, your initialization flag will be zero on the very first call. As long as you set the flag to something else when you initialize the rest of the data structure, you’ll always know if you’re dealing with a new allocation that needs to be initialized or an existing allocation that has already been initialized. Be sure to check the initialization flag in both the step function and the finalize function. There are cases when the finalize function may be called without first calling the step function, and the finalize function needs to properly deal with those cases. Example As a more in-depth example, let’s look at a weighted average aggregate. Although most aggregates take only one parameter, our wtavg() aggregate will take two. The first parameter will be whatever numeric value we’re trying to average, while the second, optional parameter will be a weighting for this row. If a row has a weight of two, its value will be considered to be twice as important as a row with a weighting of only one. A weighted average is taken by summing the product of the values and weights, and dividing by the sum of the weights. To put things in SQL terms, if our wtavg() function is used like this: SELECT wtavg( data, weight ) FROM ... It should produce results that are similar to this: SELECT ( sum( data * weight ) / sum( weight ) ) FROM ... The main difference is that our wtavg() function should be a bit more intelligent about handling invalid weight values (such as a NULL) and assign them a weight value of 1.0. To keep track of the total data values and the total weight values, we need to define an aggregate context data structure. This will hold the state data for our aggregate. The only place this structure is referenced is the two aggregate functions, so there is no need to put it in a separate header file. It can be defined in the code right along with the two functions: Aggregate Functions | 197 typedef struct wt_avg_state_s { double total_data; /* sum of (data * weight) values */ double total_wt; /* sum of weight values */ } wt_avg_state; Since the default initialization state of zero is exactly what we want, we don’t need a separate initialization flag within the data structure. In this example, I’ve made the second aggregate function parameter (the weight value) optional. If only one parameter is provided, all the weights are assumed to be one, resulting in a traditional average. This will still be different than the built-in avg() function, however. SQLite’s built-in avg() function follows the SQL standard in regard to typing and NULL handling, which might not be what you first assume. (See avg() in Appendix E for more details). Our wtavg() is a bit simpler. In addition to always returning a double (even if the result could be expressed as an integer), it simply ignores any values that can’t easily be translated into a number. First, the step function. This processes each row, adding up the value-weight products, as well as the total weight value: void wt_avg_step( sqlite3_context *ctx, int num_values, sqlite3_value **values ) { double row_wt = 1.0; int type; wt_avg_state *st = (wt_avg_state*)sqlite3_aggregate_context( ctx, sizeof( wt_avg_state ) ); if ( st == NULL ) { sqlite3_result_error_nomem( ctx ); return; } /* Extract weight, if we have a weight and it looks like a number */ if ( num_values == 2 ) { type = sqlite3_value_numeric_type( values[1] ); if ( ( type == SQLITE_FLOAT )||( type == SQLITE_INTEGER ) ) { row_wt = sqlite3_value_double( values[1] ); } } /* Extract data, if we were given something that looks like a number. */ type = sqlite3_value_numeric_type( values[0] ); if ( ( type == SQLITE_FLOAT )||( type == SQLITE_INTEGER ) ) { st->total_data += row_wt * sqlite3_value_double( values[0] ); st->total_wt += row_wt; } } Our step function uses sqlite3_value_numeric_type() to try to convert the parameter values into a numeric type without loss. If the conversion is possible, we always convert the values to a double-precision floating-point, just to keep things simple. This ap- proach means the function will work properly with text representations of numbers (such as the string '153'), but will ignore other datatypes and other strings. 198 | Chapter 9: SQL Functions and Extensions In this case, the function does not report an error, it just ignores the value. If the weight cannot be converted, it is assumed to be one. If the data value cannot be converted, the row is skipped. Once we have our totals, we need to compute the final answer and return the result. This is done in the finalize function, which is pretty simple. The main thing we need to worry about is the possibility of dividing by zero: void wt_avg_final( sqlite3_context *ctx ) { double result = 0.0; wt_avg_state *st = (wt_avg_state*)sqlite3_aggregate_context( ctx, sizeof( wt_avg_state ) ); if ( st == NULL ) { sqlite3_result_error_nomem( ctx ); return; } if ( st->total_wt != 0.0 ) { result = st->total_data / st->total_wt; } sqlite3_result_double( ctx, result ); } To use our aggregate, our application code needs to register these two functions with a database connection using sqlite3_create_function(). Since the wtavg() aggregate is designed to take either one or two parameters, we’ll register it twice: sqlite3_create_function( db, "wtavg", 1, SQLITE_UTF8, NULL, NULL, wt_avg_step, wt_avg_final ); sqlite3_create_function( db, "wtavg", 2, SQLITE_UTF8, NULL, NULL, wt_avg_step, wt_avg_final ); Here are some example queries, as seen from the sqlite3 command shell. This assumes we’ve integrated our custom aggregate into the sqlite3 code (an example of the dif- ferent ways to do this is given later in the chapter): sqlite> SELECT class, value, weight FROM t; class value weight ---------- ---------- ---------- 1 3.4 1.0 1 6.4 2.3 1 4.3 0.9 2 3.4 1.4 3 2.7 1.1 3 2.5 1.1 First, we can try things with only one parameter. This will use the default 1.0 weight for each row, resulting in a traditional average calculation: Aggregate Functions | 199 sqlite> SELECT class, wtavg( value ) AS wtavg, avg( value ) AS avg ...> FROM t GROUP BY 1; class wtavg avg ---------- ---------- ---------- 1 4.7 4.7 2 3.4 3.4 3 2.6 2.6 And finally, here is an example of the full weighted-average calculation: sqlite> SELECT class, wtavg( value, weight ) AS wtavg, avg( value ) AS avg ...> FROM t GROUP BY 1; class wtavg avg ---------- ---------------- ---------- 1 5.23571428571428 4.7 2 3.4 3.4 3 2.6 2.6 In the case of class=1, we see a clear difference, where the heavily weighted 6.4 draws the average higher. For class=2, there is only one value, so the weighted and unweighted averages are the same (the value itself). In the case of class=3, the weights are the same for all values, so again, the average is the same as an unweighted average. Collation Functions Collations are used to sort text values. They can be used with ORDER BY or GROUP BY clauses, or for defining indexes. You can also assign a collation to a table column, so that any index or ordering operation applied to that column will automatically use a specific collation. Above everything else, SQLite will always sort by datatype. NULLs will always come first, followed by a mix of integer and floating-point numeric values in their natural sort order. After the numbers come text values, followed by BLOBs. Most types have a clearly defined sort order. NULL types have no values, so they cannot be sorted. Numeric types use their natural numeric ordering, and BLOBs are always sorted using binary comparisons. Where things get interesting is when it comes to text values. The default collation is known as the BINARY collation. The BINARY collation sorts indi- vidual bytes using a simple numeric comparison of the underlying character encoding. The BINARY collation is also used for BLOBs. In addition to the default BINARY collation, SQLite includes a built-in NOCASE and RTRIM collation that can be used with text values. The NOCASE collation ignores character case for the purposes of sorting 7-bit ASCII, and would consider the expression 'A' == 'a' to be true. It does not, however, consider 'Ä' == 'ä' to be true, nor does it consider 'Ä' == 'A' to be true, as the representations of these characters are outside of the ASCII standard. The RTRIM collation (right-trim) is similar to the default BINARY collation, only it ignores trailing whitespace (that is, whitespace on the right side of the value). 200 | Chapter 9: SQL Functions and Extensions While these built-in collations offer some basic options, there are times when complex sort ordering is required. This is especially true when you get into Unicode represen- tations of languages that cannot be represented with a simple 7-bit ASCII encoding. You may also need a specialized sorting function that sorts by whole words or groups of characters if you’re storing something other than natural language text. For example, if you were storing gene sequences as text data, you might require a custom sorting function for that data. User-defined collation functions allow the developer to define a new collation by reg- istering a comparison function. Once registered, this function is used to compare strings as part of any sorting process. By defining the basic comparison operator, you essen- tially define the behavior of the whole collation. Registering a Collation To define a custom collation, an application needs to register a comparison function under a collation name. Anytime the database engine needs to sort something under that collation, it uses the comparison function to define the required ordering. You will need to reregister the collation with each database connection that requires it. There are three API calls that can be used to register a collation comparison function: int sqlite3_create_collation( sqlite3 *db, const char *name, int text_rep, void *udp, comp_func ) int sqlite3_create_collation16( sqlite3 *db, const void *name, int text_rep, void *udp, comp_func ) Registers a collation comparison function with a database connection. The first parameter is the database connection. The second parameter is the name of the custom collation encoded as a UTF-8 or UTF-16 string. The third parameter is the string encoding the comparison function expects, and can be one of SQLITE_UTF8, SQLITE_UTF16, SQLITE_UTF16BE, SQLITE_UTF16LE, or SQLITE_UTF16_ALIGNED (native UTF-16 that is 16-bit memory aligned). The fourth parameter is a generic user-data pointer that is passed to your comparison function. The last parameter is a function pointer to your comparison function (the prototype of this function is given below). You can unregister a collation by passing a NULL function pointer in under the same name and text encoding as it was originally registered. int sqlite3_create_collation_v2( sqlite3 *db, const char *name, int text_rep, void *udp, comp_func, dest_func ) This function is the same as sqlite3_create_collation(), with one additional pa- rameter. The additional sixth parameter is an optional function pointer referencing a clean-up function that is called when the collation is destroyed (the prototype of this function is given below). This allows the collation to release any resources associated with the collation (such as the user-data pointer). A NULL function pointer can be passed in if no destroy function is required. Collation Functions | 201 A collation is destroyed when the database connection is closed, when a replace- ment collation is registered, or when the collation name is cleared by binding a NULL comparison function pointer. The collation name is case-insensitive. SQLite allows multiple C sorting functions to be registered under the same name, so long as they take different text representations. If more than one comparison function is available under the same name, SQLite will pick the one that requires the least amount of conversion. If you do register more than one function under the same name, their logical sorting behavior should be the same. The format of the user-defined function pointers is given below. int user_defined_collation_compare( void* udp, int lenA, const void *strA, int lenB, const void *strB ) This is the function type of a user-defined collation comparison function. The first parameter is the user-data pointer passed into sqlite3_create_collation_xxx() as the fourth parameter. The parameters that follow pass in the length and buffer pointers for two strings. The strings will be in whatever encoding was defined by the register function. You cannot assume the strings are null-terminated. The return value should be negative if string A is less than string B (that is, A sorts before B), 0 if the strings are considered equal, and positive if string A is greater than B (A sorts after B). In essence, the return value is the ordering of A minus B. void user_defined_collation_destroy( void *udp ) This is the function type of the user-defined collation destroy function. The only parameter is the user-data pointer passed in as the fourth parameter to sqlite3_cre ate_collation_v2(). Although collation functions have access to a user-data pointer, they don’t have an sqlite3_context pointer. That means there is no way to communicate an error back to the SQLite engine. As such, if you have a complex collation function, you should try to eliminate as many error sources as you can. Specifically, that means it is a good idea to pre-allocate any working buffers you might need, as there is no way to abort a com- parison if your memory allocations fail. Since the collation function is really just a simple comparison, it is expected to work and provide an answer every time. Collations can also be dynamically registered on demand. See sqlite3_collation_nee- ded() in Appendix G for more details. Collation Example Here is a simple example of a user-defined collation. In this example, we’re defining a STRINGNUM collation that can be used to sort string representations of numeric values. Unless they’re the same length, string representations of numbers often sort in odd ways. For example, using standard text sorting rules, the string '485' will sort before the string '73' because the character '4' sorts before the character '7', just as the 202 | Chapter 9: SQL Functions and Extensions character 'D' sorts before the character 'G'. To be clear, these are text strings made up of characters that represent numeric digits, not actual numbers. The collation attempts to convert these strings into a numeric representation and then use that numeric value for sorting. Using this collation, the string '485' will sort after '73'. To keep things simple, we’re only going to deal with integer values: int col_str_num( void *udp, int lenA, const void *strA, int lenB, const void *strB ) { int valA = col_str_num_atoi_n( (const char*)strA, lenA ); int valB = col_str_num_atoi_n( (const char*)strB, lenB ); return valA - valB; } static int col_str_num_atoi_n( const char *str, int len ) { int total = 0, i; for ( i = 0; i < len; i++ ) { if ( ! isdigit( str[i] ) ) { break; } total *= 10; total += digittoint( str[i] ); } return total; } The collation attempts to convert each string into an integer value using our custom col_str_num_atoi_n() function, and then compares the numeric results. The col_ str_num_atoi_n() function is very similar to the C standard atoi() function, with the prime difference that it takes a maximum length parameter. That is required in this case, since the strings passed into our collations may not be null-terminated. We would register this collation with SQLite like this: sqlite3_create_collation( db, "STRINGNUM", SQLITE_UTF8, NULL, col_str_num ); Because the standard C function isdigit() is not Unicode aware, our collation sort function will only work with strings that are limited to 7-bit ASCII. We can then have SQL that looks like this: sqlite> CREATE TABLE t ( s TEXT ); sqlite> INSERT INTO t VALUES ( '485' ); sqlite> INSERT INTO t VALUES ( '73' ); sqlite> SELECT s FROM t ORDER BY s; 485 73 sqlite> SELECT s FROM t ORDER BY s COLLATE STRINGNUM; 73 485 Collation Functions | 203 It would also be possible to permanently associate our collation with a specific table column by including the collation in the table definition. See CREATE TABLE in Appen- dix C for more details. SQLite Extensions While custom functions are a very powerful feature, they can also introduce undesired dependencies between database files and custom SQLite environments. If a database uses a custom collation in a table definition or a custom function in a view definition, then that database can’t be opened by any application (including sqlite3) that does not have all the proper custom functions defined. This normally isn’t a big deal for a custom-built application with just a few custom features. You can simply build all the custom code directly into your application. Any- time a database file is created or opened by your application, you can create the ap- propriate function bindings and make your custom function definitions available for use by the database files. Where things get tricky is if you need to open your database files in a general purpose application, like the sqlite3 command-line shell, or one of the third-party GUI data- base managers. Without some way of bringing your custom functions and features with you, your only choice is to splice your custom-feature code into the source of whatever utilities you require, and build site-specific versions that support your SQL environ- ment. That’s not very practical in most cases—especially if the source code to the utility is unavailable. The solution is to build your custom content as an extension. Extensions come in two flavors: static and loadable (dynamic). The difference is in how the extension is built and linked into your main application. The same source can be used to build both a static extension and a loadable extension. Static extensions can be built and linked directly into an application, not unlike a static C library. Loadable extensions act as external libraries, or “plug-ins,” to the SQLite engine. If you build your extension as an external loadable extension, you can load the extension into (almost) any SQLite environment, making your custom functions and SQL environment available to sqlite3 or any other database manager. In both cases, extensions are a handy way to package a set of related functions into one deployable unit. This is particularly useful if you’re writing an SQL support library that is used by a large number of applications, or if you’re writing an SQLite interface to an existing library. Structuring your code as an extension also provides a standard way to distribute a set of custom functions to other SQLite users. By providing your code as an extension, each developer can choose to build and integrate the extension to best suit their needs, without having to worry about the format or design of the extension code. 204 | Chapter 9: SQL Functions and Extensions Even if you plan on statically linking all of your custom function and feature code directly into your application, there is still great value in packaging your code as an extension. By writing an extension, you don’t lose the ability to build and statically link your extension directly into your application, but you gain the ability to build an ex- ternal loadable module. Having your extended environment available as a loadable module allows you to re- create your application’s SQL environment in the sqlite3 command-line tool, or any other general purpose database manager. This opens up the ability to interactively examine your database files in order to design and test queries, debug problems, and track down customer support issues. This alone is a strong reason to consider writing all your custom functions as loadable extensions, even if you never plan on releasing or distributing the standalone loadable extensions. Extension Architecture Extensions are nothing more than a style of packaging your code. The SQLite API calls used to register and create custom function handlers, aggregate handlers, collations, or other custom features are completely unchanged in an extension. The only difference is in the initialization process that creates and binds your custom C functions to a database connection. The build process is also slightly different, depending if you want to build a statically linked extension or a dynamically loadable extension, but both types of extensions can be built from the same source code. Extension architecture focuses on getting dynamically loadable extensions to operate correctly across multiple platforms. The biggest challenge for the dynamic extension architecture is making sure the loadable extension is provided access to the SQLite API. Without getting into a lot of details about how the runtime linker works on different operating systems, the basic issue is that code compiled into an extension and loaded at runtime may not be able to resolve link dependencies from the loadable extension back into the application where the SQLite library sits. To avoid this problem, when an extension is initialized it is passed a large data structure that contains a C function pointer to every function in the SQLite API. Rather than calling the SQLite functions directly, an extension will dereference the required func- tion pointer and use that. This provides a means to resolve any calls into the SQLite library without depending on the linker. While this isn’t fully required for a static ex- tension, the mechanism works equally well with both static and dynamic extensions. Thankfully, the details of how this big data structure works are all well hidden from the developer by using an alternate header file and a few preprocessor macros. These macros completely hide the whole linker and function pointer issue, but with one lim- itation: all the extension code that makes calls into the SQLite API must be in the same file, along with the extension initialization function. That code may call out to other files and other libraries, just as long as that “other code” doesn’t make any direct calls to any sqlite3_xxx() function. SQLite Extensions | 205 For an SQLite extension to work correctly, every function that interacts with the SQLite library must be in the same C source file as the initial- ization function. In practice, this is rarely a significant limitation. Keeping your custom SQLite exten- sions in their own files, out of your application code, is a natural way to organize your code. Most SQLite extensions are a few hundred lines or less, especially if they are simply acting as a glue layer between SQLite and some other library. This can make them large, but usually not so large they become unmanageable as a single file. Extension Design To write an extension, we need to use the extension header file. Rather than using the more common sqlite.h file, an extension uses the sqlite3ext.h file: #include "sqlite3ext.h" SQLITE_EXTENSION_INIT1; /* required by SQLite extension header */ The SQLite extension header defines two macros. The first of these is SQLITE_EXTEN SION_INIT1, and should be referenced at the top of the C file that holds the extension source. This macro defines a file-scoped variable that holds a pointer to the large API structure. Each extension needs to define an entry point. This acts as an initialization function for the extension. The entry point function looks like this: int ext_entry_point( sqlite3 *db, char **error, const sqlite3_api_routines *api ) This is the prototype of an extension entry point. The first parameter is the database connection that is loading this extension. The second parameter can be used to pass back a reference to an error message, should the extension be unable to prop- erly initialize itself. The last parameter is used to convey a block of function pointers to assist in the linking process. We’ll see how this is used in a moment. This function is called by the SQLite engine when it loads a static or dynamic extension. Typically, this function will create and register any custom functions or other custom extensions with the database connection. The entry point has two main jobs. The first job is to finish the initialization process by calling the second extension macro. This should be done as the first bit of code in the entry point (the macro expands into a line of code, so if you’re working in pure C you will need to put any function-scope variables before the initialization macro). It must be done before any sqlite3_xxx() calls are made, or the application will crash: int ext_init( sqlite3 *db, char **error, const sqlite3_api_routines *api ) { /* local variable definitions */ 206 | Chapter 9: SQL Functions and Extensions SQLITE_EXTENSION_INIT2(api); /* ... */ } This macro is the only time you should need to directly reference the api parameter. Once the entry function has finished the extension API initialization, it can proceed with its second main job, which is registering any and all custom functions or features provided by this extension. Unlike a lot of functions, the name of the entry point function is somewhat important. When a dynamic extension is loaded, SQLite needs to ask the runtime linker to return a function pointer to the entry point function. In order to do this, the name of the entry point needs to be known. As we’ll see when we look at the dynamic load functions, by default SQLite will look for an entry point named sqlite3_extension_init(). In theory, this is a good function name to use, since it will allow a dynamic extension to be loaded even if all you know is the filename. Although the same application can load multiple dynamic extensions, even if they have the same entry point name, that is not true about statically linked extensions. If you need to statically link more than one extension into your application, the entry points must have unique names or the linker won’t be able to properly link in the extensions. As a result, it is customary to name the entry point something that is unique to the extension, but fairly easy to document and remember. The entry point often shares the same name as the extension itself, possibly with an _init suffix. The example extension we’ll be looking at is named sql_trig, so good choices for the entry point would be sql_trig() or sql_trig_init(). Example Extension: sql_trig For our example extension, we will be creating a pair of SQL functions that expose some simple trigonometric functions from the standard C math library. Since this is just an example, we’ll only be creating two SQL functions, but you could use the same basic technique to build SQL functions for nearly every function in the standard math library. The first half of our sql_trig.c source file contains the two functions we will be defining in our example extension. The functions themselves are fairly simple, extracting one double-precision floating-point number, converting from degrees to radians, and then returning the result from the math library. I’ve also shown the top of the file with the required #include statements and initialization macros: /* sql_trig.c */ #include "sqlite3ext.h" SQLITE_EXTENSION_INIT1; SQLite Extensions | 207 #include /* this bit is required to get M_PI out of MS headers */ #if defined( _WIN32 ) #define _USE_MATH_DEFINES #endif /* _WIN32 */ #include static void sql_trig_sin( sqlite3_context *ctx, int num_values, sqlite3_value **values ) { double a = sqlite3_value_double( values[0] ); a = ( a / 180.0 ) * M_PI; /* convert from degrees to radians */ sqlite3_result_double( ctx, sin( a ) ); } static void sql_trig_cos( sqlite3_context *ctx, int num_values, sqlite3_value **values ) { double a = sqlite3_value_double( values[0] ); a = ( a / 180.0 ) * M_PI; /* convert from degrees to radians */ sqlite3_result_double( ctx, cos( a ) ); } You’ll notice these are declared as static functions. Making them static hides them from the linker, eliminating any possible name conflicts between this extension and other extensions. As long as the extension entry point is in the same file (which, as we’ve already discussed, is required for other reasons), the entry point will still be able to properly register these functions. Declaring these functions static is not strictly re- quired, but doing so is a good practice and can eliminate linking conflicts. We then need to define our entry point. Here is the second part of the sql_trig.c file: int sql_trig_init( sqlite3 *db, char **error, const sqlite3_api_routines *api ) { SQLITE_EXTENSION_INIT2(api); sqlite3_create_function( db, "sin", 1, SQLITE_UTF8, NULL, sql_sin, NULL, NULL ); sqlite3_create_function( db, "cos", 1, SQLITE_UTF8, NULL, sql_cos, NULL, NULL ); return SQLITE_OK; } This entry point function should not be declared static. Both the static linker (in the case of a static extension) and the dynamic linker (in the case of a loadable extension) need to be able to find the entry point function for the extension to work correctly. Making the function static would hide the function from the linker. These two blocks of code make up our entire sql_trig.c source file. Let’s look at how to build that file as either a static extension or a dynamically loadable extension. 208 | Chapter 9: SQL Functions and Extensions Building and Integrating Static Extensions To statically link an extension into an application, you can simply build the extension source file into the application, just like any other .c file. If your application code was contained in the file application.c, you could build and link our example sql_trig ex- tension using the commands shown here. In the case of most Linux, Unix, and Mac OS X systems, our trig example requires that we explicitly link in the math library (libm). In some cases, the standard C library (libc) is also required. Windows includes the math functions in the standard runtime libraries, so linking in the math library is not required. Unix and Mac OS X systems (with math lib): $ gcc -o application application.c sqlite3.c sql_trig.c -lm Windows systems, using the Visual Studio compiler: > cl /Feapplication application.c sqlite3.c sql_trig.c These commands should produce an executable named application (or applica tion.exe under Windows). Just linking the code together doesn’t magically make it integrate into SQLite. You still need to make SQLite aware of the extension so that the SQLite library can initialize the extension correctly: int sqlite3_auto_extension( entry_point_function ); Registers an extension entry point function with the SQLite library. Once this is done, SQLite will automatically call an extension’s entry point function for every database connection that is opened. The only parameter is a function pointer to the entry point. This function only works with statically linked extensions and does not work with dynamically loadable extensions. This function can be called as many times as is necessary to register as many unique entry points as are required. This function is called by an application, typically right after calling sqlite3_initial ize(). Once an extension’s entry point is registered with the SQLite library, SQLite will initialize the extension for each and every database connection it opens or creates. This effectively makes your extension available to all database connections managed by your application. The only odd thing about sqlite3_auto_extension() is the declaration of the entry point function. The auto extension API call declares the function pointer to have a type of void entry_point( void ). That defines a function that takes no parameters and returns no value. As we’ve already seen, the actual extension entry point has a slightly more complex prototype. SQLite Extensions | 209 The code that actually calls the extension first casts the provided function pointer to the correct type, so the fact that the types don’t match is only an issue for setting the pointer. Extensions typically don’t have header files, since the entry point function would typically be the only thing in a header. To get everything working, you can either provide the proper prototype for the entry point and then cast back to what the API is expecting, or you can simply declare the function prototype incorrectly, and let the linker match things up. Pure C doesn’t type-check function parameters when it links, so this will work, even if it isn’t the most elegant approach. Here’s what the proper prototype with a cast might look like in our application code: /* declare the (correct) function prototype manually */ int sql_trig_init( sqlite3 *db, char **error, const sqlite3_api_routines *api ); /* ... */ sqlite3_auto_extension( (void(*)(void))sql_trig_init ); /* needs cast */ /* ... */ Or, if you’re working in pure C, you can just declare a different prototype: /* declare the (wrong) function prototype manually */ void sql_trig_init(void); /* ... */ sqlite3_auto_extension( sql_trig_init ); /* ... */ As long as the actual sql_trig_init() function is in a different file, this will compile and link correctly, resulting in the desired behavior. If you want a quick practical example of how to add a static extension to an existing application, we can add our sql_trig extension to the sqlite3 shell with a minimum number of changes. We’ll need our sql_trig.c file, which contains the two SQL trig functions, plus the sql_trig_init() entry function. We’ll also need the shell.c source code for the sqlite3 command-line application. First, we need to add some initialization hooks into the sqlite3 source. Make a copy of the shell.c file as shell_trig.c. Open your new copy and search for the phrase “int main(” to quickly locate the starting point of the application. Right before the main function, in global file scope, add a prototype for our sql_trig_init() entry point: /* ... */ void sql_trig_init(void); /* insert this line */ int main(int argc, char **argv){ /* ... */ Then, inside the existing main() function, search for a call to “open_db(” to find a good spot to insert our code. Right before the small block of code (and comments) that contains the first call to open_db(), add this line: sqlite3_auto_extension( sql_trig_init ); 210 | Chapter 9: SQL Functions and Extensions With those two edits, you can save and close the shell_trig.c file. We can then recompile our modified shell_trig.c source into a custom sqlite3trig utility that has our extension built into it. Unix/Linux and Mac OS X: $ gcc -o sqlite3trig sqlite3.c shell_trig.c sql_trig.c -lm Windows: > cl /Fesqlite3trig sqlite3.c shell_trig.c sql_trig.c Our new sqlite3trig application now has our extension built directly into it, and our functions are accessible from any database that is opened with our modified sqlite3trig utility: $ ./sqlite3trig SQLite version 3.X.XX Enter ".help" for instructions Enter SQL statements terminated with a ";" sqlite> SELECT sin( 30 ); 0.5 sqlite> SELECT cos( 30 ); 0.866025403784439 Although we had to modify the source, the modifications were fairly small. While we needed to modify and recompile the main application (sqlite3trig, in this case) to integrate the extension, you can see how easy it would be to add additional extensions. Using Loadable Extensions Dynamic extensions are loaded on demand. An application can be built without any knowledge or understanding of a specific extension, but can still load it when requested to do so. This means you can add new extensions without having to rebuild or recom- pile an application. Loadable extension files are basically shared libraries in whatever format is appropriate for the platform. Loadable extensions package compiled code into a format that the operating system can load and link into your application at runtime. Table 9-1 provides a summary of the appropriate file formats on different platforms. Table 9-1. Summary of loadable extension file format Platform File type Default file extension Linux and most Unix Shared object file .so Mac OS X Dynamic library .dylib Windows Dynamically linked library .DLL SQLite Extensions | 211 Loadable extensions are not supported on all platforms. Loadable extensions depend on the operating system having a well-supported runtime linker, and not all handheld and embedded devices offer this level of support. In general, if a platform supports some type of dynamic or shared library for application use, there is a reasonable chance the loadable extension interface will be available. If the platform does not support dynamic or shared libraries, you may be limited to statically linked extensions. How- ever, in most embedded environments this isn’t a major limitation. Although the file formats and extensions are platform dependent, it is not uncommon to pick a custom file extension that is used across all your supported platforms. Using a common file extension is not required, but it can keep the cross-platform C or SQL code that is responsible for loading the extensions a bit simpler. Like database files, there is no official extension for an SQLite loadable extension, but .sqlite3ext is some- times used. That’s what I’ll use in our examples. Building Loadable Extensions Generally, building a loadable extension is just like building a dynamic or shared li- brary. The code must first be compiled into an object file (a .o or .obj file) and that file must be packaged into a shared or dynamic library. The process of building the object file is exactly the same for both static and dynamic libraries. You can build the object file directly with one of these commands. Mac OS X and Unix/Linux: $ gcc -c sql_trig.c Windows: > cl /c sql_trig.c Once you have the object file, that needs to be converted into a dynamic or shared library using the linker. The commands for that are a bit more platform dependent. First, the Unix and Linux command, which builds a shared object file and links in the standard math library: $ ld -shared -o sql_trig.sqlite3ext sql_trig.o -lm Mac OS X, which uses dynamic libraries, rather than shared object files: $ ld -dylib -o sql_trig.sqlite3ext sql_trig.o -lm And finally, Windows, where we need to build a DLL file. In this case, we need to tell the linker which symbols we want exported. For an extension, only the entry point needs to be exported, so we just include that on the command-line: > link /dll /out:sql_trig.sqlite3ext /export:sql_trig_init sql_trig.obj You can test out your dynamic extension in sqlite3 using the .load command. The command takes two parameters. The first is the filename of your loadable extension, and the second is the name of the entry point function: 212 | Chapter 9: SQL Functions and Extensions $ sqlite3 SQLite version 3.X.XX Enter ".help" for instructions Enter SQL statements terminated with a ";" sqlite> SELECT sin( 60 ); Error: no such function: sin sqlite> .load sql_trig.sqlite3ext sql_trig_init sqlite> SELECT sin( 60 ); 0.866025403784439 As you can see, when we first start sqlite3, it has no awareness of our extension or the SQL functions it contains. The .load command is used to dynamically load the exten- sion. Once loaded, our custom trig functions are available without any need to recom- pile or rebuild the sqlite3 utility. Loadable Extension Security There are some minor security concerns associated with loadable extensions. Because an extension might contain just about any code, a loadable extension might be used to override application values for the SQLite environment. In specific, if an application uses an authorization function to protect against certain types of queries or modifica- tions, a loadable extension could clear the authorization callback function, eliminating any authorization step (see sqlite3_set_authorizer() in Appendix G for more details). To prevent this, and other possible issues, an application must explicitly enable the ability to load external extensions. This has to be done each time a database connection is established. int sqlite3_enable_load_extension( sqlite3 *db, int onoff ) Enables or disables the ability to load dynamic extensions. Loadable extensions are off by default. The first parameter is the database connection to set. The second parameter should be true (nonzero) to enable extensions, or false (zero) to disable them. This function always returns SQLITE_OK. Most general purpose applications, including the sqlite3 shell, automatically enable loadable extensions for every database they open. If your application will support loadable extensions, you will need to enable this as well. Extension loading needs to be enabled for each database connection, every time the database connection is opened. Loading Loadable Extensions There are two ways to load an extension. One is through a C API call, and one is through an SQL function that calls down into the same code as the C API function. In both cases, you provide a filename and, optionally, the name of the entry point function. SQLite Extensions | 213 int sqlite3_load_extension( sqlite3 *db, const char *ext_name, const char *entry_point, char **error ) Attempts to load a loadable extension and associate it to the given database con- nection. The first parameter is the database connection to associate with this extension. The second parameter is the filename of the extension. The third pa- rameter is the name of the entry point function. If the entry point name is NULL, the entry point sqlite3_extension_init is used. The fourth parameter is used to pass back an error message if something goes wrong. This string buffer should be released with sqlite3_free(). This last parameter is optional and can be set to NULL. This will return either SQLITE_OK, to indicate the extension was loaded and the initialization function was successfully called, or it may return SQLITE_ERROR to indicate something went wrong. If an error condition is returned, there may or may not be a valid error string. This function is typically called as soon as a database connection is opened, before any statements are prepared. Although it is legal to call sqlite3_load_extension() at any time, any API calls made by the extension entry point and initialization function are subject to standard restrictions. In specific, that means any calls to sqlite3_create_func tion() made by the extension entry point function will fail to redefine or delete a func- tion if there are any executing SQL statements. The other way to load a loadable extension is with the built-in SQL function load_extension(). load_extension( 'ext_name' ) load_extension( 'ext_name', 'entry_point' ) This SQL function loads the extension with the given filename. If an entry point name is given, that is used as the initialization function. If not, the name sqlite3_extension_init will be used. This function is similar to the C sqlite3_load_extension() call, with one major limi- tation. Because this is an SQL function, when it is called there will be, by definition, an SQL statement executing when the extension is loaded. That means that any extension loaded with the load_extension() SQL function will be completely unable to redefine or delete a custom function, including the specialized set of like() functions. To avoid this problem while testing your loadable extensions in the sqlite3 shell, use the .load command. This provides direct access to the C API call, allowing you to get around the limitations in the SQL function. See .load in Appendix B for more details. No matter which mechanism you use to load a loadable extension, you’ll need to do it for each database connection your application opens. Unlike the sqlite3_auto_exten sion() function, there is no automatic way to import a set of loadable extensions for each and every database. 214 | Chapter 9: SQL Functions and Extensions The only way to completely unload a loadable extension is to close the database connection. Multiple Entry Points Although most extensions have only a single entry point function, there is nothing that says this must be true. It is perfectly acceptable to define multiple entry points in a single extension—just make sure they each call SQLITE_EXTENSION_INIT2(). Multiple entry points can be used to control the number of imported functions. For example, if you have a very large extension that defines a significant number of func- tions in several different subcategories, you would likely define one main entry point that imports every extension, aggregation, collation, and other features with one call. You could also define an entry point for each subcategory of functionality, or one entry point for all the functions, one for all the collations, etc. You might also define one entry point to bind UTF-8 functions, and another for UTF-16. No matter how you want to mix and match things, this allows an extension user to import just the functionality they need. There is no danger in redefining a function from two different entry points (assuming all of the entry points register similar functions in similar ways), so different entry points can register overlapping sets of functions with- out concern. Even if your extension is not large and doesn’t really justify multiple entry points, a second one can still be handy. Some extensions define a “clear” entry point, for exam- ple, sql_trig_clear(). This would typically be very similar to the _init() entry point function, but rather than binding all the function pointers into a database connection, it would bind all NULL pointers. This effectively “unloads” the extension from the SQL environment—or at least removes all the functions it created. The extension file may still be in memory, but the SQL functions are no longer available to that database connection. The only thing to remember about a _clear() entry point is that it cannot be called while an SQL statement is being executed, because of the redefine/delete rules for functions like sqlite3_create_function(). This also means you cannot call a _clear() entry point using the SQL function load_extension(). Chapter Summary Custom functions, aggregations, and collations can be an extremely powerful means to extend and expand the SQL environment to fit your own needs and designs. Exten- sions make for a relatively painless way to modularize and compartmentalize those custom features. This makes the extension code easier to test, support, and distribute. In the next chapter, we’ll look at one of the more powerful customizations in SQLite: virtual tables. Virtual tables allow a developer to merge the SQLite environment to just about any data source. Like other custom features, the easiest way to write a virtual table is to use an extension. SQLite Extensions | 215 CHAPTER 10 Virtual Tables and Modules This chapter looks at how to use and write virtual tables. A virtual table is a custom extension to SQLite that allows a developer to define the structure and contents of a table through code. To the database engine, a virtual table looks like any other table— a virtual table can be queried, updated, and manipulated using the same SQL com- mands that are used on other tables. The main difference is where the table data comes from. When processing a normal table, the SQLite library might access a database file to retrieve row and column values. In the case of a virtual table, the SQLite library calls in to your code to perform these functions and return data values. Your functions, in turn, can gather or calculate the requested data from whatever sources you want. Developing a virtual table implementation, which is known as an SQLite module, is a fairly advanced feature. This chapter should give you a good idea of what virtual tables are capable of doing and the basics of how to write your own module. We’ll be walking through the code for two different modules. The first is a fairly simple one that exposes some internal SQLite data as a table. The second example will allow read-only access to standard Apache httpd server logs. This chapter should provide a solid starting point. However, if you find yourself having to write a more robust module, you may need to dig a bit deeper into the development documentation, found at http://www.sqlite.org/vtab.html. I would also suggest having a look at the source code to some of the other modules out there (including those that ship with SQLite) to get a better idea of how the advanced features work. Modules are complex enough that sometimes it is easier to modify an existing module, rather than implementing everything from the ground up. Like the last chapter, this type of advanced development is usually best learned hands- on. To help with this, the full source to both examples is available in the book down- load. See “Example Code Download” on page xvi for more details. 217 Introduction to Modules Virtual tables are typically used to link the SQLite database engine to an alternate data source. There are two general categories of virtual tables: internal and external. There aren’t any implementation differences between the categories, they just provide a rough way to define a module’s functionality. Internal Modules Internal modules are self-contained within the database. That is, the virtual table acts as a fancy front-end to more traditional database tables that are created and maintained by the virtual table module. These back-end tables are sometimes known as shadow tables. Most importantly, all the data used by the module is still stored within the database file. These types of modules typically provide a specialized type of indexing or lookup feature that is not well suited to the native database indexes. Internal virtual tables may require multiple shadow tables to efficiently operate. The two largest modules included in the SQLite distribution (FTS3 and R*Trees) are both internal style modules. Both of these modules create and configure several standard tables to store and index the data they’ve been asked to maintain. Generally, internal modules are used to improve or extend the data manipulation facilities of a database. In most cases, an internal virtual table isn’t doing anything an SQL developer couldn’t do on their own, the module is just making it easier or faster (or both). Internal modules often play the role of an abstract “smart view” that offers highly optimized access patterns to specific types of data or specific structures of data. Both the Full Text Search module and the R*Tree module are prime examples of mod- ules that provide highly specialized searches on specific types and structures of data. External Modules The other major category of modules are external modules. These are modules that interface with some type of external data source. That data source might be something as simple as an external file. For example, a module could expose a CSV file or Excel file as an SQL table within the database. Pretty much any structured file can be exposed this way. An external module can also be used to present other data sources to the SQLite database engine. You could actually write an SQLite module that exposed tables from a MySQL database to the SQLite database engine. Or, for something a bit more unusual, have the query SELECT ip FROM dns WHERE hostname = 'www.oreilly.com' go out and process a DNS request. External modules can get quite exotic. In the case of external modules, it is important to understand that the data is not im- ported or copied into the SQLite database. Rather than loading the data into standard tables and allowing you to access it from there, an external module acts as a real-time 218 | Chapter 10: Virtual Tables and Modules translator between the SQLite data structures and whatever external data source you wish to access. Modules will typically reflect changes to their data source in real time. Of course, you can use an external module as an importer by copying the data from a virtual table to a standard table with an INSERT...SELECT statement. If the module has full read/write support, you can even use it as an exporter by copying data from the database into the virtual table. By using this technique, I’ve seen cases of SQLite being used as a “universal translator” for several different external data formats. By writing a virtual table module that can speak to each file format, you can easily and quickly move data between supported formats. Example Modules To help explain how modules work, we’re going to work through two examples. The first example is a very simple internal module that exposes the output of the PRAGMA database_list command as a full-blown table. This allows you to run SELECT queries (including WHERE constraints) against the current database list. Although this module is read-only and extremely simple, it should serve as a good first introduction to the module system. The second example is a bit more in-depth. We’ll be looking at building an external module that exposes Apache httpd server logs to the database engine. This allows a webmaster to run SQL queries directly against a logfile (including the active logfile) without having to first import the data into a traditional database table. SQL for Anything As we’ll see with the webserver logfile example, developing an external SQLite module can be an easy way to provide generic search-and-report services to arbitrary data for- mats. In the case of webserver logs, many server administrators have a stash of scripts and utilities they use for logfile analysis. While these can work quite well for clearly defined tasks, such scripts often require significant code modifications to alter search parameters or report formats. This can make custom scripts difficult to modify and somewhat inflexible. Webserver log analysis is a common enough problem that there are some extremely powerful general purpose packages available for download. Some of these packages are quite robust and impressive, but to use them effectively requires understanding and experience with the package and the tools that it provides. With the external data module, you can simply attach the SQLite engine directly to your logfiles, making the logs appear as a big (and constantly updating) table. Once this is done, you have the whole power of the relational database engine at your dis- posal. Best of all, the queries and searches are all defined in SQL, a language that many web administrators already know. Report generation becomes a snap and, when combined with the sqlite3 command-line utility, the module will enable real-time Introduction to Modules | 219 interaction with the live log data. This allows a system administrator faced with a se- curity or performance incident to quickly formulate and execute arbitrary searches and summary reports interactively, in a language and environment they’re already com- fortable using. This is one of the more compelling uses of virtual tables. While there are many instances of applications that can take advantage of the custom index formats and improved search features offered by some virtual tables, the true magic happens with external modules. The ability to integrate any regular data source into a full SQL environment makes for an extremely powerful and enabling tool, especially in cases where there is a need to directly interact with the data in real time. The next time you’re thinking about clobbering together some scripts to scan or filter a structured data source, ask yourself how hard it would be to write an SQLite module instead. Modules can definitely be tricky to write, but once you have a working module, you also have the full power of the SQL language at your hands. Module API The virtual table API is one of the more advanced SQLite APIs. In specific, it does very little hand-holding and will often fool those that make assumptions. The functions you need to write are often required to do a very specific set of operations. If you fail to do any one of those, or forget to initialize a data structure field, the result might very well be a bus error or segmentation fault. That street goes two ways, however. While the SQLite core expects you to do your job, it does a very good job of always doing its job in a very predictable and documented way. Most of this code operates fairly deeply in the SQLite core, and SQLite does a solid job of protecting your code against odd user behavior. For example, none of the example code checks for NULL parameter values, as you can be sure SQLite will never allow a NULL database pointer (or some equally critical parameter) to be passed into your function. Implementing a virtual table module is a bit like developing an aggregate function, only a lot more complex. You must write a series of functions that, taken together, define the behavior of the module. This block of functions is then registered under a module name. int sqlite3_create_module( sqlite3 *db, const char *name, const sqlite3_module *module, void *udp ) Creates and registers a virtual table module with a database connection. The sec- ond parameter is the name of the module. The third parameter is a block of function pointers that implements the virtual table. This pointer must remain valid until the SQLite library is shut down. The final parameter is a generic user-data pointer that is passed to some of the module functions. 220 | Chapter 10: Virtual Tables and Modules int sqlite3_create_module_v2( sqlite3 *db, const char *name, const sqlite3_module *p, void *udp, destroy_callback ) The v2 version of this function is identical to the original function, except for an additional fifth parameter. This version adds a destroy callback of the form void callback(void *udp). This function can be used to release or otherwise clean up the user-data pointer, and is called when the entire module is unloaded. This is done when the database is shut down, or when a new module with the same name is registered in place of this one. The destroy function pointer is optional, and can be set to NULL. Function pointers are passed in through an sqlite3_module structure. The main reason for this is that there are nearly 20 functions that define a virtual table. All but a few of those functions are mandatory. A module defines a specific type of virtual table. Once a module has been successfully registered, an actual table instance of that type must be created using the SQL com- mand CREATE VIRTUAL TABLE. A single database may have multiple instances of the same type of virtual table. A single database may also have different types of virtual tables, just as long as all the modules are properly registered. The syntax for the CREATE VIRTUAL TABLE command looks something like this: CREATE VIRTUAL TABLE table_name USING module_name( arg1, arg2, ... ) A virtual table is named, just like any other table. To define the table, you must provide the module name and any arguments the module requires. The argument block is op- tional, and the exact meaning of the arguments is up to the individual module imple- mentations. It is the responsibility of the module to define the actual structure (column names and types) of the table. The arguments have no predefined structure and do not need to be valid SQL expressions or column definitions. Each argument is passed as a literal text value to the module, with only the leading and trailing whitespace trimmed. Everything else, including whitespace within the argument, is passed as a single text value. Here is a quick overview of the different functions that are defined by sqlite3_module structure. When we look at the example modules, we’ll go back through these one at a time in much more detail. The module functions are divided up into three rough groups. The first set of functions operate on table instances. The second set includes the functions that scan a table and return data values. The last group of functions deals with implementing transaction control. To implement a virtual table module, you will need to write a C function that performs each of these tasks. Functions that deal with individual table instances include: xCreate() Required. Called when a virtual table instance is first created with the CREATE VIRTUAL TABLE command. Module API | 221 xConnect() Required, but frequently the same as xCreate(). Very similar to xCreate(), this is called when a database with an existing virtual table instance is loaded. Called once for each table instance. xDisconnect() Required. Called when a database containing a virtual table instance is detached or closed. Called once for each table instance. xDestroy() Required, but frequently the same as xDisconnect(). Very similar to xDisconnect(), this is called when a virtual table instance is destroyed with the DROP TABLE command. xBestIndex() Required. Called, sometimes several times, when the database engine is preparing an SQL statement that involves a virtual table. This function is used to determine how to best optimize searches and queries made against the table. This information helps the optimizer understand how to get the best performance out of the table. xUpdate() Optional. Called to modify (INSERT, UPDATE, or DELETE) a table row. If this function is not defined, the virtual table will be read-only. xFindFunction() Optional. Called when preparing an SQL statement that uses virtual table values as parameters to an SQL function. This function allows the module to override the default implementation of any SQL function. This is typically used in conjunction with the SQL functions like() or match() to define module-specific versions of these functions (and, from that, module-specific versions of the LIKE and MATCH SQL expressions). xRename() Required. Called when a virtual table is renamed using the ALTER TABLE...RENAME command. The second group of functions deals with processing table scans. These functions operate on a table cursor, which holds all of the state information required to perform a table scan. As the database engine scans a table and steps through each individual row, the cursor is responsible for keeping track of which row is being processed. A single virtual table instance may be involved in more than one table scan at a time. To function correctly, the module must keep all state information in the table cursor, and cannot use user-data pointers or static variables. Consider, for example, a virtual table instance that is self-joined, and must have more than one scan active at the same time. 222 | Chapter 10: Virtual Tables and Modules Cursor functions include: xOpen() Required. Called to create and initialize a table cursor. xClose() Required. Called to shut down and release a table cursor. xFilter() Required. Called to initiate a table scan and provide information about any specific conditions put on this particular table scan. Conditions typically come from WHERE constraints on the query. The xFilter() function is designed to work in conjunction with xBestIndex() to allow a virtual table to pre-filter as many rows as it can. After readying the module for a table scan, xFilter() should also look up the first row. This may be called more than once between xOpen() and xClose(). xNext() Required. Called to advance a table cursor to the next row. xEof() Required. Called to see if a table cursor has reached the end of the table or not. EOF is traditional shorthand for end-of-file. This function is always called right after a call to xFilter() or xNext(). xRowid() Required. Called to extract the virtual ROWID of the current row. xColumn() Required. Called to extract a column value for the current row. Normally called multiple times per row. Finally, we have the transaction control functions. These allow external data sources to take part in the transaction control process, and include: xBegin() Optional. Called when a transaction is started. xSync() Optional. Called to start committing a transaction. xCommit() Optional. Called to finalize a database transaction. xRollback() Optional. Called to roll back a database transaction. If this sounds confusing, don’t give up just yet. As we start to work through the code examples, we will go back through each function and take a closer look at all the details. You may be surprised to see that the transactional functions are optional. The reason for this is that internal modules don’t need or require their own transactional control. When an internal module modifies any standard table in response to a virtual table operation, the normal transactional engine is already protecting those changes and Module API | 223 updates. Additionally, external read-only modules don’t require transactional control because they aren’t driving any modifications to their external data sources. The only type of module that really needs to implement transactional control are those that provide transaction-safe read/write support to external data sources. Simple Example: dblist Module The first example takes the output of the PRAGMA database_list command and presents it as a table. Since the output from the PRAGMA command is already in the same structure as a table, this conversion is fairly simple. The main reason for doing this is to use the full SELECT syntax, including WHERE conditions, against the virtual table. This is not possible with the PRAGMA command. The PRAGMA database_list command normally returns three columns: seq, name, and file. The seq column is a sequence value that indicates which “slot” the database is attached to. The name column is the logical name of the database, such as main or temp, or whatever name was given to the ATTACH DATABASE command. (See ATTACH DA- TABASE in Appendix C). The file column displays the full path to the database file, if such a file exists. In-memory databases, for example, do not have any associated filenames. To keep things simple, the module uses the seq value as our virtual ROWID value. The seq value is an integer value and is unique across all of the active databases, so it serves this purpose quite well. Create and Connect The first set of functions we’ll be looking at are used to create or connect a virtual table instance. The functions you need to provide are: int xCreate( sqlite3 *db, void *udp, int argc, char **argv, sqlite3_vtab **vtab, char **errMsg ) Required. This function is called by SQLite in response to a CREATE VIRTUAL TABLE command. This function creates a new instance of a virtual table and initi- alizes any required data structures and database objects. The first parameter is the database connection where the table needs to be created. The second parameter is the user-data pointer passed into sqlite3_create_mod ule(). The third and fourth parameters pass in a set of creation arguments. The fifth parameter is a reference to a virtual table (vtab) structure pointer. Your func- tion must allocate and return one of these structures. The final parameter is a reference to an error message. This allows you to pass back a custom error message if something goes wrong. If everything works as planned, this function returns SQLITE_OK. If the return code is anything other than SQLITE_OK, the vtab structure should not be allocated. 224 | Chapter 10: Virtual Tables and Modules Every module is passed at least three arguments. The variable argv[0] will always contain the name of the module used to create the virtual table. This allows the same xCreate() function to be used with similar modules. The logical name of the database (main, temp, etc.) is passed in as argc[1], and argv[2] contains the user- provided table name. Any additional arguments given to the CREATE VIRTUAL TABLE statement will be passed in, starting with argv[3], as text values. int xConnect( sqlite3 *db, void *udp, int argc, char **argv, sqlite3_vtab **vtab, char **errMsg ) Required. The format and parameters of this function are identical to xCreate(). The main difference is that xCreate() is only called when a virtual table instance is first created. xConnect(), on the other hand, is called any time a database is opened. The function still needs to allocate and return a vtab structure, but it should not need to initialize any database objects. If no object creation step is required, many modules use the same function for both xCreate() and xConnect(). These functions bring a virtual table instance into being. For each virtual table, only one of these functions will be called over the lifetime of a database connection. The create and connect functions have two major tasks. First, they must allocate an sqlite3_vtab structure and pass that back to the SQLite engine. Second, they must define the table structure with a call to sqlite3_declare_vtab(). The xCreate() call must also create and initialize any storage, be it shadow tables, external files, or what- ever is required by the module design. The order of these tasks is not important, so long as all of the tasks are accomplished before the xCreate() or xConnect() function returns. Allocate the vtab structure The xCreate() and xConnect() functions are responsible for allocating and passing back an sqlite3_vtab structure. That structure looks like this: struct sqlite3_vtab { const sqlite3_module *pModule; /* module used by table */ int nRef; /* SQLite internal use only */ char *zErrMsg; /* Return error message */ }; The module is also (eventually) responsible for deallocating this structure, so you can technically use whatever memory management routines you want. However, for max- imum compatibility, it is strongly suggested that modules use sqlite3_malloc(). This will allows the module to run in any SQLite environment. The only field of interest in the sqlite3_vtab structure is the zErrMsg field. This field allows a client to pass a custom error message back to the SQLite core if any of the functions (other than xCreate() or xConnect()) return an error code. The xCreate() and Simple Example: dblist Module | 225 xConnect() functions return any potential error message through their sixth parameter, since they are unable to allocate and pass back a vtab structure (including the zErrMsg pointer) unless the call to xCreate() or xConnect() was successful. The xCreate() and xConnect() functions initialize the vtab error message pointer to NULL after allocating the vtab structure. Typically, a virtual table needs to carry around a lot of state. In many systems, this is done with some kind of user-data pointer or other generic pointer. While the sqlite3_create_module() function does allow you to pass in a user-data pointer, that pointer is only made available to the xCreate() and xConnect() functions. Additionally, the same user-data pointer is provided to every table instance managed by a given module, so it isn’t a good place to keep instance-specific data. The standard way of providing instance-specific state data is to wrap and extend the sqlite3_vtab structure. For example, our dblist module will define a custom vtab data structure that looks like this: typedef struct dblist_vtab_s { sqlite3_vtab vtab; /* this must go first */ sqlite3 *db; /* module-specific fields then follow */ } dblist_vtab; By defining a custom data structure, the module can effectively extend the standard sqlite3_vtab structure with its own data. This will only work if the sqlite3_vtab struc- ture is the first field, however. It must also be a vanilla C struct, and not a C++ class or some other managed object. Also, note that the vtab field is a full instance of the sqlite3_vtab structure, and not a pointer. That is, the custom vtab structure “contains- a” sqlite3_vtab structure, and does not “references-a” sqlite3_vtab structure. It might seem a bit ugly to append module instance data to the sqlite3_vtab structure in this fashion, but this is how the virtual table interface is designed to work. In fact, this is the whole reason why the xCreate() and xConnect() functions are responsible for allocating the memory required by the sqlite3_vtab structure. By having the module allocating the memory, it can purposely overallocate the structure for its own means. In the case of the dblist module, the only additional parameter the module requires is the database connection. Most modules require a lot more information. In specific, the xCreate() and xConnect() functions are the only time the module code will have access to the user-data pointer, the database connection pointer (the sqlite3 pointer), the database name, or the virtual table name. If the module needs access to these later, it needs to stash copies of this data in the vtab structure. The easiest way to make copies of these strings is with sqlite3_mprintf(). (See sqlite3_mprintf() in Appendix G.) Most internal modules will need to make a copy of the database connection, the name of the database that contains the virtual module, and the virtual table name. Not only is this information required to create any shadow tables (which happens in xCreate()), but this information is also required to prepare any internal SQL statements (typically in xOpen()), as well as deal with xRename() calls. One of the most common 226 | Chapter 10: Virtual Tables and Modules bugs in module design is to assume there is only one database attached to the database connection, and that the virtual table is living in the main database. Be sure to test your module when virtual tables are created and manipulated in other databases that have been opened with ATTACH DATABASE. The dblist module doesn’t have any shadow tables, and the data we need to return is specific to the database connection, not any specific database. The module still needs to keep a copy of the database connection, so that it can prepare the PRAGMA statement, but that’s it. As a result, the dblist_vtab structure is much simpler than most internal structures. Define the table structure The other major responsibility of the xCreate() and xConnect() functions is to define the structure of the virtual table. int sqlite3_declare_vtab( sqlite3 *db, const char *sql ) This function is used to declare the format of a virtual table. This function may only be called from inside a user-defined xCreate() or xConnect() function. The first parameter is the database connection passed into xCreate() or xConnect(). The second parameter is a string that should contain a single, properly formed CREATE TABLE statement. Although the module must provide a table name in the CREATE TABLE statement, the table name (and database name, if provided) is ignored. The given name does not need to be the name of the virtual table instance. In addition, any constraints, default values, or key definitions within the table definition are also ignored—this includes any defi- nition of an INTEGER PRIMARY KEY as a ROWID alias. The only parts of the CREATE TABLE statement that really matters are the column names and column types. Everything else is up to the virtual table module to enforce. Like standard tables, virtual tables have an implied ROWID column that must be unique across all of the rows in the virtual table. Most of the virtual table operations reference rows by their ROWID, so a module will need some way to keep track of that value or generate a unique ROWID key value for every row the virtual table manages. The dblist virtual table definition is quite simple, reflecting the same structure as the PRAGMA database_list output. Since the table structure is also completely static, the code can just define the SQL statement as a static string: static const char *dblist_sql = "CREATE TABLE dblist ( seq INTEGER, name TEXT, file TEXT );"; /* ... */ sqlite3_declare_vtab( db, dblist_sql ); Depending on the design requirements, a module might need to dynamically build the table definition based off the user-provided CREATE VIRTUAL TABLE arguments. Simple Example: dblist Module | 227 It was already decided that the dblist example will simply use the seq values returned by the PRAGMA command as the source of both the seq output column and the ROWID values. Virtual tables have no implicit way of aliasing a standard column to the ROWID column, but a module is free to do this explicitly in the code. It makes sense for a virtual table to define its own structure, rather than having it defined directly by the CREATE VIRTUAL TABLE statement. This allows the application to adapt to fit its own needs, and tends to greatly simplify the CREATE VIRTUAL TABLE statements. There is one drawback, however, in that if you want to look up the structure of a virtual table, you cannot simply look in the sqlite_master system table. Each virtual table instance will have an entry in this table, but the only thing you’ll find there is the original CREATE VIRTUAL TABLE statement. If you want to look up the column names and types of a virtual table instance, you’ll need to use the command PRAGMA table_info( table _name ). This will provide a full list of all the column names and types in a table, even for a virtual table. See table_info in Appendix F for more details. Storage initialization If a virtual table module manages its own storage, the xCreate() function needs to allocate and initialize the required storage structure. In the case of an internal module that uses shadow tables, the module will need to create the appropriate tables. Only the xCreate() function needs to do this. The next time the database is opened, xCon nect(), and not xCreate(), will be called. The xConnect() function may want to verify the correct shadow tables exist in the correct database, but it should not create them. If you’re writing an internal module that uses shadow tables, it is customary to name the shadow tables after the virtual table. In most cases you’ll also want to be sure to create the shadow tables in the same database as the virtual table. For example, if your module requires three shadow tables per virtual table instance, such as Data, IndexA, and IndexB, a typical way to create the tables within your xCreate() function would be something like this (see sqlite3_mprintf() in Appendix G for details on the %w format): sql_cmd1 = sqlite3_mprintf( "CREATE TABLE \"%w\".\"%w_Data\" (...)", argv[1], argv[2],... ); sql_cmd2 = sqlite3_mprintf( "CREATE TABLE \"%w\".\"%w_IndexA\" (...)", argv[1], argv[2],... ); sql_cmd3 = sqlite3_mprintf( "CREATE TABLE \"%w\".\"%w_IndexB\" (...)", argv[1], argv[2],... ); This format will properly create per-instance shadow tables in the same database as the virtual table. The double quotes also ensure you can handle nonstandard identifier names. You should use a similar format (with a fully quoted database name and table name) in every SQL statement your module may generate. If you’re writing an external module that manages its own files, or something similar, you should try to follow some similar convention. Just remember not to use the database name (argv[1]) in your naming convention, as this can change, depending on how the database was open, or attached to the current database connection. 228 | Chapter 10: Virtual Tables and Modules Create/connect dblist example Although the dblist module could be considered an internal module, the module does not manage storage for any of the data it uses. This means there is no requirement to create shadow tables. This allows the module to use the same function for both the xCreate() and xConnect() function pointers. Here is the full dblist create and connect function: static int dblist_connect( sqlite3 *db, void *udp, int argc, const char *const *argv, sqlite3_vtab **vtab, char **errmsg ) { dblist_vtab *v = NULL; *vtab = NULL; *errmsg = NULL; if ( argc != 3 ) return SQLITE_ERROR; if ( sqlite3_declare_vtab( db, dblist_sql ) != SQLITE_OK ) { return SQLITE_ERROR; } v = sqlite3_malloc( sizeof( dblist_vtab ) ); /* alloc our custom vtab */ *vtab = (sqlite3_vtab*)v; if ( v == NULL ) return SQLITE_NOMEM; v->db = db; /* stash this for later */ (*vtab)->zErrMsg = NULL; /* initalize this */ return SQLITE_OK; } The create/connect function walks through the required steps point by point. We verify the argument count (in this case, only allowing the standard three arguments), define the table structure, and finally allocate and initialize our custom vtab structure. Remember that you should not pass back an allocated vtab structure unless you’re returning an SQLITE_OK status. Disconnect and Destroy Not surprisingly, the xCreate() and xConnect() functions each have their own counterparts: int xDisconnect( sqlite3_vtab *vtab ) Required. This is the counterpart to xConnect(). It is called every time a database that contains a virtual table is detached or closed. This function should clean up any process resources used by the virtual table implementation and release the vtab data structure. int xDestroy( sqlite3_vtab *vtab ) Required. This is the counterpart to xCreate(), and is called in response to a DROP TABLE command. If an internal module has created any shadow tables to store module data, this function should call DROP TABLE on those tables. As with Simple Example: dblist Module | 229 xDisconnect(), this function should also release any process resources and release the virtual table structure. Many modules that do not manage their own storage use the same function for xDisconnect() and xDestroy(). As with the xCreate() and xConnect() functions, only one of these functions will be called within the context of a given database connection. Both functions should release the memory allocated to the vtab pointer. The xDestroy() function should also delete, drop, or deallocate any storage used by the virtual table. Make sure you use fully quali- fied and quoted database and table names. The dblist version of this function—which covers both xDisconnect() and xDestroy()—is very simple: static int dblist_disconnect( sqlite3_vtab *vtab ) { sqlite3_free( vtab ); return SQLITE_OK; } The code frees the vtab memory, and that’s about it. Query Optimization Virtual tables present a challenge for the query optimizer. In order to optimize SELECT statements and choose the most efficient query plan, the optimizer must weigh a num- ber of factors. In addition to understanding the constraints on the query (such as WHERE conditions), optimization also requires some understanding of how large a table is, what columns are indexed, and how the table can be sorted. There is no automatic way for the query optimizer to deduce this information from a virtual table. Virtual tables cannot have traditional indexes, and if the internal virtual table implements a fancy custom indexing system, the optimizer has no way of knowing about it or how to best take advantage of it. While every query could perform a full table scan on a virtual table, that largely defeats the usefulness of many internal modules that are specifically designed to provide an optimized type of lookup. The solution is to allow the query optimizer to ask the virtual table module questions about the cost and performance of different kinds of lookups. This is done through the xBestIndex() function: int xBestIndex( sqlite3_vtab *vtab, sqlite3_index_info *idxinfo ) Required. When an SQL statement that references a virtual table is prepared, the query optimizer calls this function to gather information about the structure and capabilities of the virtual table. The optimizer is basically asking the virtual table a series of questions about the most efficient access patterns, indexing abilities, and natural ordering provided by the module. This function may be called several times when a statement is prepared. 230 | Chapter 10: Virtual Tables and Modules Communication between the query optimizer and the virtual table is done through the sqlite3_index_info structure. This data structure contains an input and output section. The SQLite library fills out the input section (input to your function), essen- tially asking a series of questions. You can fill out the output section of the structure, providing answers and expense weightings to the optimizer. If xBestIndex() sounds complicated, that’s because it is. The good news is that if you ignore the optimizer, it will revert to a full table scan for all queries and perform any constraint checking on its own. In the case of the dblist module, we’re going to take the easy way out, and more or less ignore the optimizer: static int dblist_bestindex( sqlite3_vtab *vtab, sqlite3_index_info *info ) { return SQLITE_OK; } Given how simple the module is, and the fact that it will never return more than 30 rows (there is an internal limit on the number of attached databases), this is a fair trade off between performance and keeping the code simple. Even with fairly large datasets, SQLite does a pretty good job at processing full table scans with surprising speed. Of course, not all modules—especially internal ones—can get away with this. Most larger modules should try to provide an intelligent response to the optimizer’s ques- tions. To see how this is done, we’ll take a more in-depth look at this function later on in the chapter. See “Best Index and Filter” on page 262. Custom Functions As much as possible, a good module will attempt to make virtual table instances look and act exactly like standard tables. Functions like xBestIndex() help enforce that abstraction, so that virtual tables can interact with the optimizer to correctly produce more efficient lookups—especially in the case of an internal module trying to provide a better or faster indexing method. There are a few other cases when SQLite needs a bit of help to hide the virtual table abstraction from the database user and other parts of the SQLite engine. SQL function calls and expression processing is one such area. The xFindFunction() allows a module to override an existing function and provide its own implementation. In most cases, this is not needed (or even recommended). The major exceptions are the SQL functions like() and match(), which are used to imple- ment the SQL expressions LIKE and MATCH (see Appendix D for more details). A text-search engine is likely to override the like() and match() functions to provide an implementation that can directly access the search string and base its index opti- mization off the provided arguments. Without the ability to override these functions, it would be very difficult to optimize text searches, as the standard algorithm would require a full stable scan, extracting each row value and doing an external comparison. Simple Example: dblist Module | 231 int xFindFunction( sqlite3_vtab *vtab, int arg, const char *func_name, custom_function_ref, void **udp_ref) Optional. This function allows a module to override an existing function. It is called when preparing an SQL statement that uses a virtual table column as the first parameter in an SQL function (or the second, in the case of like(), glob(), match(), or regexp()). The first parameter is the vtab structure for this table in- stance. The second parameter indicates how many parameters are being passed to the SQL function, and the third parameter holds the name of the function. The fourth parameter is a reference to a scalar function pointer (see “Scalar Func- tions” on page 182), and the fifth parameter is a reference to a user-data pointer. Using data from the first three parameters, a virtual table module needs to decide if it wants to override the existing function or not. If the module does not want to override the function, it should simply return zero. If the module does want to provide a custom function, it needs to set the function pointer reference (the fourth parameter) to the scalar function pointer of its choice and set the user-data pointer reference to a user-data pointer. The new function (and user-data pointer) will be called in the same context as the original function. Most modules will not need to implement this function, and those that do should only need to override a few key functions. The dblist module does not provide an implementation. Table Rename Many modules, especially internal modules, key specific information off the name of the virtual table. This means that if the name of the virtual table is changed, the module needs to update any references to that name. This is done with the xRename() function. int xRename( sqlite3_vtab *vtab, const char *new_name ) Required. This function is called in response to the SQL command ALTER TABLE...RENAME. The first parameter is the table instance being renamed, and the second parameter is the new table name. In the case of an internal module, the most likely course of action is to rename any shadow tables to match the new name. Doing this properly will require knowing the original table name, as well as the database (main, temp, etc.), that was passed into xCreate() or xConnect(). In the case of an external module, this function can usually just return SQLITE_OK, unless the table name has significance to some external data mapping. As with any virtual table function that deals with table names, the module needs to properly qualify any SQL operation with a full database and table name, both properly quoted. 232 | Chapter 10: Virtual Tables and Modules The dblist module has a very short xRename() function: static int dblist_rename( sqlite3_vtab *vtab, const char *newname ) { return SQLITE_OK; } The dblist module does not use the table name for anything, so it can safely do nothing. Opening and Closing Table Cursors We will now look at the process of scanning a table and retrieving the rows and column values from the virtual table. This is done by opening a table cursor. The cursor holds all the state data required for the table scan, including SQL statements, file handles, and other data structures. After the cursor is created, it is used to step over each row in the virtual table and extract any required column values. When the module indicates that no more rows are available, the cursor is either reset or released. Virtual table cursors can only move forward through the rows of a table, but they can be reset back to the beginning for a new table scan. A cursor is created using the xOpen() function and released with the xClose() function. Like the vtab structure, it is the responsibility of the module to allocate an sqlite3_vtab_cursor structure and return it back to the SQLite engine. int xOpen( sqlite3_vtab *vtab, sqlite3_vtab_cursor **cursor ) Required. This function must allocate, initialize, and return a cursor. int xClose( sqlite3_vtab_cursor *cursor ) Required. This function must clean up and release the cursor structure. Basically, it should undo everything done by xOpen(). The native sqlite3_vtab_cursor structure is fairly minimal, and looks like this: struct sqlite3_vtab_cursor { sqlite3_vtab *pVtab; /* pointer to table instance */ }; As with the sqlite3_vtab structure, a module is expected to extend this structure with whatever data the module requires. The custom dblist cursor looks like this: typedef struct dblist_cursor_s { sqlite3_vtab_cursor cur; /* this must go first */ sqlite3_stmt *stmt; /* PRAGMA database_list statement */ int eof; /* EOF flag */ } dblist_cursor; For the dblist module, the only cursor-specific data that is needed is an SQLite state- ment pointer and an EOF flag. The flag is used to indicate when the module has reached the end of the PRAGMA database_list output. Simple Example: dblist Module | 233 Outside of allocating the dblist_cursor, the only other task the dblist xOpen() function needs to do is prepare the PRAGMA SQL statement: static int dblist_open( sqlite3_vtab *vtab, sqlite3_vtab_cursor **cur ) { dblist_vtab *v = (dblist_vtab*)vtab; dblist_cursor *c = NULL; int rc = 0; c = sqlite3_malloc( sizeof( dblist_cursor ) ); *cur = (sqlite3_vtab_cursor*)c; if ( c == NULL ) return SQLITE_NOMEM; rc = sqlite3_prepare_v2( v->db, "PRAGMA database_list", -1, &c->stmt, NULL ); if ( rc != SQLITE_OK ) { *cur = NULL; sqlite3_free( c ); return rc; } return SQLITE_OK; } As with xCreate() and xConnect(), no sqlite3_vtab_cursor should be allocated or passed back unless an SQLITE_OK is returned. There is no need to initialize the pVtab field of the cursor—SQLite will take care of that for us. The dblist version of xClose() is very simple. The module must make sure the prepared statement is finalized before releasing the cursor structure: static int dblist_close( sqlite3_vtab_cursor *cur ) { sqlite3_finalize( ((dblist_cursor*)cur)->stmt ); sqlite3_free( cur ); return SQLITE_OK; } You may be wondering why the module puts the statement pointer into the cursor. This requires the module to reprepare the PRAGMA statement for each cursor. Wouldn’t it make more sense to put the statement pointer in the vtab structure? That way it could be prepared only once, and then reused for each cursor. At first, that looks like an attractive option. It would be more efficient and, in most cases, work just fine—right up to the point were SQLite needs to create more than one cursor on the same table instance at the same time. Since the module depends on the statement structure to keep track of the position in the virtual table data (that is, the output of PRAGMA database_list), the module design needs each cursor to have its own statement. The easiest way to do this is simply to prepare and store the statement with the cursor, binding the statement lifetime to the cursor lifetime. 234 | Chapter 10: Virtual Tables and Modules Filtering Rows The xFilter() function works in conjunction with the xBestIndex() function, provid- ing the SQLite query engine a means to communicate any specific constraints or con- ditions put on the query. The xBestIndex() function is used by the query optimizer to ask the module questions about different lookup patterns or limits. Once SQLite decides what to do, the xFilter() function is used to tell the module which plan of action is being taken for this particular scan. int xFilter( sqlite3_vtab_cursor *cursor, int idx_num, const char *idx_str, int argc, sqlite3_value **argv ) Required. This function is used to reset a cursor and initiate a new table scan. SQLite will communicate any constraints that have been placed on the current cursor. The module may choose to skip over any rows that do not meet these constraints. All of the parameters are determined by actions taken by the xBest Index() function. The first row of data must also be fetched. The idea is to allow the module to “pre-filter” as many rows as it can. Each time the SQLite library asks the module to advanced the table cursor to the next row, the module can use the information provided to the xFilter() function to skip over any rows that do not meet the stated criteria for this table scan. The xBestIndex() and xFilter() functions can also work together to specify a specific row ordering. Normally, SQLite makes no assumptions about the order of the rows returned by a virtual table, but xBestIndex() can be used indicate the ability to support one or more specific orderings. If one of those orderings is passed into xFilter(), the table is required to return rows in the specified order. To get any use out of the xFilter() function, a module must also have a fully imple- mented xBestIndex() function. The xBestIndex() function sets up the data that is passed to the xFilter() function. Most of the data passed into xFilter() has no specific meaning to SQLite, it is simply based off code agreements between xBestIndex() and xFilter(). Implementing all this can be quite cumbersome. Thankfully, as with xBestIndex(), it is perfectly valid for a module to ignore the filtering system. If a user query applies a set of conditions on the rows it wants returned from a virtual table, but the module does not filter those out, the SQLite engine will be sure to do it for us. This greatly simplifies the module code, but with the trade-off that any operation against that mod- ule turns into a full table scan. Full table scans may be acceptable for many types of external modules, but if you’re developing a customized index system, you have little choice but to tackle writing robust xBestIndex() and xFilter() functions. To get a better idea on how to do this, see “Best Index and Filter” on page 262. Simple Example: dblist Module | 235 Even if the actual filtering process is ignored, the xFilter() function is still required to do two important tasks. First, it must reset the cursor and prepare it for a new table scan. Second, xFilter() is responsible for fetching the first row of output data. Since the dblist module doesn’t utilize the filtering system, these are pretty much the only things the xFilter() function ends up doing: static int dblist_filter( sqlite3_vtab_cursor *cur, int idxnum, const char *idxstr, int argc, sqlite3_value **value ) { dblist_cursor *c = (dblist_cursor*)cur; int rc = 0; rc = sqlite3_reset( c->stmt ); /* start a new scan */ if ( rc != SQLITE_OK ) return rc; c->eof = 0; /* clear EOF flag */ dblist_get_row( c ); /* fetch first row */ return SQLITE_OK; } Although the dblist module does not utilize the xBestIndex() data, there are still im- portant things to do. The xFilter() function must first reset the statement. This “rewinds” the pragma statement, putting our cursor at the head of the table. There are situations where sqlite3_reset() may be called on a freshly prepared (or freshly reset) statement, but that is not a problem. There are other calling sequences which may require xFilter() to reset the statement. Because both xFilter() and the xNext() function (which we’ll look at next) need to fetch row data, we’ve broken that out into its own function: static int dblist_get_row( dblist_cursor *c ) { int rc; if ( c->eof ) return SQLITE_OK; rc = sqlite3_step( c->stmt ); if ( rc == SQLITE_ROW ) return SQLITE_OK; /* we have a valid row */ sqlite3_reset( c->stmt ); c->eof = 1; return ( rc == SQLITE_DONE ? SQLITE_OK : rc ); /* DONE -> OK */ } The main thing this function does is call sqlite3_step() on the cursor’s SQL statement. If the module gets a valid row of data (SQLITE_ROW), everything is good. If the module gets anything else (including SQLITE_DONE) it considers the scan done. In that case, the module resets the statement before returning SQLITE_OK (if it got to the end of the table) or the error. Although the module could wait for xFilter() to reset the statement or xClose() to finalize it, it is best to reset the statement as soon as we know we’ve reached the end of the available data. 236 | Chapter 10: Virtual Tables and Modules Extracting and Returning Data We now, finally, get to the core of any virtual table implementation. Once the module has a valid cursor, it needs to be able to advanced that cursor over the virtual table data and return column values. This core set of four functions is used to do just that: int xNext( sqlite3_vtab_cursor *cursor ) Required. This function is used to advance the cursor to the next row. When the SQLite engine no longer needs data from the current row, this is called to advance the virtual table scan to the next row. If a virtual table is already on the last row of the table when xNext() is called, it should not return an error. Note that xNext() truly is a “next” function and not a “get row” function. It is not called to fetch the first row of data. The first row of data should be fetched and made available by the xFilter() function. If the module is filtering rows via xBestIndex() and xFilter(), it is legitimate for xNext() to skip over any rows in the virtual table that do not meet the conditions put forth to xFilter(). Additionally, if xBestIndex() indicated an ability to return the data in a specific order, xNext() is obligated to do so. Otherwise, xNext() may return rows in any order it wishes, so long as they are all returned. int xEof( sqlite3_vtab_cursor *cursor ) Required. This function is used to determine if the virtual table has reached the end of the table. Every call to xFilter() and xNext() will immediately be followed by a call to xEof(). If the previous call to xNext() advanced the cursor past the end of the table, xEof() should return a true (nonzero) value, indicating that the end of the table has been reached. If the cursor still points to a valid table row, xEof() should return false (zero). xEof() is also called right after xFilter(). If a table is empty or will return no rows under the conditions defined by xFilter(), then xEof() needs to return true at this time. There is no guarantee xNext() will keep being called until xEof() returns true. The query may decide to terminate the table scan at any time. int xRowid( sqlite3_vtab_cursor *cursor, sqlite_int64 *rowid ) Required. This function is used to retrieve the ROWID value of the current row. The ROWID value should be passed back through the rowid reference provided as the second parameter. int xColumn( sqlite3_vtab_cursor *cursor, sqlite3_context *ctx, int cidx ) Required. This function is used to extract column values from the cursor’s current row. The parameters include the virtual table cursor, an sqlite3_context structure, and a column index. Values should be returned using the sqlite3_context and the sqlite3_result_xxx() functions. The column index is zero-based, so the first col- umn defined in the virtual table definition will have a column index of zero. This function is typically called multiple times between calls to xNext(). Simple Example: dblist Module | 237 The first two functions, xNext() and xEof(), are used to advance a cursor through the virtual table data. A cursor can only be advanced through the data, it cannot be asked to back up, save for a full reset back to the beginning of the table. Unless xBest Index() and xFilter() agreed on a specific filtering or ordering, xNext() is under no obligation to present the data in a specific order. The only requirement is that contin- uous calls to xNext() will eventually visit each row exactly once. At each row, xRowid() and xColumn() can be used to extract values from the current row. xRowid() is used to extract the virtual ROWID value, while xColumn() is used to extract values from all the other columns. While a cursor is at a specific row, the xRowid() and xColumn() functions may be called any number of times in any order. Since the dblist module depends on executing the PRAGMA statement to return data, most of these functions are extremely simple. For example, the dblist xNext() function calls the dblist_get_row() function, which in turn calls sqlite3_step() on the cursor’s statement: static int dblist_next( sqlite3_vtab_cursor *cur ) { return dblist_get_row( (dblist_cursor*)cur ); } The dblist xEof() function returns the cursor EOF flag. This flag is set by dblist_get_row() when the module reaches the end of the PRAGMA database_list data. The flag is simply returned: static int dblist_eof( sqlite3_vtab_cursor *cur ) { return ((dblist_cursor*)cur)->eof; } The data extraction functions for the dblist module are also extremely simple. The dblist module uses the seq column from the PRAGMA database_list output as its virtual ROWID. This means that it can return the value of the seq column as our ROWID. As it happens, the seq column is the first column, so it has an index of zero: static int dblist_rowid( sqlite3_vtab_cursor *cur, sqlite3_int64 *rowid ) { *rowid = sqlite3_column_int64( ((dblist_cursor*)cur)->stmt, 0 ); return SQLITE_OK; } The xColumn() function is nearly as simple. Since there is a one-to-one mapping between the output columns of the PRAGMA statement and the dblist virtual table columns, the module can extract values directly from the PRAGMA output and pass them back as col- umn values for our virtual table: static int dblist_column( sqlite3_vtab_cursor *cur, sqlite3_context *ctx, int cidx ) { dblist_cursor *c = (dblist_cursor*)cur; sqlite3_result_value( ctx, sqlite3_column_value( c->stmt, cidx ) ); 238 | Chapter 10: Virtual Tables and Modules return SQLITE_OK; } In most cases, these functions would be considerably more complex than what the dblist module has here. The fact that the dblist module depends on only a single SQL command to return all of the required data makes the design of these functions quite simple—even more so, since the output of the SQL command exactly matches the data format we need. To get a better idea of what a more typical module might look like, have a look at the implementation of these functions in the other example module (“Advanced Example: weblog Module” on page 246). Virtual Table Modifications As with any other table, modules support the ability to make modifications to a virtual table using the standard INSERT, UPDATE, and DELETE commands. All three operations are supported by the xUpdate() function. This is a table-level function that operates on a table instance, not a cursor. int xUpdate( sqlite3_vtab *vtab, int argc, sqlite3_value **argv, sqlite_int64 *rowid ) Optional. This call is used to support all virtual table modifications. It will be called in response to any INSERT, UPDATE, or DELETE command. The first parameter is the table instance. The second and third parameters pass in a series of database values. The fourth parameter is a reference to a ROWID value, and is used to pass back the newly defined ROWID when a new row is inserted. The argv parameter will have a valid sqlite3_value structure for each argument, although some of those values may have the type SQLITE_NULL. Rows are always inserted or updated as whole sets. Even if the SQL UPDATE command only updates a single col- umn of a row, the xUpdate() command will always be provided with a value for every column in a row. If only a single argument is provided, this is a DELETE request. The sole argument (argv[0]) will be an SQLITE_INTEGER that holds the ROWID of the row that needs to be deleted. In all other cases, exactly n+2 arguments will be provided, where n is the number of columns, including HIDDEN ones (see “Create and Connect” on page 248) in the table definition. The first argument (argv[0]) is used to refer to existing ROWID values, while the second (argv[1]) is used to refer to new ROWID values. These two arguments will be followed by a value for each column in a row, starting with argv[2]. Essentially, the arguments argv[1] through argv[n+1] represent a whole set of row values starting with the implied ROWID column followed by all of the declared columns. Simple Example: dblist Module | 239 If argv[0] has the type SQLITE_NULL, this is an INSERT request. If the INSERT statement provided an explicit ROWID value, that value will be in argv[1] as an SQLITE_INTEGER. The module should verify the ROWID is appropriate and unique before using it. If no explicit ROWID value was given, argv[1] will have a type of SQLITE_NULL. In this case, the module should assign an unused ROWID value and pass it back via the rowid reference pointer in the xUpdate() parameters. If argv[0] has the type SQLITE_INTEGER, this is an UPDATE. In this case, both argv[0] and argv[1] will be SQLITE_INTEGER types with ROWID values. The existing row indicated in argv[0] should be updated with the values supplied in argv[1] through argv[n+1]. In most cases, argv[0] will be the same as argv[1], indicating no change in the ROWID value. However, if the UPDATE statement includes an explicit update to the ROWID column, it may be the case that the first two arguments do not match. In that case, the row indi- cated by argv[0] should have its ROWID value changed to argv[1]. In either case, all the other columns should be updated with the additional arguments. It is the module’s responsibility to enforce any constraints or typing requirements on the incoming data. If the data is invalid or otherwise cannot be inserted or updated into the virtual table, xUpdate() should return an appropriate error, such as SQLITE_CON STRAINT (constraint violation). There may be times when a modification (including a DELETE) happens while an active cursor is positioned at the row in question. The module design must be able to handle this situation. The xUpdate() function is optional. If no implementation is provided, all virtual table instances provided by this module will be read-only. That is the case with our dblist module, so there is no implementation for xUpdate(). Cursor Sequence Most cursor functions have very specific tasks. Some of these, like xEof(), are typically very small, while others, like xNext(), can be quite complex. Regardless of size or com- plexity, they all need to work together to perform the proper tasks and maintain the state of the sqlite3_vtab_cursor structure, including any extensions your module might require. In order to maintain the proper state, it is important to understand which functions can be called, and when. Figure 10-1 provides a sequence map of when cursor functions can be called. Your module needs to be prepared to properly deal with any of these transitions. Some of the call sequences can catch people by surprise, such as having xClose() called before xEof() returns true. This might happen if an SQL query has a LIMIT clause. Also, it is possible for xRowid() to be called multiple times between calls to xNext(). Similarly, xColumn() may be called multiple times with the same column index between calls to 240 | Chapter 10: Virtual Tables and Modules xNext(). It is also possible that neither xRowid() nor xColumn() (nor both) may be called at all between calls to xNext(). In addition to cursor functions, some table-level functions may also be called through- out this sequence. In specific, xUpdate() may be called at any time, possibly altering the row a cursor is currently processing. Generally, this happens by having an update statement open a cursor, find the row it is looking to modify, and then call xUpdate() outside of the cursor context. Figure 10-1. The lifespan of a virtual table cursor. This shows the possible calling sequences for the cursor functions of a virtual table module. It can be tricky to test your module and confirm that everything is working properly. The only advice I can offer is to test your module with a known and relatively small set of data, running it through as many query types as possible. Try to include different variations of GROUP BY, ORDER BY, and any number of join operations (including self- joins). When you’re first starting to write a module, it might also help to put simple printf() or other debug statements at the top of each function. This will assist in understanding the call patterns. Transaction Control Like any other database element, virtual tables are expected to be aware of database transactions and support them appropriately. This is done through four optional func- tions. These functions are table-level functions, not cursor-level functions. int xBegin( sqlite3_vtab *vtab ) Optional. Indicates the start of a transaction involving the virtual table. Any return code other than SQLITE_OK will cause the transaction to fail. int xSync( sqlite3_vtab *vtab ) Optional. Indicates the start of a transactional commit that involves the virtual table. Any return code other than SQLITE_OK will cause the whole transaction to automatically be rolled back. Simple Example: dblist Module | 241 int xCommit( sqlite3_vtab *vtab ) Optional. Indicates the finalization of a transactional commit that involves the virtual table. The return code is ignored—if xSync() succeeded, this function must succeed. int xRollback( sqlite3_vtab *vtab ) Optional. Indicates that a transaction involving the virtual table is being rolled back. The module should revert its state to whatever state it was in prior to the call to xBegin(). The return code is ignored. These functions are optional and are normally only required by external modules that provide write capabilities to external data sources. Internal modules that record their data into standard tables are covered by the existing transaction engine (which will automatically begin, commit, or roll back under the control of the user SQL session). Modules that are limited to read-only functionality do not need transactional control, since they are not making any modifications. Internal modules (modules that store all their data in shadow database tables) do not need to implement transaction control functions. The existing, built-in transaction system will automatically be applied to any changes made to standard database tables. If you do need to support your own transactions, it is important to keep the program flow in mind. xBegin() will always be the first function to be called.* Typically, there will be calls to xUpdate() followed by a two-step sequence of calls to xSync() and xCommit() to close and commit the transaction. Once xBegin() has been called, it is also possible to get a call to xRollback() to roll the transaction back. The xRollback() func- tion can also be called after xSync() (but before xCommit()) if the sync step fails. Full transactions do not nest, and virtual tables do not support save-points. Once a call to xBegin() has been made, you will not get another one until either xCommit() or xRollback() has been called. In keeping with the ACID properties of a transaction, any modifications made between calls to xBegin() and xSync() should only be visible to this virtual table instance in this database connection (that is, this vtab structure). This can be done by delaying any writes or modifications to external data sources until the xSync() function is called, or by somehow locking the data source to ensure other module instances (or other appli- cations) cannot modify or access the data. If the data is written out in xSync(), the data source still needs to be locked until a call to xCommit() or xRollback() is made. If xSync() returns SQLITE_OK, it is assumed that any call to xCommit() will succeed, so you want to try to make your modifications in xSync() and verify and release them in xCommit(). * In theory. Currently, calls are made directly to xSync() and xCommit() following the call to xCreate(). It isn’t clear if this is considered a bug or not, so this behavior may change in future versions of SQLite. 242 | Chapter 10: Virtual Tables and Modules Proper transactional control is extremely hard, and making your transactions fully atomic, consistent, isolated, and durable is no small task. Most external modules that attempt to implement transactions do so by locking the external data source. You still need to support some type of rollback ability, but exclusive access eliminates any iso- lation concerns. Since the dblist module is read-only, it does not need to provide any transactional functions. Register the Module Now that we’ve had a look at all the functions required to define a module, we need to register them. As we’ve already seen, this is done with the sqlite3_create_ module() function. To register the module, we need to fill out an sqlite3_module struc- ture and pass that to the create function. You may have noticed that all of our module functions were marked static. This is because the module was written as an extension (see the section “SQLite Exten- sions” on page 204). By structuring the code that way, we can easily build our virtual table module into an application, or we can create a dynamic extension. Here is the initialization function for our extension: static sqlite3_module dblist_mod = { 1, /* iVersion */ dblist_connect, /* xCreate() */ dblist_connect, /* xConnect() */ dblist_bestindex, /* xBestIndex() */ dblist_disconnect, /* xDisconnect() */ dblist_disconnect, /* xDestroy() */ dblist_open, /* xOpen() */ dblist_close, /* xClose() */ dblist_filter, /* xFilter() */ dblist_next, /* xNext() */ dblist_eof, /* xEof() */ dblist_column, /* xColumn() */ dblist_rowid, /* xRowid() */ NULL, /* xUpdate() */ NULL, /* xBegin() */ NULL, /* xSync() */ NULL, /* xCommit() */ NULL, /* xRollback() */ NULL, /* xFindFunction() */ dblist_rename /* xRename() */ }; int dblist_init( sqlite3 *db, char **error, const sqlite3_api_routines *api ) { int rc; SQLITE_EXTENSION_INIT2(api); Simple Example: dblist Module | 243 /* register module */ rc = sqlite3_create_module( db, "dblist", &dblist_mod, NULL ); if ( rc != SQLITE_OK ) { return rc; } /* automatically create an instance of the virtual table */ rc = sqlite3_exec( db, "CREATE VIRTUAL TABLE temp.sql_database_list USING dblist", NULL, NULL, NULL ); return rc; } The most important thing to notice is that the sqlite3_module structure is given a static allocation. The SQLite library does not make a copy of this structure when the module is registered, so the sqlite3_module structure must remain valid for the duration of the database connection. In this case, we use a file-level global that is statically initialized with all the correct values. The extension entry point function is a bit unique, in that it not only defines the module, but it also goes ahead and creates an instance of a dblist virtual table. Normally, an extension initialization function wouldn’t (and shouldn’t) do something like this, but in this case it makes sense. Like any other table, virtual tables are typically bound to a specific database. But the active database list we get from the PRAGMA database_list command is a function of the current state of the database connection (and all attached databases), and isn’t really specific to a single database. If you were to create a dblist table in multiple databases that were all attached to the same database connection, they would all return the same data. It is the database connection (and not a specific data- base) that is the real source of the data. So, in the somewhat unique case of the dblist module, we only need one instance of the virtual table per database connection. Ideally, it would always be there, no matter which databases are attached. It would also be best if the table wasn’t “left behind” in a database file after that database was closed or detached. Not only would this tie the database file to our module, it is also unnecessary since a dblist table instance doesn’t have any state beyond the database connection. To meet all these needs, the module goes ahead and just makes a single instance of the table in the temporary database. Every database connection has a temporary database, and it is always named temp. This makes the table instance easy to find. Creating it in the temp database also keeps the table instance out of any “real” database files, and ties the lifetime of the table to the lifetime of the database connection. All in all, it is a perfect, though somewhat unusual, fit for this specific module. The end result is that if you load the dblist extension, it will not only register the dblist module, it will also create an instance of the dblist virtual table at temp.sql_data base_list. System tables in SQLite have the prefix sqlite_, but those names are re- served and the extension cannot create a table with that prefix. The name sql_data base_list gets the idea across, however. 244 | Chapter 10: Virtual Tables and Modules Example Usage After all that work, what do we get? First, lets have a look at what the PRAGMA data base_list does by itself. Here is some example output: sqlite> PRAGMA database_list; seq name file ---------- ---------- ----------------------------- 0 main /Users/jak/sqlite/db1.sqlite3 1 temp 2 memory 3 two /Users/jak/sqlite/db2.sqlite3 In this case, I ran the sqlite3 utility with the file db1.sqlite3, created an empty tem- porary table (so the temp database shows up), attached an in-memory database as memory, and finally attached a second database file as two. Now let’s load our module extension and see what we get: sqlite> .load dblist.sqlite3ext dblist_init sqlite> SELECT * FROM sqlite3_database_list; seq name file ---------- ---------- ----------------------------- 0 main /Users/jak/sqlite/db1.sqlite3 1 temp 2 memory 3 two /Users/jak/sqlite/db2.sqlite3 And we get…the exact same thing! Actually, that’s a good thing—that was the whole point. The key thing is that, unlike the PRAGMA command, we can do this: sqlite> SELECT * FROM sqlite3_database_list WHERE file == ''; seq name file ---------- ---------- ----------------------------- 1 temp 2 memory This shows us all of the databases that have no associated filename. You’ll note that PRAGMA database_list (and hence the dblist module) returns an empty string, and not a NULL, for a database that does not have an associated database file. Perhaps most useful, we can also make queries like this to figure out what the logical database name is for a particular database file: sqlite> SELECT name FROM sqlite3_database_list ...> WHERE file LIKE '%db2.sqlite3'; name ---------- two I’d be the first to admit this isn’t exactly ground-breaking work. Parsing the direct output of PRAGMA database_list isn’t that big of a deal—for a program or for a human. The main point of this example wasn’t to show the full power of virtual tables, but to give us a problem to work with where we could focus on the functions and interface Simple Example: dblist Module | 245 required by the virtual table system, rather than focusing on the complex code required to implement it. Now that you’ve seen the module interface and have a basic idea of how things work, we’re going to shift our focus to something a bit more practical, and a bit more complex. Advanced Example: weblog Module Now that we’ve seen a very simple example of a virtual table module, you should have some idea of how they work. Although our dblist module was a good introduction to how virtual tables operate, it isn’t a very representative example. To provide a more advanced and realistic example, we’re going to look at a second example module. This module is known as weblog, and is designed to parse Apache httpd server logs and present them to the database engine as a virtual table. It will parse the default Apache combine or common logfile formats, or any other logfile that matches this format. Apache logfiles are cross-platform and reasonably common. Many people have access to logfiles with a decent amount of interesting data, allowing this example to be a bit more hands-on. Be warned that some of the code explanations will be a bit brief. Although the functions are larger, much of the code involves rather basic housekeeping-type tasks, such as string scanning. Rather than focus on these parts, most of the descriptions will focus on how the code interacts with the SQLite library. Many of the housekeeping details will be, as they say, left as an exercise for the reader. The weblog module is designed as an external read-only module. The module gets all of its data directly from a web server logfile, making it dependent on external resources to provide data. The module does not let you modify those data sources, however. A weblog virtual table would be created with an SQL command like this: CREATE VIRTUAL TABLE current USING weblog( /var/log/httpd/access.log ); Notice that the filename has no single quotes and is not a string literal. Table parameters include everything between the commas (of which we have none, since there is only one argument), so if you need to reference a file with spaces, you can do something like this: CREATE VIRTUAL TABLE log USING weblog( /var/log/httpd/access log file.txt ); The first example will create a table instance current, and bind it to the data found in the /var/log/httpd/access.log file. The second example will bind the SQL table log to the file access log file.txt in the same directory. Briefly, the Apache common log format contains seven fields. The first field is the IP address of the client. In rare situations this might be a hostname, but most servers are configured to simply record the IP address in dot-decimal format. The second field is a legacy ident field. Most web servers do not support this, and record only a single dash. 246 | Chapter 10: Virtual Tables and Modules The third field records the username, if given. If not given, this field is also recorded as a single dash. The fourth field is a timestamp, surrounded by square brackets ( [ ] ). The fifth is the first line of the HTTP request, in double quotes. This contains the HTTP operation (such as GET or POST) as well as the URL. In the sixth column is the HTTP result code (e.g., 200 for OK, 404 for missing resource), with the number of payload bytes returned in the seventh field. The combine file format adds two more fields. The eighth field is the referrer header, which contains a URL. The ninth field is the user-agent header, also in double quotes. Count Logfile field Meaning 1 Client Address IP or hostname of HTTP client 2 Ident Legacy field, not used 3 Username Client-provided username 4 Timestamp Time of transaction 5 HTTP Request HTTP operation and URL 6 Result Code Result status of HTTP request 7 Bytes Payload bytes 8 Referrer URL of referrer page 9 User Agent Client software identifier The weblog module is designed to read the combine file format. However, if given a common logfile that lacks the last two fields, these extra fields will simply be NULL. Although the logfile has seven or nine columns, the weblog virtual table will have more than nine columns. The virtual table adds a number of additional columns that present the same data in different ways. For example, the IP address will be returned in one column as a text value that holds the traditional dotted notation. Another column will provide a raw integer represen- tation. The text column is easier for humans to understand, but the integer column allows for faster searches, especially over ranges. The underlying data is the same: the two columns just return the data in different formats. Similarly, the timestamp column can return the string value from the logfile, or it can return separate integer values for the year, month, day, etc. If this were a fully supported SQLite extension, it would likely include more than just the weblog module. Ideally, it would also include a number of utility functions, such as a function that converted text values containing dot-decimal IP addresses to and from integer values. (Then again, if this were a fully supported module, it would include decent error messages and other polish that this example lacks. I’m trying to keep the line counts as small as possible.) Some of these functions would reduce the need for extra columns, since you could just convert the data using SQL, but there are still times when having the extra columns is extremely useful. Advanced Example: weblog Module | 247 Create and Connect Since the weblog module is an external module, there isn’t any data to initialize. This means that, like the dblist, we can use the same function for both xCreate() and xConnect(). Before we get into the function, let’s have a quick look at our augmented vtab structure. Since this module does not use the table name for anything, the only data we need to keep around is the logfile filename: typedef struct weblog_vtab_s { sqlite3_vtab vtab; char *filename; } weblog_vtab; The weblog create/connect function is a bit longer than the dblist version, but still fairly easy to follow. First, it verifies that we have exactly four arguments. Remember that the first three arguments are always the module name, the database name, and the table name. The fourth argument is the first user-provided argument, which in this case is the log filename. The function tries to open that file for read-only access, just to verify the file is there and can be opened it for reading. This test isn’t foolproof, but it is a nice check. The module then allocates the vtab structure, stashes a copy of the filename, and declares the table definition: static int weblog_connect( sqlite3 *db, void *udp, int argc, const char *const *argv, sqlite3_vtab **vtab, char **errmsg ) { weblog_vtab *v = NULL; const char *filename = argv[3]; FILE *ftest; if ( argc != 4 ) return SQLITE_ERROR; *vtab = NULL; *errmsg = NULL; /* test to see if filename is valid */ ftest = fopen( filename, "r" ); if ( ftest == NULL ) return SQLITE_ERROR; fclose( ftest ); /* allocate structure and set data */ v = sqlite3_malloc( sizeof( weblog_vtab ) ); if ( v == NULL ) return SQLITE_NOMEM; ((sqlite3_vtab*)v)->zErrMsg = NULL; /* need to init this */ v->filename = sqlite3_mprintf( "%s", filename ); if ( v->filename == NULL ) { sqlite3_free( v ); return SQLITE_NOMEM; } v->db = db; 248 | Chapter 10: Virtual Tables and Modules sqlite3_declare_vtab( db, weblog_sql ); *vtab = (sqlite3_vtab*)v; return SQLITE_OK; } The table definition contains 20 columns total. The first 9 map directly to the fields within the logfile, while the extra 11 columns provide different representations of the same data. The last column represents the whole line of the logfile, without modifications: const static char *weblog_sql = " CREATE TABLE weblog ( " " ip_str TEXT, " /* 0 */ " login TEXT HIDDEN, " /* 1 */ " user TEXT, " /* 2 */ " time_str TEXT, " /* 3 */ " req TEXT, " /* 4 */ " result INTEGER, " /* 5 */ " bytes INTEGER, " /* 6 */ " ref TEXT, " /* 7 */ " agent TEXT, " /* 8 */ #define TABLE_COLS_SCAN 9 " ip_int INTEGER, " /* 9 */ " time_day INTEGER, " /* 10 */ " time_mon_s TEXT, " /* 11 */ " time_mon INTEGER, " /* 12 */ " time_year INTEGER, " /* 13 */ " time_hour INTEGER, " /* 14 */ " time_min INTEGER, " /* 15 */ " time_sec INTEGER, " /* 16 */ " req_op TEXT, " /* 17 */ " req_url TEXT, " /* 18 */ " line TEXT HIDDEN " /* 19 */ " ); "; #define TABLE_COLS 20 You may have noticed a few of the columns have the keyword HIDDEN. This keyword is only valid for virtual table definitions. Any column marked HIDDEN will not be returned by SELECT * FROM... style queries. You can explicitly request the column, but it is not returned by default. This is very similar in behavior to the ROWID column found in standard tables. In our case, we’ve marked the login and line columns as HIDDEN. The login column almost never contains valid data, while the line column is redundant (and large). The columns are there if you need them, but in most cases people aren’t interested in seeing them. To keep the general output cleaner, I’ve chosen to hide them. Disconnect and Destroy As with xConnect() and xCreate(), the weblog xDisconnect() and xDestroy() functions share the same implementation: Advanced Example: weblog Module | 249 static int weblog_disconnect( sqlite3_vtab *vtab ) { sqlite3_free( ((weblog_vtab*)vtab)->filename ); sqlite3_free( vtab ); return SQLITE_OK; } Free up the memory used for the filename, free up the memory used by the vtab struc- ture, and return. Simple and easy. Other Table Functions The last set of table-level functions includes xBestIndex(), xFindFunction(), xRename(), and xUpdate(), as well as the four transactional functions, xBegin(), xSync(), xCommit(), and xRollback(). The xFindFunction() is optional, and the weblog module has no use for it, so there is no implementation of this function. Since this is a read- only module, same is true of xUpdate(). Similarly, the transactional functions are also optional and not required for read-only modules. For table-level functions, that leaves only xRename() and xBestIndex(). The xRename() function is required, but since the module makes no use of the virtual table instance name, it is basically a no-op: static int weblog_rename( sqlite3_vtab *vtab, const char *newname ) { return SQLITE_OK; } In the case of the weblog module, once you set the name of the external logfile when creating a virtual table, there is no way to alter it, other than dropping and re-creating the table. The last function, xBestIndex(), is required, but it isn’t actually returning any useful data: static int weblog_bestindex( sqlite3_vtab *vtab, sqlite3_index_info *info ) { return SQLITE_OK; } Since the module has no indexing system, it can’t offer any optimized search patterns. The logfile is always scanned start to finish anyway, so every query is a full table scan. Open and Close We can now move on to the cursor functions. The first thing to look at is the weblog cursor structure. The weblog cursor is a bit more complex than the dblist example, as it needs to read and scan the data values from the logfile. 250 | Chapter 10: Virtual Tables and Modules There are three basic sections to this structure. The first is the base sqlite3_vtab_ cursor structure. As always, this must come first, and must be a full instance of the structure: #define LINESIZE 4096 typedef struct weblog_cursor_s { sqlite3_vtab_cursor cur; /* this must be first */ FILE *fptr; /* used to scan file */ sqlite_int64 row; /* current row count (ROWID) */ int eof; /* EOF flag */ /* per-line info */ char line[LINESIZE]; /* line buffer */ int line_len; /* length of data in buffer */ int line_ptrs_valid; /* flag for scan data */ char *(line_ptrs[TABLE_COLS]); /* array of pointers */ int line_size[TABLE_COLS]; /* length of data for each pointer */ } weblog_cursor; The second block deals with the data we need to scan the logfile. The weblog module uses the standard C library f functions (such as fopen()) to open and scan the logfile. Each weblog cursor needs a unique FILE pointer, just as each dblist cursor required a unique statement structure. The module uses the FILE structure to keep track of its location within the file, so each cursor needs its own unique FILE structure. The cursor needs to keep track of the number of lines it has read from the file, as this value is used as the ROWID. Finally, the cursor needs an EOF flag to indicate when it has reached the end of the file. Having a unique FILE pointer for each cursor means the module needs to reopen the file for each table scan. In the case of the weblog module, this is actually an advantage, as each table scan will reassociate itself with the correct file. This can be important in a web server environment, where logfiles may roll frequently. The third section of the weblog_cursor structure holds everything the cursor needs to know about the current line. The cursor has a buffer to hold the text and length of the current line. There are also a series of pointers and length counters that are used to scan the line. Since scanning the line is fairly expensive, and must be done all at once, the module delays scanning the line until it’s sure the data is needed. Once scanned, the module will keep the scan data around until it reads a new line. To keep track of when a line has been scanned, the cursor contains a “valid” flag. As we go through the rest of the module functions, you’ll see how these fields are used. You might be thinking that a 4 KB line buffer seems a bit large, but frequently it is not enough. CGI scripts that use extensive query strings can generate very long logfile lines. Another issue is that many referrer URLs, especially those from search engines, can be extremely large. While most lines are only a hundred characters or so, it is best if the Advanced Example: weblog Module | 251 module can try to deal with the longer ones as well. Even with a 4 KB buffer, you’ll need to properly deal with potential buffer overflows. Now that we’ve seen what the cursor looks like, let’s have a look at how it is opened and created. When the module needs to create a new cursor, it will first attempt to open the correct logfile. Assuming that succeeds, it will allocate the cursor structure and initialize the basic data: static int weblog_open( sqlite3_vtab *vtab, sqlite3_vtab_cursor **cur ) { weblog_vtab *v = (weblog_vtab*)vtab; weblog_cursor *c; FILE *fptr; *cur = NULL; fptr = fopen( v->filename, "r" ); if ( fptr == NULL ) return SQLITE_ERROR; c = sqlite3_malloc( sizeof( weblog_cursor ) ); if ( c == NULL ) { fclose( fptr ); return SQLITE_NOMEM; } c->fptr = fptr; *cur = (sqlite3_vtab_cursor*)c; return SQLITE_OK; } The open function doesn’t need to initialize the line data, as this will all be reset when we read the first line from the data file. The xClose() function is relatively simple: static int weblog_close( sqlite3_vtab_cursor *cur ) { if ( ((weblog_cursor*)cur)->fptr != NULL ) { fclose( ((weblog_cursor*)cur)->fptr ); } sqlite3_free( cur ); return SQLITE_OK; } Close the file, release the memory. Filter Since the weblog module chooses to ignore the xBestIndex() function, it largely ignores xFilter() as well. The file is reset to the beginning, just to be sure, and the module reads the first line of data: 252 | Chapter 10: Virtual Tables and Modules static int weblog_filter( sqlite3_vtab_cursor *cur, int idxnum, const char *idxstr, int argc, sqlite3_value **value ) { weblog_cursor *c = (weblog_cursor*)cur; fseek( c->fptr, 0, SEEK_SET ); c->row = 0; c->eof = 0; return weblog_get_line( (weblog_cursor*)cur ); } The weblog_get_line() function reads in a single line from the logfile and copies it into our line buffer. It also verifies that it got a full line. If it didn’t get a full line, the function keeps reading (but discards the input) to make sure the file location is left at the be- ginning of the next valid line. We can reduce how often this happens by making the line buffer bigger, but no matter how big we make the buffer, it is always a good idea to make sure a whole line is consumed, even if the tail is discarded: static int weblog_get_line( weblog_cursor *c ) { char *cptr; int rc = SQLITE_OK; c->row++; /* advance row (line) counter */ c->line_ptrs_valid = 0; /* reset scan flag */ cptr = fgets( c->line, LINESIZE, c->fptr ); if ( cptr == NULL ) { /* found the end of the file/error */ if ( feof( c->fptr ) ) { c->eof = 1; } else { rc = -1; } return rc; } /* find end of buffer and make sure it is the end a line... */ cptr = c->line + strlen( c->line ) - 1; /* find end of string */ if ( ( *cptr != '\n' )&&( *cptr != '\r' ) ) { /* overflow? */ char buf[1024], *bufptr; /* ... if so, keep reading */ while ( 1 ) { bufptr = fgets( buf, sizeof( buf ), c->fptr ); if ( bufptr == NULL ) { /* found the end of the file/error */ if ( feof( c->fptr ) ) { c->eof = 1; } else { rc = -1; } break; } bufptr = &buf[ strlen( buf ) - 1 ]; if ( ( *bufptr == '\n' )||( *bufptr == '\r' ) ) { break; /* found the end of this line */ Advanced Example: weblog Module | 253 } } } while ( ( *cptr == '\n' )||( *cptr == '\r' ) ) { *cptr-- = '\0'; /* trim new line characters off end of line */ } c->line_len = ( cptr - c->line ) + 1; return rc; } Besides reading a full line, this function also resets the scan flag (to indicate the line buffer has not had the individual fields scanned) and adds one (1) to the line count. At the end, the function also trims off any trailing newline or carriage return characters. Rows and Columns We only have a few functions left. In specific, the module only needs to define the two row-handling functions, xNext() and xEof(). We also need the two column functions, xRowid() and xColumn(). Three of these four functions are quite simple. The xNext() function can call weblog_get_line(), just as the xFilter() function did. The xEof() and xRowid() func- tions return or pass back values that have already been calculated elsewhere: static int weblog_next( sqlite3_vtab_cursor *cur ) { return weblog_get_line( (weblog_cursor*)cur ); } static int weblog_eof( sqlite3_vtab_cursor *cur ) { return ((weblog_cursor*)cur)->eof; } static int weblog_rowid( sqlite3_vtab_cursor *cur, sqlite3_int64 *rowid ) { *rowid = ((weblog_cursor*)cur)->row; return SQLITE_OK; } The interesting function is the xColumn() function. If you’ll recall, in addition to the line buffer, the weblog_cursor structure also had an array of character pointers and length values. Each of these pointers and lengths corresponds to a column value in the defined table format. Before the module can extract those values, it needs to scan the input line and mark all the columns by setting the pointer and length values. Using a length value means the module doesn’t need to insert termination characters into the original string buffer. That’s good, since several of the fields overlap. Using terminating characters would require making private copies of these data fields. In the end, a length value is quite useful anyway, as most of SQLite’s value-handling routines utilize length values. 254 | Chapter 10: Virtual Tables and Modules The function that sets up all these pointers and length calculations is weblog_scan line(). We’ll work our way through this section by section. At the top are, of course, the variable definitions. The start and end pointers will be used to scan the line buffer, while the next value keeps track of the terminating character for the current field: static int weblog_scanline( weblog_cursor *c ) { char *start = c->line, *end = NULL, next = ' '; int i; /* clear pointers */ for ( i = 0; i < TABLE_COLS; i++ ) { c->line_ptrs[i] = NULL; c->line_size[i] = -1; } With the variables declared, the first order of business is to reset all of the column pointers and sizes. Next, the scan function loops over the native data fields in the line. This scans up to nine fields from the line buffer. These fields correspond to all the primary fields in a combine format logfile. If the logfile is a common format file (with only seven fields) or if the line buffer was clipped off, fewer fields are scanned. Any fields that are not prop- erly scanned will eventually end up returning NULL SQL values: /* process actual fields */ for ( i = 0; i < TABLE_COLS_SCAN; i++ ) { next = ' '; while ( *start == ' ' ) start++; /* trim whitespace */ if (*start == '\0' ) break; /* found the end */ if (*start == '"' ) { next = '"'; /* if we started with a quote, end with one */ start++; } else if (*start == '[' ) { next = ']'; /* if we started with a bracket, end with one */ start++; } end = strchr( start, next ); /* find end of this field */ if ( end == NULL ) { /* found the end of the line */ int len = strlen ( start ); end = start + len; /* end now points to '\0' */ } c->line_ptrs[i] = start; /* record start */ c->line_size[i] = end - start; /* record length */ while ( ( *end != ' ' )&&( *end != '\0' ) ) end++; /* find end */ start = end; } This loop attempts to scan one field at a time. The first half of the loop figures out the ending character of the field. In most cases it is a space, but it can also be a double- quote or square bracket. Once it knows what it’s looking for, the string is scanned for the next end marker. If the marker isn’t found, the rest of the string is used. Advanced Example: weblog Module | 255 When this loop exits, the code has attempted to set up the first nine column pointers. These make up the native fields of the logfile. The next step is to set up pointers and lengths for the additional 11 columns that represent subfields and alternate represen- tations. The first additional value is the IP address, returned as an integer. This function doesn’t do data conversions, so a direct copy of pointer and length from the first column can be made: /* process special fields */ /* ip_int - just copy */ c->line_ptrs[9] = c->line_ptrs[0]; c->line_size[9] = c->line_size[0]; Next, all of the date field pointers and lengths are set up. This section of code makes some blatant assumptions about the format of the timestamp, but there isn’t much choice. The code could scan the individual fields, but it would still be forced to make assumptions about the ordering of the fields. In the end, it is easiest to just assume the format is consistent and hardcode the field lengths. This example ignores the time zone information: /* assumes: "DD/MMM/YYYY:HH:MM:SS zone" */ /* idx: 012345678901234567890... */ if (( c->line_ptrs[3] != NULL )&&( c->line_size[3] >= 20 )) { start = c->line_ptrs[3]; c->line_ptrs[10] = &start[0]; c->line_size[10] = 2; c->line_ptrs[11] = &start[3]; c->line_size[11] = 3; c->line_ptrs[12] = &start[3]; c->line_size[12] = 3; c->line_ptrs[13] = &start[7]; c->line_size[13] = 4; c->line_ptrs[14] = &start[12]; c->line_size[14] = 2; c->line_ptrs[15] = &start[15]; c->line_size[15] = 2; c->line_ptrs[16] = &start[18]; c->line_size[16] = 2; } After the date fields, the next step is to extract the HTTP operation and URL. These are extracted as the first two subfields of the HTTP Request log field. The code plays some games to be sure it doesn’t accidentally pass a NULL pointer into strchr(), but otherwise it just finds the first two spaces and considers those to be the ending of the two fields it is trying to extract: /* req_op, req_url */ start = c->line_ptrs[4]; end = ( start == NULL ? NULL : strchr( start, ' ' ) ); if ( end != NULL ) { c->line_ptrs[17] = start; c->line_size[17] = end - start; start = end + 1; } end = ( start == NULL ? NULL : strchr( start, ' ' ) ); if ( end != NULL ) { c->line_ptrs[18] = start; c->line_size[18] = end - start; } 256 | Chapter 10: Virtual Tables and Modules The final column represents the full contents of the line buffer. We also need to set the valid flag to indicate the field pointers are valid and ready for use: /* line */ c->line_ptrs[19] = c->line; c->line_size[19] = c->line_len; c->line_ptrs_valid = 1; return SQLITE_OK; } Once this function has been called, all the fields that could be scanned will have a valid pointer and length value. With the data scanned, this and subsequent calls to xCol umn() can use the relevant values to pass back their database values. Let’s return to looking at xColumn(). The first thing the xColumn() code does is making sure the line has already been scanned. If not, the code calls weblog_scanline() to set up all the field pointers: static int weblog_column( sqlite3_vtab_cursor *cur, sqlite3_context *ctx, int cidx ) { weblog_cursor *c = (weblog_cursor*)cur; if ( c->line_ptrs_valid == 0 ) { weblog_scanline( c ); /* scan line, if required */ } if ( c->line_size[cidx] < 0 ) { /* field not scanned and set */ sqlite3_result_null( ctx ); return SQLITE_OK; } Next, if the requested column doesn’t have a valid set of values, the module passes back an SQL NULL for the column. The code then processes columns with specific conversion needs. Any column that needs special processing or conversion will be caught by this switch statement. The first specialized column is the integer version of the IP address. This block of code converts each octet of the IP address into an integer value. The only issue is that all integer values within SQLite are signed, so the code needs to be careful about con- structing the value into a 64-bit integer. For maximum compatibility, it avoids using shift operations: switch( cidx ) { case 9: { /* convert IP address string to signed 64 bit integer */ int i; sqlite_int64 v = 0; char *start = c->line_ptrs[cidx], *end, *oct[4]; for ( i = 0; i < 4; i++ ) { oct[i] = start; end = ( start == NULL ? NULL : strchr( start, '.' ) ); if ( end != NULL ) { start = end + 1; } Advanced Example: weblog Module | 257 } v += ( oct[3] == NULL ? 0 : atoi( oct[3] ) ); v *= 256; v += ( oct[2] == NULL ? 0 : atoi( oct[2] ) ); v *= 256; v += ( oct[1] == NULL ? 0 : atoi( oct[1] ) ); v *= 256; v += ( oct[0] == NULL ? 0 : atoi( oct[0] ) ); sqlite3_result_int64( ctx, v ); return SQLITE_OK; } The next specialized column is one of the two month fields. In the logfile, the month value is given as a three-character abbreviation. One column returns this original text value, while another returns a numeric value. To convert from the abbreviation to the numeric value, the code simply looks for constants in the month string. If it can’t find a match, the code breaks out. As we’ll see, if the code breaks out it will eventually end up returning the text value: case 12: { int m = 0; if ( strncmp( c->line_ptrs[cidx], "Jan", 3 ) == 0 ) m = 1; else if ( strncmp( c->line_ptrs[cidx], "Feb", 3 ) == 0 ) m = 2; else if ( strncmp( c->line_ptrs[cidx], "Mar", 3 ) == 0 ) m = 3; else if ( strncmp( c->line_ptrs[cidx], "Apr", 3 ) == 0 ) m = 4; else if ( strncmp( c->line_ptrs[cidx], "May", 3 ) == 0 ) m = 5; else if ( strncmp( c->line_ptrs[cidx], "Jun", 3 ) == 0 ) m = 6; else if ( strncmp( c->line_ptrs[cidx], "Jul", 3 ) == 0 ) m = 7; else if ( strncmp( c->line_ptrs[cidx], "Aug", 3 ) == 0 ) m = 8; else if ( strncmp( c->line_ptrs[cidx], "Sep", 3 ) == 0 ) m = 9; else if ( strncmp( c->line_ptrs[cidx], "Oct", 3 ) == 0 ) m = 10; else if ( strncmp( c->line_ptrs[cidx], "Nov", 3 ) == 0 ) m = 11; else if ( strncmp( c->line_ptrs[cidx], "Dec", 3 ) == 0 ) m = 12; else break; /* give up, return text */ sqlite3_result_int( ctx, m ); return SQLITE_OK; } There are a number of additional columns (including some of the “native” ones) that are returned as integers. None of these columns require special processing, other than the string-to-integer conversion. The standard atoi() function is used for this conver- sion. Although the string pointers are not null-terminated, the atoi() function will automatically return once it encounters a non-numeric character. Since all of these fields are bound by spaces or other characters, this works out exactly the way we want: case 5: /* result code */ case 6: /* bytes transfered */ case 10: /* day-of-month */ case 13: /* year */ case 14: /* hour */ case 15: /* minute */ case 16: /* second */ sqlite3_result_int( ctx, atoi( c->line_ptrs[cidx] ) ); return SQLITE_OK; default: break; } 258 | Chapter 10: Virtual Tables and Modules sqlite3_result_text( ctx, c->line_ptrs[cidx], c->line_size[cidx], SQLITE_STATIC ); return SQLITE_OK; } Finally, any field that did not require special processing is returned as a text value. Although the line buffer will be overwritten when the next line is read, the data pointer passed into sqlite3_result_text() only needs to stay valid until the next call to xNext(). This allows the module to use the SQLITE_STATIC flag. With that, we’ve defined all the required functions for our weblog module. Register the Module Now that we’ve seen how all the module functions are implemented, the last thing to do is register the weblog module as part of the extension initialization function: static sqlite3_module weblog_mod = { 1, /* iVersion */ weblog_connect, /* xCreate() */ weblog_connect, /* xConnect() */ weblog_bestindex, /* xBestIndex() */ weblog_disconnect, /* xDisconnect() */ weblog_disconnect, /* xDestroy() */ weblog_open, /* xOpen() */ weblog_close, /* xClose() */ weblog_filter, /* xFilter() */ weblog_next, /* xNext() */ weblog_eof, /* xEof() */ weblog_column, /* xColumn() */ weblog_rowid, /* xRowid() */ NULL, /* xUpdate() */ NULL, /* xBegin() */ NULL, /* xSync() */ NULL, /* xCommit() */ NULL, /* xRollback() */ NULL, /* xFindFunction() */ weblog_rename /* xRename() */ }; int weblog_init( sqlite3 *db, char **error, const sqlite3_api_routines *api ) { SQLITE_EXTENSION_INIT2(api); return sqlite3_create_module( db, "weblog", &weblog_mod, NULL ); } Since there is no attempt to create an instance of a weblog table, this initialization function is a bit simpler than the previous dblist example. Example Usage Now that we’ve worked through the whole example, let’s see what the code can do. Here are a few different examples that show off the power of the weblog module. Advanced Example: weblog Module | 259 While doing these types of queries is not a big deal for people that are comfortable with SQL, realize that we can run all of these queries without having to first import the logfile data. Not only does that make the whole end-to-end process much faster, it means we can run these types of queries against active, “up to the second” logfiles. To show off how this module works, the server administrators of http://oreilly.com/ were nice enough to provide me with some of their logfiles. The file referred to as oreilly.com_access.log is an Apache combine logfile with 100,000 lines of data. Once compiled and built into a loadable module, we can import the weblog module and create a virtual table that is bound to this file using these commands: sqlite> .load weblog.sqlite3ext weblog_init sqlite> CREATE VIRTUAL TABLE log USING weblog( oreilly.com_access.log ); We then issue queries to look at different aspects of the file. For example, if we want to know what the most common URL is, we run a query like this: sqlite> SELECT count(*) AS Count, req_url AS URL FROM log ...> GROUP BY 2 ORDER BY 1 DESC LIMIT 8; Count URL ----- ---------------------------------------- 2490 /images/oreilly/button_cart.gif 2480 /images/oreilly/button_acct.gif 2442 /styles/all.css 2348 /images/oreilly/888-line.gif 2233 /styles/chrome.css 2206 /favicon.ico 1975 /styles/home2.css 1941 /images/oreilly/satisfaction-icons.gif It is fairly common to see favicon.ico very near the top, along with any site-wide CSS and image files. In the case of smaller sites that have a lot less traffic, it isn’t uncommon for the most requested URL to be /robots.txt, which is used by search engines. We can also see what the most expensive items on the website are, in terms of bytes moved: sqlite> SELECT sum(bytes) AS Bytes, count(*) AS Count, req_url AS URL ...> FROM log WHERE result = 200 GROUP BY 3 ORDER BY 1 DESC LIMIT 8; Bytes Count URL -------- ----- --------------------------------------------- 46502163 1137 /images/oreilly/mac_os_x_snow_leopard-148.jpg 40780252 695 / 37171328 2384 /styles/all.css 35403200 2180 /styles/chrome.css 31728906 781 /catalog/assets/pwr/engine/js/full.js 31180460 494 /catalog/9780596510046/index.html 21573756 88 /windows/archive/PearPC.html 21560154 3 /catalog/dphotohdbk/chapter/ch03.pdf 260 | Chapter 10: Virtual Tables and Modules We see that some of these items are not that large, but are requested frequently. Other items have only a small number of requests, but are big enough to make a noticeable contribution to the total number of served bytes. Here is one final example. This shows what IP addresses are downloading the most number of unique items. Since this is from live data, I’ve altered the IP addresses: sqlite> SELECT count(*) AS Uniq, sum(sub_count) AS Ttl, ...> sum(sub_bytes) AS TtlBytes, sub_ip AS IP ...> FROM (SELECT count(*) AS sub_count, sum(bytes) AS sub_bytes, ...> ip_str AS sub_ip FROM log GROUP BY 3, req_url) ...> GROUP BY 4 ORDER BY 1 DESC LIMIT 8; Uniq Ttl TtlBytes IP ---- ---- ---------- ------------------ 1295 1295 31790418 10.5.69.83 282 334 13571771 10.170.13.97 234 302 4234382 10.155.7.28 213 215 3089112 10.155.7.77 163 176 2550477 10.155.7.29 159 161 4279779 10.195.137.175 153 154 2292407 10.23.146.198 135 171 2272949 10.155.7.71 For each IP address, the first column is the number of unique URLs requested, while the second column is the total number of requests. The second column should always be greater than or equal to the first column. The third column is the total number of bytes, followed by the (altered) IP address in question. Exactly how this query works is left as an exercise for the reader. There are countless other queries we could run. For anyone that has ever imported log data into an SQL database and played around with it, none of this is particularly inspiring. But consider this for a moment: the query time for the first two of these examples is a bit less than five seconds on an economy desktop system that is several years old. The third query was a bit closer to eight seconds. Five seconds to scan a 100,000-row table might not be blazingly fast, but remember that those five seconds are the grand total for everything, including data “import.” Using the virtual table module allows us to go from a raw logfile with 100,000 lines to a query answer in just that amount of time—no data staging, no format conversions, no data imports. That’s important, since importing involves a lot of I/O and can be a slow process. For example, importing the same file into a standard SQLite table by more traditional means takes nearly a minute and that doesn’t even include any queries! Now consider that we enable all this functionality with less than 400 lines of C code. Accessing the original data, rather than importing it into standard tables, allows the end-to-end data analysis process to be much faster, and allows you to query the data, as it is recorded by the web server, in real time. As an added bonus, the virtual table can also be used as an importer, by using the CREATE TABLE... AS or INSERT... SELECT SQL commands. Advanced Example: weblog Module | 261 If you find yourself faced with the task of writing a script to analyze, search, or sum- marize some structured source of data, you might consider writing an SQLite module instead. A basic, read-only module is a fairly minor project, and once you’ve got that in place you have the complete power of the SQLite database engine at your disposal (plus an added data importer!). That makes it easy to write, test, and tune whatever queries you need in just a few lines of SQL. Best Index and Filter Let’s take a closer look at the xBestIndex() and xFilter() functions. Both of our ex- ample modules were fairly simple and didn’t use them, but proper use of these functions is critical for internal modules that implement some types of high-performance index- ing system. Purpose and Need By default, the only way to get data out of a table—virtual or otherwise—is to do a full table scan. This can be quite expensive, especially if the table is large and the query is trying to extract a small number of rows. Standard tables have ways of boosting retrieval speeds, such as using indexes. The query optimizer can use other hints found in a standard table definition, such as knowing which columns are unique or have other constraints on them. Virtual tables lack these features. You cannot create an index on a virtual table, and the query optimizer has no knowledge of the structure or format of a virtual table, other than the column names. The only known constraint on a virtual table is that each virtual row must have a unique, integer ROWID. Without any additional information, it is very difficult to optimize a query that involves a virtual table. This is true for both the query planner and the virtual table itself. For the best performance, the query optimizer needs to understand what types of lookups the virtual table is best suited to doing. Conversely, the virtual table module needs to understand the nature of the user query, including any constraints, so that it can use any internal indexes or lookup optimizations to the best of its ability. The purpose of the xBestIndex() and xFilter() functions is to bridge this gap. When an SQL statement is prepared, the query optimizer may call xBestIndex() several times, presenting several different query possibilities. This allows the module to formulate its own query plan and pass back an approximate cost metric to the query optimizer. The query optimizer will use this information to pick a specific query plan. When the query statement is executed, the SQLite library uses xFilter() to commu- nicate back to the module which query plan was actually chosen. The module can use this information to optimize its internal data lookups, as well as skip over any rows 262 | Chapter 10: Virtual Tables and Modules that are not relevant to the query at hand. This allows a virtual table to implement more targeted data lookups and retrievals, not unlike an index on a traditional table. xBestIndex() If you’ll recall, the xBestIndex() function is a table-level function that looks like this: int xBestIndex( sqlite3_vtab *vtab, sqlite3_index_info *idxinfo ); The whole key to this function is the sqlite3_index_info structure. This structure is divided into two sections. The first section provides a series of inputs to your function, allowing SQLite to propose a query plan to the module. The input section should be treated as read-only. The second section is the output section. A module uses this second section to com- municate back to the query optimizer information about which constraints the virtual table is prepared to enforce, and how expensive the proposed query might be. The module is also given a chance to associate an internal query plan or other data to this particular proposal. The query optimizer will then use this data to select a specific query plan. The input section consists of two size values and two arrays. The nConstraint integer indicates how many elements are in the aConstraint[] array. Similarly, the nOrder By[] integer indicates how many elements are in the aOrderBy[] array: struct sqlite3_index_info { /**** Inputs ****/ int nConstraint; /* Number of entries in aConstraint */ struct sqlite3_index_constraint { int iColumn; /* Column on lefthand side of constraint */ unsigned char op; /* Constraint operator */ unsigned char usable; /* True if this constraint is usable */ int iTermOffset; /* Used internal - xBestIndex should ignore */ } *aConstraint; /* Table of WHERE clause constraints */ int nOrderBy; /* Number of terms in the ORDER BY clause */ struct sqlite3_index_orderby { int iColumn; /* Column number */ unsigned char desc; /* True for DESC. False for ASC. */ } *aOrderBy; /* The ORDER BY clause */ The aConstraint[] array communicates a series of simple constraints that a query may put on the virtual table. Each array element defines one query constraint by passing values for a column index (aConstraint[i].iColumn) and a constraint operator (aCon straint[i].op). The column index refers to the columns of the virtual table, with a zero signifying the first column. An index of ‒1 indicates the constraint is being applied to the virtual ROWID column. The specific constraint operator is indicated with one of these constants. The referenced column (the column index) is always assumed to be on the lefthand side. These are the only operators that can be optimized by a virtual table: Best Index and Filter | 263 SQLITE_INDEX_CONSTRAINT_EQ /* COL = Expression */ SQLITE_INDEX_CONSTRAINT_GT /* COL > Expression */ SQLITE_INDEX_CONSTRAINT_LE /* COL <= Expression */ SQLITE_INDEX_CONSTRAINT_LT /* COL < Expression */ SQLITE_INDEX_CONSTRAINT_GE /* COL >= Expression */ SQLITE_INDEX_CONSTRAINT_MATCH /* COL MATCH Expression */ For example, if one of the aConstraint elements had the values: aConstraint[i].iColumn = -1; aConstraint[i].op = SQLITE_INDEX_CONSTRAINT_LE; That would roughly translate to a WHERE clause of: ...WHERE ROWID <= ? The parameter on the right side of the expression may change from query to query, but will remain constant for the any given table scan, just as if it were a statement parameter with a bound value. Each aConstraint[] element also contains a usable element. Some constraints may not be usable by the optimizer due to joins or other external conditions put on the query. Your code should only pay attention to those constraints where the usable field is nonzero. The second array of the input section, aOrderBy[], communicates a set of requested ORDER BY sortings (it may also be generated by columns in a GROUP BY clause). Each ordering element is defined by a column index and a direction (ascending or descend- ing). The column indexes work the same way, with defined columns starting at 0 and ‒1 referring to the ROWID. The ordering elements should be treated as a series of ORDER BY arguments, with the whole data set being sorted by the first ordering, then subsets of equal values being sorted by the second ordering, and so on. The output section contains the data that is passed back to the SQLite optimizer. It consists of a constraint array and a set of values. The aConstraintUsage[] array will always be the same size as the aConstraint[] array (that is, will always have nCon straint elements). SQLite will always zero out the memory used by the output section. This is why it is safe to ignore the structure in simplified implementations of xBest Index()—the structure is basically preset to an answer of, “this module cannot optimize anything.” In that case, every virtual table query will require a full table scan: /**** Outputs ****/ struct sqlite3_index_constraint_usage { int argvIndex; /* If >0,constraint is part of argv to xFilter */ unsigned char omit; /* Do not code a test for this constraint */ } *aConstraintUsage; int idxNum; /* Number used to identify the index */ char *idxStr; /* Application-defined string */ int needToFreeIdxStr; /* Free idxStr using sqlite3_free() if true */ int orderByConsumed; /* True if output is already ordered */ double estimatedCost; /* Estimated cost of using this index */ }; 264 | Chapter 10: Virtual Tables and Modules If a module is able to optimize some part of the query, this is indicated to the query optimizer by modifying the output section of the sqlite3_index_info structure to in- dicate what query optimizations the module is willing and capable of performing. Each element of the aConstraintUsage[] array corresponds to the same ordered element of the aConstraint[] array. For each constraint described in an aConstraint[] element, the corresponding aConstraintUsage[] element is used to describe how the module wants the constraint applied. The argvIndex value is used to indicate to SQLite that you want the expression value of this constraint (that is, the value on the righthand side of the constraint) to be passed to the xFilter() function as one of the argv parameters. The argvIndex values are used to determine the sequence of each expression value. Any aConstraintUsage[] element can be assigned any index value, just as long as the set of assigned index values starts with one and has no gaps. No aConstraintUsage[] elements should share the same nonzero argvIndex value. If the default argvIndex value of zero is returned, the expres- sion value is not made available to the xFilter() function. Exactly how this is used will make more sense when we look more closely at xFilter(). The omit field of an aConstraintUsage[] element is used to indicate to the SQLite library that the virtual table module will take the responsibility to enforce this constraint and that SQLite can omit the verification process for this constraint. By default, SQLite verifies the constraints of every row returned by a virtual table (e.g., every row xNext() stops at). Setting the omit field will cause SQLite to skip the verification process for this constraint. Following the constraint array is a series of fields. The first three fields are used to communicate to the xFilter() function. The fields idxNum and idxStr can be used by the module however it wishes. The SQLite engine makes no use of these fields, other than to pass them back to xFilter(). The third field, needToFreeIdxStr, is a flag that indicates to the SQLite library that the memory pointed to by idxStr has been dynam- ically allocated by sqlite3_malloc(), and the SQLite library should free that memory with sqlite3_free() if the library decides it is no longer required. This flag is needed to prevent memory leaks. Remember that xBestIndex() may be called several times as part of the prepare process for an SQL statement. The module will usually pass back a unique idxStr value for each proposed query plan. Only one of these idxStr values will be passed to xFilter(), however, and the rest must be dis- carded. That means that any string (or other memory block) you provide to idxStr needs to either be static memory, or the memory needs to be allocated with sqlite3_mal loc() and the needToFreeIdxStr flag needs to be set. This allows the SQLite library to properly clean up any unused idxStr allocations. The orderByConsumed field is used to indicate that the module is able to return the data presorted in the order defined by the aOrderBy array. This is an all-or-nothing flag. If three aOrderBy elements are given, but the module can only sort the output by the first column, it must return a false value. Best Index and Filter | 265 Finally, the estimatedCost field is used to communicate a cost value back to the SQLite library. If this is an external module, this number should approximate the total number of disk accesses required to return all rows that meet the specified constraints. If this is an internal module, it can be an approximation of the number of sqlite3_step() and sqlite3_column_xxx() calls. In situations where a full table scan is required, it can es- timate the number of rows in the virtual table. The exact measurement is not extremely meaningful, other than the relative values between different calls to xBestIndex(). xFilter() The xFilter() function provides a way for the SQLite library to notify the module, within the context of a specific table cursor, exactly what constraints and ordering should be applied to the next table scan. Recall that the xFilter() prototype looks like this: int xFilter( sqlite3_vtab_cursor *cursor, int idxNum, const char *idxStr, int argc, sqlite3_value **argv ) The first argument is the table cursor that requires these constraints. The idxNum and idxStr values are the same values that were passed back by the module in a prior call to xBestIndex(). These mean whatever the module wants, just as long as the code in xBestIndex() and xFilter() agrees on what they are and what the values represent. Finally, the last two arguments are derived from the aConstraintUsage[].argvIndex values passed back by the module. The argv parameter is an array of sqlite3_value structures, while the argc parameter indicates how many elements are in the argv array. Going back to our prior example, consider an sqlite3_index_info structure with an aConstraint[i] element, where iColumn=-1 and op=SQLITE_INDEX_CONSTRAINT_LE (indi- cating a constraint of ROWID >= ?). If the module’s xBestIndex() function set aCon straintUsage[i].argIndex to a value of 1, the argv[0] value passed into xFilter() will have the value found on the righthand side of the expression. Notice that the argument indexes between xBestIndex() and xFilter() are off by one. Because sqlite3_index_info considers an aConstraintUsage[].argvIndex value of 0 to indicate an invalid index, the argvIndex values start at 1. The actual argv indexes will all be one less, however, as they start at 0. Using the idxNum, idxStr, and argv values, it is the responsibility of xFilter() to configure this table cursor to provide the correct constraints and ordering that were promised by the corresponding sqlite3_index_info block. 266 | Chapter 10: Virtual Tables and Modules Typical Usage The design of xBestIndex() and xFilter() functions is strongly focused on optimizing internal style modules. These are modules that are going to use one or more SQL state- ments that operate over a set of internal tables to produce the virtual table data. This is similar to how the dblist module works, but normally involves more complex SQL commands. A module is free to do whatever it wants with the idxNum and idxStr values, but most internal modules use them to pass off pre-built SQL command strings. Each time xBestIndex() is called, the module tries to figure out how it would service the query constraints and ordering constraints, by adding conditions, constraints, and ORDER BY parameters to the internal SQL statements used to generate the virtual table data. The xBestIndex() function marks the constraints it can use and builds the required SQL command strings, complete with statement parameters. These SQL commands are passed back with the idxStr value. The idxNum can be used to pass back a string length, or some other index value or bit flags or whatever the module wants. The argvIndex values of the aConstraintUsage elements are set to the corresponding statement pa- rameter index values. In essence, the xBestFilter() function will build the SQL com- mand strings that query the virtual table data in such a way that the required constraints and ordering are already “baked in” to the behind-the-scenes queries. When xFilter() is called, the idxStr value will have the relevant SQL command strings for that query configuration. The SQL command strings can then be prepared, and the constraint expressions pass in via the argv array, and can be bound to any statement parameters. The xFilter() function starts to step through the prepared statements, generating the first row. Like the dblist internal module, subsequent calls to xNext() continue to step through any internal statements, returning additional rows. As long as xBestIndex() can derive a reasonable set of SQL command strings that are capable of expressing the required internal query (or queries), this is all reasonably straightforward. If necessary, multiple SQL command strings can be passed into xFilter() by defining them one after another in a large string, and using the tail pa- rameter of sqlite3_prepare_xxx() to prepare multiple statements, one after another. Things can be difficult when dealing with external modules. Very often external mod- ules can’t define complex query conditions or sort ordering with a simple string. Although the idxStr pointer can be used to pass in some type of data structure, it can be difficult to encode all the constraint information. This is one of the reasons why many modules, and especially external modules, forego the use of xBestIndex() and xFilter(), and just depend on full table scans for all operations. Full table scans might be slower, but they still work. Best Index and Filter | 267 That might sound bad, but remember that even on a standard table with a standard index, you typically don’t start to see really good returns on using the index unless a constraint and appropriate index are able to eliminate 80% or better of the rows. Spending a lot of time to build a constraint handler that only filters out a small per- centage of rows is normally a losing proposition. While that can be the whole point of internal modules, the primary goal of most external modules is to simply provide data connectivity. If you’re working on an external module, get the basic data translation working first, and then worry about possibly implementing more efficient lookup patterns. Wrap-Up If you’ve made it through the whole chapter, you should have a pretty good idea of what the virtual table system can do and how it works. A well-written internal module can bring a whole new feature set to the database, allowing an application to shape and manipulate data structures that SQLite might not otherwise be able to store or process in an efficient manner. External modules can provide a data conduit, providing both speed and flexibility in the use of external data sources. They can also double as pow- erful data importers. That power comes at a cost, however. Without getting into filesystems and storage engines, a module is the most complex bit of extension code most people will ever write for SQLite. A good module depends on a solid design that will properly track all of the necessary states and perform all of the required functions. To avoid frustration, it is often a good idea to start with the simple base cases and expand your code and design to cover the more complex situations. 268 | Chapter 10: Virtual Tables and Modules APPENDIX A SQLite Build Options SQLite has a fair number of compile-time options and build directives. Many of these are used to change default values, behaviors, or limits, while others are used to enable optional modules or disable existing features. One of the main reasons you may find yourself recompiling SQLite is to alter the compiler directives. All of these directives are standard C #define flags. You can often define them in your IDE’s build environment, as environment variables (if you’re compiling from a command-line) or by passing one or more flags to your compiler. The defaults are reasonable for most modern desktop systems. If you’re developing for a mobile system, you may want to pick and choose your configuration more carefully. Some of the advanced extensions can add a considerable amount of bulk to the SQLite library, as well as demand additional resources at runtime. While these changes may not be significant to a desktop system, they may cause problems in more restricted environments. Shell Directives There is only one compiler directive that is specific to the shell.c source file and the sqlite3 command-line tool. This directive only needs to be defined when building the shell.c source. The actual SQLite core does not recognize this directive. See also SQLITE_ENABLE_IOTRACE later in this appendix, which enables the .iotrace command in sqlite3, but must be defined when compiling both shell.c and sqlite3.c. ENABLE_READLINE Enable the GNU Readline library Common Usage ENABLE_READLINE=1 Default undefined (unused) 269 Description A value of 1 enables use of the GNU Readline library. When this is enabled, the Readline library must be linked in as part of the build process. Default Values The following compiler directives are used to alter the default value for different data- base parameters. These directives alter the default startup values. Most of these parameters can be changed at runtime. In general, if you alter a default value for your application, it is also a good idea to alter the default values for any tools you might use (e.g., the sqlite3 command-line tool). Having everything using the same set of defaults reduces the chance that the configu- ration values inherited by a new database will be out of sync with the rest of the system. Conversely, if your application has critical dependencies on specific settings, it would be best to explicitly set those values at runtime through an appropriate SQL command or API call, even if you also configure the appropriate defaults when building the library. This guarantees that the values will remain correct, regardless of the SQLite library configuration. SQLITE_DEFAULT_AUTOVACUUM Default auto-vacuum mode Common Usage SQLITE_DEFAULT_AUTOVACUUM=<0|1|2> Default 0 (auto-vacuum disabled) Description Sets the default auto-vacuum mode for new databases. The default value of 0 completely disables the auto-vacuum functionality. Value Mode Meaning 0 None Auto-vacuum disabled 1 Full Auto-vacuum enabled, running 2 Incremental Auto-vacuum enabled, deferred See Also auto_vacuum [PRAGMA, Ap F] SQLITE_DEFAULT_AUTOVACUUM 270 | Appendix A: SQLite Build Options SQLITE_DEFAULT_CACHE_SIZE Default maximum cache size Common Usage SQLITE_DEFAULT_CACHE_SIZE=number-of-pages Default 2000 pages Description Sets the default maximum cache size, in pages, of a new database. See Also default_cache_size [PRAGMA, Ap F], cache_size [PRAGMA, Ap F] SQLITE_DEFAULT_FILE_FORMAT Default file format Common Usage SQLITE_DEFAULT_FILE_FORMAT=<1|4> Default 1 (original/legacy format) Description Sets the default file format. SQLite 3.3.0 introduced a new file format (4) that understands descending indexes and uses a more efficient encoding for Boolean values (integer values 0 and 1). All versions of SQLite after 3.3.0 can read and write both file formats. SQLite versions before 3.3.0 can only understand the original (1) format. As the number of pre-3.3.0 users declines, it is expected that the default version will be changed to the newer file format. See Also legacy_file_format [PRAGMA, Ap F] SQLITE_DEFAULT_JOURNAL_SIZE_LIMIT Default size limit for persistent journals Common Usage SQLITE_DEFAULT_JOURNAL_SIZE_LIMIT=bytes Default Undefined (unlimited) Description Sets the default size limit for persistent journal files. When this directive is undefined, there is no limit. This limit may be set at runtime. SQLITE_DEFAULT_JOURNAL_SIZE_LIMIT Appendix A: SQLite Build Options | 271 See Also journal_size_limit [PRAGMA, Ap F] SQLITE_DEFAULT_MEMSTATUS Memory status configuration Common Usage SQLITE_DEFAULT_MEMSTATUS=<0|1> Default 1 (enabled and available) Description Sets the default configuration (enabled or disabled) of the memory status system. Memory status must be available for the sqlite3_memory_used(), sqlite3_memory_highwater(), sqlite3_soft_heap_limit(), or sqlite3_status() API functions to provide valid data. The memory status system can be enabled or disabled at runtime using sqlite3_config(). See Also sqlite3_config() [C API, Ap G] SQLITE_DEFAULT_PAGE_SIZE Default database page size Common Usage SQLITE_DEFAULT_PAGE_SIZE=bytes Default 1024 bytes Description Sets the default page size of a database in bytes. Values must be a power-of-two value between 512 and SQLITE_MAX_PAGE_SIZE, which defaults to 32 KB (and is the maximum value suppor- ted). Possible values include: 512, 1024, 2048, 4096, 8192, 16384, or 32768. See Also SQLITE_MAX_PAGE_SIZE, page_size [PRAGMA, Ap F] SQLITE_DEFAULT_TEMP_CACHE_SIZE Temporary cache size Common Usage SQLITE_DEFAULT_TEMP_CACHE_SIZE=number-of-pages Default 500 pages SQLITE_DEFAULT_MEMSTATUS 272 | Appendix A: SQLite Build Options Description Sets the default maximum size of any temporary page cache. Temporary page caches are used to store intermediate results and other transitory data. Temporary databases use a standard page cache and do not use this value. There is no way to modify this value at runtime. YYSTACKDEPTH Maximum parser stack depth Common Usage YYSTACKDEPTH=max-depth Default 100 Description Sets the maximum stack-depth of the parser used to process SQL statements. It is very unusual for applications to require a value greater than 25. If you find yourself adjusting this value upwards, you most likely need to reconsider your SQL style. Sizes and Limits In moving from SQLite 2 to SQLite 3, one of the major design goals was to eliminate as many limitations as possible. Algorithms and data structures were designed to scale to extreme values. Unfortunately, this “knows no bounds” approach was in conflict with the desire to have comprehensive testing. In many cases it was difficult, if not impossible, to test the extreme limits. This lead to the uncomfortable situation where SQLite was “known” to be reliable, as long as your usage patterns didn’t push too far beyond the tested limits. If you ventured too far out into uncharted territory, the sta- bility and reliability was less known and less understood, and over time a number of bugs were found when things were pushed beyond sane values. As a result, many of the default maximum values have been reined in from the absurd to the extremely unlikely. You’ll notice that most of these values are fairly large and should not present a practical limit for nearly any properly formed database. In fact, if you find yourself making these values larger, you should likely take a moment to con- sider what design decisions lead to the need for adjustment. In many cases, a better solution will be found in adjusting the design of the database, rather than adjusting the limits. If you adjust these limits at all, it might be best to make them smaller. This will help catch overflow and runaway systems. Despite their comfortable size, these limits are small enough to allow testing up to and including the maximum values. This allows SQLite to provide the same level of confi- dence and stability across the entire operational domain. YYSTACKDEPTH Sizes and Limits | 273 Although these compile-time directives are used to define absolute maximums, all of these limits can be lowered at runtime with the sqlite3_limit() API call. See the entry sqlite3_limit() in Appendix G for more information. SQLITE_MAX_ATTACHED Max number of attached databases Common Usage SQLITE_MAX_ATTACHED=number-of-databases Default 10 databases Description The maximum number of databases that can be attached to a single session. There is a hard limit of 30. SQLITE_MAX_COLUMN Max number of columns in any structure Common Usage SQLITE_MAX_COLUMN=number-of-columns Default 2000 columns Description The maximum number of columns on any database table, index, or view, as well as any transient tables found in a SELECT. There is a hard limit of 32,767. SQLITE_MAX_COMPOUND_SELECT Max terms in compound statement Common Usage SQLITE_MAX_COMPOUND_SELECT=number-of-select-terms Default 500 terms Description The maximum number of SELECT terms in a compound SELECT statement (a statement that uses UNION, UNION ALL, INTERSECT, or EXCEPT). SQLITE_MAX_ATTACHED 274 | Appendix A: SQLite Build Options SQLITE_MAX_DEFAULT_PAGE_SIZE Upper bound on automatic page size Common Usage SQLITE_MAX_DEFAULT_PAGE_SIZE=bytes Default 8192 bytes Description Normally, SQLite creates a database with pages that are SQLITE_DEFAULT_PAGE_SIZE in size. However, if the filesystem driver indicates a larger size may offer better performance, SQLite may choose a different default page size. In those situations where SQLite chooses a nondefault value, the actual page size will be limited to this size or smaller. See Also SQLITE_DEFAULT_PAGE_SIZE SQLITE_MAX_EXPR_DEPTH Max SQL expression tree depth Common Usage SQLITE_MAX_EXPR_DEPTH=stack-depth Default 1000 levels Description Maximum depth of an SQL expression tree. This is used to limit the stack-space used when parsing a large SQL expression. A value of 0 represents no limit. SQLITE_MAX_FUNCTION_ARG Max number of function arguments Common Usage SQLITE_MAX_FUNCTION_ARG=number-of-arguments Default 127 arguments Description The maximum number of arguments in an SQL function. There is a hard upper limit of 1,000. SQLITE_MAX_FUNCTION_ARG Appendix A: SQLite Build Options | 275 SQLITE_MAX_LENGTH Max row size Common Usage SQLITE_MAX_LENGTH=bytes Default 1000000000 (1,000,000,000) bytes Description The maximum length of a row, including any BLOB values. The maximum supported value is 2,147,483,647 bytes (2 GB). SQLITE_MAX_LIKE_PATTERN_LENGTH Max search pattern length Common Usage SQLITE_MAX_LIKE_PATTERN_LENGTH=bytes Default 50000 bytes Description The maximum length of a LIKE or GLOB search pattern. A maliciously crafted search pattern can consume a large number of resources, so if you allow arbitrary user search patterns, you might consider significantly lowering this value. SQLITE_MAX_PAGE_COUNT Max number of pages in a database Common Usage SQLITE_MAX_PAGE_COUNT=number-of-pages Default 1073741823 (1,073,741,823) pages, or one giga-page Description The maximum number of pages in a single database. This, along with the current page size, defines the maximum size for a database file. With the default page size of 1024 bytes, this limits database files to one terabyte. With the maximum page size of 32 KB, this limits database files to 32 terabytes. In theory, this number can be raised as high as 4,294,967,296 (largest 32-bit unsigned integer), but there should be little need to increase this value. See Also SQLITE_DEFAULT_PAGE_SIZE, max_page_count [PRAGMA, Ap F], page_size [PRAGMA, Ap F] SQLITE_MAX_LENGTH 276 | Appendix A: SQLite Build Options SQLITE_MAX_PAGE_SIZE Max database page size Common Usage SQLITE_MAX_PAGE_SIZE=bytes Default 32768 bytes Description The maximum size of a database page. The default 32K value is the maximum allowed by the internal architecture of SQLite. This value can only be lowered. See Also SQLITE_DEFAULT_PAGE_SIZE SQLITE_MAX_SQL_LENGTH Max SQL statement length Common Usage SQLITE_MAX_SQL_LENGTH=bytes Default 1000000 (1,000,000) bytes Description The maximum length of an SQL statement. SQLITE_MAX_TRIGGER_DEPTH Max trigger recursion depth Common Usage SQLITE_MAX_TRIGGER_DEPTH=depth Default 1000 Description The maximum recursion level for triggers. This value is only meaningful when recursive trig- gers are enabled. SQLITE_MAX_VARIABLE_NUMBER Max bind variables in SQL statement Common Usage SQLITE_MAX_VARIABLE_NUMBER=number-of-variables Default 999 variables SQLITE_MAX_VARIABLE_NUMBER Appendix A: SQLite Build Options | 277 Description The maximum number of bind variables in a prepared SQL statement. Operation and Behavior These directives alter some of the fundamental behaviors of SQLite. Most of these are related to getting SQLite working on platforms with limited support. SQLITE_CASE_SENSITIVE_LIKE Define LIKE case sensitivity Common Usage SQLITE_CASE_SENSITIVE_LIKE Default Undefined (not case-sensitive) Description If defined, the LIKE operator will be case-sensitive by default. See Also LIKE [SQL Expr, Ap D], case_sensitive_like [PRAGMA, Ap F] SQLITE_HAVE_ISNAN Use system isnan() function Common Usage SQLITE_HAVE_ISNAN Default Undefined (use internal function) Description If defined, SQLite will use the system isnan() function to determine if a floating-point value is a valid or not. Normally, SQLite uses an internal version of this function. SQLITE_OS_OTHER Override default OS detection Common Usage SQLITE_OS_OTHER=<0|1> Default 0 SQLITE_CASE_SENSITIVE_LIKE 278 | Appendix A: SQLite Build Options Description SQLite has an SQLITE_OS_* directive for each operating system it natively supports. Normally, SQLite will try to determine what operating system it is running on by examining various automatic compiler directives. If you’re cross-compiling, you can manually set SQLITE_OS_ OTHER to 1. This will override all other SQLITE_OS_* flags and disable the default operating system interfaces. This directive is mainly of interest to people working on embedded systems. SQLITE_SECURE_DELETE Enable secure delete Common Usage SQLITE_SECURE_DELETE Default Undefined (no overwrites) Description If defined, this directive sets the default secure delete option to on. SQLite will zero out deleted rows, as well as zero out and write back any recovered database pages before adding them to the free list. This prevents someone from recovering deleted data by examining the database file. Be aware that SQLite cannot securely delete information from the underlying storage device. If the write operation causes the filesystem to allocate a new device-level block, the old data may still exist on the raw device. There is also a slight performance penalty associated with this directive. See Also secure_delete [PRAGMA, Ap F] SQLITE_THREADSAFE Specify thread mode Common Usage SQLITE_THREADSAFE=<0|1|2> Default 1 (serialized mode) Description Sets the default thread mode. SQLite supports three thread modes. Value Mode Meaning 0 Single thread Disables thread support 1 Fully serialized Full thread support 2 Multithread Basic thread support SQLITE_THREADSAFE Appendix A: SQLite Build Options | 279 Single thread mode disables all mutexes and thread support. In this mode, the locking code is completely removed from the build and the mode cannot be changed. All interaction with the SQLite library must be done from a single thread. Serialized allows a database connection to be used across multiple threads. Multithread allows the SQLite library to be used by multiple threads, but a database connection can only be used by one thread at a time. If an application is built with thread support, it can switch between thread-safe modes (1 or 2) at application startup. See Also sqlite3_threadsafe() [C API, Ap G] SQLITE_TEMP_STORE Specify temporary storage location Common Usage SQLITE_TEMP_STORE=<0|1|2|3> Default 1 (use files, allow override) Description This directive controls how temporary files are stored. Temporary files may either be stored on disk or in memory. Rollback journals and master journals are always stored on disk. This parameter applies to temporary databases (used to store temporary tables), materialized views and subqueries, transient indexes, and transient databases used by VACUUM. It is perfectly safe to use memory based temporary storage. Value Temp. storage location 0 Always on disk 1 Defaults to on disk, allows override 2 Defaults to in memory, allows override 3 Always in memory See Also temp_store [PRAGMA, Ap F], sqlite3_config() [C API, Ap G] Debug Settings SQLite includes a small number of directives used for enabling various debugging facilities. They can be activated by simply defining the directive. No specific value is required. Normally, these debug directives are only used for testing and development purposes, as they add significant overhead and make everything run noticeably slower. SQLITE_TEMP_STORE 280 | Appendix A: SQLite Build Options All of these directives are undefined by default. Simply defining the directive will enable the feature. SQLITE_DEBUG General debugging and sanity checking Description This directive is used for debugging SQLite. Defining this directive turns on a great number of assert tests, as well as some other debugging facilities. SQLITE_MEMDEBUG Debug the memory allocator Description This directive is used to debug the memory allocator. If defined, an instrumented memory allocator is in place of the normal dynamic memory system. Enable Extensions This section includes compiler directives that can be used to turn various SQLite ex- tensions on or off. A number of these are fairly simple changes that can be used to tailor SQLite to a specific operating environment. Others are more dramatic, and can be used to enable some major extension modules. While there should be little risk in enabling these extensions, many of them are not as vigorously tested as the core system. You should feel free to enable what you need, but you may want to refrain from enabling an extension unless you actually need it. All of these directives are undefined by default. Simply defining the directive will enable the feature. SQLITE_ENABLE_ATOMIC_WRITE Enable atomic write support Description Enables a runtime check for filesystems that support atomic write operations. The process SQLite uses to write out changes is significantly simpler on filesystems that support atomic writes. If enabled, this parameter will cause SQLite to query the filesystem to see if it supports atomic writes and, if so, use them. This type of filesystem is still relatively rare, however, and there are costs associated with the check, so the whole thing is turned off by default. SQLITE_ENABLE_ATOMIC_WRITE Appendix A: SQLite Build Options | 281 SQLITE_ENABLE_COLUMN_METADATA Enable column metadata Description Enables the retrieval of column metadata required to support the following C API calls: • sqlite3_column_database_name() • sqlite3_column_database_name16() • sqlite3_column_table_name() • sqlite3_column_table_name16() • sqlite3_column_origin_name() • sqlite3_column_origin_name16() • sqlite3_table_column_metadata() SQLITE_ENABLE_FTS3 Enable full-text search module Description Enables full-text search module. See “Full-Text Search Module” on page 169 for more details. SQLITE_ENABLE_FTS3_PARENTHESIS Enable FTS extended syntax Description Enables the extended FTS3 query syntax, including nested parenthetical queries and the AND and NOT keywords. SQLITE_ENABLE_ICU Enable the ICU extension Description Enables International Components for Unicode (http://www.icu-project.org/) support. This li- brary allows SQLite to deal with non-English and non-ASCII more intelligently. Using this directive requires the ICU library to be available and linked in any build. See “ICU Interna- tionalization Extension” on page 167 for more details. SQLITE_ENABLE_IOTRACE Enable I/O trace debugging Description Enables I/O debug output. If both the SQLite library and the sqlite3 utility are compiled with this option, an .iotrace command will be included in the utility. This command will cause SQLite to log all I/O transactions to a trace file. This directive should only be used when building the sqlite3 utility. SQLITE_ENABLE_COLUMN_METADATA 282 | Appendix A: SQLite Build Options SQLITE_ENABLE_LOCKING_STYLE Enable extended file locking Description Enables extended locking styles. Normally, SQLite running on a Unix-like system will use POSIX fcntl() based locks for all files. If this parameter is defined, SQLite will change its locking style depending on the type of filesystem it is using. This flag only has significant effects under Mac OS X, where it is enabled by default. See the SQLite website for more details (http://sqlite.org/compile.html#enable_locking_style). SQLITE_ENABLE_MEMORY_MANAGEMENT Enable extended memory management Description Enables extended memory management and tracking that allows SQLite to release unused memory upon request. This parameter is required for sqlite3_release_memory() and sqlite3_soft_heap_limit() functions to have any effect. SQLITE_ENABLE_MEMSYS3 Enable alternate memory allocator Description Enables one of two alternate memory allocators. Only one can be enabled at a time. The alternate allocators are used when SQLite is in “heap” mode. This allows the application to provide a static chunk of memory that SQLite will use for all of its internal allocations. This is most commonly done with embedded systems where memory usage must be carefully con- trolled. MEMSYS3 uses a hybrid allocation algorithm based off dlmalloc(). See Also SQLITE_ENABLE_MEMSYS5 SQLITE_ENABLE_MEMSYS5 Enable alternate memory allocator Description Enables one of two alternate memory allocators. Only one can be enabled at a time. MEM SYS5 is essentially the same as MEMSYS3, only it uses a power-of-two, first-fit algorithm. MEM SYS3 tends to be more frugal with memory, while MEMSYS5 is better at preventing fragmentation. Determining which module is best for your specific needs is largely an issue of testing and measuring. See Also SQLITE_ENABLE_MEMSYS3 SQLITE_ENABLE_MEMSYS5 Appendix A: SQLite Build Options | 283 SQLITE_ENABLE_RTREE Enable R*Tree spatial index Description Enables the R*Tree extension module. R*Trees are designed for range queries and are com- monly used to store durations, coordinates, or geospatial data. See “R*Trees and Spatial In- dexing Module” on page 171 for more details. SQLITE_ENABLE_STAT2 Enable extended ANALYZE statistics Description Enables the calculation of additional statistics by the ANALYZE command. If present, a 10- sample histogram will be calculated for any analyzed index. The query planner can use this data to estimate how many rows will be filtered by a range condition. SQLITE_ENABLE_UPDATE_DELETE_LIMIT Enable extended SQL syntax Description Enables an extended syntax for the UPDATE and DELETE commands that allows the inclusion of ORDER BY and LIMIT clauses. Because support for directives requires altering the SQL parser, this directive can only be used when building from a full development tree. It will not work if building from an amalgamation or a source distribution. SQLITE_ENABLE_UNLOCK_NOTIFY Enable extended locking API Description Enables the sqlite3_unlock_notify() C API. This function allows an application to register a callback function that will be called when a shared-cache table lock is released. This build directive is only applicable to applications using the shared-cache functionality. YYTRACKMAXSTACKDEPTH Enable parser stack tracking Description Enables depth tracking code in the SQL parser. This can be used to determine sensible values for the YYSTACKDEPTH directive. SQLITE_ENABLE_RTREE 284 | Appendix A: SQLite Build Options Limit Features In addition to enabling extensions that are normally disabled, there are also a small number of features that are normally enabled, but can be optionally disabled. Generally you should only need to disable these features on embedded systems that lack the required filesystem support. All of these directives are undefined by default. Simply defining the directive will disable the feature. SQLITE_DISABLE_LFS Disable large file support Description Disables large file support. If disabled, all file operations will be limited to 32-bit offsets. This limits files to two gigabytes. SQLITE_DISABLE_DIRSYNC Disable directory synchronization Description Disables directory syncs. Normally, SQLite will request that the operating system synchronize the parent directory of a deleted file to ensure the directory entries are immediately updated on disk. This directive disables that synchronization. SQLITE_ZERO_MALLOC Disable memory allocator Description Disables the memory allocator and replaces it with a stub allocator. This stub allocator always fails, making the SQLite library unusable. An application can provide its own memory man- agement functions and scratch memory blocks at library startup using the sqlite3_config() API. This build option allows SQLite to be built on platforms that do not have native memory management routines. Omit Core Features In addition to all the other build directives, SQLite has a fair number of SQLITE_ OMIT_* compile-time directives. These are designed to remove core features from the build in an effort to make the core database library as small and compact as possible. For example, SQLITE_OMIT_ANALYZE eliminates all code support for the ANALYZE com- mand (and subsequent query optimizations), while SQLITE_OMIT_VIRTUALTABLE elimi- nates the entire virtual table facility. SQLITE_ZERO_MALLOC Omit Core Features | 285 In general, these directives should only be of interest to embedded systems developers that are counting every byte. Along with any relevant omit flags, you should make sure the compiler is set to build with any “optimize for size” type features enabled. In order to use most of these omit directives, you need to be building SQLite from the development sources found in the source control tree. Most omit directives won’t work correctly when applied to a source distribution or to the pre-built amalgamation. Also be aware that these compile-time directives are not officially supported, in the sense that they are not part of the official testing chain. For any given version of SQLite, there may be both compile problems and runtime issues if arbitrary sets of omit flags are enabled. Use (and test) at your own risk. For a full list of the omit compiler directives, see the SQLite website (http://sqlite.org/ compile.html#omitfeatures). SQLITE_ZERO_MALLOC 286 | Appendix A: SQLite Build Options APPENDIX B sqlite3 Command Reference The sqlite3 program is a command-line interface, or shell, that allows the user to interactively issue SQL commands and display the results. This can be used to try out queries, test existing databases, debug problems, or just play around and learn SQL. The sqlite3 program is similar to the mysql application for MySQL, the pgsql appli- cation for PostgreSQL, the sqlplus application for Oracle, or the sqlcmd application for SQL Server. Once sqlite3 has started up and opened a database, the main sqlite> prompt is dis- played. At this point, SQL statements can be entered. SQL statements should be ter- minated with a semicolon, and will be executed immediately. The results (if any) will then be displayed. When the database is ready to execute a new statement, the main command prompt will appear. Longer SQL statements can be entered on multiple lines. In this case, additional lines will display the continue ...> prompt, which indicates this line is continuation of the previous line or lines. Remember to enter a semicolon to terminate and execute an SQL statement. The sqlite3 source code is included in most SQLite distributions as the source file shell.c. This, along with the amalgamation files, are all that is required to build the sqlite3 application. For more information on building sqlite3, and what options are available, see “Building” on page 21 and Appendix A. A number of other third-party, general-purpose, shell-type programs exist for SQLite, including a number of GUI-based applications. See the SQLite website (http://www .sqlite.org/cvstrac/wiki?p=ManagementTools) for references to other interactive utilities and tools. 287 Command-Line Options The sqlite3 tool understands the following command-line format: sqlite3 [options...] [database [SQL_string]] Options are given first, followed by an optional database filename. This database will be opened as the main database. If the database file does not exist, it will be created. If no database filename is given (or an empty string is given), a file-backed temporary database will be created. If the database name :memory: is given, an in-memory database will be created. If a database filename is given, an optional SQL string can be provided. This string may consist of one or more SQL statements separated by semicolons. The SQL statements need to be passed as a single argument, so they will most likely need to be enclosed in quotes. If present, sqlite3 will open the database file, execute the provided SQL state- ments, and exit. Dot-commands cannot be included in the SQL string. If no SQL string is given, sqlite3 will provide an interactive prompt and accept either SQL or dot- commands from the terminal interface. The interactive startup sequence will attempt to locate and open the .sqliterc init file in the current user’s home directory. If the file exists, lines will be read and executed before any other processing (including the command-line SQL string). The init file may contain both dot-commands and SQL statements. All SQL statements must end in a semicolon. The init file is processed before the command-line options. This allows the command-line options to override any dot-commands that are in the init file. Recognized command-line options are: -bail Turns the bail flag on. Batch files will stop processing if an error is encountered. See the .bail command for more details. -batch Forces batch-style I/O. This will suppress things like the welcome banner and command prompts. -column Sets the output mode to column. See the .mode command for more details -csv Sets the output mode to csv. See the .mode command for more details. -echo Turns echo mode on. See the .echo command for more details. -header -noheader Turns headers on or off. See the .headers command for more details. -init filename If this is an interactive session, use this file as an init file rather than the .sqliterc file. 288 | Appendix B: sqlite3 Command Reference -interactive Forces interactive style I/O. This includes things like the welcome banner and command prompts. -help Displays a summary of the command-line options and exits. -html Sets the output mode to html. See the .mode command for more details. -line Sets the output mode to line. See the .mode command for more details. -list Sets the output mode to list. See the .mode command for more details. -nullvalue string Sets the NULL display string. See the .nullvalue command for more details. -separator string Sets the separator string. See the .separator command for more details. -version Prints the SQLite version and exits. Options will also be recognized if they are prefaced with two dashes, rather than just one. Interactive Dot-Commands This section covers the sqlite3 dot-commands. Dot-commands control the mode and configuration of the sqlite3 utility and, in some cases, the underlying SQLite library. Normally, any input to the sqlite3 utility is assumed to be an SQL statement and is passed to the SQLite library for processing. To distinguish these commands from SQL statements, all of the dot-commands start with a period, or “dot.” Hence, the name. Unlike SQL statements, dot-commands do not end in a semicolon. The whole com- mand must be given on a single line of input. It is important to understand that the dot-commands are implemented by the sqlite3 program itself, and not the underlying SQLite library. Dot-commands cannot be issued to the SQLite statement APIs. For example, the core SQLite library has no import capability. The functionality of the .import command is provided by the sqlite3 code, using the standard SQLite API calls. If you want to include an import function in your application, you can either write your own, using the standard SQLite API calls, or you can rip the code out of shell.c and attempt to adapt it for your own purposes. Several of the dot-commands turn configuration flags on or off. If a number is given, 0 is false, or off, and all other numbers are taken to be true, or on. The strings on and yes are recognized to be true, or on, and all other values are taken to be false, or off. Interactive Dot-Commands | 289 .backup Perform a low-level copy of a database to file Common Usage .backup [database_name] filename Description The .backup command performs a low-level copy of an open or attached database to the provided filename. This command is safe to run against an active source database (although it may not succeed). If the database name is not provided, the main database will be backed up. This command is frequently used to write-back an active in-memory database to a data- base file. This is the opposite of the .restore command. .bail Stop if an error is encountered Common Usage .bail switch Description The .bail command controls the error-handling behaviors. It only affects noninteractive ses- sions, when SQL commands are being read from a file or from standard input. If the bail flag is set, any error will cause the processing to stop. The default is off. .databases List all of the currently attached databases Common Usage .databases Description The .databases command generates a table of all the currently attached databases. The format of the table is: Column name Column type Meaning seq Integer Database sequence number name Text Logical database name file Text Path and name of database file The first database (sequence number 0) should always have the name main. This is the database that was first opened. The second database (sequence number 1), if present, should always be the temp database, where any temporary objects are created. Any attached databases will be listed after that. .backup 290 | Appendix B: sqlite3 Command Reference .dump Produce an SQL dump file Common Usage .dump [table-pattern ...] Description The .dump command generates a series of SQL commands suitable to recreate one or more tables from the database. If no parameters are given, every table in the main database will be dumped. If one or more parameters are given, only those tables in the main database that match the provided LIKE patterns will be dumped. To record the .dump output to a file, use the .output command. A dump will preserve the values in any ROWID alias column (an INTEGER PRIMARY KEY col- umn), but it will not preserve more general ROWID values. .echo Turn command echoing on or off Common Usage .echo switch Description The .echo command turns the SQL command echoing on or off. If set to on, any SQL com- mands will be printed before being executed. This is most useful for noninteractive sessions. The default is off. .exit Quit and exit the sqlite3 application Common Usage .exit Description The .exit command quits and exits the sqlite3 application. .explain Format output for EXPLAIN SQL command Common Usage .explain [switch] Description The .explain command sets several sqlite3 parameters (such as mode and column width), making it easier to view the output from any EXPLAIN SQL commands. When turned off, the previous settings are restored. If no parameter is given, the explain mode will be turned on. Turning explain on when it is already on resets all shell parameters to the default explain mode. The default is off. .explain Appendix B: sqlite3 Command Reference | 291 .headers Control display of column names and headers Common Usage .headers switch Description The .headers command controls the display of column names and headers. If on, sqlite3 will generate a row of column names before displaying any output. In some modes, a separator between the column names and the resulting data will also be displayed. The default is off. .help Display help Common Usage .help Description The .help command displays each sqlite3 dot-command along with a brief summary of the command syntax and purpose. .import Import an external data file into a table Common Usage .import filename table-name Description The .import command attempts to import data from an external file and insert it into the specified table. The table must already exist. The file should be a text file with one row per line. Each line should consist of a value separated by the current separator string. The values in each line must match the number and order of columns returned by the command SELECT * FROM table-name. The import feature does not understand any form of value quotation or character escape, so any instance of the separator string within a value will cause an error. Any attempt to import a file that uses quotations will result in the quotations being taken as a literal part of the value. The built-in import functionality is extremely simple, and is not designed to work with robust file formats. If you have a need to frequently import a specific data format (including so-called universal formats, such as CSV), it may be simpler and easier to write your own importer. This can often be done reasonably quickly in the scripting language of your choice, or by writing a read-only external module. .headers 292 | Appendix B: sqlite3 Command Reference .indices Display all of the indexes associated with one or more tables Common Usage .indices [table-pattern] Description The .indices command displays the name of different indexes in the main database. If no parameter is given, the name of every index will be displayed. If an optional table matching pattern is given, the name of any index associated with a table that matches the provided LIKE pattern will be displayed. Only one table pattern is permitted. Index names will be returned one per line, with no additional information. .iotrace Direct I/O trace information to a file Common Usage .iotrace [filename|-] Description The .iotrace command controls I/O trace debugging. If a filename is given, I/O trace will be turned on in the main library and directed to the file. If a single dash is given, I/O trace will be redirected to standard output. If no parameters are given, I/O trace will be disabled. The .iotrace command is only available if both the SQLite library and sqlite3 are compiled with the directive SQLITE_ENABLE_IOTRACE. .load Load a dynamic extension Common Usage .load filename [entry-point] Description The .load command attempts to load and link an SQLite dynamic extension. If no entry point is given, the default sqlite3_extension_init name will be used. Extensions that redefine existing SQL functions (including built-in functions) must be loaded using this command. Using the SQL function load_extension() will not work, as an extension cannot redefine an existing function while an SQL statement is being processed (for example, the statement executing the SQL function load_extension()). .load Appendix B: sqlite3 Command Reference | 293 .log Turn logging on or off Common Usage .log (filename|stdout|stderr|off) Description The .log command designates an output file for messages sent to sqlite3_log(). The SQLite library logs most error conditions, to assist with debugging. Custom extensions (including custom SQL functions, collations, and virtual tables) may also log messages. .mode Set the output mode Common Usage .mode (column[s]|csv|html|insert|line[s]|list|tabs|tcl) [table-name] Description The .mode command sets the output mode. This determines how the output data is formatted and presented. The optional table name is only used by the insert mode. The default mode is list. Supported modes include: column Output is in a tabular format, with one row per line. The width of each column is defined by the values provided by the .width command. Output will be clipped to fit into the column width. If headers are on, the first two lines will be the column names, followed by separator dashes. In most cases, this is the easiest format for humans to read, assuming the column widths are set to something useful. csv Output is in comma-separated values, with one row per line. Each value is separated by a single comma character with no trailing space. If headers are on, the first line will be the set of column names. If a value contains a comma, the value will be enclosed in double quotes. If the value contains a double quote character, the SQL rules for escaping quotes will be used and the literal double quote will be replaced with two double quote characters in sequence. This is not what many CSV importers expect, however. html Output is in an HTML table format. Each database row will be output as a element (table row). If headers are on, the first table row element will contain the column names. For older versions of HTML, the output is suitable to place within a element. insert Output is a series of SQL INSERT statements. If a table name is given as part of the .mode command, that name will be used as the table in the INSERT statement. If no table name is given, the name table will be used. The header flag has no effect in this mode. .log 294 | Appendix B: sqlite3 Command Reference line Output is one value per line in the format column_name = value (with some possible leading white space before the column name). Each row is separated by a blank line. The header flag has no effect in this mode. list Output consists of a sequence of values, with one row per line. Each value is separated by the current separator string (default is |, a bar or pipe). No separator will appear after the final value. There is no provision to quote or escape the separator string if it appears in a value. If headers are on, the first line will contain the column names. In general, this is the easiest format to parse, assuming a unique separator is used. tabs Output is in tab-separated values, with one row per line. There is no provision to quote or escape tab characters included in values. If headers are on, the first line will contain the column names. tcl Output consists of a sequence of values, with one row per line. Each value is separated by the current separator string (default is |, a bar or pipe). A separator will be included after the final value. All values will be put in double quotes, although there is no provision to quote or escape double quote characters included in values. If headers are on, the first line will contain the column names. This mode is designed to be used with the Tcl script- ing language. If the output of sqlite3 is fed directly into a script or other automated system, be very sure you understand how values are delimited and that there are provisions to quote, escape, or avoid any separators within the returned values. Of specific note, be aware that the CSV format is not nearly as universal as many people think. While the format works well for numbers and simple data, if text values require quoting or escape sequences, compatibility should be tested before being used. .nullvalue Set the string used to represent a NULL output Common Usage .nullvalue string Description The .nullvalue command sets the string used to represent NULL outputs in any output mode other than insert. The default is an empty string. .nullvalue Appendix B: sqlite3 Command Reference | 295 .output Set the output destination Common Usage .output (filename|stdout) Description The .output command sets the output destination. By default, any command output is di- rected to the terminal interface (in the case of interactive sessions) or the program output (in the case of batch sessions). Given a filename, the .output command will redirect any command output to a file. The file will be created if it does not exist. If the file does exist, it will be truncated. If the output is being redirected to a file, it will not also be displayed to the terminal. Commands and command prompts will not be output to the file, making this a suitable way to generate SQL dump files using the .dump command. To reset the output to the terminal interface, set the output to stdout (standard output). .prompt Set the command prompt Common Usage .prompt main [continue] Description The .prompt command modifies the sqlite3 command prompts. The main prompt is used for most commands, while the continue prompt is used for multiline SQL commands. The default main prompt is sqlite> and the default continue prompt is ...>. .quit Quit and exit the sqlite3 application Common Usage .quit Description The .quit command quits and exits the sqlite3 application. .read Execute SQL commands from a file Common Usage .read filename Description The .read command reads and executes dot-commands and SQL statements from a file. .output 296 | Appendix B: sqlite3 Command Reference .restore Perform a low-level copy of a database file to a database Common Usage .restore [database_name] filename Description The .restore command performs a low-level copy of a database file into an open or attached database. If the database name is not provided, the main database will be populated. This command is frequently used to populate an active in-memory database from a database file. This command is safe to run against an active source database (although it may not succeed). This is the opposite of the .backup command. .schema Display SQL creation commands for schema Common Usage .schema [table-pattern] Description The .schema command displays the SQL commands used to create the schema (tables, views, indexes, etc.). If no parameter is given, the CREATE command for every object in the main and temp databases will be displayed. A single parameter can also be given, in which case only those objects associated with tables that match the LIKE pattern will be displayed. .separator Define the string used as a column separator Common Usage .separator string Description The .separator command defines the separator string used between columns when the list output mode is set. The list mode is the default mode and the default separator string is | (a pipe or bar). .show Display current sqlite3 settings Common Usage .show Description The .show command displays the current values for a small number of the sqlite3 settings, including the state of the echo flag, the explain flag, the headers flag, the current output mode, the NULL string, the output destination, the separator string, and any configured column widths. .show Appendix B: sqlite3 Command Reference | 297 .tables Display the list of table and view names Common Usage .tables [table-pattern] Description The .tables command displays the list of names for all of the table and view objects that are found in the main and temp databases. .timeout Set a lock retry timer Common Usage .timeout milliseconds Description The .timeout command is used to set a retry timer. If a timer value is set and a locked database is encountered, rather than immediately returning a “database locked” error, in most cases SQLite will keep attempting to reacquire the lock until the timer expires. A value of 0 or less disables the timeout value. .timer Enable or disable CPU time measurements Common Usage .timer switch Description The .timer command can be used to enable or disable CPU time measurements. If enabled, SQLite will track and display the user time and system time required to process each request. .width Set the display width for each column Common Usage .width numb [numb ...] Description The .width command is used to set the default width of each column. The width values are used to format the display when the output mode is set to column. The first value is used for the width of the first column, the second value is used for the second column, and so on. If no values have been set, a column width of 10 will be used. Up to 100 column widths can be specified. .tables 298 | Appendix B: sqlite3 Command Reference APPENDIX C SQLite SQL Command Reference This appendix lists the SQL commands and syntax that are supported by SQLite. SQL statements consist of a single command and any required parameters. Command state- ments are separated by a semicolon. Technically, standalone statements do not need to be terminated with a semicolon, but most interactive environments require the use of a semicolon to indicate that the current command statement is complete and should be executed. For example, the C API sqlite3_exec() does not require that command statements end with a semicolon, but interactive use of sqlite3 requires ending each statement with a semicolon. In most situations where a table name is called for, a view name can be used instead. As noted in the syntax diagrams, in most instances where any object identifier is used (table name, view name, etc.), the name can be qualified with a logical database name to prevent any ambiguity between objects in different databases that share a similar name (see ATTACH DATABASE in this appendix). If the object is unqualified, it will be searched for in the temp database, followed by the main database, followed by each attached database, in order. If an unqualified identifier appears in a CREATE statement, the object will be created in the main database, unless the statement contains some type of CREATE TEMPORARY syntax. Object identifiers that use nonstandard characters must be quoted. See “Basic Syntax” on page 30 for more info. The SELECT, UPDATE, and DELETE commands contain clauses that are used to define search criteria on table rows. These table references can include the nonstandard phrases INDEXED BY or NOT INDEXED, to indicate whether the query optimizer should (or should not) attempt to use an index to satisfy the search condition. These extensions are in- cluded in SQLite to assist with testing, debugging, and hand-tuning queries. Their use in production code is not recommended, and therefore they are not included in the syntax diagrams or command explanations found in this appendix. For more infor- mation, see the SQLite website (http://www.sqlite.org/lang_indexedby.html). 299 Finally, be aware that the syntax diagrams presented with each command should not be taken as the definitive specification for the full command syntax. Some rarely used, nonstandard syntax (such as the INDEXED BY extension discussed in the previous para- graph) are not included in these diagrams. Similarly, there are possible syntax combi- nations that the diagrams will indicate are possible, but do not actually form logical statements. For example, according to the syntax diagrams, a JOIN operator can contain both a prefixed NATURAL condition, as well as a trailing ON or USING condition. This isn’t possible in practice, as a join is limited to only one type of condition. While it would have been possible to present the diagram with only the allowed syntax, the diagram would have become much larger and much more complex. In such situations, it was decided that making the diagram easy to understand was more important than making it walk an absolute line on what was allowed or not allowed. Thankfully, such situations are reasonably rare. Just don’t assume that because the parser can parse it means that the command makes sense to the database engine. SQLite SQL Commands The following SQL commands and syntax are supported by SQLite. ALTER TABLE Modify an existing table Syntax Common Usage ALTER TABLE database_name.table_name RENAME TO new_table_name; ALTER TABLE database_name.table_name ADD COLUMN column_def...; Description The ALTER TABLE command modifies an existing table without performing a full dump and reload of the data. The SQLite version of ALTER TABLE supports two basic operations. The RENAME variant is used to change the name of a table, while ADD COLUMN is used to add a new column to an existing table. Both versions of the ALTER TABLE command will retain any existing data in the table. ALTER TABLE 300 | Appendix C: SQLite SQL Command Reference RENAME The RENAME variant is used to “move” or rename an existing table. An ALTER TABLE...RENAME command can only modify a table in place, it cannot be used to move a table to another database. A database name can be provided when specifying the original table name, but only the table name should be given when specifying the new table name. Indexes and triggers associated with the table will remain with the table under the new name. If foreign key support is enabled, any foreign keys that reference this table will also be updated. View definitions and trigger statements that reference the table by name will not be modified. These statements must be dropped and recreated, or a replacement table must be created. ADD COLUMN The ADD COLUMN variant is used to add a new column to the end of a table definition. New columns must always go at the end of the table definition. The existing table rows are not actually modified, meaning that the added columns are implied until a row is modified. This means the ALTER TABLE...ADD COLUMN command is quite fast, even for large tables, but it also means there are some limitations on the columns that can be added. The added column: • Cannot have a PRIMARY KEY constraint • Cannot have a UNIQUE constraint • Must have a literal, non-NULL default value if a NOT NULL constraint is given • Cannot have a default of CURRENT_TIME, CURRENT_DATE, or CURRENT_TIMESTAMP If foreign key constraints are enabled and the added column is defined as a foreign key (it has a REFERENCES clause), the new column must have a default of NULL. Additionally, if the new column has a CHECK constraint, that constraint will only be applied to new values. This can lead to data that is inconsistent with the CHECK. There is no way to remove a column once it has been added. See Also CREATE TABLE ANALYZE Compute index meta-data Syntax ANALYZE Appendix C: SQLite SQL Command Reference | 301 Common Usage ANALYZE; ANALYZE database_name; ANALYZE database_name.table_name; Description The ANALYZE command computes and records statistical data about database indexes. If avail- able, this data is used by the query optimizer to compute the most efficient query plan. If no parameters are given, statistics will be computed for all indexes in all attached databases. You can also limit analysis to just those indexes in a specific database, or just those indexes associated with a specific table. The statistical data is not automatically updated as the index values change. If the contents or distribution of an index changes significantly, it would be wise to reanalyze the appropriate database or table. Another option would be to simply delete the statistical data, as no data is usually better than incorrect data. Data generated by ANALYZE is stored in one or more tables named sqlite_stat#, starting with sqlite_stat1. These tables cannot be manually dropped, but the data inside can be altered with standard SQL commands. Generally, this is not recommended, except to delete any ANALYZE data that is no longer valid or desired. By default, the ANALYZE command generates data on the number of entries in an index, as well as the ratio of unique values to total values. This ratio is computed by dividing the total number of entries by the number of unique values, rounding up to the nearest integer. This data is used to compute the cost difference between a full-table scan and an indexed lookup. If SQLite is compiled with the SQLITE_ENABLE_STAT2 directive, then ANALYZE will also generate an sqlite_stat2 table that contains a histogram of the index distribution. This is used to compute the cost of targeting a range of values. There is one known issue with the ANALYZE command. When generating the sqlite_stat1 table, ANALYZE must calculate the number of unique values in an index. To accomplish this, the ANALYZE command uses the standard SQL test for equivalence between index values. This means that if a single-column index contains multiple NULL entries, they will each be con- sidered a nonequivalent, unique value (since NULL != NULL). As a result, if an index contains a large number of NULL values, the ANALYZE data will incorrectly consider the index to have more uniqueness than it actually does. This can incorrectly influence the optimizer to pick an index-based lookup when a full-table scan would be less expensive. Due to this behavior, if an index contains a noticeable percentage of NULL entries (say, 10 to 15% or more) and it is common to ask for all of the NULL (or non-NULL) rows, it is recommended that the ANALYZE data for that index is discarded. See Also CREATE INDEX, SELECT ANALYZE 302 | Appendix C: SQLite SQL Command Reference ATTACH DATABASE Attach a database file Syntax Common Usage ATTACH DATABASE 'filename' AS database_name; Description The ATTACH DATABASE command associates the database file filename with the current database connection under the logical database name database_name. If the database file filename does not exist, it will be created. Once attached, all references to a specific database are done via the logical database name, not the filename. All database names must be unique, but (when shared cache mode is not enabled) attaching the same filename multiple times under different database names is properly supported. The database name main is reserved for the primary database (the one that was used to create the database connection). The database name temp is reserved for the database that holds temporary tables and other temporary data objects. Both of these database names exist for every database connection. If the filename :memory: is given, a new in-memory database will be created and attached. Multiple in-memory databases can be attached, but they will each be unique. If an empty filename is given ( '' ), a temporary file-backed database will be created. Like an in-memory database, each database is unique and all temporary databases are automatically deleted when they are closed. Unlike an in-memory database, file-based temporary databases can grow to large sizes without consuming excessive memory. All databases attached to a database connection must share the same text encoding as the main database. If you attempt to attach a database that has a different text encoding, an SQLite logic error will be returned. If the main database was opened with sqlite3_open_v2(), each attached database will be opened with the same flags. If the main database was opened read-only, all attached databases will also be read-only. Associating more than one database to the same database connection enables the execution of SQL statements that reference tables from different database files. Transactions that involve multiple databases are atomic, assuming the main database is not an in-memory database. In that case, transactions within a given database file continue to be atomic, but operations that bridge database files may not be atomic. If any write operations are performed on any database, a master journal file will be created in association with the main database. If the main database is located in a read-only area, the master journal file cannot be created and the operation will fail. If some databases are read- only and some are read/write, make sure the main database is one of the databases that is located in a read/write area. ATTACH DATABASE Appendix C: SQLite SQL Command Reference | 303 Any place SQLite expects a table name, it will accept the format database_name.table_name. This can be used to refer to a table within a specific database that might otherwise be ambiguous. See Also DETACH DATABASE, encoding [PRAGMA, Ap F], temp_store [PRAGMA, Ap F], sqlite3_open() [C API, Ap G] BEGIN TRANSACTION Open an explicit transaction Syntax Common Usage BEGIN; BEGIN EXCLUSIVE TRANSACTION; Description The BEGIN TRANSACTION command starts an explicit transaction. Once started, an explicit transaction will remain open until the transaction is either committed using COMMIT TRANSAC TION, or rolled back using ROLLBACK TRANSACTION. Transactions are used to group multiple discrete commands (such as a sequence of INSERT commands that insert rows into cross- referenced tables) into a single logical command. Transactions are ACID-compliant, in that they are atomic, consistent, isolated, and durable. They’re an important part of correct database design and database use. For more information, see “Transaction Control Language” on page 51. All changes and modifications to a database are done within a transaction. Normally, SQLite is in autocommit mode. In this mode, each and every statement that might modify the data- base is automatically wrapped in its own transaction. Each command begins a transaction, executes the given command statement, and attempts to commit the changes. If any error occurs, the wrapper transaction is rolled back. The BEGIN command turns off autocommit mode, opening a transaction and leaving it open until it is explicitly committed or rolled back. This allows multiple commands (and multiple modifications) to be packaged into a single transaction. Once a COMMIT or ROLLBACK is issued, the database connection is put back into autocommit mode. BEGIN TRANSACTION 304 | Appendix C: SQLite SQL Command Reference Transactions cannot be nested. For that functionality, use SAVEPOINT. Executing a BEGIN com- mand while the database connection is already in a transaction will result in an error, but will not change the state of the preexisting transaction. There is a significant cost associated with committing a transaction. In autocommit mode, this cost is seen by every command. In some situations, it can be prudent to wrap several commands into a single transaction. This helps amortize the transaction cost across several statements. When doing large operations, such as bulk inserts, it is not unusual to wrap several hundred, or even a thousand or more INSERT commands into a single transaction. The only caution in doing this is that a single error can cause the whole transaction to rollback, so you need to be prepared to re-create all of the rolled back INSERT statements. In SQLite, transactions are controlled by locks. You can specify the locking behavior you want with the modifier DEFERRED, IMMEDIATE, or EXCLUSIVE. The default mode is DEFERRED, in which no locks are acquired until they are needed. This allows the highest level of concurrency, but also means the transaction may find itself needing a lock it cannot acquire, and may require a rollback. The IMMEDIATE mode attempts to immediately acquire the reserved lock, allowing other connections to continue to read from the database, but reserving the right to elevate itself to write status at any time. Starting an EXCLUSIVE transaction will attempt to grab the exclusive lock, allowing full access to the database, but denying access by any other database connection. Higher locking levels means greater certainty that a transaction can be success- fully committed at the cost of lower levels of concurrency. For more information on SQLite’s locking and concurrency model, see http://sqlite.org/lock ingv3.html. See Also COMMIT TRANSACTION, ROLLBACK TRANSACTION, SAVEPOINT, RELEASE SAVEPOINT COMMIT TRANSACTION Finish and commit a transaction Syntax Common Usage COMMIT; Description The COMMIT TRANSACTION command attempts to close and commit any changes made during the current transaction. The alias END TRANSACTION may also be used. If a COMMIT command is made while SQLite is in autocommit mode, an error will be issued. COMMIT TRANSACTION Appendix C: SQLite SQL Command Reference | 305 If the COMMIT is successful, the database will be synchronized and any modifications made within the transaction will become a permanent part of the database record and the database connection will be put back in autocommit mode. If the commit is not successful, the transaction may or may not be rolled back, depending on the type of error. If the transaction is not rolled back, you can usually just reissue the COMMIT command. If the transaction is rolled back, all modifications made as part of the trans- action are lost. You can determine the specific state of the database connection using the sqlite3_get_autocommit() API call, or by trying to issue the BEGIN command. If the database returns a logical error as a result of the BEGIN command, the database is still in a valid trans- action. You can also issue the ROLLBACK command, which will either roll back the transaction if it is still in place, or return an error if the transaction was already rolled back. There is a significant cost associated with committing a transaction. See BEGIN TRANSACTION for more details on how to reduce this cost. See Also BEGIN TRANSACTION, ROLLBACK TRANSACTION, END TRANSACTION, sqlite3_get_autocommit() [C API, Ap G] CREATE INDEX Define and create a new table index Syntax Common Usage CREATE INDEX index_name ON table_name ( column_name COLLATE NOCASE ); CREATE UNIQUE INDEX database_name.index_name ON table_name ( col1, col2 ,... ); CREATE INDEX 306 | Appendix C: SQLite SQL Command Reference Description The CREATE INDEX command creates a user-defined index. Upon creation, the index is popu- lated from the existing table data. Once created, the index will be automatically maintained, so that modifications to the referenced table will be reflected in the index. The query optimizer will automatically consider any indexes that have been created. Indexes cannot be created on virtual tables or views. An index can reference multiple columns, but all of the columns must be from the same table. In the case of multicolumn indexes, the index will be built in the same order as the column listing. For performance-related indexes, the column ordering can be very important. See “Order Matters” on page 109 for more details. The table must be in the same database as the index. To create an index on a temporary table, create the index in the temp database. If a table is dropped, all associated indexes are also dropped. A user-defined index may also be explicitly dropped with the DROP INDEX command. If the optional UNIQUE clause is included, the index will not allow inclusion of equivalent index entries. An index entry includes the whole set of indexed columns, taken as a group, so you may still find duplicate column values in a unique multicolumn index. As usual, NULL entries are considered unique from each other, so multiple NULL entries may exist even in a unique single-column index. An optional collation can be provided for each column. By default, the column’s native col- lation will be used. If an alternate collation is provided, the index can only be used in queries that also specify that collation. Additionally, each indexed column can specify an ascending (ASC) or descending (DESC) sort order. By default, all indexed columns will be sorted in an ascending order. Use of descending indexes requires a modern file format. If the database is still using the legacy file format, descending indexes will not be supported and the DESC keyword will be silently ignored. SQLite indexes include a full copy of the indexed data. Be cautious of your database size when indexing columns that consist of large text or BLOB values. Generally, indexes should only be created on columns that have a relatively unique set of values. If any single value appears in more than 10% to 15% of the rows, an index is usually inadvisable. It is almost always unwise to index a Boolean column, or any similar column that holds relatively few values. There is a cost associated with maintaining indexes, so they should only be created when they serve some purpose. Creating an index that already exists will normally generate an error. If the optional IF NOT EXISTS clause is provided, this error is silently ignored. This leaves the original definition (and data) in place. See Also DROP INDEX, ANALYZE, REINDEX, CREATE TABLE, COLLATE [SQL Expr, Ap D] CREATE INDEX Appendix C: SQLite SQL Command Reference | 307 CREATE TABLE Define and create a new table Syntax column-def: type-name: CREATE TABLE 308 | Appendix C: SQLite SQL Command Reference column-constraint: CREATE TABLE Appendix C: SQLite SQL Command Reference | 309 table-constraint: CREATE TABLE 310 | Appendix C: SQLite SQL Command Reference foreign-key-clause: conflict-clause: CREATE TABLE Appendix C: SQLite SQL Command Reference | 311 Common Usage CREATE TABLE database_name.table_name ( c1_name c1_type, c2_name c2_type... ); CREATE TABLE database_name.table_name AS SELECT * FROM... ; CREATE TABLE tbl ( a, b, c ); CREATE TABLE people ( people_id INTEGER PRIMARY KEY, name TEXT ); CREATE TABLE employee ( employee_id INTEGER PRIMARY KEY NOT NULL, name TEXT NOT NULL, start_date TEXT NOT NULL DEFAULT CURRENT_DATE, parking_spot INTEGER UNIQUE ); Description The CREATE TABLE command is used to define a new table. It is one of the most complex SQL commands understood by SQLite, though nearly all of the syntax is optional. A new table can be created in a specific database by qualifying the table name with an explicit database name. If one of the optional keywords TEMP or TEMPORARY is present, any database name given as part of the table name will be ignored, and the new table will be created in the temp database. Creating a table that already exists will normally generate an error. If the optional IF NOT EXISTS clause is provided, this error is silently ignored. This leaves the original definition (and data) in place. There are two variations of CREATE TABLE. The difference is in how the columns are defined. The least common variation uses a simple AS SELECT subquery to define the structure and initial contents of the table. The number of columns and the column names will be taken from the result set of the subquery. The rows of the result set will be loaded into the table as part of the table creation process. Because this variation provides no way to define column affinities (typical datatypes), keys, or constraints, it is typically limited to defining “quick and dirty” temporary tables. To quickly create and load structured data, it is often better to create a table using the standard notation and then use an INSERT...SELECT command to load the table. The standard notation explicitly defines a list of columns and table constraints. Basic format The most common way to define a table structure is to provide a list of column definitions. Column definitions consist of a name and a type, plus zero or more column-level constraint definitions. The list of column definitions is followed by a list of table-level constraints. For the most part, column-level constraints and table-level constraints are very similar. The main difference is that column constraints apply to the values found in a single column, while table constraints can deal with one or more columns. It is possible to define most column constraints as table- level constraints that only reference a single column. For example, a multicolumn primary key must be defined as a table constraint, but a single-column primary key can be defined as either a table constraint or a column constraint. The column name is a standard identifier. If nonstandard characters (such as a space or a hyphen) are used, the identifier must be quoted in the CREATE TABLE statement as well as any other reference. CREATE TABLE 312 | Appendix C: SQLite SQL Command Reference The column name is followed by a type indicator. In SQLite, the type is optional, since nearly any column can hold any datatype. SQLite columns do not technically have types, but rather have type affinities. An affinity describes the most favored type for the column and allows SQLite to do implicit conversions in some cases. An affinity does not limit a column to a specific type, however. The use of affinities also accounts for the fact that the type format is extremely flexible, allowing type names from nearly any dialect of SQL. For more specifics on how type affinities are determined and used, see “Column types” on page 36. If you want to make sure a specific affinity is used, the most straightforward type names are INT, REAL, TEXT, or BLOB. SQLite does not use precision or size limits internally. All integer values are signed 64-bit values, all floating-point values are 64-bit values, and all text and BLOB values are variable length. All tables have an implied root column, known as ROWID, that is used internally by the database to index and store the database table structure. This column is not normally displayed or returned in queries, but can be accessed directly using the name ROWID, _ROWID_, or OID. The alternate names are provided for compatibility with other database engines. Generally, ROWID values should never be used or manipulated directly, nor should the ROWID column be directly used as a table key. To use a ROWID as a key value, it should be aliased to a user-defined column. See “PRIMARY KEY constraint” on page 314. Column constraints Each column definition can include zero or more column constraints. Column constraints follow the column type indicator; there is no comma or other delimiter between basic column definitions and the column constraints. The constraints can be given in any order. Most of the column constraints are easy to understand. The PRIMARY KEY constraint is a bit unique, however, and is discussed below, in its own section. The NOT NULL constraint prohibits the column from containing NULL entries. The UNIQUE constraint requires all the values of the column to be unique. An automatic unique index will be created on the column to enforce this constraint. Be aware that UNIQUE does not imply NOT NULL, and unique columns are allowed to have more than one NULL entry. This means there is a tendency for columns with a UNIQUE constraint to also have a NOT NULL constraint. The CHECK constraint provides an arbitrary user-defined expression that must remain true. The expression can safely access any column in the row. The CHECK constraint is very useful to enforce specific data formats, ranges or values, or even specific datatypes. For example, if you want to be absolutely sure nothing but integer values are entered into a column, you can add a constraint such as: CHECK ( typeof( column_name ) == 'integer' ) The DEFAULT constraint defines a default value for the column. This value is used when an INSERT statement does not include a specific value for this column. A DEFAULT can either be a literal value or, if enclosed in parentheses, an expression. Any expression must evaluate to a constant value. You can also use the special values CURRENT_TIME, CURRENT_DATE, or CURRENT_ TIMESTAMP. These will insert an appropriate text value indicating the time the row was first created. If no DEFAULT constraint is given, the default value will be NULL. CREATE TABLE Appendix C: SQLite SQL Command Reference | 313 The COLLATION constraint is used to assign a specific collation to a column. This not only defines the sort order for the column, it also defines how values are tested for equality (which is important for things such as UNIQUE constraints). SQLite includes three built-in collations: BINARY (the default), NOCASE, and RTRIM. BINARY treats all values as binary data that must match exactly. NOCASE is similar to binary, only it is case-insensitive for ASCII text values (in specific, character codes < 128). Also included is RTRIM (right-trim), which is like BINARY, but will trim any trailing whitespace from TEXT values before doing comparisons. Finally, columns can contain a REFERENCES foreign key constraint. If given as a column con- straint, the foreign table reference can contain no more than one foreign column name. If no column references are given, the foreign table must have a single-column primary key. For more information on foreign keys, see the section “Foreign Keys” on page 89. Note that a column-level foreign key constraint does not actually contain the words FOREIGN KEY. That syntax is for table-level foreign key constraints. Table constraints Generally, the table-level constraints are the same as the column-level constraints, except that they operate across more than one column. In most cases, table-level constraints have similar syntax to their column-level counterparts, with the addition of a list of columns that are applied to the constraint. The UNIQUE table constraint requires that each group of column values must be UNIQUE from all the other groups within the table. In the case of a multicolumn UNIQUE constraint, any individual column is allowed to have duplicate values, it is only the group of column values, taken as a whole, that must remain unique. Both UNIQUE and PRIMARY KEY multicolumn con- straints can define individual column collations and orderings that are different from the individual column collations. The table-level CHECK constraint is identical to the column-level CHECK constraint. Both forms are allowed to use an arbitrary expression that references any column in the row. Finally, multicolumn foreign keys are defined with the FOREIGN KEY constraint. The list of local table columns must be the same size, and in the same order, as the foreign column list provided by the REFERENCES clause. For more information on foreign keys, see “Foreign Keys” on page 89. PRIMARY KEY constraint The PRIMARY KEY constraint is used to define the primary key (or PK) for the table. From a database design and theory standpoint, it is desirable for every table to have a primary key. The primary key defines the core purpose of the table by defining the specific data points that make each row a unique and complete record. From a practical standpoint, SQL does not require a table to have a PK. In fact, SQL does not require that rows within a table be unique. Nonetheless, there are some advantages to defining a primary key, especially when using foreign keys. In most cases a foreign key in one table will refer to the primary key of another table, and explicitly defining a primary key can make it easier to establish this relationship. SQLite also provides some additional features for single- column primary keys. There can be only one PRIMARY KEY constraint per table. It can be defined at either the column level or the table level, but each table can have only one. A PRIMARY KEY constraint implies a CREATE TABLE 314 | Appendix C: SQLite SQL Command Reference UNIQUE constraint. As with a standalone UNIQUE constraint, this will cause the creation of an automatic unique index (with one exception). In most database systems, PRIMARY KEY also implies NOT NULL, but due to a long-standing bug, SQLite allows the use of NULL entries in a primary key column. For proper behavior, be sure to define at least one column of the primary key to be NOT NULL. If a column has the type identifier INTEGER (it can be upper- or lowercase, but must be the exact word “integer”), an ascending collation (the default), and has a single-column PRIMARY KEY constraint, then that column will become an alias for the ROWID column. Behind the scenes, this makes an INTEGER PRIMARY KEY column the root column, used internally to index and store the database table. Using a ROWID alias allows for very fast row access without requiring a secondary index. Additionally, SQLite will automatically assign an unused ROWID value to any row that is inserted without an explicit column value. Columns defined as INTEGER PRIMARY KEY can really truly hold only integer values. Addition- ally, unlike other primary key columns, they have an inherent NOT NULL constraint. Default values are assigned using the standard ROWID allocation algorithm. This algorithm will auto- matically assign a value that is one larger than the largest currently used ROWID value. If the maximum value is met, a random (unused) ROWID value will be chosen. As rows are added and removed from a table, this allows ROWID values to be recycled. While recycling values is not a problem for internal ROWID values, it can cause problems for reference values that might be lurking elsewhere in the database. To avoid problems, the keyword AUTOINCREMENT can be used with an INTEGER PRIMARY KEY to indicate that automati- cally generated values should not be recycled. Default values assigned by AUTOINCREMENT will be one larger than the largest ROWID value that was ever used, but don't depend on each and every value being used. If the maximum value is reached, an error is returned. When using a ROWID alias to automatically generate keys, it is a common practice to insert a new row and call the SQL function last_insert_rowid(), or the C function sqlite3_last_insert_rowid(), to retrieve the ROWID value that was just assigned. This value can be used to insert or update rows that reference the newly inserted row. It is also always possible to insert a row with a specific ROWID (or ROWID alias) value. Conflict clause Nearly every column constraint and table constraint can have an optional conflict resolution clause. This clause can be used to specify what action SQLite takes if a command attempts to violate that particular constraint. Constraint violations most commonly happen when at- tempting to insert or update invalid row values. The default action is ON CONFLICT ABORT, which will attempt to back-out any changes made by the command that caused the constraint violation, but will otherwise attempt to leave any current transaction in place and valid. For more information on the other conflict resolution choices, see UPDATE. Note that the conflict resolution clause in UPDATE and INSERT applies to the actions taken by the UPDATE and INSERT commands themselves. Any conflict resolution clause found in a CREATE TABLE statement is applied to any command operating on the table. See Also DROP TABLE, INSERT, UPDATE, CREATE INDEX CREATE TABLE Appendix C: SQLite SQL Command Reference | 315 CREATE TRIGGER Create a new trigger action Syntax CREATE TRIGGER 316 | Appendix C: SQLite SQL Command Reference Common Usage CREATE TRIGGER database_name.trigger_name BEFORE INSERT ON table_name FOR EACH ROW BEGIN stmt1; stmt2; END; CREATE TRIGGER access_audit BEFORE UPDATE ON access FOR EACH ROW BEGIN INSERT INTO audit_trail VALUES ( OLD.level, NEW.level, CURRENT_TIMESTAMP ); END; Description The CREATE TRIGGER command creates a trigger and binds it to a table or view. When the conditions defined by the trigger are met, the trigger will “fire,” automatically executing statements found in the trigger body (the part between BEGIN and END). A table or view may have any number of triggers associated with it, including multiple triggers of the same type. If the optional TEMP or TEMPORARY keyword is present, a trigger will be created in the temp database. A trigger can also be made temporary by qualifying the trigger name with the da- tabase name temp. If the trigger name is qualified with a database name, specifying TEMP or TEMPORARY will result in an error, even if the given database name is temp. Temporary triggers can be attached to either temporary or standard (nontemporary) tables. A specific table instance can be chosen by qualifying the table name with a database name. In all other cases, the trigger and the table should be in the same database. Either the trigger name or the table name can be qualified with a database name (or both, if they match). Triggers associated with tables may be BEFORE or AFTER triggers. If no time is specified, BEFORE is used. The timing indicates if the trigger fires before or after the defined trigger action. In both cases, the action is verified before the trigger is fired. For example, a BEFORE INSERT trigger will not fire if the insert will cause a constraint violation. The trigger action can be either a DELETE, INSERT, or UPDATE statement that gets run against the trigger’s table. In the case of UPDATE, the trigger can fire when any column is updated, or only when one or more columns from the specified list is updated. Triggers associated with views must be INSTEAD OF triggers. The default timing for views is still BEFORE, so the INSTEAD OF must be specified. As the name indicates, INSTEAD OF triggers fire in the place of the defined action. Although views are read-only in SQLite, defining one or more INSTEAD OF DELETE, INSERT, or UPDATE trigger will allow those commands to be run against the view. Very often, views will have a whole series of INSTEAD OF triggers to deal with different combinations of column updates. The SQL standard defines both FOR EACH ROW as well as FOR EACH STATEMENT triggers. SQLite only supports FOR EACH ROW triggers, which fire once for each row affected by the specified condition. This makes the FOR EACH ROW clause optional in SQLite. Some popular databases that support both types of triggers will default to FOR EACH STATEMENT triggers, however, so explicit use of the FOR EACH ROW clause is recommended. Triggers also have an optional WHEN clause that is used to control whether the trigger actually fires or not. Don’t underestimate the WHEN clause. In many cases, the logic in the WHEN clause is more complex than the trigger body. CREATE TRIGGER Appendix C: SQLite SQL Command Reference | 317 The trigger body itself consists of one or more INSERT, UPDATE, DELETE, or SELECT statements. The first three commands can be used in the normal way. A SELECT statement can be used to call user-defined functions. Any results returned by a standalone SELECT statement will be ignored. Table identifiers within the trigger body cannot be qualified with a database name. All table identifiers must be from the same database as the trigger table. Both the WHEN clause and the trigger body have access to some additional column qualifiers. Columns associated with the trigger table (or view) may be qualified with the pseudo-identifier NEW (in the case of INSERT and UPDATE triggers) or OLD (in the case of UPDATE and DELETE triggers). These represent the before and after values of the row in question and are only valid for the current row that caused the trigger to fire. Commands found in a trigger body can also use the RAISE expression to raise an exception. This can be used to ignore, roll back, abort, or fail the current row in an error situation. For more information, see RAISE and UPDATE. There are some additional limits on trigger bodies. Within a trigger body, UPDATE and DELETE commands cannot use index overrides (INDEXED BY, NOT INDEXED), nor is the ORDER BY...LIMIT syntax supported (even if support has been properly enabled). The INSERT...DEFAULT VALUES syntax is also unsupported. If a trigger is fired as the result of a command with an explicit ON CONFLICT clause, the higher-level conflict resolution will override any ON CONFLICT clause found in a trigger body. If a trigger modifies rows from the same table it is attached to, the use of AFTER triggers is strongly recommended. If a BEFORE trigger modifies the rows that are part of the original statement (the one that caused the trigger to fire) the results can be undefined. Also, the NEW.ROWID value is not available to BEFORE INSERT triggers unless an explicit value has been provided. If a table is dropped, all of its triggers are automatically dropped. Similarly, if a table is renamed (via ALTER TABLE), any associated triggers will be updated. However, dropping or altering a table will not cause references found in a trigger body to be updated. If a table is dropped or renamed, make sure any triggers that reference it are updated as well. Failing to do so will cause an error when the trigger is fired. Creating a trigger that already exists will normally generate an error. If the optional IF NOT EXISTS clause is provided, this error is silently ignored. This leaves the original definition (and data) in place. One final note. Some of the syntax and many of the functional limitations of CREATE TRIG GER are checked at execution, not at creation. Just because the CREATE TRIGGER command returned without error doesn’t mean the trigger description is valid. It is strongly suggested that all triggers are verified and tested. If a trigger encounters an error, that error will be bubbled up to the statement that caused the trigger to fire. This can cause perplexing results, such as commands producing errors about tables or columns that are not part of the original statement. If a command is producing an unexplained or odd error, check to make sure there are no faulty triggers associated with any of the tables referenced by the command. See Also DROP TRIGGER, CREATE TABLE, CREATE VIEW, INSERT, UPDATE, DELETE, RAISE [SQL Expr, Ap D] CREATE TRIGGER 318 | Appendix C: SQLite SQL Command Reference CREATE VIEW Create a new view Syntax Common Usage CREATE VIEW database_name.view_name AS SELECT...; Description The CREATE VIEW statement establishes a new view within the named database. A view acts as a prepackaged subquery statement, and can be accessed and referenced as if it were a table. A view does not actually instance the data, but is dynamically generated each time it is accessed. If the optional TEMP or TEMPORARY keyword is present, the view will be created in the temp database. Specifying either TEMP or TEMPORARY in addition to an explicit database name will result in an error, unless the database name is temp. Temporary views may access tables from other attached databases. All nontemporary views are limited to referencing data sources from within their own database. Creating a view that already exists will normally generate an error. If the optional IF NOT EXISTS clause is provided, this error is silently ignored. This leaves the original definition in place. See Also DROP VIEW, CREATE TABLE, SELECT CREATE VIEW Appendix C: SQLite SQL Command Reference | 319 CREATE VIRTUAL TABLE Create a new virtual table Syntax Common Usage CREATE VIRTUAL TABLE database_name.table_name USING weblog( access.log ); CREATE VIRTUAL TABLE database_name.table_name USING fts3( ); Description The CREATE VIRTUAL TABLE command creates a virtual table. Virtual tables are data sources that are defined by code and can represent highly optimized data sources or external data sources. The standard SQLite distribution includes virtual table implementations for Full Text Search, as well as an R*Tree-based indexing system. Virtual tables are covered in detail in Chapter 10. A virtual table is removed with the standard DROP TABLE command. See Also sqlite3_create_module() [C API, Ap G], DROP TABLE DELETE Delete rows from a table Syntax Common Usage DELETE FROM database_name.table_name; DELETE FROM database_name.table_name WHERE id = 42; Description The DELETE command permanently removes rows from a table. Any row that satisfies the WHERE expression will be removed. A WHERE condition that causes no rows to be deleted is not considered an error. If no WHERE condition is provided, it is assumed to always be true, and every row in the table will be deleted. CREATE VIRTUAL TABLE 320 | Appendix C: SQLite SQL Command Reference A DELETE command with no WHERE clause will delete every row in a table. If no WHERE clause is provided, there are some situations when SQLite can simply truncate the whole table. This is much faster than deleting every row individually, but it skips any per-row processing. Truncation will only happen if the table has no triggers and is not part of a foreign key relationship (assuming foreign key support is enabled). Truncation can also be disabled by having an authorizer return SQLITE_IGNORE for the delete operation (see sqlite3_set_au- thorizer()). If the SQLite library has been compiled with the optional SQLITE_ENABLE_UPDATE_DELETE_ LIMIT directive, an optional ORDER BY...LIMIT clause may be used to delete a specific number of rows. See the SQLite website for more details. When a DELETE appears within a trigger body, additional limitations apply. See CREATE TRIGGER. Deleting data from a table will not decrease the size of the database file unless auto-vacuum mode is enabled. To recover space previously taken up by deleted data, the VACUUM command must be run. See Also INSERT, UPDATE, VACUUM, auto_vacuum [PRAGMA, Ap F], CREATE TRIGGER DETACH DATABASE Detach a database file Syntax Common Usage DETACH DATABASE database_name; Description The DETACH DATABASE command detaches and dissociates a named database from a database connection. If the same file has been attached multiple times, this will only detach the named attachment. If the database is an in-memory or temporary database, the database will be destroyed and the contents will be lost. You cannot detach the main or temp databases. The DETACH command will fail if issued inside a transaction. See Also ATTACH DATABASE DETACH DATABASE Appendix C: SQLite SQL Command Reference | 321 DROP INDEX Delete a table index from a database Syntax Common Usage DROP INDEX database_name.index_name; Description The DROP INDEX command deletes an explicitly created index. The index and all the data it contains is deleted from the database. The table the index references is not modified. You cannot drop automatically generated indexes, such as those that enforce unique constraints declared in table definitions. Dropping an index that does not exist normally generates an error. If the optional IF EXISTS clause is provided, this error is silently ignored. See Also CREATE INDEX, CREATE TABLE, DROP TABLE DROP TABLE Delete a table from a database Syntax Common Usage DROP TABLE database_name.table_name; Description The DROP TABLE command removes a table from a database. The table and all the data it contains are permanently removed from the database. Any associated indexes and triggers are also removed. Views that might reference the table are not removed. Delete triggers will not be fired. The DROP TABLE command may also be used to remove virtual tables. In that case, a destroy request is sent to the table module, which is free to do as it sees fit. DROP INDEX 322 | Appendix C: SQLite SQL Command Reference If foreign keys are enabled, the DROP TABLE command will perform the equivalent of a DELETE for each row in the table. This happens after any associated triggers have been dropped, so this will not cause any delete triggers to fire. If any immediate key constraints are violated, the DROP TABLE command will fail. If any deferred constraints are violated, an error will be returned when the transaction is committed. Unless the database is in auto-vacuum mode, dropping a table will not cause the database file to shrink in size. The database pages used to hold the table data will be placed on the free list, but they will not be released. The database must be vacuumed to release the free database pages. Dropping a table that does not exist normally generates an error. If the optional IF EXISTS clause is provided, this error is silently ignored. See Also CREATE TABLE, ALTER TABLE, DROP INDEX, DROP TRIGGER, auto_vacuum [PRAGMA, Ap F], VACUUM DROP TRIGGER Delete a trigger action from a database Syntax Common Usage DROP TRIGGER database_name.trigger_name; Description The DROP TRIGGER command removes a trigger from the database. A trigger will also be removed when the associated table is removed. Dropping a trigger that does not exist normally generates an error. If the optional IF EXISTS clause is provided, this error is silently ignored. See Also CREATE TRIGGER, DROP TABLE DROP VIEW Delete a view from a database Syntax DROP VIEW Appendix C: SQLite SQL Command Reference | 323 Common Usage DROP VIEW database_name.view_name; Description The DROP VIEW command removes a view from the database. Although the view will no longer be available, none of the referenced data is altered in any way. Dropping a view will also remove any associated triggers. Dropping a view that does not exist will normally generate an error. If the optional IF EXISTS clause is provided, this error is silently ignored. See Also CREATE VIEW, CREATE TABLE, DROP TRIGGER END TRANSACTION Finish and commit a transaction See: COMMIT TRANSACTION EXPLAIN Explain the query plan Syntax Common Usage EXPLAIN INSERT ...; EXPLAIN QUERY PLAN SELECT ...; Description The EXPLAIN command offers insight into the internal database operation. Placing EXPLAIN in front of any SQL statement (other than itself) returns information about how SQLite would execute the given SQL statement. The SQL statement is not actually executed. By itself, EXPLAIN will return a result set that describes the internal VDBE instruction sequence used to execute the provided SQL statement. A fair amount of knowledge is required to understand the output. The full EXPLAIN QUERY PLAN variant will return a high-level summary of the query plan using English-like explanations of how the query will be assembled and if data will be accessed by a table scan or by index lookup. The EXPLAIN QUERY PLAN command is most useful for tuning SELECT statements and adjusting index placement. These commands exist to help database administrators and developers understand how SQLite is processing SQL commands. They are designed for interactive tuning and adjustments. It is END TRANSACTION 324 | Appendix C: SQLite SQL Command Reference not recommended that applications programmatically query or utilize this data, as both the data and the format may change from one version of SQLite to the next. INSERT Insert new rows into a table Syntax Common Usage INSERT INTO database_name.table_name ( col1, col2 ) VALUES ( val1, val2 ); INSERT INTO table_name VALUES ( val1, val2, val3... ); INSERT INTO table_name ( col1, col2 ) SELECT c1, c2 FROM...; INSERT INTO table_name DEFAULT VALUES; INSERT OR IGNORE INTO table_name ( col1, col2 ) VALUES ( val1, val2 ); REPLACE INTO table_name ( col1, col2 ) VALUES ( val1, val2 ); INSERT Appendix C: SQLite SQL Command Reference | 325 Description The INSERT command adds new rows to tables. An individual INSERT command can only insert rows into one table, and in most cases can only insert one row at a time. There are several variants of the INSERT command. The primary differences relate to how the columns and values are specified. The basic format of the INSERT command starts with the command word INSERT, followed by an optional conflict resolution clause (OR ROLLBACK, etc.). The INSERT conflict resolution clause is identical to the one found in the UPDATE command. See UPDATE for more information on the different conflict resolution options and how they behave. The command word REPLACE can also be used, which is just a shortcut for INSERT OR REPLACE. This is discussed in more detail below. After the conflict clause, the command declares which table it is acting upon. This is generally followed by a list of columns in parentheses. This column list defines which columns will have values set in the newly inserted row. If no column list is given, a default column list is assumed to include all of the table’s columns, in order, as they appear in the table definition. A default list will not include the raw ROWID column, but any ROWID alias (INTEGER PRIMARY KEY) column is included. If an explicit list of columns is given, the list may contain any column (including the ROWID column) in any order. Following the column list is a description of the values to insert into the new row. The values are most commonly defined by the keyword VALUES, followed by an explicit list of values in parentheses. The values are matched to columns by position. No matter how the column list is defined, the column list and the value list must have the same number of entries so that one value can be matched to each column in the column list. The INSERT values can also be defined with a subquery. Using a subquery is the only case when a single INSERT command can insert more than one row. The result set generated by the sub- query must have the same number of columns as the column list. As with the value list, the values of the subquery result set are matched to the insert columns by position. Any table column that does not appear in the insert column list is assigned a default value. Unless the table definition says otherwise, the default value is a NULL. If you have a ROWID alias column that you want assigned an automatic value, you must use an explicit column list, and that list must not include the ROWID alias. If a column is contained in the column list, either explicitly, or by default, a value must be provided for that column. There is no way to specify a default value except by leaving the column out of the column list, or knowing what the default value is and explicitly inserting it. Alternatively, the column list and value list can be skipped all together. The DEFAULT VALUES variant provides neither a column list nor a source of values and can be used to insert a new row that consists entirely of default values. Because a typical INSERT command only allows a single row to be inserted, it is not uncommon to have a string of INSERT commands that are used to bulk-load or otherwise populate a new table. Like any other command, each INSERT is normally wrapped in its own transaction. Committing and synchronizing this transaction can be quite expensive, often limiting the number of inserts to a few dozen a second. INSERT 326 | Appendix C: SQLite SQL Command Reference Due of the expense associated with committing a transaction, it is very common to batch multiple inserts into a single transaction. Especially in the case of bulk inserts, it is not un- common to batch 1,000 or even 10,000 or more INSERT commands into a single transaction, allowing a much higher insert rate. Just understand that if an error is encountered, there are situations where the whole transaction will be rolled back. While this may be acceptable for loading bulk data from a file (that can simply be rerun), it may not be acceptable for data that is inserted from a real-time data stream. Batch transactions greatly speed things up, but they can also make it significantly more difficult to recover from an error. One final word on the INSERT OR REPLACE command. This type of command is frequently used in event-tracking systems where a “last seen” timestamp is required. When an event is pro- cessed, the system needs to either insert a new row (if this type of event has never been seen before) or it needs to update an existing record. While the INSERT OR REPLACE variant seems perfect for this, it has some specific limitations. Most importantly, the command truly is an “insert or replace” and not an “insert or update.” The REPLACE option fully deletes any old rows before processing the INSERT request, making it ineffective to update a subset of columns. In order to effectively use the INSERT OR REPLACE option, the INSERT needs to be capable of completely replacing the existing row, and not simply updating a subset of columns. See Also UPDATE, BEGIN TRANSACTION PRAGMA Look up or modify an SQLite configuration Syntax Common Usage PRAGMA page_size; PRAGMA cache_size = 5000; PRAGMA table_info( table_name ); Description The PRAGMA command tunes and configures SQLite’s internal components. It is a bit of a catchall command, used to configure or query configuration parameters for both the database engine and database files. It can also be used to query information about a database, such as a list of tables, indexes, column information, and other aspects of the schema. PRAGMA Appendix C: SQLite SQL Command Reference | 327 The PRAGMA command is the only command, outside of SELECT, that may return multiple rows. Appendix F covers the different PRAGMA commands in detail. See Also Appendix F REINDEX Rebuild an index from source data Syntax Common Usage REINDEX collation_name; REINDEX database_name.table_name; REINDEX database_name.index_name; Description The REINDEX command deletes the data within an index and rebuilds the index structure from the source table data. The table referenced by the index is not changed. REINDEX is most frequently used when the definition of a collation sequence has changed and all of the indexes that use that collation must be rebuilt. This ensures that the index order correctly matches the order defined by the collation. If a collation name is provided, all indexes that use that collation, in all attached databases, will be reindexed. If a table name is given, all the indexes associated with that table will be reindexed. If a specific index name is given, just that index will be rebuilt. See Also CREATE INDEX, DROP INDEX RELEASE SAVEPOINT Remove and release save-point from transaction log Syntax REINDEX 328 | Appendix C: SQLite SQL Command Reference Common Usage RELEASE savepoint_name; Description The RELEASE SAVEPOINT command removes a save-point from the transaction log. It indicates that any modifications made since the named save-point was set have been accepted by the application logic. Normally, a RELEASE will not alter the database or transaction log, other than removing the save-point marker. Releasing a save-point will not commit any modifications to disk, nor will it make those changes available to other database connections accessing the same database. Changes bounded by a released save-point may still be lost if the transaction is rolled back to a prior save-point, or if the whole transaction is rolled back. The one exception to this rule is if the named save-point was set outside of a transaction, causing an implicit transaction to be started. In that case, releasing the save-point will cause the whole transaction to be committed. See Also SAVEPOINT, ROLLBACK TRANSACTION, COMMIT TRANSACTION, BEGIN TRANSACTION REPLACE Delete and reinsert an existing row Description The REPLACE command is an alias for the INSERT OR REPLACE variant of the INSERT command. See: INSERT ROLLBACK TRANSACTION Undo part or all of the current transaction Syntax Common Usage ROLLBACK; ROLLBACK TO SAVEPOINT savepoint_name; ROLLBACK TRANSACTION Appendix C: SQLite SQL Command Reference | 329 Description The ROLLBACK command is used to roll back a transaction state. This is analogous to an undo function for transactions. There are two forms of ROLLBACK. The most basic form has no TO clause and causes the entire transaction to be rolled back. Any and all changes and modifications made to the database as part of the transaction are reverted, the transaction is released, and the database connection is put back into autocommit mode with no active transaction. If a TO clause is provided, the transaction is rolled back to the state it was in just after the named save-point was created. The named save-point will remain on the save-point stack. You can roll back to any save-point, but if more than one save-point exists with the same name, the most recent save-point will be used. After rolling back to a save-point, the original transaction is still active. If the named save-point was created outside of a transaction (causing an implicit transaction to be started) the whole transaction will be rolled back, but the save-point and transaction will remain in place. See Also BEGIN TRANSACTION, COMMIT TRANSACTION, SAVEPOINT, RELEASE SAVEPOINT SAVEPOINT Place a save-point marker in the transaction command sequence Syntax Common Usage SAVEPOINT savepoint_name; Description The SAVEPOINT command creates a save-point marker in the transaction log. If there is no active transaction in progress, the save-point will be marked and an implicit BEGIN DEFERRED TRANS ACTION will be issued. Save-points allow subsections of a transaction to be rewound and partially rolled back without losing the entire transaction state. A transaction can have multiple active save-points. Con- ceptually, these save-points act as if they were on a stack. Save-point names do not need to be unique. Save-points are useful in multistep processes where each step needs to be attempted and verified before proceeding to the next step. By creating save-points before starting each step, it may be possible to revert just a single step, rather than the whole transaction, when a logical error case is encountered. SAVEPOINT 330 | Appendix C: SQLite SQL Command Reference See Also RELEASE SAVEPOINT, ROLLBACK TRANSACTION, BEGIN TRANSACTION SELECT Query data from the database Syntax result-column: SELECT Appendix C: SQLite SQL Command Reference | 331 join-source: single-source: SELECT 332 | Appendix C: SQLite SQL Command Reference ordering-term: compound-operator: Common Usage SELECT * FROM tbl; SELECT name FROM employees WHERE employee_id = 54923; SELECT 5 + 6; Description The SELECT command is used to query the database and return a result. The SELECT command is the only SQL command capable of returning a user-generated result, be it a table query or a simple expression. Most consider SELECT to be the most complex SQL command. Although the basic format is fairly easy to understand, it does take some experience to understand its full power. All of Chapter 5 is devoted to the SELECT command. Basic format The core SELECT command follows a simple pattern that can be roughly described as SELECT output FROM input WHERE filter. The output section describes the data that makes up the result set, the input section describes what tables, views, subqueries, etc., will act as data sources, and the filter section sets up conditions on which rows are part of the result set and which rows are filtered out. A SELECT can either be SELECT ALL (default) or SELECT DISTINCT. The ALL keyword returns all rows in the result set, regardless of their composition. The DISTINCT keyword will force the select statement to eliminate duplicate results in the result set. There is usually a considerable performance penalty for calculating larger DISTINCT results. SELECT Appendix C: SQLite SQL Command Reference | 333 The result set columns are defined by a series of comma-separated expressions. Every SELECT statement must have at least one result expression. These expressions often consist of only a source column name, although they can be any general expression. The character * means “return all columns,” and will include all standard table columns from all of the source tables. All the standard columns of a specific source table can be returned with the format table_name.*. In both cases, the ROWID column will not be included, although ROWID alias columns will be included. Virtual tables can also mark some columns as hidden. Like the ROWID column, hidden columns will not be returned by default, but can be explicitly named as a result column. Result columns can be given explicit names with an optional AS clause (the actual AS keyword is optional as well). Unless an AS clause is given, the name of the output column is at the discretion of the database engine. If an application depends on matching the names of specific output columns, the columns should be given explicit names with an AS clause. The FROM clause defines where the data comes from and how it is shuffled together. If no FROM clause is given, the SELECT statement will return only one row. Each source is joined together with a comma or a JOIN operation. The comma acts as an unconditional CROSS JOIN. Different sources, including tables, subqueries, or other JOIN statements, can be grouped together into a large transitory table, which is fed through the rest of the SELECT statement, and ultimately used to produce the result set. For more information on the specific JOIN operators, see “FROM Clause” on page 63. Each data source, be it a named table or a subquery, can be given an optional AS clause. Similar to result set columns, the actual AS keyword is optional. The AS clause allows an alias to be assigned to a given source. This is important to disambiguate table instances (for example, in a self-join). The WHERE clause is used to filter rows. Conceptually, the FROM clause, complete with joins, is used to define a large table that consists of every possible row combination. The WHERE clause is evaluated against each of those rows, passing only those rows that evaluate to true. The WHERE clause can also be used to define join conditions, by effectively having the FROM clause produce the Cartesian product of the two tables, and use the WHERE clause to filter out only those rows that meet the join condition. Additional clauses Beyond SELECT, FROM, and WHERE, the SELECT statement can do additional processing with GROUP BY, HAVING, ORDER BY, and LIMIT. The GROUP BY clause allows sets of rows in the result set to be collapsed into single rows. Groups of rows that share equivalent values in all of the expressions listed in the GROUP BY clause will be condensed to a single row. Normally, every source column reference in the result set expressions should be a column or expression included in the GROUP BY clause, or the column should appear as a parameter of an aggregate function. The value of any other source column is the value of the last row in the group to be processed, effectively making the result undefined. If a GROUP BY expression is a literal integer, it is assumed to be a column index to the result set. For example, the clause GROUP BY 2 would group the result set using the values in the second result column. SELECT 334 | Appendix C: SQLite SQL Command Reference A HAVING clause can only be used in conjunction with a GROUP BY clause. Like the WHERE clause, a HAVING expression is used as a row filter. The key difference is that the HAVING expression is applied after any GROUP BY manipulation. This sequence allows the HAVING expression to filter aggregate outputs. Be aware that the WHERE clause is usually more efficient, since it can elim- inate rows earlier in the SELECT pipeline. If possible, filtering should be done in the WHERE clause, saving the HAVING clause to filter aggregate results. The ORDER BY clause sorts the result set into a specific order. Typically, the output ordering is not defined. Rows are returned as they become available, and no attempt is made to return the results in any specific order. The ORDER BY clause can be used to enforce a specific output ordering. Output is sorted by each expression in the clause, in turn, from most specific to least specific. The fact that the output of a SELECT can be ordered is one of the key differences between an SQL table and a result set. As with GROUP BY, if one of the ORDER BY expressions is a literal integer, it is assumed to be a column index to the result set. Finally, the LIMIT clause can be used to control how many rows are returned, starting at a specific offset. If no offset is provided, the LIMIT will start from the beginning of the result set. Note that the two syntax variations (comma or OFFSET) provide the parameters in the opposite order. Since the row order of a result is undefined, a LIMIT is most frequently used in conjunction with an ORDER BY clause. Although it is not strictly required, including an ORDER BY brings some meaning to the limit and offset values. There are very few cases when it makes sense to use a LIMIT without some type of imposed ordering. Compound statements Compound statements allow one or more SELECT...FROM...WHERE...GROUP BY...HAVING sub- statements to be brought together using set operations. SQLite supports the UNION, UNION ALL, INTERSECT, and EXCEPT compound operators. Each SELECT statement in a compound SELECT must return the same number of columns. The names of the result set columns will be taken from the first SELECT statement. The UNION operator returns the union of the SELECT statements. By default, the UNION operator is a proper set operator and will only return distinct rows (including those from a single table) . UNION ALL, by contrast, will return the full set of rows returned by each SELECT. The UNION ALL operator is significantly less expensive than the UNION operator, so the use of UNION ALL is encouraged, when possible. The INTERSECT command will return the set of rows that appear in both SELECT statements. Like UNION, the INTERSECT operator is a proper set operation and will only return one instance of each unique row, no matter how many times that row appears in both result sets of the individual SELECT statements. The EXCEPT compound operator acts as a set-wise subtraction operator. All unique rows in the first SELECT that do not appear in the second SELECT will be returned. See Also CREATE TABLE, INSERT, UPDATE, DELETE SELECT Appendix C: SQLite SQL Command Reference | 335 UPDATE Modify existing rows in a table Syntax Common Usage UPDATE database_name.table_name SET col5 = val5, col2 = val2 WHERE id = 42; Description The UPDATE command modifies one or more column values within existing table rows. The command starts out with a conflict resolution clause, followed by the name of the table that contains the rows we’re updating. The table name is followed by a list of column names and new values. The final WHERE clause determines which rows are updated. A WHERE condition that causes no rows to be updated is not considered an error. If no WHERE clause is given, every row in the table is updated. An UPDATE command with no WHERE clause will update every row in a table. The values that are used to update a row may be given as expressions. These expressions are evaluated within the context of the original row values. This allows the value expression to refer to the old row value. For example, to increment a column value by one, you might find SQL similar to UPDATE...SET col = col + 1 WHERE.... Columns and values can be given in any order, as long as they appear in pairs. Any column, including ROWID, may be updated. Any columns that do not appear in the UPDATE command remain unmodified. There is no way to return a column to its default value. UPDATE 336 | Appendix C: SQLite SQL Command Reference If the SQLite library has been compiled with the optional SQLITE_ENABLE_UPDATE_DELETE_ LIMIT directive, an optional ORDER BY...LIMIT clause may be used to update a specific number of rows. See the SQLite website (http://www.sqlite.org/lang_update.html) for more details. The optional conflict resolution clause found at the beginning of the UPDATE (or INSERT) com- mand is a nonstandard extension that controls how SQLite reacts to a constraint violation. For example, if a column must be unique, any attempt to update the value of that column to a value that is already in use by another row would cause a constraint violation. The constraint resolution clause determines how this situation is resolved. ROLLBACK A ROLLBACK is immediately issued, rolling back any current transaction. An SQLITE_CONSTRAINT error is returned to the calling process. If no explicit transaction is currently in progress, the behavior will be identical to an ABORT. ABORT This is the default behavior. Changes caused by the current command are undone and SQLITE_CONSTRAINT is returned, but no ROLLBACK is issued. For example, if a constraint violation is encountered on the fourth of ten possible row updates, the first three rows will be reverted, but the current transaction will remain active. FAIL The command will fail and return SQLITE_CONSTRAINT, but any rows that have been pre- viously modified will not be reverted. For example, if a constraint violation is encountered on the fourth of ten possible row updates, the first three modifications will be left in place and further processing will be cut short. The current transaction will not be committed or rolled back. IGNORE Any constraint violation error is ignored. The row update will not be processed, but no error is returned. For example, if a constraint violation is encountered on the fourth of ten possible rows, not only are the first three row modifications left in place, processing continues with the remaining rows. REPLACE The specific action taken by a REPLACE resolution depends on what type of constraint is violated. If a UNIQUE constraint is violated, the existing rows that are causing the constraint violation will first be deleted, and then the UPDATE (or INSERT) is allowed to be processed. No error is returned. This allows the command to succeed, but may result in one or more rows being deleted. In this case, any delete triggers associated with this table will not fire unless recursive triggers are enabled. Currently, the update hook is not called for automatically deleted rows, nor is the change counter incremented. These two behaviors may change in a future version, however. If a NOT NULL constraint is violated, the NULL is replaced by the default value for that column. If no default value has been defined, the ABORT resolution is used. If a CHECK constraint is violated, the IGNORE resolution is used. UPDATE Appendix C: SQLite SQL Command Reference | 337 Any conflict resolution clause found in an UPDATE (or INSERT) command will override any conflict resolution clause found in a table definition. See Also INSERT, DELETE, CREATE TABLE VACUUM Recover free space and optimize database Syntax Common Usage VACUUM; Description The VACUUM command recovers free space from the database file and releases it to the filesys- tem. VACUUM can also defragment database structures and repack individual database pages. VACUUM can only be run against the main database (the database used to create the database connection). VACUUM has no effect on in-memory databases. When data objects (rows, whole tables, indexes, etc.) are deleted or dropped from a database, the file size remains unchanged. Any database pages that are recovered from deleted objects are simply marked as free and available for any future database storage needs. As a result, under normal operations the database file can only grow in size. Additionally, as rows are inserted and deleted from the database, the tables and indexes can become fragmented. In a dynamic database that normally experiences a high number of in- serts, updates, and deletes, it is common for free pages to be scattered all across the database file. If a table or index requires additional pages for more storage, these will first be allocated off the free list. This means the actual parts of the database file that hold a particular table or index may become scattered and mixed all across the database file, lowering seek performance. Finally, as rows are inserted, updated, and deleted, unused data blocks and other “holes” may appear within the individual database pages. This reduces the number of records that can be stored in a single page, increasing the total number of pages required to hold a table. In effect, this increases the storage overhead for the table, increasing read/write times and decreasing cache performance. The vacuum process addresses all three of these issues by copying all the data within a database file to a separate, temporary database. This data transfer is done at a fairly high level, dealing with the logical elements of the database. As a result, individual database pages are “re- packed,” data objects are defragmented, and free pages are ignored. This optimizes storage space, reduces seek times, and recovers any free space from the database file. Once all this is done, the content of the temporary database file is copied back to the original file. VACUUM 338 | Appendix C: SQLite SQL Command Reference As the VACUUM command rebuilds the database file from scratch, VACUUM can also be used to modify many database-specific configuration parameters. For example, you can adjust the page size, file format, default encoding, and a number of other parameters that normally become fixed once a database file is created. To change something, just set the default new database pragma values to whatever you wish, and vacuum the database. Be warned that this behavior is not always desirable. For example, if you have a database with a nondefault page size or file format, you need to be sure that you explicitly set all the correct pragma values before you vacuum the database. If you fail to do this, the database will be re- created with the default configuration values, rather than the original values. If you work with database files that have any nonstandard parameters, it is best to explicitly set all of these configuration values before you vacuum the database. VACUUM will re-create the database using the current default values. For example, if you have a database that uses a custom page size and you wish to maintain that page size, you must issue the appropriate PRAGMA page_size command before issuing the VACUUM command. Failing to do so will result in the database being rebuilt with the default page size. Logically, the database contents should remain unchanged from a VACUUM. The one exception is ROWID values. Columns marked INTEGER PRIMARY KEY will be preserved, but unaliased ROWID values may be reset. Also, indexes are rebuilt from scratch, rather than copied, so VACUUM does the equivalent of a REINDEX for each index in the database. Generally, any reasonably dynamic database should vacuumed periodically. A good rule of thumb is to consider a full VACUUM any time 30% to 40% of the database content changes. It may also be helpful to VACUUM the database after a large table or index is dropped. Be warned that the VACUUM process requires exclusive access to the database file and can take a significant amount of time to complete. VACUUM also requires enough disk space to hold the original file, plus the optimized copy of the database, plus a rollback journal that can be as large as the original file. SQLite also supports an auto-vacuum mode, which enables portions of the vacuum process to be done automatically. It has some significant limitations, however, and even if auto- vacuum is enabled, it is still advisable to do a full manual VACUUM from time to time. See Also auto_vacuum [PRAGMA, Ap F], temp_store [PRAGMA, Ap F], temp_store_directory [PRAGMA, Ap F], REINDEX VACUUM Appendix C: SQLite SQL Command Reference | 339 APPENDIX D SQLite SQL Expression Reference Like most computer languages, SQL has a fairly flexible expression syntax that can be used to combine and compute values. Nearly any time that an SQL command requires a result, conditional, or other value, a full expression can be used to express that value. In addition to literal values, basic arithmetic operations, and function calls, expressions also contain column references, and complex operator expressions. These can be com- bined and mixed to create some fairly complex expressions and behaviors. Many times an expression is used to define a conditional, such as which rows are re- turned in a result. In these contexts, an expression need only return a logical true or false value. In other situations, such as defining the result set of a SELECT statement, expressions that return diverse values are more appropriate. The following sections each cover a specific category of expression. Although the dif- ferent types of operators and expression formats have been divided up to make their descriptions easier to organize, remember that any expression type can be used in any other expression type. If you want to play around with an operator or expression construction outside of a larger query, remember that you can execute any arbitrary expression by simply placing it in a SELECT statement: SELECT 1 + 1; This is extremely useful for testing specific conditions, value conversions, and other situations that may be causing problems within some larger SQL statement. 341 Literal Expressions The simplest type of expression is a literal, or inline value. These are specific values that are expressed directly in the SQL command. SQLite supports a number of literal forms, including one for each major datatype. Each supported datatype has a specific literal representation. This allows the expression processor to understand the desired datatype as well as the specific value. NULL A NULL is represented by the bare keyword NULL. NULL Integer An integer number is represented by a bare sequence of numeric digits. All integers must be given in base-10. A prefix of zero digits does not represent octal numbers, nor are hexadecimal integers supported. No magnitude separators (such as a comma after the thousands digit) are allowed. The number can be prefaced with a + or - to represent the sign of the number: 8632 -- Eight thousand, six hundred, thirty-two 0032 -- Thirty-two -5 -- Negative five +17 -- Seventeen 342 | Appendix D: SQLite SQL Expression Reference Real or floating-point A real number is represented by a bare sequence of numeric digits, followed by a period (decimal point), followed by another sequence of numeric digits. Either of the number sequences on the left or right of the decimal point can be omitted, but not both. SQLite always uses a period for the decimal point, regardless of interna- tionalization settings. The number can be prefaced with a + or - to represent the sign of the number. The initial set of numbers can be followed by an optional exponent, used to rep- resent scientific notation. This is represented with the letter E (upper- or lowercase) followed by an optional + or -, followed by a sequence of numeric digits. The number does not need to be normalized. If an exponent is included and the number group to the right of the decimal point is omitted, the decimal point may also be omitted. This is the only situation when the decimal point may be omitted: 32.4 -- 32.4 -535. -- -535.0 .43 -- 0.43 4.5e+1 -- 45.0 78.34E-5 -- 0.0007834 7e2 -- 700.0 Text or string A text value is represented by a string of characters enclosed in single quotes ( ' ' ). Double quotes ( " " ) are used to enclose identifiers, and should not be used to enclose literal values. To escape a single quote inside of a text literal, use two single quote characters in a row. The backslash character ( \ ), used in C and many other languages as an escape character, is not considered special by SQL and cannot be used to escape quote characters within text literals. A zero-length text value is not the same as a NULL: 'Jim has a dog.' Jim has a dog. 'Jim''s dog is big.' Jim's dog is big. 'C:\data\' C:\data\ '' (zero-length text value) BLOB A BLOB value is represented as an X (upper- or lowercase) followed by a text literal consisting of hexadecimal characters (0–9, A–F, a–f). Two hex characters are re- quired for each full byte, so there must be an even number of characters. The length of the BLOB (in bytes) will be the number of hex characters divided by two. Like text values, the byte values are given in order: X'7c' X'8A26E855' x'' Literal Expressions | 343 Be aware that these are input formats that are recognized by the SQL command parser. They are not necessarily the output format used to display the values. The display format is up to the SQL environment, such as the sqlite3 utility. To output values as valid SQL literals, see the quote() SQL function. In addition to explicit literals, SQLite supports three named literals that can be used to insert the current date or time. When an expression is evaluated, these named tags will be converted into literal text expressions of the appropriate value. Supported tags are: CURRENT_TIME A text value in the format HH:MM:SS. CURRENT_DATE A text value in the format YYYY-MM-DD. CURRENT_TIMESTAMP A text value in the format YYYY-MM-DD HH:MM:SS. All times and dates are in UTC, not your local time zone. Lastly, in any place that SQLite will accept a literal expression, it will also accept a statement parameter. Statement parameters are placeholders, similar to external vari- ables. When using the C API, a statement can be prepared, values can then be bound to the parameters, and the statement can be executed. The statement can be reset, new values can be bound, and the statement can be executed again. Statement parameters allow frequently reused statements (such as many INSERT statements) to be prepared once and used over and over again by simply binding new values to the statement parameters. There are a number of performance and security benefits from this process, but it is only applicable for those using a programming interface to SQLite. SQLite supports the following syntax for statement parameters: ? A single question mark character. SQLite will automatically assign an index to each parameter. ?numb A single question mark character followed by a number. The number will become the parameter index. The same index may be used more than once. :name A single colon character followed by a name. The API provides a way to look up the index based off the name. The same name may be used more than once. @name A single at (@) character followed by a name. The API provides a way to look up the index based off the name. The same name may be used more than once. This variation is nonstandard. 344 | Appendix D: SQLite SQL Expression Reference $name A single dollar sign character followed by a name. The API provides a way to look up the index based off the name. The same name may be used more than once. This variation is nonstandard and understands a special syntax that is designed to be used with Tcl variables. Statement parameters can only be used to replace literals. They cannot be used to replace identifiers, such as table or column names. See the section “Bound Parame- ters” on page 133 for more details on how to use statement parameters with the C API. Logic Representations SQL and SQLite have a fair number of logic operations. Many of the operators, such as <= (test for less-than or equal) perform some type comparison or search between parameter expressions and return a logic value—that is, true or false. A number of SQL commands use these logic values to control how commands are applied to the database. For example, the WHERE clause of a SELECT command ultimately computes a logic value to determine if a given row is included in the result set or not. Despite the fact that logic values are commonly used, SQLite does not have a native logic datatype (such as a Boolean). Rather, logic values are expressed as integer values. Zero is used to represent false, while one (or any other nonzero integer value) is used to represent true. For other datatypes, SQLite uses the standard conversion rules to translate a text or BLOB value into an integer, before considering if the value is true or false. This means that most nonnumeric strings (such as 'abc') will translate to false. The one exception is NULL. Next to true and false, NULL is considered a third logic state. This necessitates the concept of three valued logic, or 3VL. For the purpose of three valued logic, NULL is considered to be an unknown or undefined state. For more details, see “Three-Valued Logic” on page 31. In general, this means that once a NULL is introduced into an expression, it tends to be propagated through the expression. Nearly all unary and binary operators will return NULL if any of the parameter ex- pressions are NULL. Unary Expressions Unary operators are the simplest type of expression operator. They take a single (or unitary) parameter expression and modify or alter that expression in some way. In all cases, if the parameter expression is NULL, the operator will also return NULL. Unary Expressions | 345 SQLite supports the following unary expression operators: - Sign negation A unary negative sign will invert the sign of a numeric expression, and is equivalent to being multiplied by ‒1. Positive expressions become negative, while negative expressions become positive. Any non-NULL parameter expression will be con- verted into a numeric type before the conversion. + Positive sign Logically, this operator is a nonoperation. It does not force numbers to be positive (use the abs() SQL function for that), it simply maintains the current sign. It can be used with any datatype, including text and BLOB types, and will simply return the value, without conversion. Although this operator does not alter the value of the parameter expression, the result expression is still considered a “computed” expression. Applying this oper- ator to a column identifier will dissociate the resulting expression from the source table. This alters the way the query optimizer considers the expression. For exam- ple, the optimizer won’t attempt to use any indexes associated with the source column or a computed result column. ~ Bit inversion Inverts or negates all of the bits of the parameter expression. Any non-NULL parameter expression will be converted into an integer before the bit inversion. NOT Logic inversion The NOT operator is used to invert the meaning of any logic expression. Any non- NULL expression will be converted to an integer value. All nonzero values will return 0, while a 0 value will return 1. Don’t confuse this unary operator with the optional NOT found in some binary operators. The end result is the same, but the syntax ordering is a bit different. Along with the COLLATE expression, these operators have the highest precedence. Binary Expressions Binary operators take two values as parameter expressions and combine or compare them in some way that produces an output value. This section includes all of the op- erators with nonkeyword representations, plus AND and OR. With the exception of AND and OR, all of these operators will produce a NULL result if either parameter expression is NULL. 346 | Appendix D: SQLite SQL Expression Reference SQLite supports the following binary expression operators: || String concatenation The || operator is used to concatenate text values. This operator is defined by the SQL standard. Although some databases support the nonstandard + operator for concatenating text values, SQLite does not. Non-NULL parameter expressions will first be converted to text values. * Multiplication Standard numeric multiplication of two numbers. Non-NULL parameter expres- sions will first be converted to numeric values. If both expressions are integers, the result will also be an integer. / Division Standard numeric division of two numbers. The result will be the lefthand operator divided by the righthand operator. Non-NULL parameter expressions will first be converted to numeric values. If both expressions are integers, the integer division will be used and the result will also be an integer. % Modulo or remainder Standard numeric modulo of two numbers. The expression value will be the remainder of the left operator divided by the right. If either parameter expression is a real number, the result will be a real number. The result will be a whole number between 0 and one less than the value of the converted righthand expression. Non- NULL parameter expressions will first be converted to numeric values. + Addition Standard numeric addition of two numbers. If both parameter expressions are in- tegers, the result will also be an integer. Non-NULL parameter expressions will first be converted to numeric values. - Subtraction Standard numeric subtraction of two numbers. If both parameter expressions are integers, the result will also be an integer. Non-NULL parameter expressions will first be converted to numeric values. << >> Bit shifts Binary bit shift. The lefthand expression is shifted right (>>) or left (<<) by the number of bits indicated in the right operator. Any non-NULL parameter expres- sions will first be converted into integers. These operators should be familiar to C programmers, but are nonstandard operators in SQL. & | Binary AND, OR Binary AND and OR bitwise operations. Any non-NULL parameter expression will first be converted into integers. Logic expressions should not use these oper- ators, but should use AND or OR instead. As with bit shifts, these are nonstandard operators. Binary Expressions | 347 < <= => > Greater-than, less-than variations Compares the parameter expressions and returns a logic value of 0, 1, or NULL, depending on which expression is greater-than, less-than, or equal-to. Parameter expressions do not need to be numeric, and will not be converted. In fact, they don’t even need to be the same type. The results may depend on the collations associated with the parameter expressions. = == Equal Compares the operands for equality and returns a logic value of 0, 1, or NULL. Like most logic operators, equal is bound by the rules of three valued logic. In specific, NULL == NULL will result in a NULL, not true (1). The specific definition of equal for text values depends on the collations associated with the parameter expressions. Both forms (single or double equal sign) are the exact same. != <> Not equal Compares the expressions for inequality and returns a logic value of 0, 1, or NULL. Like equal, not equal is also bound by the rules of three valued logic ,so NULL != NULL is NULL, not false (0). The specific definition of not equal depends on the collations associated with the parameter expressions. Both forms are the same. AND OR Logical AND, OR Logical and and or operators. These can be used to string together complex logic expressions. The AND and OR operators are some of the only operators that may return an integer logic value when one of their parameter expressions is NULL. See “Three-Valued Logic” on page 31 for more details on how AND and OR operate under three valued logic. These operators are listed in order of precedence. However, a number of operators have the same precedence as their neighbors. As with most expression languages, SQL allows subexpressions to be enclosed in parentheses to enforce a specific evaluation ordering. Function Calls In addition to operators, SQLite supports both built-in and user-defined function calls. There are two categories of function calls. Scalar functions are called with a specific set of parameters and return a value, just like functions in almost any other expression language. Scalar functions can be used in just about any context in any SQLite expres- sion. An example of a scalar function is abs(), which returns the absolute value of a numeric parameter. 348 | Appendix D: SQLite SQL Expression Reference There are also aggregate functions, which are used to collapse or summarize groups of rows. Aggregate functions can only be used in expressions that define the result set or HAVING clause of a SELECT statement. Aggregates are often used in conjunction with GROUP BY clauses. In essence, an aggregate function is called many times with different input values, but returns only one value per dataset. An example of an aggregate func- tion is avg(), which computes the average value for a sequence of numeric inputs. The syntax for a function call looks like this: As with many expression languages, a function can be called by naming the function and providing a list of zero or more comma-separated parameter expressions within a set of parentheses. In some contexts, the special syntax of a single * character can also be used in place of a parameter list. Like the result set definition of a SELECT statement, this has an implied meaning of “everything.” The optional keyword DISTINCT can also be included before the first parameter. This is only relevant for aggregate functions. If present, it will verify that each set of parameters passed in to an aggregate will be unique and distinct. The keyword has no effect when used with scalar functions. For a full listing of all the built-in functions that SQLite supports, see Appendix E. Column Names One of the most common types of expressions is the column name. The general format is fairly simple, consisting of just the column name. If there is any ambiguity between different tables, you can prefix the column name with an optional table name or a database and table name. Identifiers (database names, table names, or column names) that include nonstandard characters can be enclosed in double quotes ( "" ) or square brackets ( [] ) to escape them. For example [table name].[column name]. Column Names | 349 Column name expressions are always evaluated in some type of context. For example, if you’re formulating a WHERE expression that is part of a SELECT statement, the expression defined there will be evaluated once for each possible row in the result set. As each row is processed, the value of the column for that row will be put into the expression and the expression will be evaluated. The context defines what column references are avail- able to any particular expression. In addition to actual table columns, many expressions within a SELECT statement can also reference columns from the result set by referencing the alias assigned in an AS clause. Similarly, if a source table in the FROM clause is assigned a table alias, this alias must be used in any table reference. The use of table aliases is especially common when formulating join condition expressions on self-joins (a table joined to itself), and other situations when you need to refer to a specific instance of a specific table. A table alias can also be assigned to a nameless subquery. General Expressions This section includes all the keyword expressions. Many of these have unique formats with one or more variations. Syntax diagrams have been provided to help understand the expression format. AND Logical AND See: “Binary Expressions” on page 346, AND BETWEEN Inclusion within a range Syntax Description The BETWEEN expression checks to see if the value of a test expression is between a minimum expression and a maximum expression, inclusive. The expression is logically equivalent to (test >= min AND test <= max), although the test expression is only evaluated once. The BETWEEN operator will return a logic value of 0, 1, or NULL, under the rules of three valued logic. See Also IN, EXISTS AND 350 | Appendix D: SQLite SQL Expression Reference CASE Conditional expression evaluation Syntax Example Usage CASE WHEN col IS NULL THEN 'null' WHEN NOT col THEN 'false' -- convert to logic value WHEN NOT NOT col THEN 'true' -- convert to logic value ELSE 'unknown' END Description The CASE expression is similar to the C switch statement, or a series of if-then statements. The expression consists of an optional test expression followed by one or more WHEN...THEN clauses. The statement is finished with an optional ELSE clause and a required END keyword. Each WHEN...THEN clause is evaluated in order. The first WHEN expression that is found to be equivalent to the test expression will cause the whole CASE expression to take the value of the corresponding return expression. If no equivalent WHEN expression is found, the default ex- pression value is used. If no ELSE clause is provided, the default expression is assumed to be NULL. If no test expression is given, the first WHEN expression that evaluates to true will be used. CAST Force a type conversion Syntax CAST Appendix D: SQLite SQL Expression Reference | 351 Description The CAST operator forces the cast expression to the datatype described by type name. The actual representation (integer, real, text, BLOB) will be derived from the type name by search- ing (case-insensitively) for these specific substrings. The first string that matches will deter- mine the type. If no matches are found, the cast expression will be converted into the most appropriate numeric value. Substring Datatype INT Integer CHAR Text CLOB Text TEXT Text BLOB BLOB REAL Float FLOA Float DOUB Float Note that the type string FLOATING POINT will be interpreted as an integer, and not a real, since the first substring listed will match the INT at the end of the name. This first entry takes precedence over subsequent substrings, resulting in an unexpected cast. It is best to use somewhat standard database types to ensure correct conversion. Once a specific datatype has been determined, the actual conversion is done using the stand- ard rules. See Table 7-1 for specifics. See Also CREATE TABLE [SQL Cmd, Ap C] COLLATE Associate a specific collation to an expression Syntax Description The COLLATE operator associates a specific collation with an expression. The COLLATE operator does not alter the value of the expression, but it does change how equality and ordering are tested in the enclosing expression. If two text expressions are involved in an equality or order test, the collation is determined with the following rules: COLLATE 352 | Appendix D: SQLite SQL Expression Reference 1. If the lefthand (first) expression has an explicit collation, that is used. 2. If the righthand (second) expression has an explicit collation, that is used. 3. If the lefthand (first) expression is a direct reference to a column with a collation, the column collation is used. 4. If the righthand (second) expression is a direct reference to a column with a col- lation, the column collation is used. 5. The default BINARY collation is used. For example: 'abc' == 'ABC' => 0 (false) 'abc' COLLATE NOCASE == 'ABC' => 1 (true) 'abc' == 'ABC' COLLATE NOCASE => 1 (true) 'abc' COLLATE NOCASE == 'ABC' COLLATE BINARY => 1 (true) 'abc' COLLATE BINARY == 'ABC' COLLATE NOCASE => 0 (false) For more information on collations, see “Collation Functions” on page 200. See Also CREATE TABLE, sqlite3_create_collation() [C API, Ap G] EXISTS Test if a row exists Syntax Description The EXISTS operator returns a logic value (integer 0 or 1) depending on whether the SELECT statement returns any rows. The number of columns and their values are irrelevant. As long as the SELECT statement returns at least one row consisting of at least one column, the operator will return true—even if the column value is a NULL. See Also IN, BETWEEN GLOB Match text values using patterns Syntax GLOB Appendix D: SQLite SQL Expression Reference | 353 Description The GLOB operator is used to match text values against a pattern. If the search expression can be matched to the pattern expression, the GLOB operator will return true (1). All non-NULL parameter expressions will be converted to text values. GLOB is case sensitive, so 'a' GLOB 'A' is false. The syntax of the pattern expression is based off common command-line wildcards, also known as file-globbing. The * character in the pattern will match zero or more characters in the search expression. The ? character will match exactly one of any character, while the list wildcard ( [] ) will match exactly one character from its set of characters. All other characters within the pattern are taken as literals. GLOB patterns have no escape character. For example, the pattern '*.xml' will match all search expressions that end in the four char- acters .xml, while the pattern '??' will match any text value that is exactly two characters long. The pattern '[abc]' will match any single a, b, or c character. The list wildcard allows a range of characters to be specified. For example [a-z] will match any single lowercase alphabetic character, while [a-zA-Z0-9] will match any single alpha- numeric character, uppercase or lowercase. You can match any character except those indicated in a list by placing a ^ at the beginning of the list. For example, the pattern [^0-9] will match any single character except a numeric character. To match a literal *, ?, or [, put them in a list wildcard; for example [*] or [[]. To match a ^ inside a list, don’t put it first. To match (or not match) a literal ] inside a list, make it the first character after the opening [ or [^. To match a literal - inside a range, make it the last character in the list. The GLOB operator is implemented by the glob() SQL function. As such, its behavior can be overridden by registering a new glob() function. See Also glob() [SQL Func, Ap E], LIKE, MATCH, REGEXP IN Test if value is in set Syntax IN 354 | Appendix D: SQLite SQL Expression Reference Common Usage col IN ( test1, test2, test3 ) col IN ( SELECT c FROM t ) col IN temp.in_test Description The IN operator tests to see if the test expression equal to (or not equal to) any of the values found on the righthand side of the expression. This is a three valued logic operator and will return 0, 1, or NULL. A NULL will be returned if a NULL is found on the lefthand side, or if a NULL appears anywhere in an unmatched test group. There are three ways to define the test group. First, an explicit series of zero or more expres- sions can be given. Second, a subquery can be provided. This subquery must return a single column. The test expression will be evaluated against each row returned by the subquery. Both of these formats require parentheses. The last way to define the test group is by providing a table name. The table must consist of only a single column. You cannot provide a table and column, it must be a single-column table. This final style is most frequently used with temporary tables. If you need to execute the same test multiple times, it can be more efficient to build a temporary table (for example, with CREATE TEMP TABLE...AS SELECT), and use it over and over, rather than using a subquery as part of the IN expression. See Also BETWEEN, EXISTS IS Equality test, including NULLs Syntax Description The IS and IS NOT operators are very similar to the equal ( = or == ) and not equal ( != or <> ) operators. The main difference is how NULLs are handled. The standard equality tests are subject to the rules of three valued logic, which will result in a NULL if either parameter expression is NULL. The IS operator considers NULL to be “just another value,” and will always return a 0 or 1. For example, the expression NULL IS NULL is considered to be true (1) and the expression NULL IS 6 is considered to be false (0). See Also ISNULL IS Appendix D: SQLite SQL Expression Reference | 355 ISNULL Test for NULL Syntax Description The ISNULL and other operators shown here are used to test for NULL or non-NULL expres- sions. They are syntactic variations on IS NULL and IS NOT NULL, which are properly formed IS expressions. See Also IS LIKE Match text values using patterns Syntax Description The LIKE operator is used to match text values against a pattern. If the search expression can be matched to the pattern expression, the LIKE operator will return true (1). All non-NULL parameter expressions will be converted to text values. LIKE is a standardized SQL operator. By default, LIKE is not case sensitive, so 'a' LIKE 'A' is true. However, this case-insensitivity only applies to standard 7-bit ASCII characters. By default, LIKE is not Unicode aware. See “ICU Internationalization Extension” on page 167 for more info. The case sensitivity of LIKE can be adjusted with the PRAGMA case_sensitive_like. ISNULL 356 | Appendix D: SQLite SQL Expression Reference The LIKE pattern syntax supports two wildcards. The % character will match zero or more characters, while the _ character will match exactly one. All other characters in the pattern will be taken as literals. A literal % or _ can be matched by proceeding it with the character defined in the optional escape expression. The first character of this expression will be used as the escape character. There is no default escape character (in specific, the C-style \ character is not a default). For example, the pattern '%.xml' will match all search expressions that end in the four char- acters .xml, while the pattern '__' will match any text value that is exactly two characters long. It is not possible to match a literal % or _ without defining an escape character. The LIKE operator is implemented by the like() SQL function. As such, its behavior can be overridden by registering a new like() function. See Also like() [SQL Func, Ap E], case_sensitive_like [PRAGMA, Ap F], GLOB, MATCH, REGEXP MATCH Match text values using patterns Syntax Description The purpose of the MATCH operator is to support a user-defined pattern-matching algorithm. No default implementation exists, however, so any attempt to use the MATCH operator without first defining a match() SQL function will result in an error. See Also match() [SQL Func, Ap E], LIKE, GLOB, REGEXP NOTNULL Test for non-NULL See: ISNULL OR Logical OR See: “Binary Expressions” on page 346, OR OR Appendix D: SQLite SQL Expression Reference | 357 RAISE Indicate an error condition Syntax Description The RAISE expression isn’t an expression in the traditional sense. Rather than producing a value, it provides a means to raise an error exception. The RAISE expression can only be used in the body of a trigger. It is typically used to flag a constraint violation, or some similar problem. It is common to use a RAISE expression in conjunction with a CASE expression, or some other expression that selectively executes subexpressions depending on the logic of the SQL statement. See Also CREATE TRIGGER [SQL Cmd, Ap C] REGEXP Match text values using patterns Syntax Description The purpose of the REGEXP operator is to support a user-defined regular expression text- matching algorithm. No default implementation exists, however, so any attempt to use the REGEXP operator without first defining a regexp() SQL function will result in an error. This is typically done using a third-party regular expression library. See Also regex() [SQL Func, Ap E], LIKE, GLOB, MATCH RAISE 358 | Appendix D: SQLite SQL Expression Reference SELECT Extract expression value from database Syntax Description A SELECT expression is a subquery within parentheses. The SELECT statement must return only one column. The value of the expression becomes the value of the first row (and only the first row). If no rows are returned, the expression is NULL. SQLite supports several different types of subqueries, so you need to be careful about which style you’re using. Several expressions, such as IN, allow a direct subquery as part of their syntax. Similarly, several SQL commands, such as CREATE TABLE, support a subquery syntax. In these cases, the subquery is returning a set of data, not a single value. When you use this form of a subquery as a standalone expression, it will only return one value. This can sometimes have unexpected results. For example, consider these two expressions: col IN ( SELECT c FROM t ); col IN ( ( SELECT c FROM t ) ); The only difference between these two expressions is the extra set of parentheses around the subquery in the second line. In the first case, the IN expression sees the subquery as a direct part of the IN expression. This allows the IN to test col against each row returned by the subquery. In the second case, the extra set of inner parentheses converts the subquery into an expression. This conversion makes IN see a single-value expression list, rather than a subquery. As a result, the col expression is only tested against the first row returned by the subquery. Care must be taken when using parentheses around SELECT expressions. Extraneous paren- theses shouldn’t alter scalar expressions, but if a subquery is part of a larger expression, there can be a critical difference between a subquery and an expression derived from a subquery. See Also IN, EXISTS SELECT Appendix D: SQLite SQL Expression Reference | 359 APPENDIX E SQLite SQL Function Reference SQLite includes a number of built-in SQL functions. Many of these functions are pre- dicated by the SQL standard, while others are specific to the SQLite environment. There are two styles of functions. Scalar functions can be used as part of any SQL expression. They take zero or more parameters and calculate a return value. They can also have side-effects, just like the functions of most programming languages. Aggregate functions can only be used in SELECT header expressions and HAVING clauses. Aggregate functions combine column values from multiple rows to calculate a single return value. Often, aggregate functions are used in conjunction with a GROUP BY clause. This appendix is divided into two sections. The first section covers all the built-in scalar functions, while the second section covers all the built-in aggregate functions. You can also define your own scalar or aggregate functions. See Chapter 9 for more details. Scalar Functions abs() Compute the numerical absolute value Common Usage abs( expression ) Description The abs() function returns the absolute value of an expression. If the expression is an integer, an integer absolute value is returned. If expression is a float, the floating-point absolute value is returned. If number is NULL, a NULL is returned. All other datatypes are first converted to floating-point values, then the absolute value of the converted floating-point number is returned. 361 changes() Get the number of rows changed by the last SQL command Common Usage changes( ) Description The changes() function returns an integer that indicates the number of database rows that were modified by the most recently completed INSERT, UPDATE, or DELETE command run by this database connection. This SQL function is a wrapper around the C function sqlite3_changes(), and has all of the same limitations and conditions. See Also total_changes(), last_insert_rowid(), sqlite3_changes() [C API, Ap G] coalesce() Return first non-NULL argument Common Usage coalesce( param1, param2, ... ) Description The coalesce() function takes two or more parameters and returns the first non-NULL parameter. If all of the parameter values are NULL, a NULL is returned. See Also ifnull(), nullif() date() Decode time string into date Common Usage date( timestring, modifier, ... ) Description The date() function takes a timestring value plus zero or more modifier values and returns a text value in the format YYYY-MM-DD. It is equivalent to the call strftime( '%Y-%m-%d', timestring, ... ). See Also strftime(), datetime(), time() changes() 362 | Appendix E: SQLite SQL Function Reference datetime() Decode time string into date and time Common Usage datetime( timestring, modifier, ... ) Description The datetime() function takes a timestring value plus zero or more modifier values and returns a text value in the format YYYY-MM-DD HH:MM:SS. It is equivalent to the call strftime ( '%Y-%m-%d %H:%M:%S', timestring, ... ). See Also strftime(), date(), time() glob() Implement the GLOB operator Common Usage glob( pattern, string ) Description The glob() function implements the matching algorithm used by the SQL expression string GLOB pattern, and is normally not called directly by user-provided SQL expressions. This function exists so that it may be overridden with a user-defined SQL function, providing a user-defined GLOB operator. Note that the order of the parameters differs between the GLOB expression and the glob() function. See Also GLOB [SQL Expr, Ap D], like(), match(), regex() ifnull() Return first non-NULL argument Common Usage ifnull( param1, param2 ) Description The ifnull() function is basically a fixed two-parameter version of coalesce(). If param1 is not NULL, it is returned. If param1 is NULL, param2 is returned. See Also coalesce(), nullif() ifnull() Appendix E: SQLite SQL Function Reference | 363 hex() Dump BLOB as hexadecimal Common Usage hex( data ) Description The hex() function converts a BLOB value into a hexadecimal text representation. The pa- rameter data is assumed to be a BLOB. If it is not a BLOB, it will be converted into one. The returned text value will contain two hexadecimal characters for each byte in the BLOB. Be careful about using hex() with large BLOBs. The UTF-8 text representation is twice as big as the original BLOB value, and the UTF-16 representation is four times as large. See Also quote() julianday() Return the Julian Day number Common Usage julianday( timestring, modifier, ... ) Description The julianday() function takes a timestring value plus zero or more modifier values and returns a floating-point value representing the number of days since noon, Greenwich time, on November 24th, 4714 B.C. using the proleptic Gregorian calendar. It is equivalent to the call strftime( '%J', timestring, ... ), except that julianday() returns a floating-point value, while strftime( '%J', ... ) returns a text representation of an equivalent floating- point value. See Also strftime() last_insert_rowid() Return the ROWID of the last inserted row Common Usage last_insert_rowid( ) Description The last_insert_rowid() function returns the integer ROWID value (or ROWID alias) of the last successfully completed row INSERT. The purpose of this function is to discover an automati- cally generated ROWID, often for the purpose of inserting a foreign key that references that ROWID. For the purposes of this function, an INSERT is considered to successfully complete even if it happens inside an uncommitted transaction. hex() 364 | Appendix E: SQLite SQL Function Reference The returned value is tracked by the database connection, not the database itself. This avoids any possible race conditions between INSERT operations done from different database con- nections. It means, however, that the returned value is updated by an INSERT to any table of any attached database in the database connection. If no INSERT operations have completed with this database connection, the value 0 is returned. This SQL function is a wrapper around the C function sqlite3_last_insert_rowid(), and has all of the same limitations and conditions. See Also changes(), total_changes(), sqlite3_last_insert_rowid() [C API, Ap G] length() Return the number of characters in a string Common Usage length( data ) Description The length() function returns an integer value indicating the length of its parameter. If data is a text value, the number of characters is returned, regardless of encoding. If data is a BLOB value, the number of bytes in the BLOB are returned. If data is NULL, a NULL is returned. Numeric types are first converted to a text value. like() Implement the LIKE operator Common Usage like( pattern, string ) Description The like() function implements the matching algorithm used by the SQL expression string LIKE pattern, and is normally not called directly by user-provided SQL expressions. This function exists so that it may be overridden with a user-defined SQL function, providing a user-defined LIKE operator. Note that the order of the parameters differs between the LIKE expression and the like() function. See Also LIKE [SQL Expr, Ap D], glob(), match(), regex() like() Appendix E: SQLite SQL Function Reference | 365 load_extension() Load a dynamically linked SQLite extension Common Usage load_extension( extension ) load_extension( extension, entry_point ) Description The load_extension() function attempts to load and dynamically link the file extension as an SQLite extension. The function named entry_point is called to initialize the extension. If entry_point is not provided, it is assumed to be sqlite3_extension_init. Both parameters should be text values. This SQL function is a wrapper around the C function sqlite3_load_extension(), and has all of the same limitations and conditions. In specific, extensions loaded this way cannot redefine or delete function definitions. See Also sqlite3_load_extension() [C API, Ap G] lower() Convert all ASCII characters to lowercase Common Usage lower( text ) Description The lower() function returns a copy of text with all of the letter characters converted to lowercase. The built-in implementation of this function only works with ASCII characters (those that have a value less than 128). The ICU extension provides a Unicode-aware implementation of lower(). See Also upper() ltrim() Trim characters from the front (left) of a string Common Usage ltrim( text ) ltrim( text, extra ) Description The ltrim() function returns a copy of text that has been stripped of any prefix that consists solely of characters from extra. Characters from extra that are contained in the body of text remain untouched. load_extension() 366 | Appendix E: SQLite SQL Function Reference If extra is not provided, it is assumed to consist of the ASCII space (0x20) character. By default, tab characters and other whitespace are not trimmed. See Also rtrim(), trim() match() Implement the MATCH operator Common Usage match( pattern, string ) Description The match() function implements the matching algorithm used by the SQL expression string MATCH pattern, and is normally not called directly by user-provided SQL expressions. A default implementation of this function does not exist. To use the MATCH operator, an application-defined function must be registered. Note that the order of the parameters differs between the MATCH expression and the match() function. See Also MATCH [SQL Expr, Ap D], like(), glob(), regex() max() Return the argument with the largest value Common Usage max( param1, param2, ... ) Description Given two or more parameters, the max() function returns the parameter with the largest value. If any parameter is a NULL, a NULL will be returned. Otherwise, BLOB values are considered to have the largest value, followed by text values. These are followed by the numeric types (mixed integer values and floating-point values), sorted together in their natural order. If you want the comparison to use a specific collation, use a COLLATE expression to attach an explicit collation to the input values. For example, max( param1 COLLATE NOCASE, param2 ). There is also an aggregate version of max() that takes a single parameter. See Also min() [Agg SQL Func], max() [Agg SQL Func], COLLATE [SQL Expr, Ap D] max() Appendix E: SQLite SQL Function Reference | 367 min() Return the argument with the smallest value Common Usage min( param1, param2, ... ) Description Given two or more parameters, the min() function will return the parameter with the smallest value. If any parameter is a NULL, a NULL will be returned. Otherwise, numeric types (mixed integers and floating-point) are considered smallest, sorted together in their natural order. These are followed by text values, which are followed by BLOB values. If you want the comparison to use a specific collation, use a COLLATE expression to attach an explicit collation to the input values. For example, min( param1 COLLATE NOCASE, param2 ). There is also an aggregate version of min() that takes a single parameter. See Also max(), min(), COLLATE [SQL Expr, Ap D] nullif() Return NULL if parameters are equal Common Usage nullif( param1, param2 ) Description The nullif() function returns NULL if the parameters are equal. If the two parameters are not equal, the first parameter is returned. If both parameters are NULL, a NULL is returned. See Also ifnull(), coalesce() quote() Return the SQL literal representation of a value Common Usage quote( value ) Description The quote() function returns a text value that represents the SQL literal representation of value. Numbers are returned in their string representation with enough digits to preserve precision. Text values are returned inside single quotes with any internal single quote char- acters properly escaped. BLOBs are returned as hexadecimal literals. NULLs are returned as the string NULL. In SQLite v3.6.23.1 and earlier, this function was used internally by VACUUM. It is strongly recommended that the default implementation is not overridden. min() 368 | Appendix E: SQLite SQL Function Reference See Also hex(), VACUUM [SQL Cmd, Ap C] random() Return a random 64-bit signed integer Common Usage random( ) Description The random() function returns a pseudo-random, 64-bit signed integer value. This produces a number within the range -9,223,372,036,854,775,808 to +9,223,372,036,854,775,807. These values are generated by an automatically seeded internal pseudo-random number gen- erator to insure quality. See Also randomblob() randomblob() Return a BLOB consisting of random data Common Usage randomblob( size ) Description The randomblob() function returns a BLOB value consisting of size bytes. All of the bytes are set to pseudo-random values. These values are generated by an automatically seeded internal pseudo-random number generator to insure quality. See Also random(), zeroblob() regex() Implement the REGEX operator Common Usage regex( pattern, string ) Description The regex() function implements the matching algorithm used by the SQL expression string REGEX pattern, and is normally not called directly by user-provided SQL expressions. A default implementation of this function does not exist. To use the REGEX operator, an ap- plication defined function must be provided. regex() Appendix E: SQLite SQL Function Reference | 369 Note that the order of the parameters differs between the REGEX expression and the regex() function. See Also REGEXP [SQL Expr, Ap D], like(), glob(), match() replace() Find and replace substrings Common Usage replace( value, search, replacement ) Description The replace() function returns a copy of value with each instance of the search substring replaced by replacement. If any of the parameters are NULL, a NULL will be returned. If search is an empty string, value will be returned unmodified. Otherwise, all parameters will be converted into text values. The replace() function also works on BLOB values, although the result is returned as a text value and must be cast back to a BLOB. See Also substr() round() Round off numeric values Common Usage round( value ) round( value, precision ) Description The round() function returns a floating-point value that represents value rounded to preci sion base-10 digits after the decimal point. The value parameter is assumed to be a floating- point value, while precision is assumed to be a positive integer. Values are rounded to the closest representation, not towards zero. If either value or precision is NULL, a NULL will be returned. If precision is omitted, it is assumed to be zero. This will result in a whole number, although it will still be returned as a floating-point value. rtrim() Trim characters from the end (right) of a string Common Usage rtrim( text, extra ) rtrim( text ) replace() 370 | Appendix E: SQLite SQL Function Reference Description The rtrim() function returns a copy of text that has been stripped of any suffix that consists solely of characters from extra. Characters from extra that are contained in the body of text remain untouched. If extra is not provided, it is assumed to consist of the ASCII space (0x20) character. By default, tab characters and other whitespace are not trimmed. See Also ltrim(), trim() sqlite_compileoption_get() Access list of compile-time options used to build the SQLite library Common Usage sqlite_compileoption_get( index ) Description The sqlite_compileoption_get() function returns a text value that describes one of the build directives used when building this instance of the SQLite library. By iterating over index values (starting with zero), you can view the entire list of directives. Returned values will be in the format directive_name if the directive is a simple on/off flag, or directive_name=value if the directive is used to set a value. The SQLITE_ prefix will not be included in the returned value. If the index is negative or out of range, sqlite_compileoption_get() will return NULL. Those directives used to set default values and maximum limits will not be reported. If the directive has a value associated with it, the format will be name=value. This SQL function is a wrapper around the C function sqlite3_compileoption_get(), and has all of the same limitations and conditions. See Also sqlite_compileoption_used(), sqlite3_compileoption_get() [Ap G], sqlite3_limit() [Ap G] sqlite_compileoption_used() See if compile-time option was used to build the SQLite library Common Usage sqlite_compileoption_used( option_name ) Description Given a build directive name as a text value, the sqlite_compileoption_used() function returns a 1 if the directive was used when building this instance of the SQLite library. This function will only indicate if the directive was used, it will not indicate what (if any) value was set. If the given directive was not used, or is otherwise unrecognized, this function will return 0. The SQLITE_ prefix on the directive name is optional. sqlite_compileoption_used() Appendix E: SQLite SQL Function Reference | 371 Those directives used to set default values and maximum limits will not be reported. If the directive has a value associated with it, you can check for a specific value by searching for the full format name=value. If both the name and the value match, a 1 will be returned. This SQL function is a wrapper around the C function sqlite3_compileoption_used(), and has all of the same limitations and conditions. See Also sqlite_compileoption_get(), sqlite3_compileoption_used() [C API, Ap G] sqlite_source_id() Return the source identification value of the current SQLite library Common Usage sqlite_source_id( ) Description The sqlite_source_id() function returns a text value consisting of the check-in identifier of the source code used to build the SQLite library. The identifier consists of a date, timestamp, and an SHA1 hash of the source from the source repository. This SQL function is a wrapper around the C function sqlite3_sourceid(), and has all of the same limitations and conditions. See Also sqlite_version(), sqlite3_sourceid() [C API, Ap G] sqlite_version() Return the version string of the current SQLite library Common Usage sqlite_version( ) Description The sqlite_version() function returns a text value consisting of the version number of the SQLite library. A typical string would be something like '3.7.2'. This SQL function is a wrapper around the C function sqlite3_libversion(), and has all of the same limitations and conditions. See Also sqlite_source_id(), sqlite3_libversion() [C API, Ap G] strftime() Decode time string into any format Common Usage strftime( format, timestring, modifier1, modifier2, ... ) sqlite_source_id() 372 | Appendix E: SQLite SQL Function Reference Description The strftime() function returns a formatted string by taking a timestring (with modifiers) and formating it according to format, a printf() style format specification. If any of the pa- rameters are in an incorrect or illegal format, a NULL is returned. The format string can contain any of the following markers: • %d — day of the month (DD), 01-31 • %f — seconds with fractional part (SS.sss), 00-59 plus decimal portion • %H — hour (HH), 00-23 • %j — day of the year (NNN), 001-366 • %J — Julian Day number (DDDDDDD.ddddddd) • %m — month (MM), 01-12 • %M — minute (MM), 00-59 • %s — seconds since 1970-01-01 (Unix epoch) • %S — seconds (SS), 00-59 • %w — day of the week (N), starting with Sunday as 0 • %W — week of the year (WW), 00-53 • %Y — year (YYYY) • %% — a literal % The timestring value can be in any of these formats: • YYYY-MM-DD • YYYY-MM-DD HH:MM • YYYY-MM-DD HH:MM:SS • YYYY-MM-DD HH:MM:SS.sss • YYYY-MM-DDTHH:MM • YYYY-MM-DDTHH:MM:SS • YYYY-MM-DDTHH:MM:SS.sss • HH:MM • HH:MM:SS • HH:MM:SS.sss • now • DDDDDDD • DDDDDDD.ddddddd strftime() Appendix E: SQLite SQL Function Reference | 373 All hour values use a 24-hour clock. Any value that is not given will be implied. Any implied hour, minute, second, or subsecond is zero. Any implied month or day of month is 1. Any implied year is 2000. The full date and time formats allow either a space or a literal uppercase T between the date and time. The last two values (a single, large integer or floating-point number) provide the date (or date and time) as expressed in Julian Days. Before formatting, the timestring value is processed by one or more modifiers. The modifiers are processed one at a time in the order they are given. The date is normalized after each modifier is applied. Modifiers can be in the following formats: • [+-]NNN day[s] • [+-]NNN hour[s] • [+-]NNN minute[s] • [+-]NNN second[s] • [+-]NNN.nnn second[s] • [+-]NNN month[s] • [+-]NNN year[s] • start of month • start of year • start of day • weekday N • unixepoch • localtime • utc The first set of modifiers adds or subtracts a given unit of time from the original timestring value. For example, the call date( '2001-01-01', '+2 days' ) will return '2001-01-03'. Ma- nipulations are done at a very literal level, and the dates are normalized to legal values after each modifier is applied. For example, date( '2001-01-01', '-2 days' ) returns '2000-12-30'. The normalization process can cause some unexpected results. For example, consider date( '2001-01-31', '+1 month' ). This initially calculates the date '2001-02-31', or February 31st. Since February never has 31 days, this date is normalized into the month of March. In the case of 2001 (a nonleap year) the final result is '2001-03-03'. It is also significant that normalization is done after each modifier is applied. For example, the call date( '2001-01-31', '+1 month', '-1 month' ) will result in '2001-02-03'. strftime() 374 | Appendix E: SQLite SQL Function Reference The start of... modifiers set all time units that are smaller than the named unit to their minimum value. For example, datetime( '2001-02-28 12:30:59', 'start of month' ) will result in '2001-02-01 00:00:00' by setting everything smaller than a month (day, hour, mi- nute, second) to its minimum value. The weekday modifier shifts the current date forward in time anywhere from zero to six days, so that the day will fall on the Nth day of the week (Sunday = 0). The unixepoch modifier only works as an initial modifier, and only when the date is given as a single numeric value. This modifier forces the date to be interpreted as a Unix epoch counter, rather than the traditional Julian Day. The SQLite date/time functions do not keep track of time zone data. Unless otherwise speci- fied, all dates are assumed to be in UTC. For example, the 'now' timestring will produce date and time values in UTC. To convert a UTC timestamp to the local time, the modifier local time can be applied. Conversely, if the timestring is known to be in reference to the local time zone, the utc modifier can be used to convert the timestamp to UTC. All date and time functions are designed to operate on dates between 0000-01-01 00:00:00 and 9999-12-31 23:59:59 (Julian Day numbers 1721059.5 to 5373484.5). Any use of values outside of this range may result in undefined behavior. Unix epoch values are only valid through the date/time 5352-11-01 10:52:47. See Also date(),time(), datetime(), julianday() substr() Extract a substring Common Usage substr( string, index, count ) substr( string, index ) Description The substr() function extracts and returns a substring from string. The position of the sub- string is determined by index and its length is determined by count. If any parameter is NULL, a NULL will be returned. Otherwise, if string is not a BLOB it will be assumed to be a text value. The index and count parameters will be interpreted as integers. If count is not given, it will effectively be set to an infinitely large positive value. If index is positive, it is used to index characters from the beginning of string. The first character has an index of 1. If index is negative, it is used to index characters from the end of string. In that case, the last character has an index of ‒1. An index of 0 will result in undefined behavior. substr() Appendix E: SQLite SQL Function Reference | 375 If count is positive, then count number of characters, starting with the indexed character, will be included in the returned substring. If count is negative, the returned substring will consist of count number of characters, ending with the indexed character. The returned substring may be shorter than count if the indexed position is too close to the beginning or end of string. A count of zero will result in an empty string. This function can also be used on BLOB values. If string is a BLOB, both the index and count values will refer to bytes, rather than characters, and the returned value will be a BLOB, rather than a text value. See Also replace() time() Decode time string into time of day Common Usage time( timestring, modifier, ... ) Description The time() function takes a timestring value plus zero or more modifier values and returns a text value in the format 'HH:MM:SS'. It is equivalent to the call strftime( '%H:%M:%S', time string, ... ). See Also strftime(), datetime(), date() total_changes() Return the total number of rows changed Common Usage total_changes( ) Description The total_changes() function returns an integer that indicates the number of database rows that have been modified by any complete INSERT, UPDATE, or DELETE command run by this database connection since it was opened. This SQL function is a wrapper around the C function sqlite3_total_changes(), and has all of the same limitations and conditions. See Also changes(), last_insert_rowid(), sqlite3_total_changes() [C API, Ap G] time() 376 | Appendix E: SQLite SQL Function Reference trim() Trim characters from both ends of a string Common Usage trim( text ) trim( text, extra ) Description The trim() function returns a copy of text that has been stripped of any prefix or suffix that consists solely of characters from extra. Characters from extra that are contained in the body of text remain untouched. If extra is not provided, it is assumed to consist of the ASCII space (0x20) character. By default, tab characters and other whitespace are not trimmed. See Also ltrim(), rtrim() typeof() Return the datatype of a value Common Usage typeof( param ) Description The typeof() function returns a text value indicating the native datatype of param. Possible return values include null, integer, real, text, and blob. See Also CAST [SQL Expr, Ap D] upper() Convert all ASCII characters to uppercase Common Usage upper( text ) Description The upper() function returns a copy of text with all of the letter characters converted to uppercase. The built-in implementation of this function only works with ASCII characters (those that have a value of less than 128). The ICU extension provides a Unicode aware implementation of upper(). See Also lower() upper() Appendix E: SQLite SQL Function Reference | 377 zeroblob() Return a BLOB consisting of zeroed-out bytes Common Usage zeroblob( size ) Description The zeroblob() function returns a BLOB value consisting of size bytes. All of the bytes are set to the value zero. The actual BLOB value is not instanced into memory. This makes it safe to create BLOB values that are larger than the environment can handle. This function is most commonly used to create new BLOB values as part of an INSERT or UPDATE. Once the properly sized BLOB value has been recorded into the database, the incremental BLOB I/O API can be used to access or update subsections of the BLOB value. See Also randomblob() Aggregate Functions avg() Compute the numerical average Common Usage avg( number ) Description The avg() aggregate computes the average (mean) numeric value for all non-NULL number values. The aggregate attempts to convert any text or BLOB values into a numeric value. If a conversion is not possible, those rows will be counted with a value of 0. If all the number values are NULL, avg() will return a NULL. See Also count(), sum(), total() count() Count the number of rows Common Usage count( expression ) count( * ) zeroblob() 378 | Appendix E: SQLite SQL Function Reference Description The count() aggregate counts the number of rows with a non-NULL expression. If expres sion is given as an asterisk character (*), the total number of rows in an aggregate group will be returned, regardless of value. In some situations, the computation for the count( * ) syntax can be optimized and the result value can be computed without an actual data scan. See Also sum(), total() group_concat() Concatenate row values Common Usage group_concat( element ) group_concat( element, separator ) Description The group_contact() aggregate returns a text value that is the concatenated text representa- tions of each non-NULL element, separated by separator. If separator is not provided, it is assumed to be the single-character string “,” (comma) without any trailing space. max() Return the largest value Common Usage max( value ) Description The max() aggregate returns the largest non-NULL value. This is typically used to find the largest numeric value, but the aggregate will sort through all types of data. BLOB values are considered to have the largest value, followed by text values. These are followed by the nu- meric types (mixed integer values and floating-point values) that are sorted together in their natural order. If no non-NULL value is found, max() will return a NULL. There is also a scalar version of max() which takes two or more parameters. See Also min() [Scalar SQL Func], max() [Scalar SQL Func] max() Appendix E: SQLite SQL Function Reference | 379 min() Return the smallest value Common Usage min( value ) Description The min() aggregate returns the smallest non-NULL value. This is typically used to find the smallest numeric value, but the aggregate will sort through all types of data. Numeric types (mixed integers and floating-point) are considered smallest, and will be sorted together in their natural order. These are followed by text values, which are followed by BLOB values. If no non-NULL value is found, min() will return a NULL. There is also a scalar version of min() which takes two or more parameters. See Also max(), min() sum() Return the numerical summation Common Usage sum( number ) Description The sum() aggregate computes the sum or total of all the non-NULL number values. Any non- NULL number that is not an integer will be interpreted as a floating-point value. If all non-NULL number values are integers, sum() will return an integer. Otherwise, sum() will return a floating-point value. If no non-NULL number is found, sum() will return NULL. See Also total(), count() total() Return the numerical total Common Usage total( number ) Description The total() aggregate computes the sum or total of all the non-NULL number values. Any non-NULL number that is not an integer will be interpreted as a floating-point value. The total() aggregate is specific to SQLite. Unlike sum(), the total() aggregate will always return a floating-point value. Even if all number expressions are NULL, total() will still return a 0.0. See Also sum(),count() min() 380 | Appendix E: SQLite SQL Function Reference APPENDIX F SQLite SQL PRAGMA Reference This appendix covers all of the PRAGMA commands recognized by SQLite. Pragmas are SQLite-specific statements used to control various environmental variables and state flags within the SQLite environment. They are typically used to activate or configure many of the optional or advanced features of the SQLite library. Although some pragmas are read-only, most pragma commands can be used to query the current value or set a new value. To query the current value, just provide the name of the pragma: PRAGMA pragma_name; In most cases this will return a one-column, one-row result set with the current value. To set a new value, use a syntax like this: PRAGMA pragma_name = value; Although a few pragmas will return the updated value, in most cases the set syntax will not return anything. Any unrecognized or malformed pragma will silently return with- out error. Be very careful about spelling your pragma commands correctly, or SQLite will consider the pragma unrecognized—an application bug that can be extremely dif- ficult to find and fix. Nearly all pragmas fall into one of two categories. The first category includes pragmas that are associated with the database connection. These pragma values can modify and control how the SQLite engine, as a whole, interacts with all of the databases attached to the database connection. The second category of pragmas operate on specific databases. These pragmas will typically allow different values for each database that has been opened or attached to a database connection. Most database-specific identifiers will only last the lifetime of the database connection. If the database is detached and reattached (or closed and reopened), the pragma will assume the default value. There are a few pragmas, however, that will cause a change that is recorded into the database file itself. These values will typically persist across database connections. In both cases, the default values can typ- ically be altered with compile-time directives. 381 The syntax for database-specific pragmas is somewhat similar to database-specific identifiers used elsewhere in SQL, and uses the logical database name, followed by a period, followed by the pragma command: PRAGMA database.pragma_name; PRAGMA database.pragma_name = value; Like identifiers, if a logical database name is not given, most pragmas will assume it to be main. This is the database that was used to open the initial database connection. There are a few exceptions to this syntax, so read the documentation carefully. A number of pragma values are simple true/false Boolean values or on/off state values. In this appendix, these values are referred to as “switches.” To set a switch, several different values are recognized: True or On values False or Off values 1 0 YES NO TRUE FALSE ON OFF A switch pragma will almost always return a simple integer value of 0 or 1. Generally, any value that consists of a numeric value or constant (such as TRUE) can be given without quotes. Known identifiers, such as an attached database name or a table name, can also be given without quotes—unless they contain reserved characters. In that case, they should be given in double quotes, similar to how you would use them in any other SQL command. Other text values, such as filesystem path names, should be placed in single quotes. There are also a handful of read-only pragmas that return a full multicolumn, multirow result set. These results can be processed in code the same way a SELECT result is pro- cessed, using repeated calls to sqlite3_step() and sqlite3_column_xxx(). A few of these pragmas require additional parameters, which are put in parentheses, similar to a func- tion call. For example, here is the syntax for the table_info pragma: PRAGMA database.table_info( table_name ); For more specifics, see the individual pragma descriptions. A number of the pragmas control database values that cannot be changed once the database has been initialized. For example, the page_size pragma must be set before the database structure is written to disk. This may seem like a bit of a chicken-and-egg problem, since you need to open or attach a database in order to configure it, but opening it creates the database. 382 | Appendix F: SQLite SQL PRAGMA Reference Thankfully, the database file header is usually not initialized and actually written to disk until it is absolutely required. This delay in the initialization allows a blank data- base file to be created by opening or attaching a new file. As long as the relevant pragmas are all set before the database is used, everything works as expected. Generally, it is the first CREATE TABLE statement that triggers an initialization. The other case is attaching an additional database. The main database (the one opened to create the database con- nection) must be initialized before any other database is attached to the database connection. If the main database is uninitialized, executing an ATTACH command will first force an initialization of the main (and only the main) database. One final caution when using pragma statements from the C API. Exactly how and when a pragma takes effect is very dependent on the specific pragma and how it is being used. In some cases, the pragma will cause a change when the pragma statement is prepared. In other cases, you must step through the statement. Exactly how and when the pragmas do their thing is dependent on the particular pragma, and has been known to change from one release of SQLite to another. As a result, it is suggested that you do not pre-prepare pragma statements, like you might do with other SQL statements. Rather, any pragma statement should be prepared, stepped, and finalized in a single pass, when you want the pragma to take effect. When using a static pragma statement, it would also be perfectly safe to use sqlite3_exec(). SQLite PRAGMAs auto_vacuum Configure automatic vacuum settings Common Usage PRAGMA [database.]auto_vacuum; PRAGMA [database.]auto_vacuum = mode; Description The auto_vacuum pragma gets or sets the auto-vacuum mode. The mode can be any of the following: Values Meaning 0 or NONE Auto-vacuum is disabled 1 or FULL Auto-vacuum is enabled and fully automatic 2 or INCREMENTAL Auto-vacuum is enabled but must be manually activated The set mode can be either the name or the integer equivalent. The returned value will always be an integer. auto_vacuum Appendix F: SQLite SQL PRAGMA Reference | 383 By default, databases are created with an auto-vacuum mode of NONE. In this mode, when the contents of a database page are deleted, the page is marked as free and added to the free-page list. This is the only action that is taken, meaning that a database file will never shrink in size unless it is manually vacuumed using the VACUUM command. Auto-vacuum allows a database file to shrink as data is removed from the database. In FULL auto-vacuum mode, free pages are automatically swapped with an active page at the end of the database file. The file is then truncated to release the unused space. In FULL mode, a database should never have pages on the free list. The ability to move pages is key to the auto-vacuum system. In order to accomplish this, the database needs to maintain some extra data that allows a page to back-track references. In the event the page needs to be moved, references to the page can also be updated. Keeping all the reference data up to date consumes some storage space and processing time, but it is reasonably small. Swapping free pages and updating references also consumes processing time, and in FULL mode this is done at the end of every transaction. In INCREMENTAL mode, the reference data is maintained, but free pages are not swapped or released—they are simply put on the free list. The pages can be recovered using the incremental_vacuum pragma. This is much faster than a full VACUUM, and allows for partial recovery of space on demand. It can also be done without any additional free space. The mode can be changed between FULL and INCREMENTAL at any time. If the database is in INCREMENTAL mode and is switched to FULL mode, any pages on the free list will automatically be recovered. Because the FULL and INCREMENTAL modes require the extra data that NONE does not maintain, it is only possible to move from NONE to FULL or INCREMENTAL before the database is initialized. Once the database has been initialized, the only way to move away from NONE is to set the pragma and do a VACUUM. Similarly, the only way to move from FULL or INCREMEN TAL to NONE is with a VACUUM. In this case, the VACUUM is required even if the database is still uninitialized. Auto-vacuum has some significant limitations. Although auto-vacuum is capable of releasing free pages, it does so by swapping them with active pages. This can lead to higher levels of fragmentation within the database. Unlike the traditional VACUUM command, auto-vacuum makes no attempt to defragment the database, nor does it repack records into the individual pages. This can lead to inefficient use of space and performance degradation. Because of these limitations, it is recommended any database with a moderate transaction rate is occasionally vacuumed, even if auto-vacuum is enabled. See Also incremental_vacuum, VACUUM [SQL Cmd, Ap C] cache_size Set the size of the database page cache Common Usage PRAGMA [database.]cache_size; PRAGMA [database.]cache_size = pages; cache_size 384 | Appendix F: SQLite SQL PRAGMA Reference Description The cache_size pragma can get or temporarily set the maximum size of the in-memory page cache. The pages value represents the number of pages in the cache. Normally, each attached database has an independent cache. Adjustments made by the cache_size pragma only persist for the lifetime of the database connection. A common use of this pragma is to temporarily boost the cache size for I/O intensive operations. When building a new index on a large table, raising the cache size (sometimes as much as 100× or even 1,000× the default size) can often result in a considerable performance increase. Care must be taken when using extremely large cache sizes, however. If the cache size is so large that the cache grows to exceed the available physical memory, the overall performance is likely to be much lower than simply using a smaller cache. The actual amount of memory used by the cache is determined by the size of the database pages (plus some overhead) and the cache size (in pages). The built-in page cache has a default size of 2,000 pages and a minimum size of 10 pages. The full data cache is not allocated immediately, but grows on demand with the cache size acting as a limiter on that growth. If the cache size is made larger, the limit is simply raised. If the cache size is made smaller, the limit is lowered but the cache is not necessarily immediately flushed to recover the memory. The built-in page cache requires some overhead. The exact size of the cache overhead depends on the platform, but it is in the neighborhood of 100 bytes. The maximum memory foot-print for the cache can be approximated with the formula ( page_size + 128 ) * cache_size. See Also default_cache_size, page_size case_sensitive_like Control the case-sensitivity of the LIKE operator Common Usage PRAGMA case_sensitive_like = switch; Description The case_sensitive_like pragma controls the case-sensitivity of the built-in LIKE expression. By default, this pragma is false, indicating that the built-in LIKE operator ignores lettercase. This pragma applies to all databases attached to a database connection. This is a write-only pragma. There is no way to query for the current state, other than issuing an SQL statement such as SELECT 'A' NOT LIKE 'a';. If case_sensitive_like is true, this statement will return 1. This pragma only applies to the built-in version of LIKE. If the default behavior has been overridden by a user-defined like() SQL function, this pragma is ignored. See Also like() [SQL Func, Ap E], LIKE [SQL Expr, Ap D] case_sensitive_like Appendix F: SQLite SQL PRAGMA Reference | 385 collation_list List current collations Common Usage PRAGMA collation_list; Description The collation_list pragma lists all the active collations in the current database connection. This pragma will return a two-column table with one row per active collation. Column name Column type Meaning seq Integer Collation sequence number name Text Name of collation Unless an application has defined additional collations, the list will be limited to the built-in collations NOCASE, RTRIM, and BINARY. See Also database_list, sqlite3_create_collation() [C API, Ap G] count_changes Enable change counts for INSERT, UPDATE, and DELETE Common Usage PRAGMA count_changes; PRAGMA count_changes = switch; Description The count_changes pragma gets or sets the return value of data manipulation statements such as INSERT, UPDATE, and DELETE. By default, this pragma is false and these statements do not return anything. If set to true, each data manipulation statement will return a one-column, one-row table consisting of a single integer value. This value indicates how many rows were modified by the statement. The values returned are very similar to those returned by the SQL function changes(), with one small difference. Rows that are modified by an INSTEAD OF trigger on a view will be counted within the statement return value, but will not be counted by changes(). This pragma affects all statements processed by a given database connection. See Also changes() [SQL Func, Ap E], sqlite3_changes() [C API, Ap G] database_list List currently attached databases Common Usage PRAGMA database_list; collation_list 386 | Appendix F: SQLite SQL PRAGMA Reference Description The database_list pragma lists all of the attached databases in this database connection. This pragma will return a three-column table with one row per open or attached database. Column name Column type Meaning seq Integer Database sequence number name Text Logical database name file Text Path and name of database file Not all attached databases will have an associated file. In-memory databases and temporary databases only have an empty string in the file column. See Also ATTACH DATABASE [SQL Cmd, Ap C] default_cache_size Set default size of page cache for this database Common Usage PRAGMA [database.]default_cache_size; PRAGMA [database.]default_cache_size = pages; Description The default_cache_size pragma controls default cache size for a given database. The pages value is stored in the database file, allowing this value to persist across database connections. The default cache size may be temporarily overridden using the cache_size pragma. Setting this value also sets the limit on the current page cache. See Also cache_size, page_size encoding Control the default text encoding Common Usage PRAGMA encoding; PRAGMA encoding = format; Description The encoding pragma controls how strings are encoded and stored in a database file. Any string value that is recorded to the database is first re-encoded into this format. The format value can be one of 'UTF-8', 'UTF-16le', or 'UTF-16be'. Additionally, if the value 'UTF-16' is given, either 'UTF-16le' or 'UTF-16be' will be used, as determined by the native endian of the processor. Any version of SQLite running on any platform should be able to read a database file, regardless of the encoding. encoding Appendix F: SQLite SQL PRAGMA Reference | 387 The encoding can only be set on the main database, and only before the database is initialized. All attached databases must have the same encoding as the main database. If ATTACH is used to create a new database, it will automatically inherit the encoding of the main database. If the main database has not been initialized when ATTACH is run, it will automatically be initialized with the current default values before the ATTACH command is allowed to run. See Also ATTACH DATABASE [SQL Cmd, Ap C] foreign_keys Enable foreign key constraints Common Usage PRAGMA foreign_keys; PRAGMA foreign_keys = switch; Description The foreign_keys pragma controls the enforcement of foreign key constraints in all of the attached databases. If set to off, foreign key constraints are ignored. The default value is off. For many years, SQLite would parse foreign key constraints, but was unable to enforce them. When native support for foreign key constraints was finally added to the SQLite core, en- forcement was off by default to avoid any backwards compatibility issues. This default value may change in future releases. To avoid problems, it is recommended that an application explicitly set the pragma one way or the other. This pragma cannot be set when there is an active transaction in progress. See Also foreign_key_list foreign_key_list List all foreign keys in the given table Common Usage PRAGMA [database.]foreign_key_list( table_name ); Description The foreign_key_list pragma lists all the foreign key references that are part of the specified table. That list will contain one row for each column of each foreign key contained within the table. Column name Column type Meaning id Integer Foreign key ID number seq Integer Column sequence number for this key table Text Name of foreign table from Text Local column name foreign_keys 388 | Appendix F: SQLite SQL PRAGMA Reference Column name Column type Meaning to Text Foreign column name on_update Text ON UPDATE action on_delete Text ON DELETE action match Text Always NONE See Also foreign_keys freelist_count Return number of free pages in database file Common Usage PRAGMA [database.]freelist_count; Description The freelist_count pragma returns a single integer indicating how many database pages are currently marked as free and available (contain no valid data). These pages can be recovered by vacuuming the database. See Also auto_vacuum, incremental_vacuum, VACUUM [SQL Cmd, Ap C] full_column_names Control column-name format used in queries Common Usage PRAGMA full_column_names; PRAGMA full_column_names = switch; Description The full_column_names pragma, in conjunction with the short_column_names pragma, controls how the database connection specifies and formats column names in result sets. If full_col umn_name is enabled, output column expressions that consist of a single named table column will be expanded to the full table_name.column_name format. It is off by default. The general rules for output names are: • If an output column has an AS name clause, name is used. • If short_column_names is enabled and the output column is an unmodified source column, the result set column name is column_name. • If full_column_names is enabled and the output column is an unmodified source column, the result set column name is table_name.column_name. • The result set column name is the text of the column expression, as given. full_column_names Appendix F: SQLite SQL PRAGMA Reference | 389 If full_column_names and short_column_names are both enabled, short_column_names will over- ride full_column_names. Note that there is no guarantee the result set column names will remain consistent with future versions of SQLite. If your application depends on specific, recognizable column names, you should always use an AS clause. See Also short_column_names fullfsync Control the level of disk synchronization Common Usage PRAGMA fullfsync; PRAGMA fullfsync = switch; Description The fullfsync pragma enables the F_FULLFSYNC filesystem synchronization method. This pro- vides an extra layer of robust file syncing. Currently, Mac OS X is the only operating system that supports this method. See Also synchronous ignore_check_constraints Disable CHECK constraints Common Usage PRAGMA ignore_check_constraints; PRAGMA ignore_check_constraints = switch; Description The ignore_check_constraints pragma controls the enforcement of CHECK constraints. CHECK constraints are defined in CREATE TABLE statements as arbitrary expressions that must be true before a row can be inserted or updated. If this pragma is set, this type of constraint is ignored. Turning this pragma back on will not verify existing rows. This is an undocumented pragma. incremental_vacuum Activate an incremental vacuum Common Usage PRAGMA [database.]incremental_vacuum( pages ); Description The incremental_vacuum pragma manually triggers a partial vacuum of a database file. If the database has auto-vacuum enabled and is in incremental mode, this pragma will attempt to fullfsync 390 | Appendix F: SQLite SQL PRAGMA Reference release up to pages file pages. This is done by migrating them to the end of the file and trun- cating the file. If pages is omitted, or has a value of zero or less, the entire free list will be released. If the database does not have auto-vacuum enabled, or is not in incremental mode, this pragma has no effect. See Also auto_vacuum, freelist_count, VACUUM [SQL Cmd, Ap C] index_info List table and columns contained in an index Common Usage PRAGMA [database.]index_info( index_name ); Description The index_info pragma queries information about a database index. The result set will contain one row for each column contained in the index. Column name Column type Meaning seq Integer Column sequence number cid Integer Column index within table name Text Name of column See Also index_list, table_info index_list List all indexes associated with a table Common Usage PRAGMA [database.]index_list( table_name ); Description The index_list pragma lists all of the indexes associated with a table. The result set will contain one row for each index. Column name Column type Meaning seq Integer Index sequence number name Text Name of index unique Integer Is UNIQUE index See Also index_info, table_info index_list Appendix F: SQLite SQL PRAGMA Reference | 391 integrity_check Initiate a database file integrity verification Common Usage PRAGMA [database.]integrity_check; PRAGMA [database.]integrity_chcek( max_errors ); Description The integrity_check pragma runs a self-check on the database structure. The check runs through a battery of tests that verify the integrity of the database file, its structure, and its contents. Errors are returned as text descriptions in a single-column table. At most, max_errors are reported before the integrity check aborts. By default, max_errors is 100. If no errors are found, a single row consisting of the text value ok will be returned. Unfortunately, if an error is found, there is typically very little that can be done to fix it. Although you may be able to extract some of the data, it is best to have regular dumps or backups. See Also quick_check journal_mode Control the creation and cleanup of journal files Common Usage PRAGMA journal_mode; PRAGMA journal_mode = mode; PRAGMA database.journal_mode; PRAGMA database.journal_mode = mode; Description The journal_mode pragma gets or sets the journal mode. The journal mode controls how the journal file is stored and processed. Journal files are used by SQLite to roll back transactions due to an explicit ROLLBACK command, or because an unrecoverable error was encountered (such as a constraint violation). Normally, a journal file is required to process any SQL command that causes a modification to a database file. The active lifetime of a journal file extends from the first database modification to the end of the commit process. Journal files are required for both auto-commit transactions (in- dividual SQL commands) as well as explicit transactions. Journal files are a critical part of the transaction process. There are five supported journal modes: DELETE This is the normal behavior. At the conclusion of a transaction, the journal file is deleted. TRUNCATE The journal file is truncated to a length of zero bytes. On many filesystems, this is slightly faster than a delete. integrity_check 392 | Appendix F: SQLite SQL PRAGMA Reference PERSIST The journal file is left in place, but the header is overwritten to indicate the journal is no longer valid. On some filesystems, this can be faster still, as the file blocks remain allocated. MEMORY The journal record is held in memory, rather than on disk. This option is extremely fast, but also somewhat dangerous. If an application crashes or is killed while a transaction is in progress, it is likely to result in a corrupt database file. OFF No journal record is kept. As with the MEMORY journal mode, if the application crashes or is killed, the database will likely become corrupted. Further, if the journal is completely turned off, the database has no rollback capability. The ROLLBACK command should not be used on a database that has the journal turned off. The behavior of the journal_mode pragma depends on whether an explicit database name is given or not. If no database name is given, the pragma gets or sets the default value stored in the database connection. This is the mode that will be used for any newly attached database, and will not modify database files that are already open. Each open database also has its own unique journal mode. If a database name is given, this pragma will get or set the journal mode for that specific database. If you want to set the mode of the main database, you must explicitly use the main database name. All versions of the journal_mode pragma will return the current value as a lowercase text value. In-memory databases only support the MEMORY and OFF modes. Any attempt to set an in- memory database to a different mode will silently fail and the existing mode will be returned. See Also journal_size_limit, synchronous, ROLLBACK TRANSACTION [SQL Cmd, Ap C] journal_size_limit Set a maximum size for released journal files Common Usage PRAGMA [database.]journal_size_limit; PRAGMA [database.]journal_size_limit = max_bytes; Description The journal_size_limit pragma forces the partial deletion of large journal files that would otherwise be left in place. Although journal files are normally deleted when a transaction completes, if a database is set to use persistent journal mode, the journal file is simply left in place. In some situations, the journal file can also remain if the database is set to use exclusive locking mode. If a journal size limit has been set, SQLite will take the time to examine any journal file that would normally be left in place. If the journal file is larger than max_bytes, the file is truncated to max_bytes. This keeps very larger journal files (such as those left behind from a VACUUM) from continuing to consume excessive storage space. journal_size_limit Appendix F: SQLite SQL PRAGMA Reference | 393 When first opened or attached, each database is assigned the compile-time default. This is normally ‒1, indicating that there is no limit. Each database in a database connection can be set to a different value. Databases must be set individually, even if they are all set to the same value. Both the get and the set syntax will return the current limit. See Also journal_mode, locking_mode legacy_file_format Determine the default format for new database files Common Usage PRAGMA legacy_file_format; PRAGMA legacy_file_format = switch; Description The legacy_file_format pragma gets or sets the legacy compatibility flag. This flag is used to determine what file format will be used when creating new database files. If the flag is set to true, SQLite will create new database files in a “legacy” file format that is backwards com- patible all the way to SQLite v3.0.0. If the flag is not set, SQLite will create new files in the most up-to-date format supported by that library. By default, this pragma is on, meaning databases will be created using the legacy format. Because the default is on, the vast majority of SQLite files out there are in the legacy file format. This pragma cannot be used to determine the format of an existing file, nor can it be used to change an existing file. It only effects new files created with this database connection, and must be set before the database file is initialized. An existing legacy file can be promoted to the latest format by setting this pragma to off and manually vacuuming the file, but down- grades back to the legacy format are not supported. The main difference between the original file format and the most current file format is support for descending indexes. Descending indexes cannot be used in conjunction with the legacy file format. The newer format also uses a more efficient on-disk representation for the integer values 0 and 1 (used for Boolean values) that saves one byte per value. Please note that the use of this pragma does not guarantee backwards compatibility. If a database is created using the legacy file format, but the schema (CREATE TABLE, etc.) uses more modern SQL language features, you won’t be able to open the database with an older version of SQLite. When the latest file format (format v4) was first introduced, this flag had a default value of false. This caused a great number of problems in environments that did incremental or spora- dic upgrades. To avoid such problems, the default value was changed to true. It is conceivable that the default value for this pragma may change in the future. The v4 file format is backward compatible to SQLite version 3.3.0. See Also VACUUM [SQL Cmd, Ap C], CREATE INDEX [SQL Cmd, Ap C] legacy_file_format 394 | Appendix F: SQLite SQL PRAGMA Reference locking_mode Control how a database releases read/write locks Common Usage PRAGMA locking_mode; PRAGMA locking_mode = mode; PRAGMA database.locking_mode; PRAGMA database.locking_mode = mode; Description The locking_mode pragma controls how database file locks are managed. The mode can either be NORMAL or EXCLUSIVE. In normal mode, the database connection acquires and releases the appropriate locks with each transaction. In exclusive mode, the locks are acquired in the normal fashion, but are not released when a transaction finishes. Although exclusive locking mode prevents any other database connections from accessing the database, it also provides better performance. This may be an easy trade-off in situations such as some embedded environments, where it is very unlikely that multiple processes will ever attempt to access the same database. Exclusive locking reduces the start-up cost of a transaction by eliminating the need to acquire the appropriate locks. It also allows SQLite to skip several file reads at the start of each trans- action. For example, when SQLite starts a new transaction, it will normally reread the database header to verify that the database schema has not changed. This is not required if the database is in exclusive mode. Since temporary and in-memory databases cannot be accessed by more than one database connection, they are always in exclusive mode. Any attempt to set these database types to normal mode will silently fail. The behavior of the locking_mode pragma depends on if an explicit database is given or not. If no database name is given, the pragma gets or sets the default value stored in the database connection. This is the mode that will be used for any newly attached database. Each open database also has its own unique journal mode. If a database name is given, this pragma will get or set the journal mode for that specific database. If you want to get or set the mode of the main database, you must explicitly use the main database name. There are two ways to release the locks of a database in exclusive mode. First, the database can simply be closed (or detached). This will release any locks. The database can be reopened (or reattached). Even if the default locking mode is exclusive, the locks will not be acquired until the database goes through a transaction cycle. The second way to release the locks is to place the database into normal locked mode and execute some SQL command that will force a lock/unlock transaction cycle. Simply putting the database into normal mode will not release any locks that have already been acquired. See Also journal_mode, lock_status, ATTACH DATABASE [SQL Cmd, Ap C], DETACH DATABASE [SQL Cmd, Ap C] locking_mode Appendix F: SQLite SQL PRAGMA Reference | 395 lock_proxy_file Set the proxy locking directory Common Usage PRAGMA lock_proxy_file; PRAGMA lock_proxy_file = ':auto:'; PRAGMA lock_proxy_file = 'file_path'; Description The lock_proxy_file pragma gets or sets the file used for proxy locks. The value :auto: will set it back to the automatic default. The proxy locking system is only available on Mac OS X, and only applies to databases residing on Apple Filing Protocol shares. This is an undocumented pragma. lock_status Display lock state of each database Common Usage PRAGMA lock_status; Description The lock_status pragma lists all of the databases in a connection and their current lock status. The result set will contain one row for each attached database. Column name Column type Meaning database Text Database name status Text Lock status The following locking states may be indicated: Value Meaning unlocked Database has no locks shared Database has a shared read lock reserved Database has the reserved write-enable lock pending Database has the pending write-enable lock exclusive Database has the exclusive write lock closed Database file is not open unknown Lock state is not known The first five states correspond to the standard locks SQLite uses to maintain read and write transactions. The closed value is normally returned when a brand new database file has been created, but it has not yet been initialized. Since the locking is done through the filesystem, any database that does not have an associated file will return unknown. This includes both in- memory databases, as well as some temporary databases. lock_proxy_file 396 | Appendix F: SQLite SQL PRAGMA Reference SQLite must be compiled with the SQLITE_DEBUG directive for this pragma to be included. This is an undocumented pragma. See Also schema_version, journal_mode, journal_size_limit, ATTACH DATABASE [SQL Cmd, Ap C], DETACH DATABASE [SQL Cmd, Ap C] max_page_count Limit the size of a database Common Usage PRAGMA [database.]max_page_count; PRAGMA [database.]max_page_count = max_page; Description The max_page_count pragma gets or sets the maximum allowed page count for a database. This value is database specific, but is stored as part of the database connection and must be reset every time the database is opened. If a database attempts to grow past the maximum allowed page count, an out-of-space error will be returned, similar to when a filesystem runs out of space. Both versions of this command return the current value. If you attempt to set the value lower than the current page count, no change will be made (and the old value will be returned). The default value is 1,073,741,823 (230 ‒ 1, or one giga-page). When used in conjunction with the default 1 KB page size, this allows databases to grow up to one terabyte. There is no way to restore the default value, other than to manually set the same value (or close and reopen the database). The max_page value is limited to a 32-bit signed value. See Also page_count, page_size omit_readlock Disable read locks on read-only files Common Usage PRAGMA omit_readlock; PRAGMA omit_readlock = switch; Description The omit_readlock pragma disables read locks when accessing a read-only file. Disabling read locks will provide better performance, but can only be done safely when all processes access the file as read-only. This flag is particularly useful when accessing database files on read-only media, such as an optical disc. It can also be safely used for “reference” database files, such as dictionary files, that are distributed but never modified by the client software. omit_readlock Appendix F: SQLite SQL PRAGMA Reference | 397 Disabling read locks for all read-only files is not recommended. If one process opens a database read-only and another process opens the same database read/write, the locking system is still required for transactions to work properly. This pragma should only be considered in situa- tions where every possible process that might access a database file is doing so in a read-only fashion. This is an undocumented pragma. page_count Return total number of pages in a database Common Usage PRAGMA [database.]page_count; Description The page_count pragma returns the current number of pages in database. The page count includes all pages in the file, including free pages. The size of the database file should be page_count * page_size. See Also max_page_count, page_size, cache_size, freelist_count page_size Set the size of a database page Common Usage PRAGMA [database.]page_size; PRAGMA [database.]page_size = bytes; Description The page_size pragma gets or sets the size of the database pages. The bytes size must be a power of two. By default, the allowed sizes are 512, 1024, 2048, 4096, 8192, 16384, and 32768 bytes. This value becomes fixed once the database is initialized. The only way to alter the page size on an existing database is to set the page size and then immediately VACUUM the database. The default page size is calculated from a number of factors. The default page size starts at 1024 bytes. If the datatype driver indicates that the native I/O block of the filesystem is larger, that larger value will be used up to the maximum default size, which is normally set to 8192. These values can be altered with the SQLITE_DEFAULT_PAGE_SIZE, SQLITE_MAX_DEFAULT_PAGE_ SIZE, and SQLITE_MAX_PAGE_SIZE compiler-time directives. The end result is that the page size for file-based databases will typically be between 1 KB and 4 KB on Microsoft Windows, and 1 KB on most other systems, including Mac OS X, Linux, and other Unix systems. See Also cache_size page_count 398 | Appendix F: SQLite SQL PRAGMA Reference parser_trace Enable parser debug information Common Usage PRAGMA parser_trace = switch; Description The parser_trace pragma enables SQL parser debugging. When enabled, the SQL parser will print its state as it parses SQL commands. SQLite must be compiled with the SQLITE_DEBUG directive for parser trace functionality to be included. See Also sql_trace, vdbe_trace, vdbe_listing quick_check Initiate a partial database file integrity verification Common Usage PRAGMA [database.]quick_check; PRAGMA [database.]quick_check( max_errors ); Description The quick_check pragma runs an abbreviated integrity check on a database file. The quick_check pragma is identical to the integrity_check pragma, except that it does not verify that the contents of each index is in sync with its source-table data. By skipping this test, the time required to do the integrity test is greatly reduced. See Also integrity_check read_uncommitted Enable reads of uncommitted data from database cache Common Usage PRAGMA read_uncommitted; PRAGMA read_uncommitted = switch; Description The read_uncommitted pragma gets or sets the shared cache isolation method. If the same process opens the same database multiple times, SQLite can be configured to allow those connections to share a single cache instance. This is helpful in very low-memory situa- tions, such as low-cost embedded systems. This pragma controls which locks are required to access the shared cache. Setting this pragma relaxes some of the locking requirements and allows connections to read from the cache, even if another connection is in the middle of a transaction. This allows better concurrency, but it also means that readers may see data the writer has not committed, breaking transaction isolation. read_uncommitted Appendix F: SQLite SQL PRAGMA Reference | 399 The read_uncommitted pragma is only applicable to systems that use shared cache mode, which is normally not used on desktop systems. If you are using shared cache mode and may have a need for this pragma, please see the source code and online documentation at http://www .sqlite.org/sharedcache.html for more information. recursive_triggers Enable recursive trigger calls Common Usage PRAGMA recursive_triggers; PRAGMA recursive_triggers = switch; Description The recursive_triggers pragma gets or sets the recursive trigger functionality. If recursive triggers are not enabled, a trigger action will not fire another trigger. For many years, SQLite did not support recursive triggers. When they were added, the default for this pragma was set to off, in an effort to keep the default settings backwards compatible. The default version of this pragma may change in a future version of SQLite. See Also foreign_keys reverse_unordered_selects Reverse the order of unsorted query results Common Usage PRAGMA reverse_unordered_selects; PRAGMA reverse_unordered_selects = switch; Description The reverse_unordered_selects pragma gets or sets the reverse select flag. If this flag is set, SQLite will reverse the natural ordering of SELECT statements that do not have an explicit ORDER BY clause. SQLite, as well as the SQL standard, makes no promises about the ordering of SELECT results that do not have an explicit ORDER BY clause. Changes in the database (such as adding an index), or changes in the query optimizer can cause SQLite to return rows in a different order. Unfortunately, writing application code that is dependent on the results order is a common mistake. This pragma can be used to help find and fix those types of bugs by altering the default output order. This can be especially effective if it is turned on or off randomly before each query. recursive_triggers 400 | Appendix F: SQLite SQL PRAGMA Reference schema_version Database schema change control Common Usage PRAGMA [database.]schema_version; PRAGMA [database.]schema_version = number; Description The schema_version pragma gets or sets the schema version value that is stored in the database header. This is a 32-bit signed integer value that keeps track of schema changes. Whenever a schema-altering command is executed (for example, CREATE... or DROP...), this value is incremented. This value is used internally by SQLite to keep a number of different caches consistent, as well as keep prepared statements consistent with current database format. Manually altering this value can result in a corrupt database. See Also user_version secure_delete Overwrite deleted content with zeros Common Usage PRAGMA secure_delete; PRAGMA secure_delete = switch; PRAGMA database.secure_delete; PRAGMA database.secure_delete = switch; Description The secure_delete pragma is used to control how content is deleted from the database. Nor- mally, deleted content is simply marked as unused. If the secure_delete flag is on, deleted content is first overwritten with a series of 0 byte values, removing the deleted values from the database file. The default value for the secure delete flag is normally off, but this can be changed with the SQLITE_SECURE_DELETE build option. If a database name is given, the flag will be get or set for just that database. If no database name is given, setting the value will set it for all attached databases, while getting the value will return the current value for the main database. Newly attached databases will take on the same value as the main database. Be aware that SQLite cannot securely delete information from the underlying storage device. If the write operation causes the filesystem to allocate a new device-level block, the old data may still exist on the raw device. There is also a slight performance penalty associated with this directive. secure_delete Appendix F: SQLite SQL PRAGMA Reference | 401 The secure_delete flag is stored in the page cache. If shared cache mode is enabled, changing this flag on one database connection will cause the flag to be changed for all database con- nections sharing that cache instance. See Also SQLITE_SECURE_DELETE short_column_names Control column-name format used in queries Common Usage PRAGMA short_column_names; PRAGMA short_column_names = switch; Description The short_column_names pragma, in conjunction with the full_column_names pragma, controls how the database connection specifies and formats column names in result sets. If short_col umn_name is enabled, output column expressions that consist of a single named table column will be clipped to only include the column name. This pragma is on by default. The general rules for output names are: • If an output column has an AS name clause, name is used. • If short_column_names is enabled and the output column is an unmodified source column, the result set column name is column_name. • If full_column_names is enabled and the output column is an unmodified source column, the result set column name is table_name.column_name. • The result set column name is the text of the column expression, as given. If full_column_names and short_column_names are both enabled, short_column_names will over- ride full_column_names. Note that there is no guarantee the result set column names will remain consistent with future versions of SQLite. If your application depends on specific, recognizable column names, you should use an AS clause. See Also full_column_names sql_trace Dump SQL trace data Common Usage PRAGMA sql_trace; PRAGMA sql_trace = switch; short_column_names 402 | Appendix F: SQLite SQL PRAGMA Reference Description The sql_trace pragma dumps SQL trace results to the screen. When enabled, trace data normally passed to the sqlite3_trace() callback will be printed. SQLite must be compiled with the SQLITE_DEBUG directive for this pragma to be included. This is an undocumented pragma. See Also vdbe_trace, parser_trace, vdbe_listing synchronous Control the database disk synchronization Common Usage PRAGMA [database.]synchronous; PRAGMA [database.]synchronous = mode; Description The synchronous pragma gets or sets the current disk synchronization mode. This controls how aggressively SQLite will write data all the way out to physical storage. Because most physical storage systems (such as hard drives) are very slow when compared to processor and memory speeds, most computing environments have a large number of caches and buffers between an application and the actual, long-term physical storage system. These layers introduce a significant window of time between the time when an application is told the data was successfully written, and the time when the data is actually written to long-term storage. Three or four seconds is typical, but in some cases this window can be a dozen seconds or more. If the system crashes or suffers a power failure within that window, some of the “written” data will be lost. If that were to happen to an SQLite database file, the database would undoubtedly become corrupted. To properly enforce transactions and prevent corruption, SQLite depends on writes being permanent, even in the face of a system crash or power failure. This requires that SQLite write commands happen in order and go all the way to the physical storage. To ac- complish this, SQLite will request an immediate disk synchronization after any critical write operations. This causes the application to pause until the operating system can confirm that the data has been written to long-term storage. While this is very safe, it is also very slow. In some situations, it may be acceptable to turn off some of these protections in favor of raw speed. While this isn’t recommended for long-term situations, it can make sense for short, repeatable operations, such as bulk-loading import data. Just be sure to make a backup first. SQLite offers three levels of protection, as shown in the following table: Mode Meaning 0 or OFF No syncs at all 1 or NORMAL Sync after each sequence of critical disk operations 2 or FULL Sync after each critical disk operation synchronous Appendix F: SQLite SQL PRAGMA Reference | 403 The set mode can be either the name or the integer equivalent. The returned value will always be an integer. In FULL mode, a full synchronization is done after each and every critical disk operation. This mode is designed to avoid corruption in the face of any application crash, system crash, or power failure. It is the safest, but also the slowest. The default mode is FULL (not NORMAL). In NORMAL mode, a full synchronization is done after each sequence of critical disk operations. This mode reduces the total number of synchronization calls, but introduces a very small chance of having a system crash or power failure corrupt the database file. NORMAL mode at- tempts to strike a balance between good protection and moderate performance. In OFF mode, no attempt is made to flush or synchronize writes, leaving it up to the operating system to write out any filesystem cache at its own convenience. This mode is much, much faster than the other two modes, but leaves SQLite wide open to system crashes and power failures, even after a transaction has completed. Be aware that on some systems, a power failure can still cause database corruption, even if running in FULL mode. There are a number of disk controllers that will falsely report a syn- chronization request as successfully completed before all of the data is actually written to nonvolatile storage. If a power failure happens while data is still being held in the disk con- troller cache, a file corruption may still occur. Unfortunately, there isn’t anything SQLite or the host operating system can do about this, as there is no way for the operating system or SQLite to know the information is untrue. See Also journal_mode, fullfsync table_info List column information for a table Common Usage PRAGMA [database.]table_info( table_name ); Description The table_info pragma is used to query information about a specific table. The result set will contain one row for each column in the table. Column name Column type Meaning cid Integer Column index name Text Column name type Text Column type, as given notnull Integer Has a NOT NULL constraint dflt_value Text DEFAULT value pk Integer Is part of the PRIMARY KEY table_info 404 | Appendix F: SQLite SQL PRAGMA Reference The default value (dflt_value) will be given as a text representation of the literal value rep- resentation. For example, a default text value includes the single quotes. See Also index_list temp_store Control the temporary storage mode Common Usage PRAGMA temp_store; PRAGMA temp_store = mode; Description The temp_store pragma gets or sets the storage mode used by temporary database files. This pragma does not affect journal files. SQLite supports the following storage modes: Mode Meaning 0 or DEFAULT Use compile-time default. Normally FILE. 1 or FILE Use file-based storage 2 or MEMORY Use memory-based storage The set mode can be either the name or the integer equivalent. The returned value will always be an integer. Memory-based storage will make temporary databases equivalent to in-memory databases. File-based databases will initially be in-memory databases, until they outgrow the page cache. This means that many “file-based” temporary databases never actually make it into a file. Changing the mode will cause all temporary databases (and the data they contain) to be deleted. In some cases, this pragma can be disabled with the SQLITE_TEMP_STORE compile-time directive. Possible compile-time values include: Value Meaning 0 Always use files, ignore pragma 1 Allow pragma, default to files 2 Allow pragma, default to memory 3 Always use memory, ignore pragma temp_store Appendix F: SQLite SQL PRAGMA Reference | 405 The default value is 1. Temporary storage defaults to using files, but allows the temp_store pragma to override that choice. See Also temp_store_directory, journal_mode temp_store_directory Control the temporary file storage location Common Usage PRAGMA temp_store_directory; PRAGMA temp_store_directory = 'directory_path'; Description The temp_store_directory pragma gets or sets the location used for temporary database files. This pragma does not change the location of journal files. If the specified location is not found or is not writable, an error is generated. To revert the directory to its default value, set directory_path to a zero-length string. The default value (as well as the interpretation of this value) is OS dependent. Changing this pragma will cause all temporary databases (and the data they contain) to be deleted. Setting this pragma is not thread-safe. Ideally, this pragma should only be set as soon as the main database is opened. Many temporary databases are never actually created on the filesystem, even if they’re set to use a file for storage. Those files that are created are typically opened and then immediately deleted. Opening and deleting a file keeps the file active in the filesystem, but removes it from the directory listing. This prevents other processes from seeing or accessing the file, and guar- antees the file will be fully deleted, even if the application crashes. See Also temp_store user_version Control user-defined versioning Common Usage PRAGMA [database.]user_version; PRAGMA [database.]user_version = number; Description The user_version pragma gets or sets the user-defined version value that is stored in the database header. This is a 32-bit signed integer value that may be used for whatever purpose the developer sees fit. The value is not used by SQLite in any way. See Also schema_version temp_store_directory 406 | Appendix F: SQLite SQL PRAGMA Reference vdbe_trace Enable VDBE debug information Common Usage PRAGMA vdbe_trace; PRAGMA vdbe_trace = switch; Description The vdbe_trace pragma enables virtual database engine (VDBE) debugging. When enabled, each VDBE instruction is printed just prior to execution. SQLite must be compiled with the SQLITE_DEBUG directive for the VDBE trace functionality to be included. See the online VDBE documentation (http://www.sqlite.org/vdbe.html) for more details. See Also vdbe_listing, sql_trace, parser_trace vdbe_listing Cause the VDBE to dump each program before executing Common Usage PRAGMA vdbe_listing; PRAGMA vdbe_listing = switch; Description The vdbe_listing pragma enables VDBE debugging. When enabled, each VDBE program is printed just prior to execution. SQLite must be compiled with the SQLITE_DEBUG directive for the VDBE list functionality to be included. See the online VDBE documentation (http://www.sqlite.org/vdbe.html) for more details. See Also vdbe_trace, sql_trace, parser_trace writable_schema Allow modification of system tables Common Usage PRAGMA writable_schema; PRAGMA writable_schema = switch; Description The writable_schema pragma gets or sets the ability to modify system tables. If writable_ schema is set, tables that start with sqlite_ can be created and modified, including the sqlite_master table. This pragma makes it possible to corrupt a database using only SQL commands. Be extremely careful when using this pragma. This is an undocumented pragma. writable_schema Appendix F: SQLite SQL PRAGMA Reference | 407 APPENDIX G SQLite C API Reference This appendix covers the data structures and functions that make up the C program- ming API. The appendix is designed to act as a reference, providing specific details about each function and data structure. It does not, however, provide a high-level overview of how all the parts fit together into a useful application. To understand the ideas and patterns behind the API, it is strongly recommended you review the material in Chapter 7. Those items marked [EXP] should be considered experimental. This is not meant to imply that those functions and features are risky or untested, simply that they are newer, and there is some chance the interface may change slightly in a future version of SQLite. This is somewhat rare, but it does happen. Those functions not marked experimental are more or less set in stone. The behavior of a function may change, if it is found to have a minor bug, but the interface definition will not. The SQLite team has a very strong belief in backwards compatibility and will not change the behavior of an existing interface if there is any chance it might break existing applications. New features that require changes result in new interfaces, such as the _v2 version of some functions (for example, sqlite3_open_v2()). Generally, this makes upgrading the SQLite library used with an existing library a low-risk process. The official documentation for the complete C programming interface can be found at http://www.sqlite.org/c3ref/intro.html. 409 API Datatypes This is a partial list of C datatypes used by the SQLite API. All of the common datatypes are listed. Those that have been omitted are extremely specialized, and are only used when implementing low-level extensions. For a complete list and more information, see http://www.sqlite.org/c3ref/objlist.html or the SQLite header files. sqlite3 A database connection Description The sqlite3 structure represents a database connection to one or more database files. The structure is created with a call to sqlite3_open() or sqlite3_open_v2() and destroyed with sqlite3_close(). Nearly all data management operations must be done in the context of an sqlite3 instance. The sqlite3 structure is opaque, and an application should never access any of the data fields directly. See Also sqlite3_open(), sqlite3_open_v2(), sqlite3_close() sqlite3_backup An online backup context [EXP] Description The sqlite3_backup structure performs an online backup. This is done by making a low-level, page-by-page copy of the database image. The structure is created with a call to sqlite3_backup_init() and destroyed with sqlite3_backup_finish(). The sqlite3_backup structure is opaque, and an application should never access any of the data fields directly. See Also sqlite3_backup_init(), sqlite3_backup_finish() sqlite3_blob Incremental BLOB I/O Description The sqlite3_blob structure performs incremental I/O on a BLOB value. This allows an ap- plication to read and write BLOB values that are too large to fit in memory. The structure is created with sqlite3_blob_open() and destroyed with sqlite3_blob_close(). sqlite3 410 | Appendix G: SQLite C API Reference The sqlite3_blob structure is opaque, and an application should never access any of the data fields directly. See Also sqlite3_blob_open(), sqlite3_blob_close() sqlite3_context SQL function context Description The sqlite3_context structure acts as a data container to pass information between the SQLite library and a custom SQL function implementation. The structure is created by the SQLite library and passed into the function callbacks registered with sqlite3_create_function(). The callback functions can extract the database connection or user-data pointer from the context. The context is also used to pass back an SQL result value or error condition. An application should never create or destroy an sqlite3_context structure. The sqlite3_context structure is opaque, and an application should never access any of the data fields directly. See Also sqlite3_create_function() sqlite3_int64, sqlite3_uint64, sqlite_int64, sqlite_uint64 64-bit integer values Description Platform-, processor-, and compiler-independent 64-bit integer types. sqlite3_module Virtual table implementation [EXP] Description The sqlite3_module structure defines a virtual table. The structure contains a number of function pointers that taken together provide the implementation for a virtual table. The code for a virtual table extension typically allocates a static sqlite3_module structure. This structure is initialized with all the proper function pointers, contained within the module, and then passed into sqlite3_create_module(). See Also sqlite3_create_module() sqlite3_module Appendix G: SQLite C API Reference | 411 sqlite3_mutex A mutual exclusion lock Description The sqlite3_mutex structure provides an abstract mutual exclusion lock. An application can create its own locks using the sqlite3_mutex_alloc() call, but it is much more common to reference the lock used by the database connection. This lock can be retrieved with the sqlite3_db_mutex() call. Mutex locks are locked and unlocked using the sqlite3_mutex_ enter() and sqlite3_mutex_leave() calls. The sqlite3_mutex structure is opaque, and an application should never access any of the data fields directly. See Also sqlite3_db_mutex(), sqlite3_mutex_alloc() sqlite3_stmt A prepared statement Description The sqlite3_stmt structure holds a prepared statement. This is all the state and execution information required to build and execute an SQL statement. Statements are used to set any bound parameter values and get any result values. Statements are created with sqlite3_pre pare_xxx() and destroyed with sqlite3_finalize(). Statements are always associated with a specific database connection. The sqlite3_stmt structure is opaque, and an application should never access any of the data fields directly. See Also sqlite3_prepare_xxx(), sqlite3_finalize() sqlite3_value A database value Description The sqlite3_value structure holds a database value. The structure contains the value as well as the type information. A value might hold an integer or floating-point number, a BLOB, a text value in one of many different UTF encodings, or a NULL. Values are used as the pa- rameters to SQL function implementations. They can also be extracted from statement results. Value structures come in two types: protected and unprotected. Protected values can safely undergo standalone type conversion, while unprotected values cannot. SQL function parameters are protected, and can be passed to any form of sqlite3_value_xxx(). Values extracted from sqlite3_column_value() are not unprotected. They can safely be passed to sqlite3_bind_value() or sqlite3_result_value(), but they cannot be passed to sqlite3_ value_xxx(). To extract a native C datatype from a statement, use one of the other sqlite3_column_xxx() functions. sqlite3_mutex 412 | Appendix G: SQLite C API Reference The sqlite3_value structure is opaque, and an application should never access any of the data fields directly. See Also sqlite3_column_xxx(), sqlite3_bind_xxx(), sqlite3_value_xxx(), sqlite3_result_xxx() sqlite3_vfs A virtual file system implementation Description The sqlite3_vfs structure consists of a series of function pointers that make up a Virtual File System (VFS) module. A VFS implementation typically allocates an sqlite3_vfs structure, initializes the various fields, and passes the structure into sqlite3_vfs_register(). See Also sqlite3_vfs_register() API Functions This is a list of nearly every supported function call in the SQLite API. The only func- tions that have been omitted are those related to internal developer testing and those functions that are considered obsolete. Many of the functions return an SQLite result code. The topic of result codes is some- what complex and has a lot of history. For a full discussion, see “Result Codes and Error Codes” on page 146. The basics are pretty easy, however. Nearly every function returns SQLITE_OK if the function succeeded. Most other result codes indicate some type of error, but not all of them. For example, the codes SQLITE_ROW and SQLITE_DONE are used to indicate a specific program state. A result of SQLITE_MISUSE means that the application is using the API incorrectly. This is usually caused by the developer mis- understanding how the API works, or because of a program flow or logic error that causes an unintended call sequence. sqlite3_aggregate_context() Allocate or retrieve aggregate state memory Definition void* sqlite3_aggregate_context( sqlite3_context* ctx, int size ); ctx An SQL function context, provided by the SQLite library. size The memory allocation size, in bytes. Returns A pointer to the aggregate memory allocation. sqlite3_aggregate_context() Appendix G: SQLite C API Reference | 413 Description This function is used by an aggregate function implementation to allocate and retrieve a memory block. The first time it is called, a memory block of the requested size is allocated, all the bytes are set to zero, and the pointer is returned. Subsequent calls for the same function context will return the same block of memory. SQLite will automatically deallocate the mem- ory after the aggregate finalize function is called. This function is typically used to allocate and retrieve memory used to hold all the state information required to calculate an aggregate result. If the data structure needs to be initial- ized, a flag value is typically used. When the memory is first allocated, the flag will be set to zero (like the rest of the structure). On the first call, the structure can be initialized and the flag set to a nonzero value. Both the aggregate step and the finalize functions need to be prepared to initialize the memory. See Also sqlite3_create_function() sqlite3_auto_extension() Register an automatic extension Definition int sqlite3_auto_extension( entry_point ); void entry_point( ); entry_point A function pointer to an extension entry point. Returns An SQLite result code. Description This function registers an extension entry-point function. Once an entry point is registered, any database connection opened by the SQLite library will automatically call the entry point. Multiple entry points can be registered. It is safe to register the same entry point multiple times. In that case, the entry point will only be called once. Although the entry point is given as void entry_point( ), the actual format of an entry-point function is: int entry_point( sqlite3* db, char** error, const sqlite3_api_routines* api ); The discrepancy of the types will likely require a cast or redefinition of the entry-point function. This function can only be used to register extensions with static entry points. Dynamically loaded extensions cannot be directly registered with this function. To clear the list of auto- matic extensions, call sqlite3_reset_auto_extension(). This should be done before calling sqlite3_shutdown(). sqlite3_auto_extension() 414 | Appendix G: SQLite C API Reference For more information on extensions and entry points, see the section “SQLite Exten- sions” on page 204. See Also sqlite3_reset_auto_extension() sqlite3_backup_finish() Complete an online backup [EXP] Definition int sqlite3_backup_finish( sqlite3_backup* backup ); backup An online backup handle. Returns An SQLite result code. A value of SQLITE_OK will be returned even if the backup was not completed. Description This function releases an online backup handle. The backup handle will be released without error, even if sqlite3_backup_step() never returned SQLITE_DONE. See Also sqlite3_backup_init() sqlite3_backup_init() Start an online backup [EXP] Definition sqlite3_backup* sqlite3_backup_init( sqlite3* db_dst, const char* db_name_dst, sqlite3* db_src, const char* db_name_src ); db_dst The destination database connection. db_name_dst The destination logical database name in UTF-8. This can be main, temp, or the name given to ATTACH DATABASE. db_src The source database connection. Must be different from db_dst. db_name_src The source logical database name in UTF-8. Returns An online backup handle. sqlite3_backup_init() Appendix G: SQLite C API Reference | 415 Description This function initiates an online database backup. The online backup APIs can be used to make a low-level copy of a complete database instance without locking the database. The backup APIs can be used to perform live backups, or they can be used to copy a file-backed database to an in-memory database (or vice-versa). The application requires exclusive access to the destination database for the duration of the operation. The source database requires read-only access, but the locks are periodically re- leased to allow other processes to continue to access and modify the source database. To perform an online backup, a backup handle is created with sqlite3_backup_init(). The application continues to call sqlite3_backup_step() to transfer data, generally pausing for a short time between calls. Finally, sqlite3_backup_finish() is called to release the backup handle. For more information on using the online backup APIs, see http://www.sqlite.org/backup.html. See Also sqlite3_backup_finish(), sqlite3_backup_step() sqlite3_backup_pagecount() Get the number of pages in the source database [EXP] Definition int sqlite3_backup_pagecount( sqlite3_backup* backup ); backup An online backup handle. Returns The number of pages in the source database. Description This function gets the total number of pages in the source database. This value is updated when sqlite3_backup_step() is called, and might not reflect recent activity on the source database. See Also sqlite3_backup_remaining(), sqlite3_backup_step() sqlite3_backup_remaining() Get the number of pages remaining in a backup [EXP] Definition int sqlite3_backup_remaining( sqlite3_backup* backup ); backup An online backup handle. sqlite3_backup_pagecount() 416 | Appendix G: SQLite C API Reference Returns The number of pages remaining in the backup process. Description This function is used to get the number of pages that still need to be backed up. This value is updated when sqlite3_backup_step() is called and may not reflect recent activity on the source database. See Also sqlite3_backup_pagecount(), sqlite3_backup_step() sqlite3_backup_step() Continue an online backup [EXP] Definition int sqlite3_backup_step( sqlite3_backup* backup, int pages ); backup An online backup handle. pages The number of pages to copy. If this value is negative, all remaining pages are copied. Returns An SQLite result code. SQLITE_OK indicates the pages were successfully copied, but more pages remain. SQLITE_DONE indicates that a complete backup was made. If sqlite3_ backup_step() returns SQLITE_BUSY or SQLITE_LOCKED, the function can be safely retried at a later time. Description This function attempts to copy the specified number of pages from the source database to the destination database. A shared read lock is obtained and then released on the source database for each call to sqlite3_backup_step(). Larger page counts finish the backup more quickly, while smaller page counts allow more concurrent access. Modifications made to the source database through the same database connection that is being used for the backup are automatically reflected into the destination database, allowing the online backup to continue. Modifications made to the source database through any other database connection will cause the backup to reset and start over. This is transparent. If an application is attempting to back up a highly dynamic database it should call sqlite3_backup_step() very frequently and use a very large page count, or it risks continual reset. This could result in the backup perpetually failing to complete. See Also sqlite3_backup_init() sqlite3_backup_step() Appendix G: SQLite C API Reference | 417 sqlite3_bind_xxx() Bind values to statement parameters Definition int sqlite3_bind_blob( sqlite3_stmt* stmt, int pidx, const void* val, int bytes, mem_callback ); int sqlite3_bind_double( sqlite3_stmt* stmt, int pidx, double val ); int sqlite3_bind_int( sqlite3_stmt* stmt, int pidx, int val ); int sqlite3_bind_int64( sqlite3_stmt* stmt, int pidx, sqlite3_int64 val ); int sqlite3_bind_null( sqlite3_stmt* stmt, int pidx ); int sqlite3_bind_text( sqlite3_stmt* stmt, int pidx, const char* val, int bytes, mem_callback ); int sqlite3_bind_text16( sqlite3_stmt* stmt, int pidx, const void* val, int bytes, mem_callback ); int sqlite3_bind_value( sqlite3_stmt* stmt, int pidx, const sqlite3_value* val ); int sqlite3_bind_zeroblob( sqlite3_stmt* stmt, int pidx, int bytes ); void mem_callback( void* ptr ); stmt A prepared statement that contains parameter values. pidx The parameter index. The first parameter has an index of one (1). val The data value to bind bytes The size of the data value, in bytes (not characters). Normally, the length does not include any null terminator. If val is a null-terminated string, and this value is negative, the length will be automatically computed. mem_callback An function pointer to a memory deallocation function. This function frees the memory buffer used to hold the value. If the buffer was allocated with sqlite3_malloc(), a refer- ence to sqlite3_free() can be passed directly. The special flags SQLITE_STATIC and SQLITE_TRANSIENT can also be used. SQLITE_STATIC indicates that the application will keep value memory valid until the statement is finalized (or a new value is bound). SQLITE_TRANSIENT will cause SQLite to make an internal copy of the value buffer that will be automatically freed when it is no longer needed. Returns (sqlite3_bind_xxx()) An SQLite response code. The code SQLITE_RANGE will be returned if the parameter index is invalid. Description This family of functions is used to bind a data value to a statement parameter. Bound values remain in place until the statement is finalized (via sqlite3_finalize()) or a new value is bound to the same parameter index. Resetting the statement via sqlite3_reset() does not clear the bindings. sqlite3_bind_xxx() 418 | Appendix G: SQLite C API Reference The sqlite3_bind_zeroblob() function binds a BLOB object of the given length. All the bytes of the BLOB are set to zero. The BLOB is not actually instanced in memory, allowing the system to bind very large BLOB values. These BLOBs can be modified using the sqlite3_blob_xxx() functions. For more information on how to include statement parameters in prepared SQL statements, see “Bound Parameters” on page 133. See Also sqlite3_column_xxx(), sqlite3_result_xxx(), sqlite3_value_xxx() sqlite3_bind_parameter_count() Get the number of statement parameters Definition int sqlite3_bind_parameter_count( sqlite3_stmt* stmt ); stmt A prepared statement that contains parameter values. Returns The number of valid parameters in this statement. Description This function returns the largest valid statement parameter index in the given statement. If this function returns 6, valid parameter indexes include 1 through 6, inclusive. See Also sqlite3_bind_xxx() sqlite3_bind_parameter_index() Get the index of a named statement parameter Definition int sqlite3_bind_parameter_index( sqlite3_stmt* stmt, const char *name ); stmt A prepared statement that contains parameter values. name The parameter name, including prefix character. The name must be given in UTF-8. Returns The index of the named parameter. If the name is not found, zero will be returned. Description This function finds the index value of a named statement parameter. The returned value can be safely passed to an sqlite3_bind_xxx() function (although those functions may return SQLITE_RANGE if the name is not found). sqlite3_bind_parameter_index() Appendix G: SQLite C API Reference | 419 See Also sqlite3_bind_xxx(), sqlite3_bind_parameter_name() sqlite3_bind_parameter_name() Get the name of a statement parameter Definition const char* sqlite3_bind_parameter_name( sqlite3_stmt* stmt, int pidx ); stmt A prepared statement that contains parameter values. pidx The parameter index. The first parameter has an index of one (1). Returns The text representation of the statement parameter with the given index. Description This function looks up the original string representation of a statement parameter. This is most frequently used with named parameters, but works correctly for any explicit parameter type. The string will be UTF-8 encoded, and will include the prefix character (e.g., the pa- rameter :val will return :val, not val). The return value will be NULL if the parameter index is invalid, or if the parameter is an automatic parameter (a bare ?). See Also sqlite3_bind_xxx(), sqlite3_bind_parameter_index() sqlite3_blob_bytes() Get the size of a BLOB handle value Definition int sqlite3_blob_bytes( sqlite3_blob* blob ); blob A BLOB handle acquired from sqlite3_blob_open(). Returns The size of the BLOB value, in bytes. Description This function returns the size of the referenced BLOB value in bytes. The size of a BLOB value is fixed and cannot be modified without a call to INSERT or UPDATE. See Also sqlite3_blob_open(), sqlite3_bind_xxx() (refer to sqlite3_bind_zeroblob(), specifically), zeroblob() [Ap E] sqlite3_bind_parameter_name() 420 | Appendix G: SQLite C API Reference sqlite3_blob_close() Close a BLOB handle Definition int sqlite3_blob_close( sqlite3_blob* blob ); blob A BLOB handle acquired from sqlite3_blob_open(). It is safe to pass a NULL value. Returns An SQLite response code. Description This function closes and releases a BLOB handle acquired from sqlite3_blob_open(). The BLOB handle is always closed (and becomes invalid), even if an error code is returned See Also sqlite3_blob_open() sqlite3_blob_open() Create and open a BLOB handle Definition int sqlite3_blob_open( sqlite3* db, const char* db_name, const char* tbl_name, const char* col_name, sqlite3_int64 row_id, int flags, sqlite3_blob** blob ); db A database connection. db_name A logical database name in UTF-8. This can be main, temp, or a name given to ATTACH DATABASE. tbl_name A table name in UTF-8. col_name A column name in UTF-8 row_id A ROWID value. flags A nonzero value will open the BLOB read/write. A zero value will open the BLOB read-only. blob A reference to a BLOB handle. The new BLOB handle will be returned via this reference. The BLOB handle may be set to NULL if an error is returned. Returns An SQLite response code. sqlite3_blob_open() Appendix G: SQLite C API Reference | 421 Description This function creates a new BLOB handle used for incremental BLOB I/O. The parameters need to describe the BLOB referenced with the SQL statement: SELECT col_name FROM db_name.tbl_name WHERE ROWID = row_id; The BLOB handle remains valid until it is closed, or until it expires. A BLOB handle expires when any column of the row that contains the BLOB is modified in any way (typically by an UPDATE or DELETE). An expired BLOB handle must still be closed. When foreign key constraints are enabled, BLOB values contained in columns that are part of a foreign key can only be opened read-only. See Also sqlite3_blob_read(), sqlite3_blob_write(), sqlite3_blob_close() sqlite3_blob_read() Read data from a BLOB Definition int sqlite3_blob_read( sqlite3_blob* blob, void* buff, int bytes, int offset ); blob A BLOB handle acquired from sqlite3_blob_open(). buff A data buffer. Data is read from the BLOB into the buffer. bytes The number of bytes to read from the BLOB into the buffer. offset Offset from beginning of BLOB where read should start. Returns An SQLite return code. Attempting to read from an expired BLOB handle will return SQLITE_ABORT. Description This function reads the specified number of bytes from a BLOB value into the given buffer. The read will start from the provided offset. Any attempt to read past the end of the BLOB results in an error, meaning that the BLOB must have bytes + offset or more bytes. See Also sqlite3_blob_bytes(), sqlite3_blob_open() sqlite3_blob_write() Write data to a BLOB Definition int sqlite3_blob_write( sqlite3_blob* blob, void* buff, int bytes, int offset ); sqlite3_blob_read() 422 | Appendix G: SQLite C API Reference blob A BLOB handle acquired from sqlite3_blob_open(). buff A data buffer. Data is written from the buffer into the BLOB. bytes The number of bytes to write from the buffer into the BLOB. offset Offset from beginning of BLOB where write should start. Returns An SQLite return code. Attempting to write to an expired BLOB handle will return SQLITE_ABORT. Description This function writes the specified number of bytes from the given buffer into a BLOB value. The write starts at the provided offset. Any attempt to write past the end of the BLOB results in an error, meaning that the BLOB must have bytes + offset or more bytes. See Also sqlite3_blob_bytes(), sqlite3_blob_open() sqlite3_busy_handler() Register a busy handler Definition int sqlite3_busy_handler( sqlite3* db, busy_handler, void* udp ); int busy_handler( void* udp, int count ); db A database connection. busy_handler A function pointer to an application busy handler function. udp An application-defined user-data pointer. This value is made available to the busy handler. count The number of times the handler has been called for this lock. Returns (sqlite3_busy_handler()) An SQLite result code. Returns (busy_handler()) A nonzero return code indicates that the connection should continue to wait for the desired lock. A return code of zero indicates that the database connection should give up and return SQLITE_BUSY or SQLITE_IOERR_BLOCKED to the application. sqlite3_busy_handler() Appendix G: SQLite C API Reference | 423 Description This function registers a busy handler with a specific database connection. The busy handler is called any time the database connection encounters a locked database file. In most cases, the application can simply wait for the lock to be released before proceeding. In these situa- tions, the SQLite library will keep calling the busy handler, which can decide to keep waiting, or to give up and return an error to the application. For a full discussion of locking states and busy handlers, see the section “Database Lock- ing” on page 151. Each database connection has only one busy handler. Registering a new busy handler will replace the old one. To remove a busy handler, pass in a NULL function pointer. Setting a busy timeout with sqlite3_busy_timeout() will also reset the busy handler. See Also sqlite3_busy_timeout() sqlite3_busy_timeout() Set a busy timeout Definition int sqlite3_busy_timeout( sqlite3* db, int ms ); db A database connection. ms The total timeout duration, in milliseconds (thousandths of a second). Returns An SQLite result code. Description This function registers an internal busy handler that keeps attempting to acquire a busy lock until the total specified time has passed. Because this function registers an internal busy han- dler, any current busy handler is removed. The timeout value can be explicitly removed by setting a timeout value of zero. See Also sqlite3_busy_handler() sqlite3_changes() Get the number of changes made by an SQL statement Definition int sqlite3_changes( sqlite3* db ); db A database connection. sqlite3_busy_timeout() 424 | Appendix G: SQLite C API Reference Returns The number of rows modified by the last SQL statement that was executed. Description This function returns the number of rows that were modified by the most recent INSERT, UPDATE, or DELETE statement. Only rows directly affected by the table named in the SQL com- mand are counted. Changes made by triggers, foreign key actions, or conflict resolutions are not counted. If called from within a trigger, the count only includes modifications made at this trigger level. It will not include changes made by an enclosing scope, nor will it include changes made by subsequent trigger or foreign key actions. This function is exposed to the SQL environment as the SQL function changes(). See Also sqlite3_total_changes(), count_changes [PRAGMA, Ap F], changes() [SQL Func, Ap E] sqlite3_clear_bindings() Reset all statement parameters to NULL Definition int sqlite3_clear_bindings( sqlite3_stmt* stmt ); stmt A prepared statement. Returns An SQLite result code. Description This function resets all the statement parameters to NULL. It is equivalent to calling sqlite3_bind_null() on every valid parameter index. Many applications call this function any time sqlite3_reset() is called, to prevent parameter values from leaking from one execution to the next. See Also sqlite3_bind_xxx(), sqlite3_reset() sqlite3_close() Close a database connection Definition int sqlite3_close( sqlite3* db ); db A database connection. A NULL may be safely passed to this function (resulting in a no-op). sqlite3_close() Appendix G: SQLite C API Reference | 425 Returns An SQLite result code. Description This function closes a database connection and releases all the resources associated with it. The application is responsible for finalizing all prepared statements and closing all BLOB handles. If the database connection is currently in a transaction, it will be rolled back before the database(s) are closed. All attached databases will automatically be detached. If an application fails to finalize a prepared statement, the close will fail and SQLITE_BUSY will be returned. The statement must be finalized and sqlite3_close() must be called again. See Also sqlite3_open(), sqlite3_finalize(), sqlite3_close(), sqlite3_next_stmt() sqlite3_collation_needed() Register a collation loader Definition int sqlite3_collation_needed( sqlite3* db, void* udp, col_callback ); int sqlite3_collation_needed16( sqlite3* db, void* udp, col_callback16 ); void col_callback( void* udp, sqlite3* db, int text_rep, const char* name ); void col_callback16( void* udp, sqlite3* db, int text_rep, const void* name ); db A database connection. udp An application-defined user-data pointer. This value is made available to the collation loader callback. col_callback, col_callback16 Function pointer to an application-defined collation loader function. text_rep The desired text representation for the requested collation. This value can be one of SQLITE_UTF8, SQLITE_UTF16BE, or SQLITE_UTF16LE. name The collation name in UTF-8 or UTF-16 (native order). Returns (sqlite3_collation_needed[16]()) An SQLite result code. Description These functions register a collation loader callback function. Any time SQLite is processing an SQL statement that requires an unknown collation, the collation loader callback is called. This provides the application with an opportunity to register the required collation. If the callback is unable to register the requested collation, it should simply return. sqlite3_collation_needed() 426 | Appendix G: SQLite C API Reference Although the callback is given a desired text representation for the requested collation, the callback is under no obligation to provide a collation that uses that specific representation. As long as a collation with the proper name is provided, SQLite will perform whatever trans- lations are required. This function is most commonly used by applications that provide a very large number of collations. Rather than registering dozens, or even hundreds, of collations with each database connection, the callback allows the collations to be loaded on demand. See Also sqlite3_create_collation() sqlite3_column_xxx() Get a results column value Definition const void* sqlite3_column_blob( sqlite3_stmt* stmt, int cidx ); double sqlite3_column_double( sqlite3_stmt* stmt, int cidx ); int sqlite3_column_int( sqlite3_stmt* stmt, int cidx ); sqlite3_int64 sqlite3_column_int64( sqlite3_stmt* stmt, int cidx ); const unsigned char* sqlite3_column_text( sqlite3_stmt* stmt, int cidx ); const void* sqlite3_column_text16( sqlite3_stmt* stmt, int cidx ); sqlite3_value* sqlite3_column_value( sqlite3_stmt* stmt, int cidx ); stmt A prepared and executed statement. cidx A column index. The first column has an index of zero (0). Returns The requested value. Description These functions extract values from a prepared statement. Values are available any time sqlite3_step() returns SQLITE_ROW. If the requested type is different than the actual underlying value, the value will be converted using the conversion rules defined by Table 7-1. SQLite will take care of all memory management for the buffers returned by these functions. Pointers returned may become invalid at the next call to sqlite3_step(), sqlite3_reset(), sqlite3_finalize(), or any sqlite3_column_xxx() call on the same column index. Pointers can also become invalid because of a call to one of the sqlite3_column_bytes() functions. Be warned that sqlite3_column_int() will clip any integer values to 32 bits. If the database contains values that cannot be represented by a 32-bit signed integer, it is safer to use sqlite3_column_int64(). The buffer returned by sqlite3_column_text() and sqlite3_col umn_text16() will always be null-terminated. The structure returned by sqlite3_column_value() is an unprotected sqlite3_value object. This object can only be used in conjunction with sqlite3_bind_value() or sqlite3_result_ value(). Calling any of the sqlite3_value_xxx() functions will result in undefined behavior. sqlite3_column_xxx() Appendix G: SQLite C API Reference | 427 See Also sqlite3_column_count() [C API, Ap G], sqlite3_column_bytes() [C API, Ap G], sqlite3_col- umn_type() [C API, Ap G], sqlite3_value_xxx() [C API, Ap G] sqlite3_column_bytes() Get the size of a column buffer Definition int sqlite3_column_bytes( sqlite3_stmt* stmt, int cidx ); int sqlite3_column_bytes16( sqlite3_stmt* stmt, int cidx ); stmt A prepared and executed statement. cidx A column index. The first column has an index of zero (0). Returns The number of bytes in the column value. Description These functions return the number of bytes in a text or BLOB value. Calling these functions can cause a type conversion (invalidating buffers returned by sqlite3_column_xxx()), so care must be taken to call them in conjunction with the appropriate sqlite3_column_xxx() function. To avoid problems, an application should first extract the desired type using a sqlite3_col umn_xxx() function, and then call the appropriate sqlite3_column_bytes() function. The functions sqlite3_column_text() and sqlite3_column_blob() should be followed by a call to sqlite3_column_bytes(), while any call to sqlite3_column_text16() should be followed by a call to sqlite3_column_bytes16(). If these functions are called on a non-text or non-BLOB value, the value will first be converted to an appropriately encoded text value, then the length of that text value will be returned. See Also sqlite3_column_xxx() sqlite3_column_count() Get the number of result columns in a statement Definition int sqlite3_column_count( sqlite3_stmt* stmt ); stmt A prepared statement. Returns The number of columns returned by a prepared statement. sqlite3_column_bytes() 428 | Appendix G: SQLite C API Reference Description This function returns the number of columns available in a row result. Column indexes start at zero (0), so if this function returns 5, value column indexes are 0 to 4. The value zero (0) is returned if the statement does not return any result. See Also sqlite3_column_xxx(), sqlite3_step() sqlite3_column_database_name() Get the source database name of a result column Definition const char* sqlite3_column_database_name( sqlite3_stmt* stmt, int cidx ); const void* sqlite3_column_database_name16( sqlite3_stmt* stmt, int cidx ); stmt A prepared statement. cidx A column index. The first column has an index of zero (0). Returns The logical name of the source database for the given result column. Description These functions return the unaliased logical name of the source database that is associated with a SELECT result column. Returned pointers will remain valid until sqlite3_finalize() is called on the statement, or until one of these functions is called with the same column index. SQLite will take care of all memory management for the buffers returned by these functions. Data is only available for result columns that are derived directly from a column reference. If a result column is defined as an expression or subquery, a NULL will be returned. These functions are only available if the SQLite library was compiled with the SQLITE_ENA BLE_COLUMN_METADATA build option. See Also sqlite3_column_table_name(), sqlite3_column_origin_name(), SQLITE_ENABLE_COLUMN_META- DATA [Build Opt, Ap A] sqlite3_column_decltype() Get the declared type of a result column Definition const char* sqlite3_column_decltype( sqlite3_stmt* stmt, int cidx ); const void* sqlite3_column_decltype16( sqlite3_stmt* stmt, int cidx ); stmt A prepared statement. sqlite3_column_decltype() Appendix G: SQLite C API Reference | 429 cidx A column index. The first column has an index of zero (0). Returns The defined type name for the given result column. Description These functions return the declared type associated with a source column. This is the type that was given in the CREATE TABLE command used to define the source table column. Returned pointers will remain valid until sqlite3_finalize() is called on the statement, or until one of these functions is called with the same column index. SQLite will take care of all memory management for the buffers returned by these functions. Data is only available for result columns that are derived directly from a column reference. If a result column is defined as an expression or subquery, a NULL will be returned. See Also sqlite3_column_type(), sqlite3_column_table_name() sqlite3_column_name() Get the name of a result column Definition const char* sqlite3_column_name( sqlite3_stmt* stmt, int cidx ); const void* sqlite3_column_name16( sqlite3_stmt* stmt, int cidx ); stmt A prepared statement. cidx A column index. The first column has an index of zero (0). Returns The name of a result column. Description These functions return the name of a result column. This is the name provided by the AS clause of a select header. If no AS clause was provided, the name will be derived from the expression used to define the SELECT header. The format of derived names is not defined by the SQL standard and may change from one release of SQLite to another. If an application depends on explicit column names, it should always use AS clauses in the select header. Returned pointers will remain valid until sqlite3_finalize() is called on the statement, or until one of these functions is called with the same column index. SQLite will take care of all memory management for the buffers returned by these functions. See Also SELECT [SQL Cmd, Ap C], full_column_names [PRAGMA, Ap F], short_column_names [PRAGMA, Ap F] sqlite3_column_name() 430 | Appendix G: SQLite C API Reference sqlite3_column_origin_name() Get the source column name of a result column Definition const char* sqlite3_column_origin_name( sqlite3_stmt* stmt, int cidx ); const void* sqlite3_column_origin_name16( sqlite3_stmt* stmt, int cidx ); stmt A prepared statement. cidx A column index. The first column has an index of zero (0). Returns The name of the source column for the given result column. Description These functions return the name of the source column that is associated with a SELECT result column. Returned pointers will remain valid until sqlite3_finalize() is called on the state- ment, or until one of these functions is called with the same column index. SQLite will take care of all memory management for the buffers returned by these functions. Data is only available for result columns that are derived directly from a column reference. If a result column is defined as an expression or subquery, a NULL will be returned. These functions are only available if the SQLite library was compiled with the SQLITE_ENA BLE_COLUMN_METADATA build option. See Also sqlite3_column_name() [C API, Ap G], sqlite3_column_table_name() [C API, Ap G], sqlite3_column_database_name() [C API, Ap G], SQLITE_ENABLE_COLUMN_METADATA [Build Opt, Ap A] sqlite3_column_table_name() Get the source table name of a result column Definition const char* sqlite3_column_table_name( sqlite3_stmt* stmt, int cidx ); const void* sqlite3_column_table_name16( sqlite3_stmt* stmt, int cidx ); stmt A prepared statement. cidx A column index. The first column has an index of zero (0). Returns The name of the source table for the given result column. Description These functions return the name of the source table that is associated with a SELECT result column. Returned pointers will remain valid until sqlite3_finalize() is called on the sqlite3_column_table_name() Appendix G: SQLite C API Reference | 431 statement, or until one of these functions is called with the same column index. SQLite will take care of all memory management for the buffers returned by these functions. Data is only available for result columns that are derived directly from a column reference. If a result column is defined as an expression or subquery, a NULL will be returned. These functions are only available if the SQLite library was compiled with the SQLITE_ENA BLE_COLUMN_METADATA build option. See Also sqlite3_column_database_name() [C API, Ap G], sqlite3_column_origin_name() [C API, Ap G], SQLITE_ENABLE_COLUMN_METADATA [Build Opt, Ap A] sqlite3_column_type() Get the datatype of a result column Definition int sqlite3_column_type( sqlite3_stmt* stmt, int cidx ); stmt A prepared and executed statement. cidx A column index. The first column has an index of zero (0). Returns The native datatype code of a result value. Description This function returns the initial datatype of the value in a result column. For a given column, this value may change from one result row to the next. If this function is used, it should be called before any sqlite3_column_xxx() function. Once a type conversion takes place, the result of this function is undefined. The return value will be SQLITE_INTEGER, SQLITE_FLOAT, SQLITE_TEXT, SQLITE_BLOB, or SQLITE_NULL. See Also sqlite3_column_xxx(), sqlite3_step() sqlite3_commit_hook() Register a commit callback Definition void* sqlite3_commit_hook( sqlite3* db, commit_callback, void* udp ); int commit_callback( void* udp ); db A database connection. sqlite3_column_type() 432 | Appendix G: SQLite C API Reference commit_callback Function pointer to an application-defined commit callback function. udp An application-defined user-data pointer. This value is made available to the commit callback. Returns (sqlite3_commit_hook()) The previous user-data pointer, if applicable. Returns (commit_callback()) If nonzero, the commit is converted into a callback. Description This function registers a commit callback. This callback function is called any time the data- base performs a commit (including an autocommit). Each database connection can have only one commit callback. Registering a new commit callback will overwrite any previously reg- istered callback. To remove the commit callback, set a NULL function pointer. A commit callback must not use the associated database connection to modify any databases, nor may it call sqlite3_prepare_v2() or sqlite3_step(). If the commit callback returns a nonzero value, the commit will be canceled and the transaction will be rolled back. See Also sqlite3_rollback_hook(), sqlite3_update_hook() sqlite3_compileoption_get() Iterate over defined compile options Definition const char* sqlite3_compileoption_get( int iter ); iter An iterator index value. Returns A UTF-8 encoded string with the option name and optional value. Description This function iterates over all the build options used to compile this instance of the SQLite library. By starting with an iterator value of zero (0) and calling this function over and over with an ever-increasing iterator value, all of the defined values will be returned. If the iterator value is out of range, a NULL pointer will be returned. Only build options that were actually given will be included. Those that assumed their default values will not be given. Generally, only ENABLE, DISABLE, OMIT, and a few other directives are included in the list. DEFAULT and MAX directives that control limits or default values are not included. Most of these values can be queried with sqlite3_limit(). sqlite3_compileoption_get() Appendix G: SQLite C API Reference | 433 If the build option is a simple Boolean, only the option name will be included. If the option requires an actual value, the option will be given as name=value. In both cases, the SQLITE_ prefix will not be included. This function is exposed to the SQL environment as the SQL function sqlite_compile option_get(). Note that the SQL function has no 3 in the name. See Also sqlite3_compileoption_used(), sqlite3_compileoption_get(), sqlite_compileoption_get() [SQL Func, Ap E] sqlite3_compileoption_used() Get a compile option value Definition int sqlite3_compileoption_used( const char* name ); name A UTF-8 encoded string with the option name and, optionally, the value. Returns A Boolean that indicates if the specified compile option and, optionally, the value was used. Description This function queries if a specific compile option (and, possibly, value) was used to compile this instance of the SQLite library. This function is aware of the same build options that sqlite3_compileoption_get() returns. The function will work with or without the SQLITE_ prefix. In the case of a Boolean build option, the name should just be the build option itself. The return value will indicate if the option was used or not. In the case of build options that require an actual value, the name can either include just the name, or it can include a name-value pair in the format name=value. If just the name is given, the return value will indicate if the build option was used with any value. If the value is given, the return value will indicate if the build option was set to that specific value. This function is exposed to the SQL environment as the SQL function sqlite_compile option_used(). Note that the SQL function has no 3 in the name. See Also sqlite3_compileoption_get(), sqlite_compileoption_used() [SQL Func, Ap E] sqlite3_complete() Determine if an SQL statement is complete Definition int sqlite3_complete( const char* sql ); int sqlite3_complete16( const void* sql ); sqlite3_compileoption_used() 434 | Appendix G: SQLite C API Reference sql An SQL statement. Returns Zero (0) if the statement is incomplete, SQLITE_NOMEM if a memory allocation failed, or a nonzero value to indicate the statement is complete. Description This function determines if an SQL statement is complete. This function will return a nonzero result if the given string ends in a semicolon that is not part of an identifier, string literal, or CREATE TRIGGER statement. Although the statement is scanned, it is not fully parsed and cannot detect syntactically incorrect statements. The function sqlite3_initialize() should be called before using this function. If the SQLite library has not been initialized, this function will do so automatically. This may result in additional nonzero result codes if the initialization fails. See Also sqlite3_prepare_xxx() sqlite3_config() Advanced configuration of the SQLite library Definition int sqlite3_config( int option, ... ); option The configuration option to change. Additional parameters Additional parameters are determined by the given configuration option. Returns An SQLite result code. Description This function configures the SQLite library. It must be called before the SQLite library finishes its initialization process (either before sqlite3_initialize() returns, or after sqlite3_shut down() is called). The number of additional parameters and their types is determined by the specific configuration option. In addition to controlling the threading mode, this function can also be used to control many different aspects of SQLite’s memory usage. Setting these configuration parameters is an ad- vanced feature that is not required by the vast majority of applications. For more information on the currently supported configuration options, see http://www.sqlite .org/c3ref/c_config_getmalloc.html. See Also sqlite3_initialize(), sqlite3_db_config(), sqlite3_limit() sqlite3_config() Appendix G: SQLite C API Reference | 435 sqlite3_context_db_handle() Get a database connection from a function context Definition sqlite3* sqlite3_context_db_handle( sqlite3_context* ctx ); ctx An SQL function context, provided by the SQLite library. Returns The database connection associated with this context. Description This function returns the database connection that is associated with an SQL function context. This is called from inside a C function that implements a custom SQL function or SQL aggregate. See Also sqlite3_create_function() sqlite3_create_collation() Register a custom collation implementation Definition int sqlite3_create_collation( sqlite3* db, const char* name, int text_rep, void* udp, comp_func ); int sqlite3_create_collation16( sqlite3* db, const void* name, int text_rep, void* udp, comp_func ); int sqlite3_create_collation_v2( sqlite3* db, const char* name, int text_rep, void* udp, comp_func, dest_func ); int comp_func( void* udp, int sizeA, const void* textA, int sizeB, const void* textB ); void dest_func( void* udp ); db A database connection. name The name of the collation in UTF-8 or UTF-16, depending on the function used. text_rep The text representation expected by the comparison function. This value can be one of SQLITE_UTF8, SQLITE_UTF16 (native order), SQLITE_UTF16BE, SQLITE_UTF16LE, or SQLITE_ UTF16_ALIGNED (native order, 16-bit aligned). udp An application-defined user-data pointer. This value is made available to both the col- lation comparison function and the collation destroy function. sqlite3_context_db_handle() 436 | Appendix G: SQLite C API Reference comp_func A function pointer to the application-defined custom collation comparison function. dest_func An optional function pointer to the application-defined collation destroy function. This may be NULL. sizeA, sizeB The length of the textA and textB parameters, respectively, in bytes. textA, textB Data buffers containing the text values in the requested representation. These may not be null-terminated. Returns (sqlite3_create_collation[16][_v2]()) An SQLite result code. Returns (comp_func()) The results of the comparison. Negative for A < B, zero for A = B, positive for A > B. Description These functions define a custom collation implementation. This is done by registering a com- parison function under the given collation name and expected text representation. The only difference between sqlite3_create_collation() and sqlite3_create_colla tion16() is the text encoding used to define the collation name (the second parameter). The encoding used when calling the comparison function is determined by the third parameter. Either function can be used to register a comparison function that understands any of the given encodings. The only difference between sqlite3_create_collation() and sqlite3_create_colla tion_v2() is the addition of the destroy callback function. A collation name can be overloaded by registering multiple comparison functions under dif- ferent expected text representations. To delete a collation, register a NULL comparison func- tion pointer under the same collation name and text representation. The comparison function should return a negative value if testA is less than textB, a zero value if testA and textB are equal, or a positive value if testA is greater than textB. Conceptually, the return value represents A minus B. The destroy function is called any time a collation is overwritten or when the database con- nection is shut down. The destroy function will be passed the user-data pointer and given a chance to deallocate any relevant memory. If applicable, a reference to sqlite3_free() can be passed directly to sqlite3_create_collation_v2(). For more information on writing custom collations, see “Collation Functions” on page 200. See Also sqlite3_create_function(), sqlite3_collation_needed() sqlite3_create_collation() Appendix G: SQLite C API Reference | 437 sqlite3_create_function() Define a scalar or aggregate SQL function Definition int sqlite3_create_function( sqlite3* db, const char* name, int num_param, int text_rep, void* udp, func_func, step_func, final_func ); int sqlite3_create_function16( sqlite3* db, const void* name, int num_param, int text_rep, void* udp, func_func, step_func, final_func ); void func_func( sqlite3_context* ctx, int argc, sqlite3_value** argv ); void step_func( sqlite3_context* ctx, int argc, sqlite3_value** argv ); void final_func( sqlite3_context* ctx ); db A database connection. name The name of the collation in UTF-8 or UTF-16, depending on the function used. num_param The number of expected parameters in the SQL function. If the value is ‒1, any number of parameters will be accepted. text_rep The text representation best suited to the function(s). This value can be one of SQLITE_UTF8, SQLITE_UTF16 (native order), SQLITE_UTF16BE, SQLITE_UTF16LE, or SQLITE_ANY. udp An application-defined user-data pointer. This value can be extracted from the ctx parameter using the sqlite3_user_data() function. func_func A function pointer to an application-defined scalar SQL function implementation. If this is non-NULL, the step_func and final_func parameters must be NULL. step_func A function pointer to an application-defined aggregate step function implementation. If this is non-NULL, the func_func parameter must be NULL and the final_func parameter must be non-NULL. final_func A function pointer to an application-defined aggregate finalize function implementation. If this is non-NULL, the func_func parameter must be NULL and the step_func parameter must be non-NULL. ctx An SQL function context, provided by the SQLite library. argc The number of parameters given to the SQL function. sqlite3_create_function() 438 | Appendix G: SQLite C API Reference argv The parameter values passed into the SQL function. Returns (sqlite3_create_function[16]()) An SQLite result code. Description These functions are used to define custom SQL scalar functions or SQL aggregate functions. This is done by registering C function pointers that implement the desired SQL function. The only difference between sqlite3_create_function() and sqlite3_create_function16() is the text encoding used to define the function name (the second parameter). The encoding used when calling the registered functions is determined by the fourth parameter. Either function can be used to register functions that understands any of the given encodings. A single call to one of these functions can be used to define either an SQL scalar function or an SQL aggregate function, but not both. A scalar function is defined by providing a valid func_func parameter and setting step_func and final_func to NULL. Conversely, an aggregate function is defined by providing a valid step_func and final_func while setting func_func to NULL. By providing different values for the num_param or text_rep (or both) parameters, different functions can be registered under the same SQL function name. SQLite will choose the closest fit, first by parameter number and then by text representation. An explicit parameter number is considered a better fit than a variable length function. Text representations are judged by those that require the least amount of conversion. Both scalar and aggregate functions can be defined under the same name, assuming they accept a different number of parameters. SQLite uses this functionality internally to define both scalar and aggregate max() and min() functions. A function can be redefined by registering a new set of function pointers under the same name, parameter number, and text representation. This functionality can be used to override existing functions (including built-in functions). To unregister a function, simply pass in all NULL function pointers. Finally, there is one significant limitation to sqlite3_create_function(). Although new func- tions can be defined at any time, it is not legal to redefine or delete a function when there are active statements. Because existing statements may contain references to the existing SQL functions, all prepared statements must be invalidated when redefining or deleting a function. If there are active statements in the middle of executing, this is not possible, and will cause sqlite3_create_function() to fail. For more information on how to write custom SQL scalar and aggregate functions, see Chapter 9. See Also sqlite3_create_collation() [C API, Ap G], sqlite3_user_data() [C API, Ap G], sqlite3_con- text_db_handle() [C API, Ap G], sqlite3_value_xxx() [C API, Ap G], sqlite3_re- sult_xxx() [C API, Ap G] sqlite3_create_function() Appendix G: SQLite C API Reference | 439 sqlite3_create_module() Register a virtual table implementation [EXP] Definition int sqlite3_create_module( sqlite3* db, const char* name, const sqlite3_module* mod_struct, void* udp ); int sqlite3_create_module_v2( sqlite3* db, const char* name, const sqlite3_module* mod_struct, void* udp, dest_func ); void dest_func( void* udp ); db A database connection. name The name of the module (the virtual table type), in UTF-8. mod_struct A module structure, which includes function pointers to all of the application-defined functions required to implement a virtual table module. udp An application-defined user-data pointer. This value is made available to some of the virtual table functions, as well as the destructor function. dest_func An optional function pointer to the module destroy function. This may be NULL. Returns (sqlite3_create_module[_v2]()) An SQLite result code. Description These functions register a virtual table module. The only difference between the standard and _v2 version is the addition of the destroy function, which can be used to free any memory allocations associated with the user-data pointer. For more information on writing a virtual table module, see Chapter 10. See Also sqlite3_declare_vtab() sqlite3_data_count() Get the number of valid result columns in a statement Definition int sqlite3_data_count( sqlite3_stmt* stmt ); stmt A prepared statement. Returns The number of columns available in a prepared statement. sqlite3_create_module() 440 | Appendix G: SQLite C API Reference Description This function returns the number of columns available in a row result. Column indexes start at zero (0), so if this function returns 5, value column indexes are 0 to 4. The value zero (0) will be returned if there is no current result row. This function is almost identical to sqlite3_column_count(). The main difference is that this function will return zero (0) if no valid result row is currently available (for example, before sqlite3_step() is called), while sqlite3_column_count() will always return the number of expected columns, even if no row is currently available. See Also sqlite3_column_count() sqlite3_db_config() Advanced configuration of a database connection [EXP] Definition int sqlite3_db_config( sqlite3* db, int option, ... ); db A database connection. option The configuration option to change. Additional Parameters Additional parameters are determined by the given configuration option. Returns An SQLite result code. Description This function configures database connections. It must be called immediately following a call to one of the sqlite3_open_xxx() functions. For more information on the currently supported configuration options, see http://www.sqlite .org/c3ref/c_dbconfig_lookaside.html. See Also sqlite3_config(), sqlite3_limit() sqlite3_db_handle() Get database connection from statement Definition sqlite3* sqlite3_db_handle( sqlite3_stmt* stmt ); stmt A prepared statement. sqlite3_db_handle() Appendix G: SQLite C API Reference | 441 Returns The database connection associated with the provided statement. Description This function extracts the database connection associated with a prepared statement. See Also sqlite3_prepare_xxx() sqlite3_db_mutex() Extract a database connection mutex lock Definition sqlite3_mutex* sqlite3_db_mutex( sqlite3* db ); db A database connection. Returns The database connection mutex lock. If the library is not in the “serialized” threading mode, a NULL pointer will be returned. Description This function extracts and returns a reference to the mutual exclusion lock associated with the given database connection. The SQLite library must be in the “serialized” threading mode. If it is in any other mode, this function will return NULL. The mutex object can be used to lock the database connection for the purpose of avoiding race conditions, especially when accessing error messages or row counts that are stored as part of the database connection. See Also sqlite3_mutex_enter(), sqlite3_mutex_leave() sqlite3_db_status() Get the status of a database connection resource [EXP] Definition int sqlite3_db_status( sqlite3* db, int option, int* current, int* highest, int reset); db A database connection. option The resource option to retrieve. current The current value of the given option will be passed back through this reference. sqlite3_db_mutex() 442 | Appendix G: SQLite C API Reference highest The highest seen value of the given option will be passed back through this reference. reset If this value is nonzero, the highest seen value will be reset to the current value. Returns An SQLite result code. Description This function can be used to query information about the internal resource usage of a database connection. The function will pass back both the current and the highest seen values. This can be used to track resource consumption. For more information on the currently supported status options, see http://www.sqlite.org/ c3ref/db_status.html. See Also sqlite3_status(), sqlite3_stmt_status() sqlite3_declare_vtab() Define the schema of a virtual table [EXP] Definition int sqlite3_declare_vtab( sqlite3* db, const char *sql ); db A database connection. sql A UTF-8 encoded string that contains an appropriately formatted CREATE TABLE statement. Returns An SQLite result code. Description This function is used by a virtual table module to define the schema of the virtual table. For more information on writing a virtual table module, see Chapter 10. See Also sqlite3_create_module() sqlite3_enable_load_extension() Enable or disable dynamic extension loading Definition int sqlite3_enable_load_extension( sqlite3* db, int enabled ); sqlite3_enable_load_extension() Appendix G: SQLite C API Reference | 443 db A database connection. enabled An enable/disable flag. A nonzero value will enable dynamic modules, while a zero value will disable them. Returns An SQLite result code. Description This function enables or disables the ability to load dynamic modules. For security reasons, dynamic modules are disabled by default and must be explicitly enabled. See Also sqlite3_load_extension() sqlite3_enable_shared_cache() Enable or disable shared cache mode Definition int sqlite3_enable_shared_cache( int enabled ); enabled An enable/disable flag. A nonzero flag will enable shared cache mode, while a zero value will disable it. Returns An SQLite result code. Description This function enables or disables shared cache across database connections. By default, shared cache mode is disabled. If enabled, an application that uses multiple database connections to the same database will share page caches and other management data. This can reduce mem- ory usage and I/O, but there are also a number of minor limitations associated with having shared cache mode enabled. Altering the cache mode will not change any existing database connections. It will only affect the mode used by new database connections created with an sqlite3_open_xxx() function. For more information on shared cache mode, see http://www.sqlite.org/sharedcache.html. See Also sqlite3_open(), sqlite3_open_v2() sqlite3_errcode() Get error code for last failed API call Definition int sqlite3_errcode( sqlite3* db ); sqlite3_enable_shared_cache() 444 | Appendix G: SQLite C API Reference db A database connection. Returns An error code or extended error code. Description This function returns the error code from the last failed sqlite3_* API call associated with this database connection. If extended error codes are enabled, this function may also return an extended error code. If a failed call was followed by a successful call, the results are undefined. If the SQLite library is in “serialized” threading mode, there is a risk of a race condition between threads. To avoid problems, the current thread should use sqlite3_mutex_enter() to acquire exclusive access to the database connection before the initial API call is made. The thread can release the mutex after sqlite3_errcode() is called. In “multi-thread” mode, it is the responsibility of the application to control access to the database connection. If an API call returns SQLITE_MISUSE, it indicates an application error. In that case, the result code may or may not be available to sqlite3_errcode(). See Also sqlite3_errmsg() [C API, Ap G], sqlite3_extended_result_codes() [C API, Ap G], sqlite3_extended_errcode() [C API, Ap G], sqlite3_mutex_enter() [C API, Ap G] sqlite3_errmsg() Get an error string from last failed API call Definition const char* sqlite3_errmsg( sqlite3* db ); const void* sqlite3_errmsg16( sqlite3* db ); db A database connection. Returns Human-readable error message, in English. Description This function returns the error message from the last failed sqlite3_* API call associated with this database connection. If a failed call was followed by a successful call, the results are undefined. Most of SQLite’s built-in error messages are extremely terse and somewhat cryptic. Although the error messages are useful for logging and debugging, most applications should provide their own end-user error messages. If SQLite is being used in a threaded environment, this function is subject to the same concerns as sqlite3_errcode(). sqlite3_errmsg() Appendix G: SQLite C API Reference | 445 If an API call returns SQLITE_MISUSE, it indicates an application error. In that case, the result code may or may not be available to sqlite3_errmsg(). See Also sqlite3_errcode() sqlite3_exec() Execute SQL statements Definition int sqlite3_exec( sqlite3* db, const char* sql, exec_callback, void* udp, char** errmsg ); int exec_callback( void* udp, int c_num, char** c_vals, char** c_names ); db A database connection. sql A null-terminated, UTF-8 encoded string that contains one or more SQL statements. If more than one SQL statement is provided, they must be separated by semicolons. exec_callback An optional callback function. This application-defined callback is called once for each row in the result set. If set to NULL, no result data will be made available. udp An application-defined user-data pointer. This value is made available to the exec call- back function. errmsg An optional reference to a string pointer. If the function returns anything other than SQLITE_OK, an error message will be passed back. If no error is encountered, the pointer will be set to NULL. The reference may be NULL to ignore error messages. Error messages must be freed with sqlite3_free(). c_num The number of columns in the current result row. c_vals An array of UTF-8 strings that contain the values of the current result row. These values are generated with sqlite3_column_text() and use the same conversion rules. There is no way to determine the original datatype of the value. c_names An array of UTF-8 strings that contain the column names of the current result row. These are generated with sqlite3_column_name(). Returns (sqlite3_exec()) An SQLite result code. sqlite3_exec() 446 | Appendix G: SQLite C API Reference Returns (exec_callback()) If the callback returns a nonzero value, execution of the current statement is halted and sqlite3_exec() will immediately return SQLITE_ABORT. Description This function is a convenience function used to run SQL statements in one step. This function will prepare and step through one or more SQL statements, calling the provided callback with each result row. The callback can then process each row of the results. The callback should not attempt to free the memory used to hold the column names or value strings. All result values are given as strings. Although sqlite3_exec() is an easy way to execute standalone SQL statements, it does not allow the use of statement parameters. This can lead to performance and security concerns for statements that require application-defined values. Manually preparing and executing the statement allows the use of statement parameters and the sqlite3_bind_xxx() functions. This function has no performance advantage over the explicit sqlite3_prepare() and sqlite3_step() functions. In fact, sqlite3_exec() uses those functions internally. See Also sqlite3_column_xxx() (in specific, sqlite3_column_text()), sqlite3_column_name(), sqlite3_prepare_xxx(), sqlite3_step() sqlite3_extended_errcode() Get extended error code for last failed API call Definition int sqlite3_extended_errcode( sqlite3* db ); db A database connection. Returns An extended error code. Description This function is similar to sqlite3_errcode(), except that it will always return an extended error code, even if extended result codes are not enabled. If SQLite is being used in a threaded environment, this function is subject to the same concerns as sqlite3_errcode(). If an API call returns SQLITE_MISUSE, it indicates an application error. In that case, the result code may or may not be available to sqlite3_extended_errcode(). See Also sqlite3_errcode(), sqlite3_extended_result_codes() sqlite3_extended_errcode() Appendix G: SQLite C API Reference | 447 sqlite3_extended_result_codes() Enable or disable extended result codes Definition int sqlite3_extended_result_codes( sqlite3* db, int enabled ); db A database connection. enabled A nonzero value indicates that extended result codes should be turned on, while a zero value indicates that extended result codes should be turned off. Returns An SQLite result code. Description This function is used to enable extended result codes. Extended codes are off by default. When enabled, any API function that returns an SQLite result code may return an extended code. The extended code is always available using sqlite3_extended_errcode(). See Also sqlite3_errcode() , sqlite3_extended_errcode() sqlite3_file_control() Low-level control of database files Definition int sqlite3_file_control( sqlite3* db, const char *name, int option, void* data ); db A database connection. name The destination logical database name in UTF-8. This can be main, temp, or the logical name given to ATTACH DATABASE. option A configuration option value. data A configuration data value. The exact meaning is typically determined by the value of the option parameter. Returns An SQLite result code or other value. sqlite3_extended_result_codes() 448 | Appendix G: SQLite C API Reference Description This function allows an application to interact with the low-level file I/O layer within SQLite. The option and data parameters will be passed to the appropriate VFS (Virtual File System) driver’s xFileControl() function. See Also sqlite3_vfs_register() sqlite3_finalize() Finalize and release a prepared statement Definition int sqlite3_finalize( sqlite3_stmt* stmt ); stmt A prepared statement Returns An SQLite result code. Description This function finalizes prepared statements. Finalizing a statement releases any internal re- sources and deallocates any memory. Once a statement has been finalized, it is no longer valid and cannot be reused. See Also sqlite3_prepare_xxx(), sqlite3_step() sqlite3_free() Free a memory allocation Definition void sqlite3_free( void* ptr ); ptr A pointer to a memory allocation. Description This function releases a memory block that was previously allocated with sqlite3_malloc() or sqlite3_realloc(). This should not be used to release memory acquired through a native malloc() or new call. See Also sqlite3_malloc(), sqlite3_realloc() sqlite3_free() Appendix G: SQLite C API Reference | 449 sqlite3_free_table() Release table structure Definition void sqlite3_free_table( char** result ); result An array of table values returned by sqlite3_get_table(). Description This function properly releases the value array returned by sqlite3_get_table(). See Also sqlite3_get_table() sqlite3_get_autocommit() Get the current transaction mode Definition int sqlite3_get_autocommit( sqlite3* db ); db A database connection. Returns A nonzero value if the given database connection is in autocommit mode, a zero value if it is not. Description This function returns the current transaction mode. A nonzero return value indicates that the given database connection is in autocommit mode, meaning that no explicit transaction is open. A return value of zero means that an explicit transaction is in progress. If an error is encountered while in an explicit transaction, there may be times when SQLite is forced to roll back the transaction. This function can be used to detect if the rollback happened or not. See Also BEGIN TRANSACTION [SQL Cmd, Ap C] sqlite3_get_auxdata() Get auxiliary data from a function parameter Definition void* sqlite3_get_auxdata( sqlite3_context* ctx, int pidx ); ctx An SQL function context, provided by the SQLite library. pidx A function parameter index. The first parameter has an index of zero (0). sqlite3_free_table() 450 | Appendix G: SQLite C API Reference Returns A user-defined data pointer set with sqlite3_set_auxdata(). Description This function is used by an SQL function implementation to retrieve any application-defined auxiliary data that may have been attached to a function parameter with sqlite3_set_aux data(). If no data is available, or the data has been cleared, this function will return NULL. See Also sqlite3_set_auxdata(), sqlite3_value_xxx() sqlite3_get_table() Get a result table Definition int sqlite3_get_table( sqlite3* db, const char sql, char*** result, int* num_rows, int* num_cols, char** errmsg ); db A database connection. sql A null-terminated, UTF-8 encoded string that contains one or more SQL statements. If more than one SQL statement is provided, they must be separated by semicolons. result A reference to an array of strings. The results of the query are passed back using this reference. The value passed back must be freed with a call to sqlite3_free_table(). num_rows The number of rows in the result, not including the column names. num_cols The number of columns in the result. errmsg An optional reference to a string. If an error occurs, the reference will be set to an error message. The application is responsible for freeing the message with sqlite3_free(). If no error occurs, the reference will be set to NULL. The reference may be NULL. Returns An SQLite result code. Description This function takes an SQL statement, executes it, and passes back the full result set. If more than one SQL statement is given in an SQL string, all the statements must have a result set with the same number of columns. sqlite3_get_table() Appendix G: SQLite C API Reference | 451 The result is an array of strings with (num_rows + 1) * num_cols values. The first num_cols values are the column names, followed by individual rows. To access the value of a specific column and row, use the formula (row + 1) * num_cols + col, assuming the row and col indexes start with zero (0). See Also sqlite3_free_table(), sqlite3_free() sqlite3_initialize() Initialize the SQLite library Definition int sqlite3_initialize( ); Returns An SQLite result code. Description This function initializes the SQLite library. It can be safely called multiple times. Many stand- ard SQLite functions, such as sqlite3_open(), will automatically call sqlite3_initialize() if it was not explicitly called by the application. The SQLite library can be safely initialized and shut down repeatedly. See Also sqlite3_open(), sqlite3_shutdown() sqlite3_interrupt() Cancel any in-progress database operations. Definition void sqlite3_interrupt( sqlite3* db ); db A database connection. Description This function causes all currently running statements to abort at their earliest convenience. It is safe to call this from a different thread. Interrupted functions will return SQLITE_INTER RUPT. If the interrupted function is modifying the database file and is inside an explicit trans- action, the transaction will be rolled back. The interrupt state will continue until the active statement count reaches zero. This function should not be called if the database connection may be closed. A crash is likely if the database connection is closed while an interrupt is still outstanding. See Also sqlite3_close() sqlite3_initialize() 452 | Appendix G: SQLite C API Reference sqlite3_last_insert_rowid() Get the last inserted ROWID Definition sqlite3_int64 sqlite3_last_insert_rowid( sqlite3* db ); db A database connection. Returns The value of the last inserted ROWID. Description This function returns the ROWID of the last successfully inserted row. If no rows have been inserted since the database connection was opened, this function will return zero (0). Note that zero is a valid ROWID, but it will never be automatically assigned by SQLite. This function is typically used to get the automatically generated value for a newly inserted record. The value is often used to populate foreign keys. If the INSERT happens inside of a trigger, the inserted ROWID value is valid for the duration of the trigger. Once the trigger exits, the value returns to its previous value. If the SQLite library is in “serialized” threading mode, there is a risk of a race condition between threads. To avoid problems, the current thread should use sqlite3_mutex_enter() to acquire exclusive access to the database connection before the initial API call is made. The thread can release the mutex after sqlite3_last_insert_rowid() is called. In “multithread” mode, it is the responsibility of the application to control access to the database connection. This function is exposed to the SQL environment as the SQL function last_insert_rowid(). See Also last_insert_rowid() [SQL Func, Ap E] sqlite3_libversion() Get the SQLite library version Definition const char* sqlite3_libversion( ); Returns The SQLite library version as a UTF-8 encoded string. Description This function returns the SQLite library version as a string. The version string for SQLite v3.6.23 is "3.6.23". The version string for v3.6.23.1 is "3.6.23.1". This value is also available at compile time as the SQLITE_VERSION macro. By comparing the macro used at compile time to the value returned at runtime, an application can verify that it is linking against the correct version of the library. sqlite3_libversion() Appendix G: SQLite C API Reference | 453 This function is exposed to the SQL environment as the SQL function sqlite_version(). See Also sqlite3_libversion_number(), sqlite3_sourceid(), sqlite_version() [SQL Func, Ap E] sqlite3_libversion_number() Get the SQLite library version Definition int sqlite3_libversion_number( ); Returns The SQLite library version as a number. Description This function returns the SQLite library version as an integer. The number is in the form Mmmmppp, with M being the major version, m the minor, and p the point release. Smaller steps are not included. The version number of SQLite v3.6.23.1 (and v3.6.23) is 3006023. This value is also available at compile time as the SQLITE_VERSION_NUMBER macro. By comparing the macro used at compile time to the value returned at runtime, an application can verify that it is linking against a proper version of the library. See Also sqlite3_libversion(), sqlite3_sourceid() sqlite3_limit() Get or set SQLite library limits and maximums Definition int sqlite3_limit( sqlite3* db, int limit, int value ); db A database connection. limit The specific limit to set. value The new value. If the value is negative, the value will not be set (and the current value will be returned). Returns The previous value. Description This function gets or sets several of the database connection limits. The limits can be lowered at runtime on a per-connection basis to limit resource consumption. The hard limits are set using the SQLITE_MAX_xxx build options. If an application attempts to set the limit value higher than the hard limit, the limit will be silently set to the maximum limit. sqlite3_libversion_number() 454 | Appendix G: SQLite C API Reference For a full list of the currently supported limits, see http://www.sqlite.org/c3ref/c_limit_attached .html. See Also sqlite3_config(), sqlite3_db_config() sqlite3_load_extension() Load a dynamic extension Definition int sqlite3_load_extension( sqlite3* db, const char* file, const char* entry_point, char** errmsg ); db A database connection. file The path and filename of the extension. entry_point The name of the entry-point function. If this is NULL, the name sqlite3_exten sion_init will be used. errmsg An optional reference to a string pointer. If the function returns anything other than SQLITE_OK, and error message will be passed back. If no error is encountered, the pointer will be set to NULL. The reference may be NULL to ignore error messages. Error messages must be freed with sqlite3_free(). Returns An SQLite result code. Description This function attempts to load an SQLite dynamic extension. By default, the use of dynamic extensions are disabled, and must be enabled using the sqlite3_enable_load_extension() call. This function is also exposed as the load_extension() SQL function. Although there are no limits on when an extension may be loaded, many extensions register new functions and are thus subject to the limits of sqlite3_create_function(). For more information on dynamic extensions, please see the section “Using Loadable Exten- sions” on page 211. See Also sqlite3_enable_load_extension() [C API, Ap G], sqlite3_create_function() [C API, Ap G], load_extension() [SQL Func, Ap E] sqlite3_load_extension() Appendix G: SQLite C API Reference | 455 sqlite3_log() Log a message in the SQLite logfile [EXP] Definition void sqlite3_log( int errcode, const char* format, ... ); errcode The error code associated with this log message. format A sqlite3_snprintf() style message formatting string. Additional parameters Message formatting parameters. Description This function logs a message in the SQLite logfile. The format and additional parameters will be passed to an sqlite3_snprintf() like function for formatting. Messages are limited to a few hundred characters. Longer messages will be truncated. The sqlite3_log() function was designed to be used by extensions that do not have a normal debug or error reporting path. Normally, SQLite has no logfile and the message is simply ignored. A logfile may be configured using sqlite3_config(). See Also sqlite3_config() sqlite3_malloc() Obtain a dynamic memory allocation Definition void* sqlite3_malloc( int bytes ); bytes The size of the requested allocation, in bytes. Returns A newly allocated buffer. If the memory is not available, NULL is returned. Description This function obtains a dynamic memory allocation from the SQLite library. Memory allo- cated with sqlite3_malloc() should be released with sqlite3_free(). Allocations will always start on an 8-byte (or larger) boundary. Although many SQLite environments will simply pass memory allocation requests on to the default system memory allocator, there are some environments that configure specific buffers for the SQLite library. By using these memory handling functions, an SQLite extension or module will work correctly in any SQLite environment, regardless of how the memory allo- cator is configured. See Also sqlite3_free(), sqlite3_realloc() sqlite3_log() 456 | Appendix G: SQLite C API Reference sqlite3_memory_highwater() Get or reset the memory usage high-water mark Definition sqlite3_int64 sqlite3_memory_highwater( int reset ); reset If this value is nonzero, the high-water mark will be reset to the current in-use value. Returns The high-water mark of bytes allocated by the SQLite memory system. Description This function returns the memory usage high-water mark, or the highest seen value. The high- water mark can optionally be reset to the current value. See Also sqlite3_memory_used() sqlite3_memory_used() Get the current memory use Definition sqlite3_int64 sqlite3_memory_used( ); Returns The number of bytes currently allocated by the SQLite memory system. Description This function returns the current number of bytes allocated by the SQLite memory functions. This figure includes all of the overhead introduced by the SQLite allocator, but does not include any overhead required by the system memory handlers. See Also sqlite3_memory_highwater(), sqlite3_malloc() sqlite3_mprintf() Format and allocate a string Definition char* sqlite3_mprintf( const char* format, ... ); format An sqlite3_snprintf() style message formatting string. Additional parameters Message formatting parameters. Returns A newly allocated string built from the given parameters. sqlite3_mprintf() Appendix G: SQLite C API Reference | 457 Description This function builds a formatted string and returns it in a newly allocated memory buffer. If the required memory allocation fails, a NULL may be returned. The application is responsible for releasing the returned buffer with sqlite3_free(). This function supports the same extended formatting options as sqlite3_snprintf(). See Also sqlite3_snprintf() sqlite3_mutex_alloc() Allocate a new mutex lock Definition sqlite3_mutex* sqlite3_mutex_alloc( int type ); type The desired type of mutual exclusion lock. Returns A newly allocated and initialized mutual exclusion lock. Description This function allocates and initializes a new mutual exclusion lock of the requested type. Applications can request an SQLITE_MUTEX_RECURSIVE or SQLITE_MUTEX_FAST lock. A recursive lock must support and properly reference-count multiple enter/leave calls by the same thread. A fast mutex does not need to support anything other than simple enter/leave semantics. Depending on the threading library, a fast lock may not actually be any faster than a recursive lock. Any application mutex acquired through sqlite3_mutex_alloc() should eventually be freed with sqlite3_mutex_free(). See Also sqlite3_mutex_free(), sqlite3_db_mutex() sqlite3_mutex_enter() Acquire a mutex lock Definition void sqlite3_mutex_enter( sqlite3_mutex* mutex ); mutex A mutual exclusion lock. If a NULL value is passed, the function will return immediately. Description This function attempts to have the calling thread acquire a mutual exclusion lock. This is typically done when entering a critical section of code. The function (and thread) will block until access to the lock is granted. sqlite3_mutex_alloc() 458 | Appendix G: SQLite C API Reference Like most of the other mutex functions, this function is designed to replicate the default success behavior when given a NULL mutex. This allows the result of sqlite3_db_mutex() to be passed directly to sqlite3_mutex_enter() (or most other mutex functions) without having to test the current threading mode. See Also sqlite3_mutex_leave(), sqlite3_mutex_try(), sqlite3_db_mutex() sqlite3_mutex_free() Deallocate a mutex lock Definition void sqlite3_mutex_free( sqlite3_mutex* mutex ); mutex A mutual exclusion lock. Passing in a NULL pointer are not allowed. Description This function destroys and deallocates a mutual exclusion lock. The lock should not be held by any thread when it is freed. Applications should only free locks that the application created with sqlite3_mutex_alloc(). See Also sqlite3_mutex_alloc() sqlite3_mutex_held() Test if a mutex lock is held Definition int sqlite3_mutex_held( sqlite3_mutex* mutex ); mutex A mutual exclusion lock. If a NULL value is passed, the function should return true (nonzero). Returns A nonzero value if the given mutex is currently held, or a zero value if the mutex is not held. Description This function tests to see if the current thread holds the given mutual exclusion lock. The SQLite library only uses this function in assert() checks. Generally, this function is only available if SQLite is compiled while SQLITE_DEBUG is defined. Not all threading libraries sup- ports this function. Unsupported platforms should always return true (nonzero). Applications should limit use of this function to debug and verification code. See Also sqlite3_mutex_notheld(), sqlite3_mutex_enter(), SQLITE_DEBUG [Build Opt, Ap A] sqlite3_mutex_held() Appendix G: SQLite C API Reference | 459 sqlite3_mutex_leave() Release a mutex lock Definition void sqlite3_mutex_leave( sqlite3_mutex* mutex ); mutex A mutual exclusion lock. If a NULL value is passed, the function will simply return. Description This function allows a thread to release its hold on a mutual exclusion lock. This makes the lock available to other threads. This is typically done when leaving a critical section of code. See Also sqlite3_mutex_enter(), sqlite3_mutex_alloc(), sqlite3_db_mutex(), sqlite3_mutex_free() sqlite3_mutex_notheld() Test if a mutex lock is not held Definition int sqlite3_mutex_notheld( sqlite3_mutex* mutex ); mutex A mutual exclusion lock. If a NULL value is passed, the function should return true (nonzero). Returns A nonzero value if the given mutex is currently not held, or a zero value if the mutex is held. Description This function is essentially the opposite of sqlite3_mutex_held() and is subject to the same conditions and limitations. See Also sqlite3_mutex_held() sqlite3_mutex_try() Attempt to acquire a mutex lock Definition int sqlite3_mutex_try( sqlite3_mutex* mutex ); mutex A mutual exclusion lock. Returns The value SQLITE_OK is returned if the lock was acquired. Otherwise, SQLITE_BUSY will be returned. sqlite3_mutex_leave() 460 | Appendix G: SQLite C API Reference Description This function attempts to acquire a mutual exclusion lock for the current thread. If the lock is acquired, SQLITE_OK will be returned. If the lock is held by another thread, SQLITE_BUSY will be returned. If a NULL mutex pointer is passed in, the function will return SQLITE_OK. This function is not supported by all threading libraries. If SQLite has thread support, but this function is not supported, a valid mutex will always result in an SQLITE_BUSY return code. See Also sqlite3_mutex_enter(), sqlite3_mutex_leave() sqlite3_next_stmt() Step through a list of prepared statements Definition sqlite3_stmt* sqlite3_next_stmt( sqlite3* db, sqlite3_stmt* stmt ); db A database connection. stmt The previous statement. To get the first statement, pass in a NULL. Returns A prepared statement pointer. A NULL indicates the end of the list. Description This function iterates over the list of prepared statements associated with the given database connection. Passing in a NULL statement pointer will return the first statement. Subsequent statements can be retrieved by passing back the last returned statement value. If an application is iterating over prepared statements and finalizing them, remember that the sqlite3_finalize() function will remove the statement from the statement list. In such cases, the application should keep passing in a NULL statement value until no more statements remain. This technique is not recommended, however, as it can leave dangling statement pointers. See Also sqlite3_prepare_xxx(), sqlite3_finalize() sqlite3_open() Open a database file Definition int sqlite3_open( const char* filename, sqlite3** db_ref ); int sqlite3_open16( const void* filename, sqlite3** db_ref ); filename The path and filename of the database file as a UTF-8 or UTF-16 encoded string. sqlite3_open() Appendix G: SQLite C API Reference | 461 db_ref A reference to a database connection. If the database is successfully opened, the database connection will be passed back. Returns An SQLite result code. Description These functions open a database and create a new database connection. If the filename does not exist, it will be created. The file will be opened read/write if possible. If not, the file will be opened read-only. If the filename is :memory:, a temporary, in-memory database will be created. The database will automatically be deleted when the database connection is closed. Each call to open will create a new in-memory database. Each in-memory database is accessible from only one da- tabase connection. If the filename is NULL or an empty string (""), a temporary, file-backed database will be created. This is very similar to an in-memory database, only the database is allowed to page out to disk. This allows the database to grow to much larger sizes without worrying about memory consumption. These functions are considered legacy APIs. It is recommended that all new development use sqlite3_open_v2(). See Also sqlite3_open_v2(), sqlite3_close() sqlite3_open_v2() Open a database file Definition int sqlite3_open_v2( const char* filename, sqlite3** db_ref, int flags, const char* vfs ); filename The path and filename of the database file as a UTF-8 encoded string. db_ref A reference to a database connection. If the database is successfully opened, the database connection will be passed back. flags A series of flags that can be used to control how the database file is open. vfs The name of the VFS (Virtual File System) module to use. A NULL will result in the default module. sqlite3_open_v2() 462 | Appendix G: SQLite C API Reference Returns An SQLite result code. Description This function is very similar to sqlite3_open(), but provides better control over how the database file is opened. The flags parameter controls the state of the opened file, while the vfs parameter allows the application to specify a VFS (virtual file system) driver. The flag parameter consists of several bit-flags that can be or’ed together. The application must specify one of the following flag combinations: SQLITE_OPEN_READONLY Open the file read-only. The file must already exist. SQLITE_OPEN_READWRITE Attempt to open the file read/write. If this is not possible, open the file read-only. Opening the file read-only will not result in an error. The file must already exist. SQLITE_OPEN_READWRITE | SQLITE_OPEN_CREATE Attempt to open the file read/write. If it does not exist, create the file. If the file does exist, but permissions do not allow read/write access, open the file read-only. Opening the file read-only will not result in an error. This is the behavior of sqlite3_open(). Additionally, these optional flags may be added to one of the above flag sets: SQLITE_OPEN_NOMUTEX If the SQLite library was compiled with threading support, open the database connection in “multithread” mode. This flag cannot be used in conjunction with the SQLITE_OPEN_ FULLMUTEX flag. SQLITE_OPEN_FULLMUTEX If the SQLite library was compiled with threading support, open the database connection in “serialized” mode. This flag cannot be used in conjunction with the SQLITE_OPEN_NO MUTEX flag. SQLITE_OPEN_SHAREDCACHE Enables shared cache mode for this database connection. This flag cannot be used in conjunction with the SQLITE_OPEN_PRIVATECACHE flag. SQLITE_OPEN_PRIVATECACHE Disables shared cache mode for this database connection. This flag cannot be used in conjunction with the SQLITE_OPEN_SHAREDCACHE flag. Additional flags may be added to future versions of SQLite. See Also sqlite3_open() [Ap G], sqlite3_close() [Ap G], sqlite3_enable_shared_cache() [Ap G], sqlite3_config() [Ap G], sqlite3_vfs_find() [Ap G] sqlite3_open_v2() Appendix G: SQLite C API Reference | 463 sqlite3_overload_function() Create a stub function [EXP] Definition int sqlite3_overload_function( sqlite3* db, const char* name, int num_param ); db A database connection. name A function name, in UTF-8. num_param The number of parameters in the stub function. Returns An SQLite result code. Description This function checks to see if the named SQL function (with the specified number of param- eters) exists or not. If the function does exist, nothing is done. If the function does not already exist, a stub function is registered. This stub function will always return SQLITE_ERROR. This function exists to support virtual table modules. Virtual tables have the ability to override functions that use virtual table columns as parameters. In order to do this, the function must already exist. The virtual table could register a stub function by itself, but this risks overriding an existing function. This function ensures there is a base SQL function that can be overridden without accidentally redefining an application-defined function. See Also sqlite3_create_function(), sqlite3_create_module() sqlite3_prepare_xxx() Convert an SQL string into a prepared statement Definition int sqlite3_prepare( sqlite3* db, const char* sql, int sql_len, sqlite3_stmt** stmt_ref, const char** tail ); int sqlite3_prepare16( sqlite3* db, const void* sql, int sql_len, sqlite3_stmt** stmt_ref, const char** tail ); int sqlite3_prepare_v2( sqlite3* db, const char* sql, int sql_len, sqlite3_stmt** stmt_ref, const char** tail ); int sqlite3_prepare16_v2( sqlite3* db, const void* sql, int sql_len, sqlite3_stmt** stmt_ref, const char** tail ); db A database connection. sql One or more SQL statements in a UTF-8 or UTF-16 encoded string. If the string contains more than one SQL statement, the statements must be separated by a semicolon. If the string contains only one SQL statement, no trailing semicolon is required. sqlite3_overload_function() 464 | Appendix G: SQLite C API Reference sql_len The length of the sql buffer in bytes. If the sql string is null-terminated, the length should include the termination character. If the sql string is null-terminated but the length is not known, a negative value will cause SQLite to compute the buffer length. stmt_ref A reference to a prepared statement. SQLite will allocate and pass back the prepared statement. tail If the sql buffer contains more than one SQL statement, only the first complete statement is used. If additional statements exist, this reference will be used to pass back a pointer to the next SQL statement in the sql buffer. This reference may be set to NULL. Returns An SQLite result code. Description These functions take an SQL statement string and build a prepared statement. The prepared statement can be executed using sqlite3_step(). All prepared statements must eventually be finalized with sqlite3_finalize(). Both the original and the _v2 versions of prepare take the exact same parameters. The _v2 versions produce a slightly different statement, however. The newer statements are able to automatically recover from some errors and provide better error handling. For a more in- depth discussion of the differences, see “Prepare v2” on page 149. The original versions are considered legacy APIs, and their use in new development is not recommended. See Also sqlite3_finalize(), sqlite3_step() sqlite3_profile() Register an SQL statement profile callback Definition void* sqlite3_profile( sqlite3* db, profile_func, void* udp ); void profile_func( void* udp, const char* sql, sqlite3_uint64 time ); db A database connection. profile_func An application-defined profile callback function. udp An application-defined user-data pointer. This value is made available to the profile function. sqlite3_profile() Appendix G: SQLite C API Reference | 465 sql The text of the SQL statement that was executed, encoded in UTF-8. time The wall-clock time required to execute the statement, in nanoseconds. Most system clocks cannot actually provide nanosecond precision. Returns (sqlite3_profile()) The udp parameter of any prior profile function. If no prior profile function was registered, a NULL will be returned. Description This function allows an application to register a profiling function. The callback function is called every time a call to sqlite3_step() finishes. Only one profile function is allowed. Reg- istering a new function will override the old function. To unregister a profiling function, pass in a NULL function pointer. See Also sqlite3_trace(), sqlite3_prepare_xxx(), sqlite3_step() sqlite3_progress_handler() Register a progress handler Definition void sqlite3_progress_handler( sqlite3* db, int num_ops, progress_handler, void* udp ); int progress_handler( void* udp ); db A database connection. num_ops The number of byte-code operations the database engine should perform between calls to the progress handler. This is an approximation. progress_handler An application-defined progress handler callback function. udp An application-defined user-data pointer. This value is made available to the progress handler. Returns (progress_handler()) If the handler returns a nonzero value, the current database operations are interrupted. Description This function allows an application to register a progress handler callback. Any time the database connection is in use, the progress handler will be called. The frequency is controlled by the num_ops parameter, which is the approximate number of virtual database engine byte- code operations that will be performed between each call. sqlite3_progress_handler() 466 | Appendix G: SQLite C API Reference This function only provides a periodic callback. It cannot be used to estimate the percentage of completion. See Also sqlite3_interrupt() sqlite3_randomness() Request random value data Definition void sqlite3_randomness( int bytes, void* buffer ); bytes The desired number of random data bytes. buffer An application buffer that is large enough to hold the requested data. Description This function allows an application to request random data values from SQLite’s internal pseudo-random number generator. See Also random() [SQL Func, Ap E], randomblob() [SQL Func, Ap E] sqlite3_realloc() Resize a dynamic memory allocation Definition void* sqlite3_realloc( void* ptr, int bytes ); ptr A pointer to an existing dynamic memory buffer. May be NULL. bytes The new size of the requested buffer, in bytes. Returns The newly adjusted buffer. May be NULL if unable to allocate memory. If NULL, the old buffer will still be valid. Description This function resizes a dynamic memory allocation. It can be used to increase or decrease the size of an allocation. This may require moving the allocation. In that case, the contents of the current buffer will be copied to the beginning of the adjusted buffer. If the new buffer is smaller, some data will be dropped. Allocations will always start on an 8-byte (or larger) boundary. sqlite3_realloc() Appendix G: SQLite C API Reference | 467 If a NULL pointer is passed in, this function will act like sqlite3_malloc() and allocate a new buffer. If the bytes parameter is zero or negative, this function will act like sqlite3_free(), releasing the existing buffer and returning NULL. See Also sqlite3_malloc(), sqlite3_free() sqlite3_release_memory() Reduce memory usage Definition int sqlite3_release_memory( int bytes ); bytes The requested number of bytes to free. Returns The actual number of bytes to free. This may be more or less than the requested number. Description This function requests that the SQLite library free a specific amount of memory. SQLite will do this by deallocating noncritical memory, such as cache space (this includes the page cache). The memory that is freed may not be contiguous. See Also sqlite3_malloc(), sqlite3_free() sqlite3_reset() Reset of prepared statement Definition int sqlite3_reset( sqlite3_stmt* stmt ); stmt A prepared statement. Returns An SQLite result code. Description This function resets a prepared statement, making it available for execution. This is typically done after one or more calls to sqlite3_step(). The reset call will release any locks and other resources associated with the last execution of the statement. This function may be called any time between calls to sqlite3_prepare_xxx() and sqlite3_finalize(). This function does not clear the statement parameter bindings. An application must call sqlite3_clear_bindings() to reset the statement parameters to NULL. See Also sqlite3_prepare_xxx(), sqlite3_step(), sqlite3_finalize(), sqlite3_clear_bindings() sqlite3_release_memory() 468 | Appendix G: SQLite C API Reference sqlite3_reset_auto_extension() Remove all automatic extensions Definition void sqlite3_reset_auto_extension( ); Description This function removes all automatic extensions that were previously registered with sqlite3_auto_extension(). There is no way to remove specific extensions or to retrieve a list of the current extensions. See Also sqlite3_auto_extension() sqlite3_result_xxx() Return an result from an SQL function Definition void sqlite3_result_blob( sqlite3_context* ctx, const void* val,