xref: /sqlite-3.40.0/README.md (revision 3b328522)
1<h1 align="center">SQLite Source Repository</h1>
2
3This repository contains the complete source code for the SQLite database
4engine.  Some test scripts are also include.  However, many other test scripts
5and most of the documentation are managed separately.
6
7If you are reading this on a Git mirror someplace, you are doing it wrong.
8The [official repository](https://www.sqlite.org/src/) is better.  Go there
9now.
10
11## Obtaining The Code
12
13SQLite sources are managed using the
14[Fossil](https://www.fossil-scm.org/), a distributed version control system
15that was specifically designed to support SQLite development.
16If you do not want to use Fossil, you can download tarballs or ZIP
17archives as follows:
18
19  *  Lastest trunk check-in:
20     <https://www.sqlite.org/src/tarball/sqlite.tar.gz> or
21     <https://www.sqlite.org/src/zip/sqlite.zip>.
22
23  *  Latest release:
24     <https://www.sqlite.org/src/tarball/sqlite.tar.gz?r=release> or
25     <https://www.sqlite.org/src/zip/sqlite.zip?r=release>.
26
27  *  For other check-ins, substitute an appropriate branch name or
28     tag or hash prefix for "release" in the URLs of the previous
29     bullet.  Or browse the [timeline](https://www.sqlite.org/src/timeline)
30     to locate the check-in desired, click on its information page link,
31     then click on the "Tarball" or "ZIP Archive" links on the information
32     page.
33
34If you do want to use Fossil to check out the source tree,
35first install Fossil version 2.0 or later.
36(Source tarballs and precompiled binaries available
37[here](https://www.fossil-scm.org/fossil/uv/download.html).)
38Then run commands like this:
39
40        mkdir ~/sqlite
41        cd ~/sqlite
42        fossil clone https://www.sqlite.org/src sqlite.fossil
43        fossil open sqlite.fossil
44
45After setting up a repository using the steps above, you can always
46update to the lastest version using:
47
48        fossil update trunk   ;# latest trunk check-in
49        fossil update release ;# latest official release
50
51Or type "fossil ui" to get a web-based user interface.
52
53## Compiling
54
55First create a directory in which to place
56the build products.  It is recommended, but not required, that the
57build directory be separate from the source directory.  Cd into the
58build directory and then from the build directory run the configure
59script found at the root of the source tree.  Then run "make".
60
61For example:
62
63        tar xzf sqlite.tar.gz    ;#  Unpack the source tree into "sqlite"
64        mkdir bld                ;#  Build will occur in a sibling directory
65        cd bld                   ;#  Change to the build directory
66        ../sqlite/configure      ;#  Run the configure script
67        make                     ;#  Run the makefile.
68        make sqlite3.c           ;#  Build the "amalgamation" source file
69        make test                ;#  Run some tests (requires Tcl)
70
71See the makefile for additional targets.
72
73The configure script uses autoconf 2.61 and libtool.  If the configure
74script does not work out for you, there is a generic makefile named
75"Makefile.linux-gcc" in the top directory of the source tree that you
76can copy and edit to suit your needs.  Comments on the generic makefile
77show what changes are needed.
78
79## Using MSVC
80
81On Windows, all applicable build products can be compiled with MSVC.
82First open the command prompt window associated with the desired compiler
83version (e.g. "Developer Command Prompt for VS2013").  Next, use NMAKE
84with the provided "Makefile.msc" to build one of the supported targets.
85
86For example:
87
88        mkdir bld
89        cd bld
90        nmake /f Makefile.msc TOP=..\sqlite
91        nmake /f Makefile.msc sqlite3.c TOP=..\sqlite
92        nmake /f Makefile.msc sqlite3.dll TOP=..\sqlite
93        nmake /f Makefile.msc sqlite3.exe TOP=..\sqlite
94        nmake /f Makefile.msc test TOP=..\sqlite
95
96There are several build options that can be set via the NMAKE command
97line.  For example, to build for WinRT, simply add "FOR_WINRT=1" argument
98to the "sqlite3.dll" command line above.  When debugging into the SQLite
99code, adding the "DEBUG=1" argument to one of the above command lines is
100recommended.
101
102SQLite does not require [Tcl](http://www.tcl.tk/) to run, but a Tcl installation
103is required by the makefiles (including those for MSVC).  SQLite contains
104a lot of generated code and Tcl is used to do much of that code generation.
105The makefiles also require AWK.
106
107## Source Code Tour
108
109Most of the core source files are in the **src/** subdirectory.  But
110src/ also contains files used to build the "testfixture" test harness;
111those file all begin with "test".  And src/ contains the "shell.c" file
112which is the main program for the "sqlite3.exe" command-line shell and
113the "tclsqlite.c" file which implements the bindings to SQLite from the
114Tcl programming language.  (Historical note:  SQLite began as a Tcl
115extension and only later escaped to the wild as an independent library.)
116
117Test scripts and programs are found in the **test/** subdirectory.
118There are other test suites for SQLite (see
119[How SQLite Is Tested](http://www.sqlite.org/testing.html))
120but those other test suites are
121in separate source repositories.
122
123The **ext/** subdirectory contains code for extensions.  The
124Full-text search engine is in **ext/fts3**.  The R-Tree engine is in
125**ext/rtree**.  The **ext/misc** subdirectory contains a number of
126smaller, single-file extensions, such as a REGEXP operator.
127
128The **tool/** subdirectory contains various scripts and programs used
129for building generated source code files or for testing or for generating
130accessory programs such as "sqlite3_analyzer(.exe)".
131
132### Generated Source Code Files
133
134Several of the C-language source files used by SQLite are generated from
135other sources rather than being typed in manually by a programmer.  This
136section will summarize those automatically-generated files.  To create all
137of the automatically-generated files, simply run "make target&#95;source".
138The "target&#95;source" make target will create a subdirectory "tsrc/" and
139fill it with all the source files needed to build SQLite, both
140manually-edited files and automatically-generated files.
141
142The SQLite interface is defined by the **sqlite3.h** header file, which is
143generated from src/sqlite.h.in, ./manifest.uuid, and ./VERSION.  The
144[Tcl script](http://www.tcl.tk) at tool/mksqlite3h.tcl does the conversion.
145The manifest.uuid file contains the SHA1 hash of the particular check-in
146and is used to generate the SQLITE\_SOURCE\_ID macro.  The VERSION file
147contains the current SQLite version number.  The sqlite3.h header is really
148just a copy of src/sqlite.h.in with the source-id and version number inserted
149at just the right spots. Note that comment text in the sqlite3.h file is
150used to generate much of the SQLite API documentation.  The Tcl scripts
151used to generate that documentation are in a separate source repository.
152
153The SQL language parser is **parse.c** which is generate from a grammar in
154the src/parse.y file.  The conversion of "parse.y" into "parse.c" is done
155by the [lemon](./doc/lemon.html) LALR(1) parser generator.  The source code
156for lemon is at tool/lemon.c.  Lemon uses a
157template for generating its parser.  A generic template is in tool/lempar.c,
158but SQLite uses a slightly modified template found in src/lempar.c.
159
160Lemon also generates the **parse.h** header file, at the same time it
161generates parse.c. But the parse.h header file is
162modified further (to add additional symbols) using the ./addopcodes.awk
163AWK script.
164
165The **opcodes.h** header file contains macros that define the numbers
166corresponding to opcodes in the "VDBE" virtual machine.  The opcodes.h
167file is generated by the scanning the src/vdbe.c source file.  The
168AWK script at ./mkopcodeh.awk does this scan and generates opcodes.h.
169A second AWK script, ./mkopcodec.awk, then scans opcodes.h to generate
170the **opcodes.c** source file, which contains a reverse mapping from
171opcode-number to opcode-name that is used for EXPLAIN output.
172
173The **keywordhash.h** header file contains the definition of a hash table
174that maps SQL language keywords (ex: "CREATE", "SELECT", "INDEX", etc.) into
175the numeric codes used by the parse.c parser.  The keywordhash.h file is
176generated by a C-language program at tool mkkeywordhash.c.
177
178### The Amalgamation
179
180All of the individual C source code and header files (both manually-edited
181and automatically-generated) can be combined into a single big source file
182**sqlite3.c** called "the amalgamation".  The amalgamation is the recommended
183way of using SQLite in a larger application.  Combining all individual
184source code files into a single big source code file allows the C compiler
185to perform more cross-procedure analysis and generate better code.  SQLite
186runs about 5% faster when compiled from the amalgamation versus when compiled
187from individual source files.
188
189The amalgamation is generated from the tool/mksqlite3c.tcl Tcl script.
190First, all of the individual source files must be gathered into the tsrc/
191subdirectory (using the equivalent of "make target_source") then the
192tool/mksqlite3c.tcl script is run to copy them all together in just the
193right order while resolving internal "#include" references.
194
195The amalgamation source file is more than 100K lines long.  Some symbolic
196debuggers (most notably MSVC) are unable to deal with files longer than 64K
197lines.  To work around this, a separate Tcl script, tool/split-sqlite3c.tcl,
198can be run on the amalgamation to break it up into a single small C file
199called **sqlite3-all.c** that does #include on about five other files
200named **sqlite3-1.c**, **sqlite3-2.c**, ..., **sqlite3-5.c**.  In this way,
201all of the source code is contained within a single translation unit so
202that the compiler can do extra cross-procedure optimization, but no
203individual source file exceeds 32K lines in length.
204
205## How It All Fits Together
206
207SQLite is modular in design.
208See the [architectural description](http://www.sqlite.org/arch.html)
209for details. Other documents that are useful in
210(helping to understand how SQLite works include the
211[file format](http://www.sqlite.org/fileformat2.html) description,
212the [virtual machine](http://www.sqlite.org/vdbe.html) that runs
213prepared statements, the description of
214[how transactions work](http://www.sqlite.org/atomiccommit.html), and
215the [overview of the query planner](http://www.sqlite.org/optoverview.html).
216
217Unfortunately, years of effort have gone into optimizating SQLite, both
218for small size and high performance.  And optimizations tend to result in
219complex code.  So there is a lot of complexity in the SQLite implementation.
220
221Key files:
222
223  *  **sqlite.h.in** - This file defines the public interface to the SQLite
224     library.  Readers will need to be familiar with this interface before
225     trying to understand how the library works internally.
226
227  *  **sqliteInt.h** - this header file defines many of the data objects
228     used internally by SQLite.
229
230  *  **parse.y** - This file describes the LALR(1) grammer that SQLite uses
231     to parse SQL statements, and the actions that are taken at each step
232     in the parsing process.
233
234  *  **vdbe.c** - This file implements the virtual machine that runs
235     prepared statements.  There are various helper files whose names
236     begin with "vdbe".  The VDBE has access to the vdbeInt.h header file
237     which defines internal data objects.  The rest of SQLite interacts
238     with the VDBE through an interface defined by vdbe.h.
239
240  *  **where.c** - This file analyzes the WHERE clause and generates
241     virtual machine code to run queries efficiently.  This file is
242     sometimes called the "query optimizer".  It has its own private
243     header file, whereInt.h, that defines data objects used internally.
244
245  *  **btree.c** - This file contains the implementation of the B-Tree
246     storage engine used by SQLite.
247
248  *  **pager.c** - This file contains the "pager" implementation, the
249     module that implements transactions.
250
251  *  **os_unix.c** and **os_win.c** - These two files implement the interface
252     between SQLite and the underlying operating system using the run-time
253     pluggable VFS interface.
254
255  *  **shell.c** - This file is not part of the core SQLite library.  This
256     is the file that, when linked against sqlite3.a, generates the
257     "sqlite3.exe" command-line shell.
258
259  *  **tclsqlite.c** - This file implements the Tcl bindings for SQLite.  It
260     is not part of the core SQLite library.  But as most of the tests in this
261     repository are written in Tcl, the Tcl language bindings are important.
262
263There are many other source files.  Each has a suscinct header comment that
264describes its purpose and role within the larger system.
265
266
267## Contacts
268
269The main SQLite webpage is [http://www.sqlite.org/](http://www.sqlite.org/)
270with geographically distributed backup servers at
271[http://www2.sqlite.org/](http://www2.sqlite.org) and
272[http://www3.sqlite.org/](http://www3.sqlite.org).
273