/ Check-in [a80ae2c9]
Login

Many hyperlinks are disabled.
Use anonymous login to enable hyperlinks.

Overview
Comment:Dozens and dozens of typo fixes in comments. This change adds no value to the end product and is disruptive, so it is questionable whether or not it will ever land on trunk.
Downloads: Tarball | ZIP archive | SQL archive
Timelines: family | ancestors | typos
Files: files | file ages | folders
SHA3-256: a80ae2c98b2dcf7aea93f7c175e5f4731d09b795ea0f7d052a5dac6401c68f90
User & Date: drh 2020-07-29 16:18:46
Context
2020-07-29
16:18
Dozens and dozens of typo fixes in comments. This change adds no value to the end product and is disruptive, so it is questionable whether or not it will ever land on trunk. (Leaf check-in: a80ae2c9 user: drh tags: typos)
2020-07-28
17:29
If a writer crashes in WAL mode and leave the SHM file in an inconsistent state, subsequent transactions are now able to recover the SHM file even if there are active read transactions. (check-in: ee8a1080 user: drh tags: trunk)
Changes
Hide Diffs Unified Diffs Ignore Whitespace Patch

Changes to Makefile.in.

82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
# based on configuration. (-DSQLITE_OMIT*, -DSQLITE_ENABLE*).
# The same set of OMIT and ENABLE flags should be passed to the
# LEMON parser generator and the mkkeywordhash tool as well.
OPT_FEATURE_FLAGS = @OPT_FEATURE_FLAGS@

TCC += $(OPT_FEATURE_FLAGS)

# Add in any optional parameters specified on the make commane line
# ie.  make "OPTS=-DSQLITE_ENABLE_FOO=1 -DSQLITE_OMIT_FOO=1".
TCC += $(OPTS)

# Add in compile-time options for some libraries used by extensions
TCC += @HAVE_ZLIB@

# Version numbers and release number for the SQLite being compiled.







|







82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
# based on configuration. (-DSQLITE_OMIT*, -DSQLITE_ENABLE*).
# The same set of OMIT and ENABLE flags should be passed to the
# LEMON parser generator and the mkkeywordhash tool as well.
OPT_FEATURE_FLAGS = @OPT_FEATURE_FLAGS@

TCC += $(OPT_FEATURE_FLAGS)

# Add in any optional parameters specified on the make command line
# ie.  make "OPTS=-DSQLITE_ENABLE_FOO=1 -DSQLITE_OMIT_FOO=1".
TCC += $(OPTS)

# Add in compile-time options for some libraries used by extensions
TCC += @HAVE_ZLIB@

# Version numbers and release number for the SQLite being compiled.

Changes to Makefile.msc.

1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043

# Add the required and optional SQLite compilation options into the command
# lines used to invoke the MSVC code and resource compilers.
#
TCC = $(TCC) $(REQ_FEATURE_FLAGS) $(OPT_FEATURE_FLAGS) $(EXT_FEATURE_FLAGS)
RCC = $(RCC) $(REQ_FEATURE_FLAGS) $(OPT_FEATURE_FLAGS) $(EXT_FEATURE_FLAGS)

# Add in any optional parameters specified on the commane line, e.g.
# nmake /f Makefile.msc all "OPTS=-DSQLITE_ENABLE_FOO=1 -DSQLITE_OMIT_FOO=1"
#
TCC = $(TCC) $(OPTS)
RCC = $(RCC) $(OPTS)

# If compiling for debugging, add some defines.
#







|







1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043

# Add the required and optional SQLite compilation options into the command
# lines used to invoke the MSVC code and resource compilers.
#
TCC = $(TCC) $(REQ_FEATURE_FLAGS) $(OPT_FEATURE_FLAGS) $(EXT_FEATURE_FLAGS)
RCC = $(RCC) $(REQ_FEATURE_FLAGS) $(OPT_FEATURE_FLAGS) $(EXT_FEATURE_FLAGS)

# Add in any optional parameters specified on the command line, e.g.
# nmake /f Makefile.msc all "OPTS=-DSQLITE_ENABLE_FOO=1 -DSQLITE_OMIT_FOO=1"
#
TCC = $(TCC) $(OPTS)
RCC = $(RCC) $(OPTS)

# If compiling for debugging, add some defines.
#

Changes to README.md.

11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
...
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
...
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
[Fossil](https://www.fossil-scm.org/), a distributed version control system
that was specifically designed and written to support SQLite development.
The [Fossil repository](https://sqlite.org/src/timeline) contains the urtext.

If you are reading this on GitHub or some other Git repository or service,
then you are looking at a mirror.  The names of check-ins and
other artifacts in a Git mirror are different from the official
names for those objects.  The offical names for check-ins are
found in a footer on the check-in comment for authorized mirrors.
The official check-in name can also be seen in the `manifest.uuid` file
in the root of the tree.  Always use the official name, not  the
Git-name, when communicating about an SQLite check-in.

If you pulled your SQLite source code from a secondary source and want to
verify its integrity, there are hints on how to do that in the
................................................................................
[command-line shell](https://sqlite.org/cli.html) and
the "tclsqlite.c" file which implements the
[Tcl bindings](https://sqlite.org/tclsqlite.html) for SQLite.
(Historical note:  SQLite began as a Tcl
extension and only later escaped to the wild as an independent library.)

Test scripts and programs are found in the **test/** subdirectory.
Addtional test code is found in other source repositories.
See [How SQLite Is Tested](http://www.sqlite.org/testing.html) for
additional information.

The **ext/** subdirectory contains code for extensions.  The
Full-text search engine is in **ext/fts3**.  The R-Tree engine is in
**ext/rtree**.  The **ext/misc** subdirectory contains a number of
smaller, single-file extensions, such as a REGEXP operator.
................................................................................
(helping to understand how SQLite works include the
[file format](http://www.sqlite.org/fileformat2.html) description,
the [virtual machine](http://www.sqlite.org/opcode.html) that runs
prepared statements, the description of
[how transactions work](http://www.sqlite.org/atomiccommit.html), and
the [overview of the query planner](http://www.sqlite.org/optoverview.html).

Years of effort have gone into optimizating SQLite, both
for small size and high performance.  And optimizations tend to result in
complex code.  So there is a lot of complexity in the current SQLite
implementation.  It will not be the easiest library in the world to hack.

Key files:

  *  **sqlite.h.in** - This file defines the public interface to the SQLite







|







 







|







 







|







11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
...
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
...
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
[Fossil](https://www.fossil-scm.org/), a distributed version control system
that was specifically designed and written to support SQLite development.
The [Fossil repository](https://sqlite.org/src/timeline) contains the urtext.

If you are reading this on GitHub or some other Git repository or service,
then you are looking at a mirror.  The names of check-ins and
other artifacts in a Git mirror are different from the official
names for those objects.  The official names for check-ins are
found in a footer on the check-in comment for authorized mirrors.
The official check-in name can also be seen in the `manifest.uuid` file
in the root of the tree.  Always use the official name, not  the
Git-name, when communicating about an SQLite check-in.

If you pulled your SQLite source code from a secondary source and want to
verify its integrity, there are hints on how to do that in the
................................................................................
[command-line shell](https://sqlite.org/cli.html) and
the "tclsqlite.c" file which implements the
[Tcl bindings](https://sqlite.org/tclsqlite.html) for SQLite.
(Historical note:  SQLite began as a Tcl
extension and only later escaped to the wild as an independent library.)

Test scripts and programs are found in the **test/** subdirectory.
Additional test code is found in other source repositories.
See [How SQLite Is Tested](http://www.sqlite.org/testing.html) for
additional information.

The **ext/** subdirectory contains code for extensions.  The
Full-text search engine is in **ext/fts3**.  The R-Tree engine is in
**ext/rtree**.  The **ext/misc** subdirectory contains a number of
smaller, single-file extensions, such as a REGEXP operator.
................................................................................
(helping to understand how SQLite works include the
[file format](http://www.sqlite.org/fileformat2.html) description,
the [virtual machine](http://www.sqlite.org/opcode.html) that runs
prepared statements, the description of
[how transactions work](http://www.sqlite.org/atomiccommit.html), and
the [overview of the query planner](http://www.sqlite.org/optoverview.html).

Years of effort have gone into optimizing SQLite, both
for small size and high performance.  And optimizations tend to result in
complex code.  So there is a lot of complexity in the current SQLite
implementation.  It will not be the easiest library in the world to hack.

Key files:

  *  **sqlite.h.in** - This file defines the public interface to the SQLite

Changes to aclocal.m4.

7965
7966
7967
7968
7969
7970
7971
7972
m4_ifndef([_LT_AC_LANG_F77_CONFIG],	[AC_DEFUN([_LT_AC_LANG_F77_CONFIG])])
m4_ifndef([AC_LIBTOOL_LANG_GCJ_CONFIG],	[AC_DEFUN([AC_LIBTOOL_LANG_GCJ_CONFIG])])
m4_ifndef([_LT_AC_LANG_GCJ_CONFIG],	[AC_DEFUN([_LT_AC_LANG_GCJ_CONFIG])])
m4_ifndef([AC_LIBTOOL_LANG_RC_CONFIG],	[AC_DEFUN([AC_LIBTOOL_LANG_RC_CONFIG])])
m4_ifndef([_LT_AC_LANG_RC_CONFIG],	[AC_DEFUN([_LT_AC_LANG_RC_CONFIG])])
m4_ifndef([AC_LIBTOOL_CONFIG],		[AC_DEFUN([AC_LIBTOOL_CONFIG])])
m4_ifndef([_LT_AC_FILE_LTDLL_C],	[AC_DEFUN([_LT_AC_FILE_LTDLL_C])])








<
7965
7966
7967
7968
7969
7970
7971

m4_ifndef([_LT_AC_LANG_F77_CONFIG],	[AC_DEFUN([_LT_AC_LANG_F77_CONFIG])])
m4_ifndef([AC_LIBTOOL_LANG_GCJ_CONFIG],	[AC_DEFUN([AC_LIBTOOL_LANG_GCJ_CONFIG])])
m4_ifndef([_LT_AC_LANG_GCJ_CONFIG],	[AC_DEFUN([_LT_AC_LANG_GCJ_CONFIG])])
m4_ifndef([AC_LIBTOOL_LANG_RC_CONFIG],	[AC_DEFUN([AC_LIBTOOL_LANG_RC_CONFIG])])
m4_ifndef([_LT_AC_LANG_RC_CONFIG],	[AC_DEFUN([_LT_AC_LANG_RC_CONFIG])])
m4_ifndef([AC_LIBTOOL_CONFIG],		[AC_DEFUN([AC_LIBTOOL_CONFIG])])
m4_ifndef([_LT_AC_FILE_LTDLL_C],	[AC_DEFUN([_LT_AC_FILE_LTDLL_C])])

Changes to autoconf/INSTALL.

363
364
365
366
367
368
369
370
`--no-create'
`-n'
     Run the configure checks, but stop before creating any output
     files.

`configure' also accepts some other, not widely useful, options.  Run
`configure --help' for more details.








<
363
364
365
366
367
368
369

`--no-create'
`-n'
     Run the configure checks, but stop before creating any output
     files.

`configure' also accepts some other, not widely useful, options.  Run
`configure --help' for more details.

Changes to autoconf/Makefile.msc.

772
773
774
775
776
777
778
779
780
781
782
783
784
785
786

# Add the required and optional SQLite compilation options into the command
# lines used to invoke the MSVC code and resource compilers.
#
TCC = $(TCC) $(REQ_FEATURE_FLAGS) $(OPT_FEATURE_FLAGS) $(EXT_FEATURE_FLAGS)
RCC = $(RCC) $(REQ_FEATURE_FLAGS) $(OPT_FEATURE_FLAGS) $(EXT_FEATURE_FLAGS)

# Add in any optional parameters specified on the commane line, e.g.
# nmake /f Makefile.msc all "OPTS=-DSQLITE_ENABLE_FOO=1 -DSQLITE_OMIT_FOO=1"
#
TCC = $(TCC) $(OPTS)
RCC = $(RCC) $(OPTS)

# If compiling for debugging, add some defines.
#







|







772
773
774
775
776
777
778
779
780
781
782
783
784
785
786

# Add the required and optional SQLite compilation options into the command
# lines used to invoke the MSVC code and resource compilers.
#
TCC = $(TCC) $(REQ_FEATURE_FLAGS) $(OPT_FEATURE_FLAGS) $(EXT_FEATURE_FLAGS)
RCC = $(RCC) $(REQ_FEATURE_FLAGS) $(OPT_FEATURE_FLAGS) $(EXT_FEATURE_FLAGS)

# Add in any optional parameters specified on the command line, e.g.
# nmake /f Makefile.msc all "OPTS=-DSQLITE_ENABLE_FOO=1 -DSQLITE_OMIT_FOO=1"
#
TCC = $(TCC) $(OPTS)
RCC = $(RCC) $(OPTS)

# If compiling for debugging, add some defines.
#

Changes to autoconf/README.txt.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
This package contains:

 * the SQLite library amalgamation source code file: sqlite3.c
 * the sqlite3.h and sqlite3ext.h header files that define the C-language
   interface to the sqlite3.c library file
 * the shell.c file used to build the sqlite3 command-line shell program
 * autoconf/automake installation infrastucture for building on POSIX
   compliant systems
 * a Makefile.msc, sqlite3.rc, and Replace.cs for building with Microsoft
   Visual C++ on Windows

SUMMARY OF HOW TO BUILD
=======================







|







1
2
3
4
5
6
7
8
9
10
11
12
13
14
This package contains:

 * the SQLite library amalgamation source code file: sqlite3.c
 * the sqlite3.h and sqlite3ext.h header files that define the C-language
   interface to the sqlite3.c library file
 * the shell.c file used to build the sqlite3 command-line shell program
 * autoconf/automake installation infrastructure for building on POSIX
   compliant systems
 * a Makefile.msc, sqlite3.rc, and Replace.cs for building with Microsoft
   Visual C++ on Windows

SUMMARY OF HOW TO BUILD
=======================

Changes to autoconf/tea/Makefile.in.

170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185

#========================================================================
# Start of user-definable TARGETS section
#========================================================================

#========================================================================
# TEA TARGETS.  Please note that the "libraries:" target refers to platform
# independent files, and the "binaries:" target inclues executable programs and
# platform-dependent libraries.  Modify these targets so that they install
# the various pieces of your package.  The make and install rules
# for the BINARIES that you specified above have already been done.
#========================================================================

all: binaries libraries doc

#========================================================================







|
|







170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185

#========================================================================
# Start of user-definable TARGETS section
#========================================================================

#========================================================================
# TEA TARGETS.  Please note that the "libraries:" target refers to platform
# independent files, and the "binaries:" target includes executable programs
# and platform-dependent libraries.  Modify these targets so that they install
# the various pieces of your package.  The make and install rules
# for the BINARIES that you specified above have already been done.
#========================================================================

all: binaries libraries doc

#========================================================================

Changes to autoconf/tea/configure.ac.

82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
TEA_ADD_TCL_SOURCES([])

#--------------------------------------------------------------------
# The --with-system-sqlite causes the TCL bindings to SQLite to use
# the system shared library for SQLite rather than statically linking
# against its own private copy.  This is dangerous and leads to
# undersirable dependences and is not recommended.
# Patchs from rmax.
#--------------------------------------------------------------------
AC_ARG_WITH([system-sqlite],
 [AC_HELP_STRING([--with-system-sqlite],
   [use a system-supplied libsqlite3 instead of the bundled one])],
 [], [with_system_sqlite=no])
if test x$with_system_sqlite != xno; then
 AC_CHECK_HEADER([sqlite3.h],







|







82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
TEA_ADD_TCL_SOURCES([])

#--------------------------------------------------------------------
# The --with-system-sqlite causes the TCL bindings to SQLite to use
# the system shared library for SQLite rather than statically linking
# against its own private copy.  This is dangerous and leads to
# undersirable dependences and is not recommended.
# Patches from rmax.
#--------------------------------------------------------------------
AC_ARG_WITH([system-sqlite],
 [AC_HELP_STRING([--with-system-sqlite],
   [use a system-supplied libsqlite3 instead of the bundled one])],
 [], [with_system_sqlite=no])
if test x$with_system_sqlite != xno; then
 AC_CHECK_HEADER([sqlite3.h],

Changes to autoconf/tea/win/nmakehlp.c.

200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
...
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354

    ZeroMemory(&sa, sizeof(SECURITY_ATTRIBUTES));
    sa.nLength = sizeof(SECURITY_ATTRIBUTES);
    sa.lpSecurityDescriptor = NULL;
    sa.bInheritHandle = FALSE;

    /*
     * Create a non-inheritible pipe.
     */

    CreatePipe(&Out.pipe, &h, &sa, 0);

    /*
     * Dupe the write side, make it inheritible, and close the original.
     */

    DuplicateHandle(hProcess, h, hProcess, &si.hStdOutput, 0, TRUE,
	    DUPLICATE_SAME_ACCESS | DUPLICATE_CLOSE_SOURCE);

    /*
     * Same as above, but for the error side.
................................................................................

    ZeroMemory(&sa, sizeof(SECURITY_ATTRIBUTES));
    sa.nLength = sizeof(SECURITY_ATTRIBUTES);
    sa.lpSecurityDescriptor = NULL;
    sa.bInheritHandle = TRUE;

    /*
     * Create a non-inheritible pipe.
     */

    CreatePipe(&Out.pipe, &h, &sa, 0);

    /*
     * Dupe the write side, make it inheritible, and close the original.
     */

    DuplicateHandle(hProcess, h, hProcess, &si.hStdOutput, 0, TRUE,
	    DUPLICATE_SAME_ACCESS | DUPLICATE_CLOSE_SOURCE);

    /*
     * Same as above, but for the error side.







|





|







 







|





|







200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
...
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354

    ZeroMemory(&sa, sizeof(SECURITY_ATTRIBUTES));
    sa.nLength = sizeof(SECURITY_ATTRIBUTES);
    sa.lpSecurityDescriptor = NULL;
    sa.bInheritHandle = FALSE;

    /*
     * Create a non-inheritable pipe.
     */

    CreatePipe(&Out.pipe, &h, &sa, 0);

    /*
     * Dupe the write side, make it inheritable, and close the original.
     */

    DuplicateHandle(hProcess, h, hProcess, &si.hStdOutput, 0, TRUE,
	    DUPLICATE_SAME_ACCESS | DUPLICATE_CLOSE_SOURCE);

    /*
     * Same as above, but for the error side.
................................................................................

    ZeroMemory(&sa, sizeof(SECURITY_ATTRIBUTES));
    sa.nLength = sizeof(SECURITY_ATTRIBUTES);
    sa.lpSecurityDescriptor = NULL;
    sa.bInheritHandle = TRUE;

    /*
     * Create a non-inheritable pipe.
     */

    CreatePipe(&Out.pipe, &h, &sa, 0);

    /*
     * Dupe the write side, make it inheritable, and close the original.
     */

    DuplicateHandle(hProcess, h, hProcess, &si.hStdOutput, 0, TRUE,
	    DUPLICATE_SAME_ACCESS | DUPLICATE_CLOSE_SOURCE);

    /*
     * Same as above, but for the error side.

Changes to autoconf/tea/win/rules.vc.

704
705
706
707
708
709
710
711
!message *** Optional defines are '$(OPTDEFINES)'
!message *** Compiler version $(VCVER). Target machine is $(MACHINE)
!message *** Host architecture is $(NATIVE_ARCH)
!message *** Compiler options '$(COMPILERFLAGS) $(OPTIMIZATIONS) $(DEBUGFLAGS) $(WARNINGS)'
!message *** Link options '$(LINKERFLAGS)'

!endif








<
704
705
706
707
708
709
710

!message *** Optional defines are '$(OPTDEFINES)'
!message *** Compiler version $(VCVER). Target machine is $(MACHINE)
!message *** Host architecture is $(NATIVE_ARCH)
!message *** Compiler options '$(COMPILERFLAGS) $(OPTIMIZATIONS) $(DEBUGFLAGS) $(WARNINGS)'
!message *** Link options '$(LINKERFLAGS)'

!endif

Changes to configure.

13917
13918
13919
13920
13921
13922
13923
13924
  # would make configure fail if this is the last instruction.
  $ac_cs_success || as_fn_exit 1
fi
if test -n "$ac_unrecognized_opts" && test "$enable_option_checking" != no; then
  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: unrecognized options: $ac_unrecognized_opts" >&5
$as_echo "$as_me: WARNING: unrecognized options: $ac_unrecognized_opts" >&2;}
fi








<
13917
13918
13919
13920
13921
13922
13923

  # would make configure fail if this is the last instruction.
  $ac_cs_success || as_fn_exit 1
fi
if test -n "$ac_unrecognized_opts" && test "$enable_option_checking" != no; then
  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: unrecognized options: $ac_unrecognized_opts" >&5
$as_echo "$as_me: WARNING: unrecognized options: $ac_unrecognized_opts" >&2;}
fi

Changes to contrib/sqlitecon.tcl.

442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
...
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
  if {$px==$ix} {
    $w mark set insert $p
  }
  $w mark set anchor $p
  focus $w
}

# Find the boundry between characters that is nearest
# to $x,$y
#
proc sqlitecon::nearestBoundry {w x y} {
  set p [$w index @$x,$y]
  set bb [$w bbox $p]
  if {![string compare $bb ""]} {return $p}
  if {($x-[lindex $bb 0])<([lindex $bb 2]/2)} {return $p}
................................................................................
proc sqlitecon::Cut w {
  if {[sqlitecon::canCut $w]==1} {
    sqlitecon::Copy $w
    $w delete sel.first sel.last
  }
}

# Do a paste opeation.
#
proc sqlitecon::Paste w {
  if {[sqlitecon::canCut $w]==1} {
    $w delete sel.first sel.last
  }
  if {[catch {selection get -displayof $w -selection CLIPBOARD} topaste]
    && [catch {selection get -displayof $w -selection PRIMARY} topaste]} {







|







 







|







442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
...
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
  if {$px==$ix} {
    $w mark set insert $p
  }
  $w mark set anchor $p
  focus $w
}

# Find the boundary between characters that is nearest
# to $x,$y
#
proc sqlitecon::nearestBoundry {w x y} {
  set p [$w index @$x,$y]
  set bb [$w bbox $p]
  if {![string compare $bb ""]} {return $p}
  if {($x-[lindex $bb 0])<([lindex $bb 2]/2)} {return $p}
................................................................................
proc sqlitecon::Cut w {
  if {[sqlitecon::canCut $w]==1} {
    sqlitecon::Copy $w
    $w delete sel.first sel.last
  }
}

# Do a paste operation.
#
proc sqlitecon::Paste w {
  if {[sqlitecon::canCut $w]==1} {
    $w delete sel.first sel.last
  }
  if {[catch {selection get -displayof $w -selection CLIPBOARD} topaste]
    && [catch {selection get -displayof $w -selection PRIMARY} topaste]} {

Changes to doc/F2FS.txt.

77
78
79
80
81
82
83
84
85
86
87

3. If the process crashes before the F2FS_IOC_COMMIT_ATOMIC_WRITE is
   completed then the file is automatically restored to the state that
   it was in before F2FS_IOC_START_ATOMIC_WRITE was called. This occurs
   before the posix advisory lock is automatically dropped - there is
   no chance that another client will be able to read the file in a
   half-committed state before the rollback operation occurs.











<
<
<
<
77
78
79
80
81
82
83





3. If the process crashes before the F2FS_IOC_COMMIT_ATOMIC_WRITE is
   completed then the file is automatically restored to the state that
   it was in before F2FS_IOC_START_ATOMIC_WRITE was called. This occurs
   before the posix advisory lock is automatically dropped - there is
   no chance that another client will be able to read the file in a
   half-committed state before the rollback operation occurs.




Changes to doc/lemon.html.

692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
...
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
is passed in on the ParseAlloc() or ParseInit() routines instead of
on Parse().

<a name='extractx'></a>
<h4>The <tt>%extra_context</tt> directive</h4>

The <tt>%extra_context</tt> directive instructs Lemon to add a 2nd parameter
to the parameter list of the ParseAlloc() and ParseInif() functions.  Lemon
doesn't do anything itself with these extra argument, but it does
store the value make it available to C-code action routines, destructors,
and so forth.  For example, if the grammar file contains:</p>

<p><pre>
    %extra_context { MyStruct *pAbc }
</pre></p>
................................................................................
just not as general.
Each of these directives must begin at the left margin.  No whitespace
is allowed between the "%" and the directive name.</p>

<p>Grammar text in between "<tt>%ifdef MACRO</tt>" and the next nested
"<tt>%endif</tt>" is
ignored unless the "-DMACRO" command-line option is used.  Grammar text
betwen "<tt>%ifndef MACRO</tt>" and the next nested "<tt>%endif</tt>" is
included except when the "-DMACRO" command-line option is used.<p>

<p>The text in between "<tt>%if</tt> <i>CONDITIONAL</i>" and its
corresponding <tt>%endif</tt> is included only if <i>CONDITIONAL</i>
is true.  The CONDITION is one or more macro names, optionally connected
using the "||" and "&amp;&amp;" binary operators, the "!" unary operator,
and grouped using balanced parentheses.  Each term is true if the







|







 







|







692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
...
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
is passed in on the ParseAlloc() or ParseInit() routines instead of
on Parse().

<a name='extractx'></a>
<h4>The <tt>%extra_context</tt> directive</h4>

The <tt>%extra_context</tt> directive instructs Lemon to add a 2nd parameter
to the parameter list of the ParseAlloc() and ParseInit() functions.  Lemon
doesn't do anything itself with these extra argument, but it does
store the value make it available to C-code action routines, destructors,
and so forth.  For example, if the grammar file contains:</p>

<p><pre>
    %extra_context { MyStruct *pAbc }
</pre></p>
................................................................................
just not as general.
Each of these directives must begin at the left margin.  No whitespace
is allowed between the "%" and the directive name.</p>

<p>Grammar text in between "<tt>%ifdef MACRO</tt>" and the next nested
"<tt>%endif</tt>" is
ignored unless the "-DMACRO" command-line option is used.  Grammar text
between "<tt>%ifndef MACRO</tt>" and the next nested "<tt>%endif</tt>" is
included except when the "-DMACRO" command-line option is used.<p>

<p>The text in between "<tt>%if</tt> <i>CONDITIONAL</i>" and its
corresponding <tt>%endif</tt> is included only if <i>CONDITIONAL</i>
is true.  The CONDITION is one or more macro names, optionally connected
using the "||" and "&amp;&amp;" binary operators, the "!" unary operator,
and grouped using balanced parentheses.  Each term is true if the

Changes to doc/pager-invariants.txt.

1
2
3
4
5
6
7
8
9
10
11
12
13
..
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
..
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
 *** Throughout this document, a page is deemed to have been synced
     automatically as soon as it is written when PRAGMA synchronous=OFF.
     Otherwise, the page is not synced until the xSync method of the VFS
     is called successfully on the file containing the page.

 *** Definition:  A page of the database file is said to be "overwriteable" if
     one or more of the following are true about the page:
 
     (a)  The original content of the page as it was at the beginning of
          the transaction has been written into the rollback journal and
          synced.
 
     (b)  The page was a freelist leaf page at the start of the transaction.
................................................................................
 
     (c)  The page number is greater than the largest page that existed in
          the database file at the start of the transaction.
 
 (1) A page of the database file is never overwritten unless one of the
     following are true:
 
     (a) The page and all other pages on the same sector are overwriteable.
 
     (b) The atomic page write optimization is enabled, and the entire
         transaction other than the update of the transaction sequence
         number consists of a single page change.
 
 (2) The content of a page written into the rollback journal exactly matches
     both the content in the database when the rollback journal was written
................................................................................
 
 (6) If a master journal file is used, then all writes to the database file
     are synced prior to the master journal being deleted.
 
 *** Definition: Two databases (or the same database at two points it time)
     are said to be "logically equivalent" if they give the same answer to
     all queries.  Note in particular the content of freelist leaf
     pages can be changed arbitarily without effecting the logical equivalence
     of the database.
 
 (7) At any time, if any subset, including the empty set and the total set,
     of the unsynced changes to a rollback journal are removed and the 
     journal is rolled back, the resulting database file will be logical
     equivalent to the database file at the beginning of the transaction.
 





|







 







|







 







|







1
2
3
4
5
6
7
8
9
10
11
12
13
..
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
..
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
 *** Throughout this document, a page is deemed to have been synced
     automatically as soon as it is written when PRAGMA synchronous=OFF.
     Otherwise, the page is not synced until the xSync method of the VFS
     is called successfully on the file containing the page.

 *** Definition:  A page of the database file is said to be "overwritable" if
     one or more of the following are true about the page:
 
     (a)  The original content of the page as it was at the beginning of
          the transaction has been written into the rollback journal and
          synced.
 
     (b)  The page was a freelist leaf page at the start of the transaction.
................................................................................
 
     (c)  The page number is greater than the largest page that existed in
          the database file at the start of the transaction.
 
 (1) A page of the database file is never overwritten unless one of the
     following are true:
 
     (a) The page and all other pages on the same sector are overwritable.
 
     (b) The atomic page write optimization is enabled, and the entire
         transaction other than the update of the transaction sequence
         number consists of a single page change.
 
 (2) The content of a page written into the rollback journal exactly matches
     both the content in the database when the rollback journal was written
................................................................................
 
 (6) If a master journal file is used, then all writes to the database file
     are synced prior to the master journal being deleted.
 
 *** Definition: Two databases (or the same database at two points it time)
     are said to be "logically equivalent" if they give the same answer to
     all queries.  Note in particular the content of freelist leaf
     pages can be changed arbitrarily without effecting the logical equivalence
     of the database.
 
 (7) At any time, if any subset, including the empty set and the total set,
     of the unsynced changes to a rollback journal are removed and the 
     journal is rolled back, the resulting database file will be logical
     equivalent to the database file at the beginning of the transaction.
 

Changes to doc/vfs-shm.txt.

54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
   RECOVER     - Held during wal-index recovery.  Used to prevent a
                 race if multiple clients try to recover a wal-index at
                 the same time.
   

A particular lock manager implementation may coalesce one or more of 
the wal-index locking states, though with a reduction in concurrency.
For example, an implemention might implement only exclusive locking,
in which case all states would be equivalent to CHECKPOINT, meaning that 
only one reader or one writer or one checkpointer could be active at a 
time.  Or, an implementation might combine READ and READ_FULL into 
a single state equivalent to READ, meaning that a writer could
coexist with a reader, but no reader or writers could coexist with a
checkpointer.








|







54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
   RECOVER     - Held during wal-index recovery.  Used to prevent a
                 race if multiple clients try to recover a wal-index at
                 the same time.
   

A particular lock manager implementation may coalesce one or more of 
the wal-index locking states, though with a reduction in concurrency.
For example, an implementation might implement only exclusive locking,
in which case all states would be equivalent to CHECKPOINT, meaning that 
only one reader or one writer or one checkpointer could be active at a 
time.  Or, an implementation might combine READ and READ_FULL into 
a single state equivalent to READ, meaning that a writer could
coexist with a reader, but no reader or writers could coexist with a
checkpointer.

Changes to doc/wal-lock.md.

80
81
82
83
84
85
86
87
88
  * when an open read-transaction is upgraded to a write-transaction.

In all other cases the blocking locks implementation should prevent clients
from having to handle SQLITE\_BUSY errors and facilitate appropriate transfer
of priorities between competing clients.

Clients that lock multiple databases simultaneously must be wary of deadlock.









<
<
80
81
82
83
84
85
86


  * when an open read-transaction is upgraded to a write-transaction.

In all other cases the blocking locks implementation should prevent clients
from having to handle SQLITE\_BUSY errors and facilitate appropriate transfer
of priorities between competing clients.

Clients that lock multiple databases simultaneously must be wary of deadlock.


Changes to ext/async/README.txt.

43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
  your program crashes or if a power loss occurs after the database
  write but before the asynchronous write thread has completed, then the
  database change might never make it to disk and the next user of the
  database might not see your change.

  You lose Durability with asynchronous I/O, but you still retain the
  other parts of ACID:  Atomic,  Consistent, and Isolated.  Many
  appliations get along fine without the Durablity.

  1.1 How it Works

    Asynchronous I/O works by creating a special SQLite "vfs" structure
    and registering it with sqlite3_vfs_register(). When files opened via 
    this vfs are written to (using the vfs xWrite() method), the data is not 
    written directly to disk, but is placed in the "write-queue" to be







|







43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
  your program crashes or if a power loss occurs after the database
  write but before the asynchronous write thread has completed, then the
  database change might never make it to disk and the next user of the
  database might not see your change.

  You lose Durability with asynchronous I/O, but you still retain the
  other parts of ACID:  Atomic,  Consistent, and Isolated.  Many
  applications get along fine without the Durability.

  1.1 How it Works

    Asynchronous I/O works by creating a special SQLite "vfs" structure
    and registering it with sqlite3_vfs_register(). When files opened via 
    this vfs are written to (using the vfs xWrite() method), the data is not 
    written directly to disk, but is placed in the "write-queue" to be

Changes to ext/async/sqlite3async.c.

1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
....
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
** If async.writerHaltNow is true, then this procedure exits
** after processing a single message.
**
** If async.writerHaltWhenIdle is true, then this procedure exits when
** the write queue is empty.
**
** If both of the above variables are false, this procedure runs
** indefinately, waiting for operations to be added to the write queue
** and processing them in the order in which they arrive.
**
** An artifical delay of async.ioDelay milliseconds is inserted before
** each write operation in order to simulate the effect of a slow disk.
**
** Only one instance of this procedure may be running at a time.
*/
static void asyncWriterThread(void){
  sqlite3_vfs *pVfs = (sqlite3_vfs *)(async_vfs.pAppData);
  AsyncWrite *p = 0;
................................................................................
    ** perform the IO, and then re-request the mutex before removing 'p' from
    ** the head of the write-op queue. The idea is to increase concurrency with
    ** sqlite threads.
    **
    **     * An ASYNC_CLOSE operation.
    **     * An ASYNC_OPENEXCLUSIVE operation. For this one, we relinquish 
    **       the mutex, call the underlying xOpenExclusive() function, then
    **       re-aquire the mutex before seting the AsyncFile.pBaseRead 
    **       variable.
    **     * ASYNC_SYNC and ASYNC_WRITE operations, if 
    **       SQLITE_ASYNC_TWO_FILEHANDLES was set at compile time and two
    **       file-handles are open for the particular file being "synced".
    */
    if( async.ioError!=SQLITE_OK && p->op!=ASYNC_CLOSE ){
      p->op = ASYNC_NOOP;







|


|







 







|







1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
....
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
** If async.writerHaltNow is true, then this procedure exits
** after processing a single message.
**
** If async.writerHaltWhenIdle is true, then this procedure exits when
** the write queue is empty.
**
** If both of the above variables are false, this procedure runs
** indefinitely, waiting for operations to be added to the write queue
** and processing them in the order in which they arrive.
**
** An artificial delay of async.ioDelay milliseconds is inserted before
** each write operation in order to simulate the effect of a slow disk.
**
** Only one instance of this procedure may be running at a time.
*/
static void asyncWriterThread(void){
  sqlite3_vfs *pVfs = (sqlite3_vfs *)(async_vfs.pAppData);
  AsyncWrite *p = 0;
................................................................................
    ** perform the IO, and then re-request the mutex before removing 'p' from
    ** the head of the write-op queue. The idea is to increase concurrency with
    ** sqlite threads.
    **
    **     * An ASYNC_CLOSE operation.
    **     * An ASYNC_OPENEXCLUSIVE operation. For this one, we relinquish 
    **       the mutex, call the underlying xOpenExclusive() function, then
    **       re-acquire the mutex before setting the AsyncFile.pBaseRead 
    **       variable.
    **     * ASYNC_SYNC and ASYNC_WRITE operations, if 
    **       SQLITE_ASYNC_TWO_FILEHANDLES was set at compile time and two
    **       file-handles are open for the particular file being "synced".
    */
    if( async.ioError!=SQLITE_OK && p->op!=ASYNC_CLOSE ){
      p->op = ASYNC_NOOP;

Changes to ext/async/sqlite3async.h.

57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
**
** If a parent VFS cannot be located, then SQLITE_ERROR is returned.
** In the unlikely event that operating system specific initialization
** fails (win32 systems create the required critical section and event 
** objects within this function), then SQLITE_ERROR is also returned.
** Finally, if the call to sqlite3_vfs_register() returns an error, then 
** the error code is returned to the user by this function. In all three
** of these cases, intialization has failed and the asynchronous IO VFS
** is not registered with SQLite.
**
** Otherwise, if no error occurs, SQLITE_OK is returned.
*/ 
int sqlite3async_initialize(const char *zParent, int isDefault);

/*







|







57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
**
** If a parent VFS cannot be located, then SQLITE_ERROR is returned.
** In the unlikely event that operating system specific initialization
** fails (win32 systems create the required critical section and event 
** objects within this function), then SQLITE_ERROR is also returned.
** Finally, if the call to sqlite3_vfs_register() returns an error, then 
** the error code is returned to the user by this function. In all three
** of these cases, initialization has failed and the asynchronous IO VFS
** is not registered with SQLite.
**
** Otherwise, if no error occurs, SQLITE_OK is returned.
*/ 
int sqlite3async_initialize(const char *zParent, int isDefault);

/*

Changes to ext/expert/sqlite3expert.c.

1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
  }

  /* Register the auth callback with dbv */
  if( rc==SQLITE_OK ){
    sqlite3_set_authorizer(pNew->dbv, idxAuthCallback, (void*)pNew);
  }

  /* If an error has occurred, free the new object and reutrn NULL. Otherwise,
  ** return the new sqlite3expert handle.  */
  if( rc!=SQLITE_OK ){
    sqlite3_expert_destroy(pNew);
    pNew = 0;
  }
  return pNew;
}







|







1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
  }

  /* Register the auth callback with dbv */
  if( rc==SQLITE_OK ){
    sqlite3_set_authorizer(pNew->dbv, idxAuthCallback, (void*)pNew);
  }

  /* If an error has occurred, free the new object and return NULL. Otherwise,
  ** return the new sqlite3expert handle.  */
  if( rc!=SQLITE_OK ){
    sqlite3_expert_destroy(pNew);
    pNew = 0;
  }
  return pNew;
}

Changes to ext/expert/sqlite3expert.h.

132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
** along with such an EXPERT_REPORT_* parameter, NULL is always returned.
**
** EXPERT_REPORT_SQL:
**   Return the text of SQL statement iStmt.
**
** EXPERT_REPORT_INDEXES:
**   Return a buffer containing the CREATE INDEX statements for all recommended
**   indexes for statement iStmt. If there are no new recommeded indexes, NULL 
**   is returned.
**
** EXPERT_REPORT_PLAN:
**   Return a buffer containing the EXPLAIN QUERY PLAN output for SQL query
**   iStmt after the proposed indexes have been added to the database schema.
**
** EXPERT_REPORT_CANDIDATES:







|







132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
** along with such an EXPERT_REPORT_* parameter, NULL is always returned.
**
** EXPERT_REPORT_SQL:
**   Return the text of SQL statement iStmt.
**
** EXPERT_REPORT_INDEXES:
**   Return a buffer containing the CREATE INDEX statements for all recommended
**   indexes for statement iStmt. If there are no new recommended indexes, NULL 
**   is returned.
**
** EXPERT_REPORT_PLAN:
**   Return a buffer containing the EXPLAIN QUERY PLAN output for SQL query
**   iStmt after the proposed indexes have been added to the database schema.
**
** EXPERT_REPORT_CANDIDATES:

Changes to ext/fts1/ft_hash.c.

139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
...
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
  if( n1!=n2 ) return 1;
  return memcmp(pKey1,pKey2,n1);
}

/*
** Return a pointer to the appropriate hash function given the key class.
**
** The C syntax in this function definition may be unfamilar to some 
** programmers, so we provide the following additional explanation:
**
** The name of the function is "hashFunction".  The function takes a
** single parameter "keyClass".  The return value of hashFunction()
** is a pointer to another function.  Specifically, the return value
** of hashFunction() is a pointer to a function that takes two parameters
** with types "const void*" and "int" and returns an "int".
................................................................................
    pH->first = pNew;
  }
  pEntry->count++;
  pEntry->chain = pNew;
}


/* Resize the hash table so that it cantains "new_size" buckets.
** "new_size" must be a power of 2.  The hash table might fail 
** to resize if sqliteMalloc() fails.
*/
static void rehash(Hash *pH, int new_size){
  struct _ht *new_ht;            /* The new hash table */
  HashElem *elem, *next_elem;    /* For looping over existing elements */
  int (*xHash)(const void*,int); /* The hash function */







|







 







|







139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
...
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
  if( n1!=n2 ) return 1;
  return memcmp(pKey1,pKey2,n1);
}

/*
** Return a pointer to the appropriate hash function given the key class.
**
** The C syntax in this function definition may be unfamiliar to some 
** programmers, so we provide the following additional explanation:
**
** The name of the function is "hashFunction".  The function takes a
** single parameter "keyClass".  The return value of hashFunction()
** is a pointer to another function.  Specifically, the return value
** of hashFunction() is a pointer to a function that takes two parameters
** with types "const void*" and "int" and returns an "int".
................................................................................
    pH->first = pNew;
  }
  pEntry->count++;
  pEntry->chain = pNew;
}


/* Resize the hash table so that it contains "new_size" buckets.
** "new_size" must be a power of 2.  The hash table might fail 
** to resize if sqliteMalloc() fails.
*/
static void rehash(Hash *pH, int new_size){
  struct _ht *new_ht;            /* The new hash table */
  HashElem *elem, *next_elem;    /* For looping over existing elements */
  int (*xHash)(const void*,int); /* The hash function */

Changes to ext/fts1/fts1.c.

1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
....
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
....
2582
2583
2584
2585
2586
2587
2588
2589
2590
2591
2592
2593
2594
2595
2596
....
2656
2657
2658
2659
2660
2661
2662
2663
2664
2665
2666
2667
2668
2669
2670
....
2950
2951
2952
2953
2954
2955
2956
2957
2958
2959
2960
2961
2962
2963
2964
** IdChar(X) will be true.  Otherwise it is false.
**
** For ASCII, any character with the high-order bit set is
** allowed in an identifier.  For 7-bit characters, 
** sqlite3IsIdChar[X] must be 1.
**
** Ticket #1066.  the SQL standard does not allow '$' in the
** middle of identfiers.  But many SQL implementations do. 
** SQLite will allow '$' in identifiers for compatibility.
** But the feature is undocumented.
*/
static const char isIdChar[] = {
/* x0 x1 x2 x3 x4 x5 x6 x7 x8 x9 xA xB xC xD xE xF */
    0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,  /* 2x */
    1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0,  /* 3x */
................................................................................
/*
** The input azIn is a NULL-terminated list of tokens.  Remove the first
** token and all punctuation tokens.  Remove the quotes from
** around string literal tokens.
**
** Example:
**
**     input:      tokenize chinese ( 'simplifed' , 'mixed' )
**     output:     chinese simplifed mixed
**
** Another example:
**
**     input:      delimiters ( '[' , ']' , '...' )
**     output:     [ ] ...
*/
static void tokenListToIdList(char **azIn){
................................................................................
** is the first term of a phrase query, go ahead and evaluate the phrase
** query and return the doclist for the entire phrase query.
**
** The result is stored in pTerm->doclist.
*/
static int docListOfTerm(
  fulltext_vtab *v,     /* The full text index */
  int iColumn,          /* column to restrict to.  No restrition if >=nColumn */
  QueryTerm *pQTerm,    /* Term we are looking for, or 1st term of a phrase */
  DocList **ppResult    /* Write the result here */
){
  DocList *pLeft, *pRight, *pNew;
  int i, rc;

  pLeft = docListNew(DL_POSITIONS);
................................................................................
    }
  }
  return -1;
}

/*
** Parse the text at pSegment[0..nSegment-1].  Add additional terms
** to the query being assemblied in pQuery.
**
** inPhrase is true if pSegment[0..nSegement-1] is contained within
** double-quotes.  If inPhrase is true, then the first term
** is marked with the number of terms in the phrase less one and
** OR and "-" syntax is ignored.  If inPhrase is false, then every
** term found is marked with nPhrase=0 and OR and "-" syntax is significant.
*/
................................................................................
    */
    sqlite3_result_blob(pContext, &c, sizeof(c), SQLITE_TRANSIENT);
  }
  return SQLITE_OK;
}

/* This is the xRowid method.  The SQLite core calls this routine to
** retrive the rowid for the current row of the result set.  The
** rowid should be written to *pRowid.
*/
static int fulltextRowid(sqlite3_vtab_cursor *pCursor, sqlite_int64 *pRowid){
  fulltext_cursor *c = (fulltext_cursor *) pCursor;

  *pRowid = sqlite3_column_int64(c->pStmt, 0);
  return SQLITE_OK;







|







 







|
|







 







|







 







|







 







|







1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
....
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
....
2582
2583
2584
2585
2586
2587
2588
2589
2590
2591
2592
2593
2594
2595
2596
....
2656
2657
2658
2659
2660
2661
2662
2663
2664
2665
2666
2667
2668
2669
2670
....
2950
2951
2952
2953
2954
2955
2956
2957
2958
2959
2960
2961
2962
2963
2964
** IdChar(X) will be true.  Otherwise it is false.
**
** For ASCII, any character with the high-order bit set is
** allowed in an identifier.  For 7-bit characters, 
** sqlite3IsIdChar[X] must be 1.
**
** Ticket #1066.  the SQL standard does not allow '$' in the
** middle of identifiers.  But many SQL implementations do. 
** SQLite will allow '$' in identifiers for compatibility.
** But the feature is undocumented.
*/
static const char isIdChar[] = {
/* x0 x1 x2 x3 x4 x5 x6 x7 x8 x9 xA xB xC xD xE xF */
    0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,  /* 2x */
    1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0,  /* 3x */
................................................................................
/*
** The input azIn is a NULL-terminated list of tokens.  Remove the first
** token and all punctuation tokens.  Remove the quotes from
** around string literal tokens.
**
** Example:
**
**     input:      tokenize chinese ( 'simplified' , 'mixed' )
**     output:     chinese simplified mixed
**
** Another example:
**
**     input:      delimiters ( '[' , ']' , '...' )
**     output:     [ ] ...
*/
static void tokenListToIdList(char **azIn){
................................................................................
** is the first term of a phrase query, go ahead and evaluate the phrase
** query and return the doclist for the entire phrase query.
**
** The result is stored in pTerm->doclist.
*/
static int docListOfTerm(
  fulltext_vtab *v,     /* The full text index */
  int iColumn,          /* column to restrict to.  No restriction if >=nColumn */
  QueryTerm *pQTerm,    /* Term we are looking for, or 1st term of a phrase */
  DocList **ppResult    /* Write the result here */
){
  DocList *pLeft, *pRight, *pNew;
  int i, rc;

  pLeft = docListNew(DL_POSITIONS);
................................................................................
    }
  }
  return -1;
}

/*
** Parse the text at pSegment[0..nSegment-1].  Add additional terms
** to the query being assembled in pQuery.
**
** inPhrase is true if pSegment[0..nSegement-1] is contained within
** double-quotes.  If inPhrase is true, then the first term
** is marked with the number of terms in the phrase less one and
** OR and "-" syntax is ignored.  If inPhrase is false, then every
** term found is marked with nPhrase=0 and OR and "-" syntax is significant.
*/
................................................................................
    */
    sqlite3_result_blob(pContext, &c, sizeof(c), SQLITE_TRANSIENT);
  }
  return SQLITE_OK;
}

/* This is the xRowid method.  The SQLite core calls this routine to
** retrieve the rowid for the current row of the result set.  The
** rowid should be written to *pRowid.
*/
static int fulltextRowid(sqlite3_vtab_cursor *pCursor, sqlite_int64 *pRowid){
  fulltext_cursor *c = (fulltext_cursor *) pCursor;

  *pRowid = sqlite3_column_int64(c->pStmt, 0);
  return SQLITE_OK;

Changes to ext/fts1/fts1_hash.c.

119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
...
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
  if( n1!=n2 ) return 1;
  return memcmp(pKey1,pKey2,n1);
}

/*
** Return a pointer to the appropriate hash function given the key class.
**
** The C syntax in this function definition may be unfamilar to some 
** programmers, so we provide the following additional explanation:
**
** The name of the function is "hashFunction".  The function takes a
** single parameter "keyClass".  The return value of hashFunction()
** is a pointer to another function.  Specifically, the return value
** of hashFunction() is a pointer to a function that takes two parameters
** with types "const void*" and "int" and returns an "int".
................................................................................
    pH->first = pNew;
  }
  pEntry->count++;
  pEntry->chain = pNew;
}


/* Resize the hash table so that it cantains "new_size" buckets.
** "new_size" must be a power of 2.  The hash table might fail 
** to resize if sqliteMalloc() fails.
*/
static void rehash(fts1Hash *pH, int new_size){
  struct _fts1ht *new_ht;          /* The new hash table */
  fts1HashElem *elem, *next_elem;  /* For looping over existing elements */
  int (*xHash)(const void*,int);   /* The hash function */







|







 







|







119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
...
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
  if( n1!=n2 ) return 1;
  return memcmp(pKey1,pKey2,n1);
}

/*
** Return a pointer to the appropriate hash function given the key class.
**
** The C syntax in this function definition may be unfamiliar to some 
** programmers, so we provide the following additional explanation:
**
** The name of the function is "hashFunction".  The function takes a
** single parameter "keyClass".  The return value of hashFunction()
** is a pointer to another function.  Specifically, the return value
** of hashFunction() is a pointer to a function that takes two parameters
** with types "const void*" and "int" and returns an "int".
................................................................................
    pH->first = pNew;
  }
  pEntry->count++;
  pEntry->chain = pNew;
}


/* Resize the hash table so that it contains "new_size" buckets.
** "new_size" must be a power of 2.  The hash table might fail 
** to resize if sqliteMalloc() fails.
*/
static void rehash(fts1Hash *pH, int new_size){
  struct _fts1ht *new_ht;          /* The new hash table */
  fts1HashElem *elem, *next_elem;  /* For looping over existing elements */
  int (*xHash)(const void*,int);   /* The hash function */

Changes to ext/fts1/fts1_porter.c.

139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
...
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
...
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
**
** A consonate is any letter other than 'a', 'e', 'i', 'o', or 'u'.
** 'Y' is a consonant unless it follows another consonant,
** in which case it is a vowel.
**
** In these routine, the letters are in reverse order.  So the 'y' rule
** is that 'y' is a consonant unless it is followed by another
** consonent.
*/
static int isVowel(const char*);
static int isConsonant(const char *z){
  int j;
  char x = *z;
  if( x==0 ) return 0;
  assert( x>='a' && x<='z' );
................................................................................
*/
static int doubleConsonant(const char *z){
  return isConsonant(z) && z[0]==z[1] && isConsonant(z+1);
}

/*
** Return TRUE if the word ends with three letters which
** are consonant-vowel-consonent and where the final consonant
** is not 'w', 'x', or 'y'.
**
** The word is reversed here.  So we are really checking the
** first three letters and the first one cannot be in [wxy].
*/
static int star_oh(const char *z){
  return
................................................................................
    z[0]!='w' && z[0]!='x' && z[0]!='y' &&
    z[1]!=0 && isVowel(z+1) &&
    z[2]!=0 && isConsonant(z+2);
}

/*
** If the word ends with zFrom and xCond() is true for the stem
** of the word that preceeds the zFrom ending, then change the 
** ending to zTo.
**
** The input word *pz and zFrom are both in reverse order.  zTo
** is in normal order. 
**
** Return TRUE if zFrom matches.  Return FALSE if zFrom does not
** match.  Not that TRUE is returned even if xCond() fails and







|







 







|







 







|







139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
...
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
...
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
**
** A consonate is any letter other than 'a', 'e', 'i', 'o', or 'u'.
** 'Y' is a consonant unless it follows another consonant,
** in which case it is a vowel.
**
** In these routine, the letters are in reverse order.  So the 'y' rule
** is that 'y' is a consonant unless it is followed by another
** consonant.
*/
static int isVowel(const char*);
static int isConsonant(const char *z){
  int j;
  char x = *z;
  if( x==0 ) return 0;
  assert( x>='a' && x<='z' );
................................................................................
*/
static int doubleConsonant(const char *z){
  return isConsonant(z) && z[0]==z[1] && isConsonant(z+1);
}

/*
** Return TRUE if the word ends with three letters which
** are consonant-vowel-consonant and where the final consonant
** is not 'w', 'x', or 'y'.
**
** The word is reversed here.  So we are really checking the
** first three letters and the first one cannot be in [wxy].
*/
static int star_oh(const char *z){
  return
................................................................................
    z[0]!='w' && z[0]!='x' && z[0]!='y' &&
    z[1]!=0 && isVowel(z+1) &&
    z[2]!=0 && isConsonant(z+2);
}

/*
** If the word ends with zFrom and xCond() is true for the stem
** of the word that precedes the zFrom ending, then change the 
** ending to zTo.
**
** The input word *pz and zFrom are both in reverse order.  zTo
** is in normal order. 
**
** Return TRUE if zFrom matches.  Return FALSE if zFrom does not
** match.  Not that TRUE is returned even if xCond() fails and

Changes to ext/fts2/fts2.c.

101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
....
2407
2408
2409
2410
2411
2412
2413
2414
2415
2416
2417
2418
2419
2420
2421
....
2576
2577
2578
2579
2580
2581
2582
2583
2584
2585
2586
2587
2588
2589
2590
2591
....
3519
3520
3521
3522
3523
3524
3525
3526
3527
3528
3529
3530
3531
3532
3533
....
3870
3871
3872
3873
3874
3875
3876
3877
3878
3879
3880
3881
3882
3883
3884
....
4644
4645
4646
4647
4648
4649
4650
4651
4652
4653
4654
4655
4656
4657
4658
....
6014
6015
6016
6017
6018
6019
6020
6021
6022
6023
6024
6025
6026
6027
6028
**   varint POS_END;        (marks end of positions for this document.
** }
**
** Here, array { X } means zero or more occurrences of X, adjacent in
** memory.  A "position" is an index of a token in the token stream
** generated by the tokenizer, while an "offset" is a byte offset,
** both based at 0.  Note that POS_END and POS_COLUMN occur in the
** same logical place as the position element, and act as sentinals
** ending a position list array.
**
** A DL_POSITIONS doclist omits the startOffset and endOffset
** information.  A DL_DOCIDS doclist omits both the position and
** offset information, becoming an array of varint-encoded docids.
**
** On-disk data is stored as type DL_DEFAULT, so we don't serialize
................................................................................
** IdChar(X) will be true.  Otherwise it is false.
**
** For ASCII, any character with the high-order bit set is
** allowed in an identifier.  For 7-bit characters, 
** sqlite3IsIdChar[X] must be 1.
**
** Ticket #1066.  the SQL standard does not allow '$' in the
** middle of identfiers.  But many SQL implementations do. 
** SQLite will allow '$' in identifiers for compatibility.
** But the feature is undocumented.
*/
static const char isIdChar[] = {
/* x0 x1 x2 x3 x4 x5 x6 x7 x8 x9 xA xB xC xD xE xF */
    0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,  /* 2x */
    1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0,  /* 3x */
................................................................................
/*
** The input azIn is a NULL-terminated list of tokens.  Remove the first
** token and all punctuation tokens.  Remove the quotes from
** around string literal tokens.
**
** Example:
**
**     input:      tokenize chinese ( 'simplifed' , 'mixed' )
**     output:     chinese simplifed mixed
**
** Another example:
**
**     input:      delimiters ( '[' , ']' , '...' )
**     output:     [ ] ...
*/
static void tokenListToIdList(char **azIn){
................................................................................
    }
  }
  return -1;
}

/*
** Parse the text at pSegment[0..nSegment-1].  Add additional terms
** to the query being assemblied in pQuery.
**
** inPhrase is true if pSegment[0..nSegement-1] is contained within
** double-quotes.  If inPhrase is true, then the first term
** is marked with the number of terms in the phrase less one and
** OR and "-" syntax is ignored.  If inPhrase is false, then every
** term found is marked with nPhrase=0 and OR and "-" syntax is significant.
*/
................................................................................
    */
    sqlite3_result_blob(pContext, &c, sizeof(c), SQLITE_TRANSIENT);
  }
  return SQLITE_OK;
}

/* This is the xRowid method.  The SQLite core calls this routine to
** retrive the rowid for the current row of the result set.  The
** rowid should be written to *pRowid.
*/
static int fulltextRowid(sqlite3_vtab_cursor *pCursor, sqlite_int64 *pRowid){
  fulltext_cursor *c = (fulltext_cursor *) pCursor;

  *pRowid = sqlite3_column_int64(c->pStmt, 0);
  return SQLITE_OK;
................................................................................
    int rc = leafWriterFlush(v, pWriter);
    if( rc!=SQLITE_OK ) return rc;
  }

  /* We must have flushed a leaf at some point. */
  assert( pWriter->has_parent );

  /* Tenatively set the end leaf blockid as the end blockid.  If the
  ** interior node can be returned inline, this will be the final
  ** blockid, otherwise it will be overwritten by
  ** interiorWriterRootInfo().
  */
  *piEndBlockid = pWriter->iEndBlockid;

  return interiorWriterRootInfo(v, &pWriter->parentWriter,
................................................................................
/* OptLeavesReader is nearly identical to LeavesReader, except that
** where LeavesReader is geared towards the merging of complete
** segment levels (with exactly MERGE_COUNT segments), OptLeavesReader
** is geared towards implementation of the optimize() function, and
** can merge all segments simultaneously.  This version may be
** somewhat less efficient than LeavesReader because it merges into an
** accumulator rather than doing an N-way merge, but since segment
** size grows exponentially (so segment count logrithmically) this is
** probably not an immediate problem.
*/
/* TODO(shess): Prove that assertion, or extend the merge code to
** merge tree fashion (like the prefix-searching code does).
*/
/* TODO(shess): OptLeavesReader and LeavesReader could probably be
** merged with little or no loss of performance for LeavesReader.  The







|







 







|







 







|
|







 







|







 







|







 







|







 







|







101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
....
2407
2408
2409
2410
2411
2412
2413
2414
2415
2416
2417
2418
2419
2420
2421
....
2576
2577
2578
2579
2580
2581
2582
2583
2584
2585
2586
2587
2588
2589
2590
2591
....
3519
3520
3521
3522
3523
3524
3525
3526
3527
3528
3529
3530
3531
3532
3533
....
3870
3871
3872
3873
3874
3875
3876
3877
3878
3879
3880
3881
3882
3883
3884
....
4644
4645
4646
4647
4648
4649
4650
4651
4652
4653
4654
4655
4656
4657
4658
....
6014
6015
6016
6017
6018
6019
6020
6021
6022
6023
6024
6025
6026
6027
6028
**   varint POS_END;        (marks end of positions for this document.
** }
**
** Here, array { X } means zero or more occurrences of X, adjacent in
** memory.  A "position" is an index of a token in the token stream
** generated by the tokenizer, while an "offset" is a byte offset,
** both based at 0.  Note that POS_END and POS_COLUMN occur in the
** same logical place as the position element, and act as sentinels
** ending a position list array.
**
** A DL_POSITIONS doclist omits the startOffset and endOffset
** information.  A DL_DOCIDS doclist omits both the position and
** offset information, becoming an array of varint-encoded docids.
**
** On-disk data is stored as type DL_DEFAULT, so we don't serialize
................................................................................
** IdChar(X) will be true.  Otherwise it is false.
**
** For ASCII, any character with the high-order bit set is
** allowed in an identifier.  For 7-bit characters, 
** sqlite3IsIdChar[X] must be 1.
**
** Ticket #1066.  the SQL standard does not allow '$' in the
** middle of identifiers.  But many SQL implementations do. 
** SQLite will allow '$' in identifiers for compatibility.
** But the feature is undocumented.
*/
static const char isIdChar[] = {
/* x0 x1 x2 x3 x4 x5 x6 x7 x8 x9 xA xB xC xD xE xF */
    0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,  /* 2x */
    1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0,  /* 3x */
................................................................................
/*
** The input azIn is a NULL-terminated list of tokens.  Remove the first
** token and all punctuation tokens.  Remove the quotes from
** around string literal tokens.
**
** Example:
**
**     input:      tokenize chinese ( 'simplified' , 'mixed' )
**     output:     chinese simplified mixed
**
** Another example:
**
**     input:      delimiters ( '[' , ']' , '...' )
**     output:     [ ] ...
*/
static void tokenListToIdList(char **azIn){
................................................................................
    }
  }
  return -1;
}

/*
** Parse the text at pSegment[0..nSegment-1].  Add additional terms
** to the query being assembled in pQuery.
**
** inPhrase is true if pSegment[0..nSegement-1] is contained within
** double-quotes.  If inPhrase is true, then the first term
** is marked with the number of terms in the phrase less one and
** OR and "-" syntax is ignored.  If inPhrase is false, then every
** term found is marked with nPhrase=0 and OR and "-" syntax is significant.
*/
................................................................................
    */
    sqlite3_result_blob(pContext, &c, sizeof(c), SQLITE_TRANSIENT);
  }
  return SQLITE_OK;
}

/* This is the xRowid method.  The SQLite core calls this routine to
** retrieve the rowid for the current row of the result set.  The
** rowid should be written to *pRowid.
*/
static int fulltextRowid(sqlite3_vtab_cursor *pCursor, sqlite_int64 *pRowid){
  fulltext_cursor *c = (fulltext_cursor *) pCursor;

  *pRowid = sqlite3_column_int64(c->pStmt, 0);
  return SQLITE_OK;
................................................................................
    int rc = leafWriterFlush(v, pWriter);
    if( rc!=SQLITE_OK ) return rc;
  }

  /* We must have flushed a leaf at some point. */
  assert( pWriter->has_parent );

  /* Tentatively set the end leaf blockid as the end blockid.  If the
  ** interior node can be returned inline, this will be the final
  ** blockid, otherwise it will be overwritten by
  ** interiorWriterRootInfo().
  */
  *piEndBlockid = pWriter->iEndBlockid;

  return interiorWriterRootInfo(v, &pWriter->parentWriter,
................................................................................
/* OptLeavesReader is nearly identical to LeavesReader, except that
** where LeavesReader is geared towards the merging of complete
** segment levels (with exactly MERGE_COUNT segments), OptLeavesReader
** is geared towards implementation of the optimize() function, and
** can merge all segments simultaneously.  This version may be
** somewhat less efficient than LeavesReader because it merges into an
** accumulator rather than doing an N-way merge, but since segment
** size grows exponentially (so segment count logarithmically) this is
** probably not an immediate problem.
*/
/* TODO(shess): Prove that assertion, or extend the merge code to
** merge tree fashion (like the prefix-searching code does).
*/
/* TODO(shess): OptLeavesReader and LeavesReader could probably be
** merged with little or no loss of performance for LeavesReader.  The

Changes to ext/fts2/fts2_hash.c.

126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
...
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
  if( n1!=n2 ) return 1;
  return memcmp(pKey1,pKey2,n1);
}

/*
** Return a pointer to the appropriate hash function given the key class.
**
** The C syntax in this function definition may be unfamilar to some 
** programmers, so we provide the following additional explanation:
**
** The name of the function is "hashFunction".  The function takes a
** single parameter "keyClass".  The return value of hashFunction()
** is a pointer to another function.  Specifically, the return value
** of hashFunction() is a pointer to a function that takes two parameters
** with types "const void*" and "int" and returns an "int".
................................................................................
    pH->first = pNew;
  }
  pEntry->count++;
  pEntry->chain = pNew;
}


/* Resize the hash table so that it cantains "new_size" buckets.
** "new_size" must be a power of 2.  The hash table might fail 
** to resize if sqliteMalloc() fails.
*/
static void rehash(fts2Hash *pH, int new_size){
  struct _fts2ht *new_ht;          /* The new hash table */
  fts2HashElem *elem, *next_elem;  /* For looping over existing elements */
  int (*xHash)(const void*,int);   /* The hash function */







|







 







|







126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
...
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
  if( n1!=n2 ) return 1;
  return memcmp(pKey1,pKey2,n1);
}

/*
** Return a pointer to the appropriate hash function given the key class.
**
** The C syntax in this function definition may be unfamiliar to some 
** programmers, so we provide the following additional explanation:
**
** The name of the function is "hashFunction".  The function takes a
** single parameter "keyClass".  The return value of hashFunction()
** is a pointer to another function.  Specifically, the return value
** of hashFunction() is a pointer to a function that takes two parameters
** with types "const void*" and "int" and returns an "int".
................................................................................
    pH->first = pNew;
  }
  pEntry->count++;
  pEntry->chain = pNew;
}


/* Resize the hash table so that it contains "new_size" buckets.
** "new_size" must be a power of 2.  The hash table might fail 
** to resize if sqliteMalloc() fails.
*/
static void rehash(fts2Hash *pH, int new_size){
  struct _fts2ht *new_ht;          /* The new hash table */
  fts2HashElem *elem, *next_elem;  /* For looping over existing elements */
  int (*xHash)(const void*,int);   /* The hash function */

Changes to ext/fts2/fts2_porter.c.

141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
...
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
...
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
**
** A consonate is any letter other than 'a', 'e', 'i', 'o', or 'u'.
** 'Y' is a consonant unless it follows another consonant,
** in which case it is a vowel.
**
** In these routine, the letters are in reverse order.  So the 'y' rule
** is that 'y' is a consonant unless it is followed by another
** consonent.
*/
static int isVowel(const char*);
static int isConsonant(const char *z){
  int j;
  char x = *z;
  if( x==0 ) return 0;
  assert( x>='a' && x<='z' );
................................................................................
*/
static int doubleConsonant(const char *z){
  return isConsonant(z) && z[0]==z[1] && isConsonant(z+1);
}

/*
** Return TRUE if the word ends with three letters which
** are consonant-vowel-consonent and where the final consonant
** is not 'w', 'x', or 'y'.
**
** The word is reversed here.  So we are really checking the
** first three letters and the first one cannot be in [wxy].
*/
static int star_oh(const char *z){
  return
................................................................................
    z[0]!='w' && z[0]!='x' && z[0]!='y' &&
    z[1]!=0 && isVowel(z+1) &&
    z[2]!=0 && isConsonant(z+2);
}

/*
** If the word ends with zFrom and xCond() is true for the stem
** of the word that preceeds the zFrom ending, then change the 
** ending to zTo.
**
** The input word *pz and zFrom are both in reverse order.  zTo
** is in normal order. 
**
** Return TRUE if zFrom matches.  Return FALSE if zFrom does not
** match.  Not that TRUE is returned even if xCond() fails and







|







 







|







 







|







141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
...
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
...
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
**
** A consonate is any letter other than 'a', 'e', 'i', 'o', or 'u'.
** 'Y' is a consonant unless it follows another consonant,
** in which case it is a vowel.
**
** In these routine, the letters are in reverse order.  So the 'y' rule
** is that 'y' is a consonant unless it is followed by another
** consonant.
*/
static int isVowel(const char*);
static int isConsonant(const char *z){
  int j;
  char x = *z;
  if( x==0 ) return 0;
  assert( x>='a' && x<='z' );
................................................................................
*/
static int doubleConsonant(const char *z){
  return isConsonant(z) && z[0]==z[1] && isConsonant(z+1);
}

/*
** Return TRUE if the word ends with three letters which
** are consonant-vowel-consonant and where the final consonant
** is not 'w', 'x', or 'y'.
**
** The word is reversed here.  So we are really checking the
** first three letters and the first one cannot be in [wxy].
*/
static int star_oh(const char *z){
  return
................................................................................
    z[0]!='w' && z[0]!='x' && z[0]!='y' &&
    z[1]!=0 && isVowel(z+1) &&
    z[2]!=0 && isConsonant(z+2);
}

/*
** If the word ends with zFrom and xCond() is true for the stem
** of the word that precedes the zFrom ending, then change the 
** ending to zTo.
**
** The input word *pz and zFrom are both in reverse order.  zTo
** is in normal order. 
**
** Return TRUE if zFrom matches.  Return FALSE if zFrom does not
** match.  Not that TRUE is returned even if xCond() fails and

Changes to ext/fts2/mkfts2amal.tcl.

61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
  set nstar [expr {60 - $n}]
  set stars [string range $s78 0 $nstar]
  puts $out "/************** $text $stars/"
}

# Read the source file named $filename and write it into the
# sqlite3.c output file.  If any #include statements are seen,
# process them approprately.
#
proc copy_file {filename} {
  global seen_hdr available_hdr out
  set tail [file tail $filename]
  section_comment "Begin file $tail"
  set in [open $filename r]
  while {![eof $in]} {







|







61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
  set nstar [expr {60 - $n}]
  set stars [string range $s78 0 $nstar]
  puts $out "/************** $text $stars/"
}

# Read the source file named $filename and write it into the
# sqlite3.c output file.  If any #include statements are seen,
# process them appropriately.
#
proc copy_file {filename} {
  global seen_hdr available_hdr out
  set tail [file tail $filename]
  section_comment "Begin file $tail"
  set in [open $filename r]
  while {![eof $in]} {

Changes to ext/fts3/README.syntax.

12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
..
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
...
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
...
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
  (in this case "t1"), or by the name of one of the columns of the fts3 
  table. <full-text-query> should be replaced by an SQL expression that 
  computes to a string containing an Fts3 query.

  If the left-hand-side of the MATCH operator is set to the name of the
  fts3 table, then by default the query may be matched against any column
  of the table. If it is set to a column name, then by default the query
  may only match the specified column. In both cases this may be overriden
  as part of the query text (see sections 2 and 3 below).

  As of SQLite version 3.6.8, Fts3 supports two slightly different query 
  formats; the standard syntax, which is used by default, and the enhanced
  query syntax which can be selected by compiling with the pre-processor
  symbol SQLITE_ENABLE_FTS3_PARENTHESIS defined.

................................................................................
  case is treated as an ordinary token. For example, the following query:

    <col> MATCH 'engineering NEAR consultancy'

  matches rows that contain both the "engineering" and "consultancy" tokens
  in the same column with not more than 10 other words between them. It does
  not matter which of the two terms occurs first in the document, only that
  they be seperated by only 10 tokens or less. The user may also specify
  a different required proximity by adding "/N" immediately after the NEAR
  operator, where N is an integer. For example:

    <col> MATCH 'engineering NEAR/5 consultancy'

  searches for a row containing an instance of each specified token seperated
  by not more than 5 other tokens. More than one NEAR operator can be used
  in as sequence. For example this query:

    <col> MATCH 'reliable NEAR/2 engineering NEAR/5 consultancy'

  searches for a row that contains an instance of the token "reliable" 
  seperated by not more than two tokens from an instance of "engineering",
  which is in turn separated by not more than 5 other tokens from an
  instance of the term "consultancy". Phrases enclosed in quotes may
  also be used as arguments to the NEAR operator.

  Similar to the NEAR operator, one or more tokens or phrases may be 
  separated by OR operators. In this case, only one of the specified tokens
  or phrases must appear in the document. For example, the query:
................................................................................
  1) Parenthesis are supported. When using the enhanced query syntax,
     parenthesis may be used to overcome the built-in precedence of the
     supplied binary operators. For example, the following query:

       <col> MATCH '(hello world) OR (simple example)'

     matches documents that contain both "hello" and "world", and documents
     that contain both "simple" and "example". It is not possible to forumlate
     such a query using the standard syntax.

  2) Instead of separating tokens and phrases by whitespace, an AND operator
     may be explicitly specified. This does not change query processing at
     all, but may be used to improve readability. For example, the following
     query is handled identically to the one above:

................................................................................
       <col> MATCH 'example NOT simple'

     As for all other operators, the NOT operator must be specified in
     upper case. Otherwise it will be treated as a regular token.

  4) Unlike in the standard syntax, where the OR operator has a higher
     precedence than the implicit AND operator, when using the enhanced
     syntax implicit and explict AND operators have a higher precedence
     than OR operators. Using the enhanced syntax, the following two
     queries are equivalent:

       <col> MATCH 'sqlite fantastic OR impressive'
       <col> MATCH '(sqlite AND fantastic) OR impressive'

     however, when using the standard syntax, the query:







|







 







|





|






|







 







|







 







|







12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
..
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
...
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
...
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
  (in this case "t1"), or by the name of one of the columns of the fts3 
  table. <full-text-query> should be replaced by an SQL expression that 
  computes to a string containing an Fts3 query.

  If the left-hand-side of the MATCH operator is set to the name of the
  fts3 table, then by default the query may be matched against any column
  of the table. If it is set to a column name, then by default the query
  may only match the specified column. In both cases this may be overridden
  as part of the query text (see sections 2 and 3 below).

  As of SQLite version 3.6.8, Fts3 supports two slightly different query 
  formats; the standard syntax, which is used by default, and the enhanced
  query syntax which can be selected by compiling with the pre-processor
  symbol SQLITE_ENABLE_FTS3_PARENTHESIS defined.

................................................................................
  case is treated as an ordinary token. For example, the following query:

    <col> MATCH 'engineering NEAR consultancy'

  matches rows that contain both the "engineering" and "consultancy" tokens
  in the same column with not more than 10 other words between them. It does
  not matter which of the two terms occurs first in the document, only that
  they be separated by only 10 tokens or less. The user may also specify
  a different required proximity by adding "/N" immediately after the NEAR
  operator, where N is an integer. For example:

    <col> MATCH 'engineering NEAR/5 consultancy'

  searches for a row containing an instance of each specified token separated
  by not more than 5 other tokens. More than one NEAR operator can be used
  in as sequence. For example this query:

    <col> MATCH 'reliable NEAR/2 engineering NEAR/5 consultancy'

  searches for a row that contains an instance of the token "reliable" 
  separated by not more than two tokens from an instance of "engineering",
  which is in turn separated by not more than 5 other tokens from an
  instance of the term "consultancy". Phrases enclosed in quotes may
  also be used as arguments to the NEAR operator.

  Similar to the NEAR operator, one or more tokens or phrases may be 
  separated by OR operators. In this case, only one of the specified tokens
  or phrases must appear in the document. For example, the query:
................................................................................
  1) Parenthesis are supported. When using the enhanced query syntax,
     parenthesis may be used to overcome the built-in precedence of the
     supplied binary operators. For example, the following query:

       <col> MATCH '(hello world) OR (simple example)'

     matches documents that contain both "hello" and "world", and documents
     that contain both "simple" and "example". It is not possible to formulate
     such a query using the standard syntax.

  2) Instead of separating tokens and phrases by whitespace, an AND operator
     may be explicitly specified. This does not change query processing at
     all, but may be used to improve readability. For example, the following
     query is handled identically to the one above:

................................................................................
       <col> MATCH 'example NOT simple'

     As for all other operators, the NOT operator must be specified in
     upper case. Otherwise it will be treated as a regular token.

  4) Unlike in the standard syntax, where the OR operator has a higher
     precedence than the implicit AND operator, when using the enhanced
     syntax implicit and explicit AND operators have a higher precedence
     than OR operators. Using the enhanced syntax, the following two
     queries are equivalent:

       <col> MATCH 'sqlite fantastic OR impressive'
       <col> MATCH '(sqlite AND fantastic) OR impressive'

     however, when using the standard syntax, the query:

Changes to ext/fts3/fts3.c.

83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
....
2628
2629
2630
2631
2632
2633
2634
2635
2636
2637
2638
2639
2640
2641
2642
....
4650
4651
4652
4653
4654
4655
4656
4657
4658
4659
4660
4661
4662
4663
4664
....
5660
5661
5662
5663
5664
5665
5666
5667
5668
5669
5670
5671
5672
5673
5674
**   }
**   varint POS_END;        (marks end of positions for this document.
** }
**
** Here, array { X } means zero or more occurrences of X, adjacent in
** memory.  A "position" is an index of a token in the token stream
** generated by the tokenizer. Note that POS_END and POS_COLUMN occur 
** in the same logical place as the position element, and act as sentinals
** ending a position list array.  POS_END is 0.  POS_COLUMN is 1.
** The positions numbers are not stored literally but rather as two more
** than the difference from the prior position, or the just the position plus
** 2 for the first position.  Example:
**
**   label:       A B C D E  F  G H   I  J K
**   value:     123 5 9 1 1 14 35 0 234 72 0
................................................................................
  ** be larger in the output than it was in the input (since the delta value
  ** may be a larger positive integer than the actual docid).
  **
  ** The space required to store the output is therefore the sum of the
  ** sizes of the two inputs, plus enough space for exactly one of the input
  ** docids to grow. 
  **
  ** A symetric argument may be made if the doclists are in descending 
  ** order.
  */
  aOut = sqlite3_malloc64((i64)n1+n2+FTS3_VARINT_MAX-1+FTS3_BUFFER_PADDING);
  if( !aOut ) return SQLITE_NOMEM;

  p = aOut;
  fts3GetDeltaVarint3(&p1, pEnd1, 0, &i1);
................................................................................
/*
** The phrase iterator passed as the second argument:
**
**   * features at least one token that uses an incremental doclist, and 
**
**   * does not contain any deferred tokens.
**
** Advance it to the next matching documnent in the database and populate
** the Fts3Doclist.pList and nList fields. 
**
** If there is no "next" entry and no error occurs, then *pbEof is set to
** 1 before returning. Otherwise, if no error occurs and the iterator is
** successfully advanced, *pbEof is set to 0.
**
** If an error occurs, return an SQLite error code. Otherwise, return 
................................................................................
    pCsr->isEof = 1;
  }

  return rc;
}

/*
** Restart interation for expression pExpr so that the next call to
** fts3EvalNext() visits the first row. Do not allow incremental 
** loading or merging of phrase doclists for this iteration.
**
** If *pRc is other than SQLITE_OK when this function is called, it is
** a no-op. If an error occurs within this function, *pRc is set to an
** SQLite error code before returning.
*/







|







 







|







 







|







 







|







83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
....
2628
2629
2630
2631
2632
2633
2634
2635
2636
2637
2638
2639
2640
2641
2642
....
4650
4651
4652
4653
4654
4655
4656
4657
4658
4659
4660
4661
4662
4663
4664
....
5660
5661
5662
5663
5664
5665
5666
5667
5668
5669
5670
5671
5672
5673
5674
**   }
**   varint POS_END;        (marks end of positions for this document.
** }
**
** Here, array { X } means zero or more occurrences of X, adjacent in
** memory.  A "position" is an index of a token in the token stream
** generated by the tokenizer. Note that POS_END and POS_COLUMN occur 
** in the same logical place as the position element, and act as sentinels
** ending a position list array.  POS_END is 0.  POS_COLUMN is 1.
** The positions numbers are not stored literally but rather as two more
** than the difference from the prior position, or the just the position plus
** 2 for the first position.  Example:
**
**   label:       A B C D E  F  G H   I  J K
**   value:     123 5 9 1 1 14 35 0 234 72 0
................................................................................
  ** be larger in the output than it was in the input (since the delta value
  ** may be a larger positive integer than the actual docid).
  **
  ** The space required to store the output is therefore the sum of the
  ** sizes of the two inputs, plus enough space for exactly one of the input
  ** docids to grow. 
  **
  ** A symmetric argument may be made if the doclists are in descending 
  ** order.
  */
  aOut = sqlite3_malloc64((i64)n1+n2+FTS3_VARINT_MAX-1+FTS3_BUFFER_PADDING);
  if( !aOut ) return SQLITE_NOMEM;

  p = aOut;
  fts3GetDeltaVarint3(&p1, pEnd1, 0, &i1);
................................................................................
/*
** The phrase iterator passed as the second argument:
**
**   * features at least one token that uses an incremental doclist, and 
**
**   * does not contain any deferred tokens.
**
** Advance it to the next matching document in the database and populate
** the Fts3Doclist.pList and nList fields. 
**
** If there is no "next" entry and no error occurs, then *pbEof is set to
** 1 before returning. Otherwise, if no error occurs and the iterator is
** successfully advanced, *pbEof is set to 0.
**
** If an error occurs, return an SQLite error code. Otherwise, return 
................................................................................
    pCsr->isEof = 1;
  }

  return rc;
}

/*
** Restart iteration for expression pExpr so that the next call to
** fts3EvalNext() visits the first row. Do not allow incremental 
** loading or merging of phrase doclists for this iteration.
**
** If *pRc is other than SQLITE_OK when this function is called, it is
** a no-op. If an error occurs within this function, *pRc is set to an
** SQLite error code before returning.
*/

Changes to ext/fts3/fts3Int.h.

23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
...
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
...
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
# undef SQLITE_ENABLE_FTS3
# undef SQLITE_ENABLE_FTS4
#endif

/*
** FTS4 is really an extension for FTS3.  It is enabled using the
** SQLITE_ENABLE_FTS3 macro.  But to avoid confusion we also all
** the SQLITE_ENABLE_FTS4 macro to serve as an alisse for SQLITE_ENABLE_FTS3.
*/
#if defined(SQLITE_ENABLE_FTS4) && !defined(SQLITE_ENABLE_FTS3)
# define SQLITE_ENABLE_FTS3
#endif

#if !defined(SQLITE_CORE) || defined(SQLITE_ENABLE_FTS3)

................................................................................
*/
#define POS_COLUMN  (1)     /* Column-list terminator */
#define POS_END     (0)     /* Position-list terminator */ 

/*
** The assert_fts3_nc() macro is similar to the assert() macro, except that it
** is used for assert() conditions that are true only if it can be 
** guranteed that the database is not corrupt.
*/
#if defined(SQLITE_DEBUG) || defined(SQLITE_TEST)
extern int sqlite3_fts3_may_be_corrupt;
# define assert_fts3_nc(x) assert(sqlite3_fts3_may_be_corrupt || (x))
#else
# define assert_fts3_nc(x) assert(x)
#endif
................................................................................

#define FTS3_EVAL_FILTER    0
#define FTS3_EVAL_NEXT      1
#define FTS3_EVAL_MATCHINFO 2

/*
** The Fts3Cursor.eSearch member is always set to one of the following.
** Actualy, Fts3Cursor.eSearch can be greater than or equal to
** FTS3_FULLTEXT_SEARCH.  If so, then Fts3Cursor.eSearch - 2 is the index
** of the column to be searched.  For example, in
**
**     CREATE VIRTUAL TABLE ex1 USING fts3(a,b,c,d);
**     SELECT docid FROM ex1 WHERE b MATCH 'one two three';
** 
** Because the LHS of the MATCH operator is 2nd column "b",







|







 







|







 







|







23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
...
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
...
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
# undef SQLITE_ENABLE_FTS3
# undef SQLITE_ENABLE_FTS4
#endif

/*
** FTS4 is really an extension for FTS3.  It is enabled using the
** SQLITE_ENABLE_FTS3 macro.  But to avoid confusion we also all
** the SQLITE_ENABLE_FTS4 macro to serve as an alias for SQLITE_ENABLE_FTS3.
*/
#if defined(SQLITE_ENABLE_FTS4) && !defined(SQLITE_ENABLE_FTS3)
# define SQLITE_ENABLE_FTS3
#endif

#if !defined(SQLITE_CORE) || defined(SQLITE_ENABLE_FTS3)

................................................................................
*/
#define POS_COLUMN  (1)     /* Column-list terminator */
#define POS_END     (0)     /* Position-list terminator */ 

/*
** The assert_fts3_nc() macro is similar to the assert() macro, except that it
** is used for assert() conditions that are true only if it can be 
** guaranteed that the database is not corrupt.
*/
#if defined(SQLITE_DEBUG) || defined(SQLITE_TEST)
extern int sqlite3_fts3_may_be_corrupt;
# define assert_fts3_nc(x) assert(sqlite3_fts3_may_be_corrupt || (x))
#else
# define assert_fts3_nc(x) assert(x)
#endif
................................................................................

#define FTS3_EVAL_FILTER    0
#define FTS3_EVAL_NEXT      1
#define FTS3_EVAL_MATCHINFO 2

/*
** The Fts3Cursor.eSearch member is always set to one of the following.
** Actually, Fts3Cursor.eSearch can be greater than or equal to
** FTS3_FULLTEXT_SEARCH.  If so, then Fts3Cursor.eSearch - 2 is the index
** of the column to be searched.  For example, in
**
**     CREATE VIRTUAL TABLE ex1 USING fts3(a,b,c,d);
**     SELECT docid FROM ex1 WHERE b MATCH 'one two three';
** 
** Because the LHS of the MATCH operator is 2nd column "b",

Changes to ext/fts3/fts3_expr.c.

649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
          p = pPrev;
        }else{
          int eType = p->eType;
          isPhrase = (eType==FTSQUERY_PHRASE || p->pLeft);

          /* The isRequirePhrase variable is set to true if a phrase or
          ** an expression contained in parenthesis is required. If a
          ** binary operator (AND, OR, NOT or NEAR) is encounted when
          ** isRequirePhrase is set, this is a syntax error.
          */
          if( !isPhrase && isRequirePhrase ){
            sqlite3Fts3ExprFree(p);
            rc = SQLITE_ERROR;
            goto exprparse_out;
          }







|







649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
          p = pPrev;
        }else{
          int eType = p->eType;
          isPhrase = (eType==FTSQUERY_PHRASE || p->pLeft);

          /* The isRequirePhrase variable is set to true if a phrase or
          ** an expression contained in parenthesis is required. If a
          ** binary operator (AND, OR, NOT or NEAR) is encountered when
          ** isRequirePhrase is set, this is a syntax error.
          */
          if( !isPhrase && isRequirePhrase ){
            sqlite3Fts3ExprFree(p);
            rc = SQLITE_ERROR;
            goto exprparse_out;
          }

Changes to ext/fts3/fts3_hash.c.

124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
...
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
  if( n1!=n2 ) return 1;
  return memcmp(pKey1,pKey2,n1);
}

/*
** Return a pointer to the appropriate hash function given the key class.
**
** The C syntax in this function definition may be unfamilar to some 
** programmers, so we provide the following additional explanation:
**
** The name of the function is "ftsHashFunction".  The function takes a
** single parameter "keyClass".  The return value of ftsHashFunction()
** is a pointer to another function.  Specifically, the return value
** of ftsHashFunction() is a pointer to a function that takes two parameters
** with types "const void*" and "int" and returns an "int".
................................................................................
    pH->first = pNew;
  }
  pEntry->count++;
  pEntry->chain = pNew;
}


/* Resize the hash table so that it cantains "new_size" buckets.
** "new_size" must be a power of 2.  The hash table might fail 
** to resize if sqliteMalloc() fails.
**
** Return non-zero if a memory allocation error occurs.
*/
static int fts3Rehash(Fts3Hash *pH, int new_size){
  struct _fts3ht *new_ht;          /* The new hash table */







|







 







|







124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
...
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
  if( n1!=n2 ) return 1;
  return memcmp(pKey1,pKey2,n1);
}

/*
** Return a pointer to the appropriate hash function given the key class.
**
** The C syntax in this function definition may be unfamiliar to some 
** programmers, so we provide the following additional explanation:
**
** The name of the function is "ftsHashFunction".  The function takes a
** single parameter "keyClass".  The return value of ftsHashFunction()
** is a pointer to another function.  Specifically, the return value
** of ftsHashFunction() is a pointer to a function that takes two parameters
** with types "const void*" and "int" and returns an "int".
................................................................................
    pH->first = pNew;
  }
  pEntry->count++;
  pEntry->chain = pNew;
}


/* Resize the hash table so that it contains "new_size" buckets.
** "new_size" must be a power of 2.  The hash table might fail 
** to resize if sqliteMalloc() fails.
**
** Return non-zero if a memory allocation error occurs.
*/
static int fts3Rehash(Fts3Hash *pH, int new_size){
  struct _fts3ht *new_ht;          /* The new hash table */

Changes to ext/fts3/fts3_porter.c.

140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
...
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
...
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
**
** A consonate is any letter other than 'a', 'e', 'i', 'o', or 'u'.
** 'Y' is a consonant unless it follows another consonant,
** in which case it is a vowel.
**
** In these routine, the letters are in reverse order.  So the 'y' rule
** is that 'y' is a consonant unless it is followed by another
** consonent.
*/
static int isVowel(const char*);
static int isConsonant(const char *z){
  int j;
  char x = *z;
  if( x==0 ) return 0;
  assert( x>='a' && x<='z' );
................................................................................
*/
static int doubleConsonant(const char *z){
  return isConsonant(z) && z[0]==z[1];
}

/*
** Return TRUE if the word ends with three letters which
** are consonant-vowel-consonent and where the final consonant
** is not 'w', 'x', or 'y'.
**
** The word is reversed here.  So we are really checking the
** first three letters and the first one cannot be in [wxy].
*/
static int star_oh(const char *z){
  return
................................................................................
    z[0]!='w' && z[0]!='x' && z[0]!='y' &&
    isVowel(z+1) &&
    isConsonant(z+2);
}

/*
** If the word ends with zFrom and xCond() is true for the stem
** of the word that preceeds the zFrom ending, then change the 
** ending to zTo.
**
** The input word *pz and zFrom are both in reverse order.  zTo
** is in normal order. 
**
** Return TRUE if zFrom matches.  Return FALSE if zFrom does not
** match.  Not that TRUE is returned even if xCond() fails and







|







 







|







 







|







140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
...
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
...
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
**
** A consonate is any letter other than 'a', 'e', 'i', 'o', or 'u'.
** 'Y' is a consonant unless it follows another consonant,
** in which case it is a vowel.
**
** In these routine, the letters are in reverse order.  So the 'y' rule
** is that 'y' is a consonant unless it is followed by another
** consonant.
*/
static int isVowel(const char*);
static int isConsonant(const char *z){
  int j;
  char x = *z;
  if( x==0 ) return 0;
  assert( x>='a' && x<='z' );
................................................................................
*/
static int doubleConsonant(const char *z){
  return isConsonant(z) && z[0]==z[1];
}

/*
** Return TRUE if the word ends with three letters which
** are consonant-vowel-consonant and where the final consonant
** is not 'w', 'x', or 'y'.
**
** The word is reversed here.  So we are really checking the
** first three letters and the first one cannot be in [wxy].
*/
static int star_oh(const char *z){
  return
................................................................................
    z[0]!='w' && z[0]!='x' && z[0]!='y' &&
    isVowel(z+1) &&
    isConsonant(z+2);
}

/*
** If the word ends with zFrom and xCond() is true for the stem
** of the word that precedes the zFrom ending, then change the 
** ending to zTo.
**
** The input word *pz and zFrom are both in reverse order.  zTo
** is in normal order. 
**
** Return TRUE if zFrom matches.  Return FALSE if zFrom does not
** match.  Not that TRUE is returned even if xCond() fails and

Changes to ext/fts3/fts3_snippet.c.

602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
  int nAppend                     /* Size of zAppend in bytes (or -1) */
){
  if( nAppend<0 ){
    nAppend = (int)strlen(zAppend);
  }

  /* If there is insufficient space allocated at StrBuffer.z, use realloc()
  ** to grow the buffer until so that it is big enough to accomadate the
  ** appended data.
  */
  if( pStr->n+nAppend+1>=pStr->nAlloc ){
    sqlite3_int64 nAlloc = pStr->nAlloc+(sqlite3_int64)nAppend+100;
    char *zNew = sqlite3_realloc64(pStr->z, nAlloc);
    if( !zNew ){
      return SQLITE_NOMEM;







|







602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
  int nAppend                     /* Size of zAppend in bytes (or -1) */
){
  if( nAppend<0 ){
    nAppend = (int)strlen(zAppend);
  }

  /* If there is insufficient space allocated at StrBuffer.z, use realloc()
  ** to grow the buffer until so that it is big enough to accommodate the
  ** appended data.
  */
  if( pStr->n+nAppend+1>=pStr->nAlloc ){
    sqlite3_int64 nAlloc = pStr->nAlloc+(sqlite3_int64)nAppend+100;
    char *zNew = sqlite3_realloc64(pStr->z, nAlloc);
    if( !zNew ){
      return SQLITE_NOMEM;

Changes to ext/fts3/fts3_unicode2.c.

152
153
154
155
156
157
158
159
160
161
162
163
164
165
166


/*
** If the argument is a codepoint corresponding to a lowercase letter
** in the ASCII range with a diacritic added, return the codepoint
** of the ASCII letter only. For example, if passed 235 - "LATIN
** SMALL LETTER E WITH DIAERESIS" - return 65 ("LATIN SMALL LETTER
** E"). The resuls of passing a codepoint that corresponds to an
** uppercase letter are undefined.
*/
static int remove_diacritic(int c, int bComplex){
  unsigned short aDia[] = {
        0,  1797,  1848,  1859,  1891,  1928,  1940,  1995, 
     2024,  2040,  2060,  2110,  2168,  2206,  2264,  2286, 
     2344,  2383,  2472,  2488,  2516,  2596,  2668,  2732, 







|







152
153
154
155
156
157
158
159
160
161
162
163
164
165
166


/*
** If the argument is a codepoint corresponding to a lowercase letter
** in the ASCII range with a diacritic added, return the codepoint
** of the ASCII letter only. For example, if passed 235 - "LATIN
** SMALL LETTER E WITH DIAERESIS" - return 65 ("LATIN SMALL LETTER
** E"). The results of passing a codepoint that corresponds to an
** uppercase letter are undefined.
*/
static int remove_diacritic(int c, int bComplex){
  unsigned short aDia[] = {
        0,  1797,  1848,  1859,  1891,  1928,  1940,  1995, 
     2024,  2040,  2060,  2110,  2168,  2206,  2264,  2286, 
     2344,  2383,  2472,  2488,  2516,  2596,  2668,  2732, 

Changes to ext/fts3/fts3_write.c.

3935
3936
3937
3938
3939
3940
3941
3942
3943
3944
3945
3946
3947
3948
3949
....
4715
4716
4717
4718
4719
4720
4721
4722
4723
4724
4725
4726
4727
4728
4729
....
5444
5445
5446
5447
5448
5449
5450
5451
5452
5453
5454
5455
5456
5457
5458
....
5575
5576
5577
5578
5579
5580
5581
5582
5583
5584
5585
5586
5587
5588
5589
** If the size of the value in blob pPrev is zero, then this is the first
** term written to the node. Otherwise, pPrev contains a copy of the 
** previous term. Before this function returns, it is updated to contain a
** copy of zTerm/nTerm.
**
** It is assumed that the buffer associated with pNode is already large
** enough to accommodate the new entry. The buffer associated with pPrev
** is extended by this function if requrired.
**
** If an error (i.e. OOM condition) occurs, an SQLite error code is
** returned. Otherwise, SQLITE_OK.
*/
static int fts3AppendToNode(
  Blob *pNode,                    /* Current node image to append to */
  Blob *pPrev,                    /* Buffer containing previous term written */
................................................................................

  sqlite3_free(root.a);
  sqlite3_free(block.a);
  return rc;
}

/*
** This function is called after an incrmental-merge operation has run to
** merge (or partially merge) two or more segments from absolute level
** iAbsLevel.
**
** Each input segment is either removed from the db completely (if all of
** its data was copied to the output segment by the incrmerge operation)
** or modified in place so that it no longer contains those entries that
** have been duplicated in the output segment.
................................................................................
  for(pDef=pCsr->pDeferred; pDef; pDef=pDef->pNext){
    fts3PendingListDelete(pDef->pList);
    pDef->pList = 0;
  }
}

/*
** Free all entries in the pCsr->pDeffered list. Entries are added to 
** this list using sqlite3Fts3DeferToken().
*/
void sqlite3Fts3FreeDeferredTokens(Fts3Cursor *pCsr){
  Fts3DeferredToken *pDef;
  Fts3DeferredToken *pNext;
  for(pDef=pCsr->pDeferred; pDef; pDef=pNext){
    pNext = pDef->pNext;
................................................................................
  return SQLITE_OK;
}
#endif

/*
** SQLite value pRowid contains the rowid of a row that may or may not be
** present in the FTS3 table. If it is, delete it and adjust the contents
** of subsiduary data structures accordingly.
*/
static int fts3DeleteByRowid(
  Fts3Table *p, 
  sqlite3_value *pRowid, 
  int *pnChng,                    /* IN/OUT: Decrement if row is deleted */
  u32 *aSzDel
){







|







 







|







 







|







 







|







3935
3936
3937
3938
3939
3940
3941
3942
3943
3944
3945
3946
3947
3948
3949
....
4715
4716
4717
4718
4719
4720
4721
4722
4723
4724
4725
4726
4727
4728
4729
....
5444
5445
5446
5447
5448
5449
5450
5451
5452
5453
5454
5455
5456
5457
5458
....
5575
5576
5577
5578
5579
5580
5581
5582
5583
5584
5585
5586
5587
5588
5589
** If the size of the value in blob pPrev is zero, then this is the first
** term written to the node. Otherwise, pPrev contains a copy of the 
** previous term. Before this function returns, it is updated to contain a
** copy of zTerm/nTerm.
**
** It is assumed that the buffer associated with pNode is already large
** enough to accommodate the new entry. The buffer associated with pPrev
** is extended by this function if required.
**
** If an error (i.e. OOM condition) occurs, an SQLite error code is
** returned. Otherwise, SQLITE_OK.
*/
static int fts3AppendToNode(
  Blob *pNode,                    /* Current node image to append to */
  Blob *pPrev,                    /* Buffer containing previous term written */
................................................................................

  sqlite3_free(root.a);
  sqlite3_free(block.a);
  return rc;
}

/*
** This function is called after an incremental-merge operation has run to
** merge (or partially merge) two or more segments from absolute level
** iAbsLevel.
**
** Each input segment is either removed from the db completely (if all of
** its data was copied to the output segment by the incrmerge operation)
** or modified in place so that it no longer contains those entries that
** have been duplicated in the output segment.
................................................................................
  for(pDef=pCsr->pDeferred; pDef; pDef=pDef->pNext){
    fts3PendingListDelete(pDef->pList);
    pDef->pList = 0;
  }
}

/*
** Free all entries in the pCsr->pDeferred list. Entries are added to 
** this list using sqlite3Fts3DeferToken().
*/
void sqlite3Fts3FreeDeferredTokens(Fts3Cursor *pCsr){
  Fts3DeferredToken *pDef;
  Fts3DeferredToken *pNext;
  for(pDef=pCsr->pDeferred; pDef; pDef=pNext){
    pNext = pDef->pNext;
................................................................................
  return SQLITE_OK;
}
#endif

/*
** SQLite value pRowid contains the rowid of a row that may or may not be
** present in the FTS3 table. If it is, delete it and adjust the contents
** of subsidiary data structures accordingly.
*/
static int fts3DeleteByRowid(
  Fts3Table *p, 
  sqlite3_value *pRowid, 
  int *pnChng,                    /* IN/OUT: Decrement if row is deleted */
  u32 *aSzDel
){

Changes to ext/fts3/fts3speed.tcl.

115
116
117
118
119
120
121
122
close $::fd

puts "Success. Created files:"
puts "  fts3speed_insert.sql"
puts "  fts3speed_select.sql"
puts "  fts3speed_select2.sql"
puts "  fts3speed_optimize.sql"








<
115
116
117
118
119
120
121

close $::fd

puts "Success. Created files:"
puts "  fts3speed_insert.sql"
puts "  fts3speed_select.sql"
puts "  fts3speed_select2.sql"
puts "  fts3speed_optimize.sql"

Changes to ext/fts3/mkfts3amal.tcl.

61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
  set nstar [expr {60 - $n}]
  set stars [string range $s78 0 $nstar]
  puts $out "/************** $text $stars/"
}

# Read the source file named $filename and write it into the
# sqlite3.c output file.  If any #include statements are seen,
# process them approprately.
#
proc copy_file {filename} {
  global seen_hdr available_hdr out
  set tail [file tail $filename]
  section_comment "Begin file $tail"
  set in [open $filename r]
  while {![eof $in]} {







|







61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
  set nstar [expr {60 - $n}]
  set stars [string range $s78 0 $nstar]
  puts $out "/************** $text $stars/"
}

# Read the source file named $filename and write it into the
# sqlite3.c output file.  If any #include statements are seen,
# process them appropriately.
#
proc copy_file {filename} {
  global seen_hdr available_hdr out
  set tail [file tail $filename]
  section_comment "Begin file $tail"
  set in [open $filename r]
  while {![eof $in]} {

Changes to ext/fts3/tool/fts3cov.sh.

9
10
11
12
13
14
15
16

for f in `ls $srcdir/ext/fts3/*.c` 
do
  f=`basename $f`
  echo -ne "$f: "
  gcov -b $f | grep Taken | sed 's/Taken at least once://'
done








<
9
10
11
12
13
14
15


for f in `ls $srcdir/ext/fts3/*.c` 
do
  f=`basename $f`
  echo -ne "$f: "
  gcov -b $f | grep Taken | sed 's/Taken at least once://'
done

Changes to ext/fts3/unicode/mkunicode.tcl.

42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
...
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
  lappend aFlag $fPrev

  puts "/*"
  puts "** If the argument is a codepoint corresponding to a lowercase letter"
  puts "** in the ASCII range with a diacritic added, return the codepoint"
  puts "** of the ASCII letter only. For example, if passed 235 - \"LATIN"
  puts "** SMALL LETTER E WITH DIAERESIS\" - return 65 (\"LATIN SMALL LETTER"
  puts "** E\"). The resuls of passing a codepoint that corresponds to an"
  puts "** uppercase letter are undefined."
  puts "*/"
  puts "static int ${::remove_diacritic}(int c, int bComplex)\{"
  puts "  unsigned short aDia\[\] = \{"
  puts -nonewline "        0, "
  set i 1
  foreach r $lRange {
................................................................................
    puts "  if( r3 ) printf(\"categories(): Problem with code %d\\n\",code);"
    puts "  else printf(\"categories(): test passed\\n\");"
  }
  puts "  return (r1 || r2 || r3);"
  puts "\}"
}

# Proces the command line arguments. Exit early if they are not to
# our liking.
#
proc usage {} {
  puts -nonewline stderr "Usage: $::argv0 ?-test? ?-fts5? "
  puts            stderr "<CaseFolding.txt file> <UnicodeData.txt file>"
  exit 1
}







|







 







|







42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
...
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
  lappend aFlag $fPrev

  puts "/*"
  puts "** If the argument is a codepoint corresponding to a lowercase letter"
  puts "** in the ASCII range with a diacritic added, return the codepoint"
  puts "** of the ASCII letter only. For example, if passed 235 - \"LATIN"
  puts "** SMALL LETTER E WITH DIAERESIS\" - return 65 (\"LATIN SMALL LETTER"
  puts "** E\"). The results of passing a codepoint that corresponds to an"
  puts "** uppercase letter are undefined."
  puts "*/"
  puts "static int ${::remove_diacritic}(int c, int bComplex)\{"
  puts "  unsigned short aDia\[\] = \{"
  puts -nonewline "        0, "
  set i 1
  foreach r $lRange {
................................................................................
    puts "  if( r3 ) printf(\"categories(): Problem with code %d\\n\",code);"
    puts "  else printf(\"categories(): test passed\\n\");"
  }
  puts "  return (r1 || r2 || r3);"
  puts "\}"
}

# Process the command line arguments. Exit early if they are not to
# our liking.
#
proc usage {} {
  puts -nonewline stderr "Usage: $::argv0 ?-test? ?-fts5? "
  puts            stderr "<CaseFolding.txt file> <UnicodeData.txt file>"
  exit 1
}

Changes to ext/fts3/unicode/parseunicode.tcl.

66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
...
197
198
199
200
201
202
203
204
205
    set iDia   [expr "0x[lindex $character_decomposition_mapping 1]"]

    # Filter out upper-case characters, as they will be mapped to their
    # lower-case equivalents before this data is used.
    if {[info exists tl_lookup_table($iCode)]} continue

    # Check if this is an indirect mapping. If so, set bIndirect to true
    # and change $iAscii to the indirectly mappped ASCII character.
    set bIndirect 0
    if {[info exists dia($iDia)] && [info exists mapping($iAscii)]} {
      set iAscii $mapping($iAscii)
      set bIndirect 1
    }

    if { ($iAscii >= 97 && $iAscii <= 122)
................................................................................

    lappend lRet [list $code $general_category]
  }

  close $fd
  set lRet
}









|







 







<
<
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
...
197
198
199
200
201
202
203


    set iDia   [expr "0x[lindex $character_decomposition_mapping 1]"]

    # Filter out upper-case characters, as they will be mapped to their
    # lower-case equivalents before this data is used.
    if {[info exists tl_lookup_table($iCode)]} continue

    # Check if this is an indirect mapping. If so, set bIndirect to true
    # and change $iAscii to the indirectly mapped ASCII character.
    set bIndirect 0
    if {[info exists dia($iDia)] && [info exists mapping($iAscii)]} {
      set iAscii $mapping($iAscii)
      set bIndirect 1
    }

    if { ($iAscii >= 97 && $iAscii <= 122)
................................................................................

    lappend lRet [list $code $general_category]
  }

  close $fd
  set lRet
}


Changes to ext/fts5/extract_api_docs.tcl.

241
242
243
244
245
246
247
248
249
250
251
252
    default {
    }
  }
}
main $data

set ::fts5_docs_output












<
<
<
<
<
241
242
243
244
245
246
247





    default {
    }
  }
}
main $data

set ::fts5_docs_output





Changes to ext/fts5/fts5Int.h.

79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
#else
# define FTS5_CORRUPT SQLITE_CORRUPT_VTAB
#endif

/*
** The assert_nc() macro is similar to the assert() macro, except that it
** is used for assert() conditions that are true only if it can be 
** guranteed that the database is not corrupt.
*/
#ifdef SQLITE_DEBUG
extern int sqlite3_fts5_may_be_corrupt;
# define assert_nc(x) assert(sqlite3_fts5_may_be_corrupt || (x))
#else
# define assert_nc(x) assert(x)
#endif







|







79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
#else
# define FTS5_CORRUPT SQLITE_CORRUPT_VTAB
#endif

/*
** The assert_nc() macro is similar to the assert() macro, except that it
** is used for assert() conditions that are true only if it can be 
** guaranteed that the database is not corrupt.
*/
#ifdef SQLITE_DEBUG
extern int sqlite3_fts5_may_be_corrupt;
# define assert_nc(x) assert(sqlite3_fts5_may_be_corrupt || (x))
#else
# define assert_nc(x) assert(x)
#endif

Changes to ext/fts5/fts5_aux.c.

604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
        **   IDF = log( (N - nHit + 0.5) / (nHit + 0.5) )
        **
        ** where "N" is the total number of documents in the set and nHit
        ** is the number that contain at least one instance of the phrase
        ** under consideration.
        **
        ** The problem with this is that if (N < 2*nHit), the IDF is 
        ** negative. Which is undesirable. So the mimimum allowable IDF is
        ** (1e-6) - roughly the same as a term that appears in just over
        ** half of set of 5,000,000 documents.  */
        double idf = log( (nRow - nHit + 0.5) / (nHit + 0.5) );
        if( idf<=0.0 ) idf = 1e-6;
        p->aIDF[i] = idf;
      }
    }







|







604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
        **   IDF = log( (N - nHit + 0.5) / (nHit + 0.5) )
        **
        ** where "N" is the total number of documents in the set and nHit
        ** is the number that contain at least one instance of the phrase
        ** under consideration.
        **
        ** The problem with this is that if (N < 2*nHit), the IDF is 
        ** negative. Which is undesirable. So the minimum allowable IDF is
        ** (1e-6) - roughly the same as a term that appears in just over
        ** half of set of 5,000,000 documents.  */
        double idf = log( (nRow - nHit + 0.5) / (nHit + 0.5) );
        if( idf<=0.0 ) idf = 1e-6;
        p->aIDF[i] = idf;
      }
    }

Changes to ext/fts5/fts5_buffer.c.

296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
** Return true if character 't' may be part of an FTS5 bareword, or false
** otherwise. Characters that may be part of barewords:
**
**   * All non-ASCII characters,
**   * The 52 upper and lower case ASCII characters, and
**   * The 10 integer ASCII characters.
**   * The underscore character "_" (0x5F).
**   * The unicode "subsitute" character (0x1A).
*/
int sqlite3Fts5IsBareword(char t){
  u8 aBareword[128] = {
    0, 0, 0, 0, 0, 0, 0, 0,    0, 0, 0, 0, 0, 0, 0, 0,   /* 0x00 .. 0x0F */
    0, 0, 0, 0, 0, 0, 0, 0,    0, 0, 1, 0, 0, 0, 0, 0,   /* 0x10 .. 0x1F */
    0, 0, 0, 0, 0, 0, 0, 0,    0, 0, 0, 0, 0, 0, 0, 0,   /* 0x20 .. 0x2F */
    1, 1, 1, 1, 1, 1, 1, 1,    1, 1, 0, 0, 0, 0, 0, 0,   /* 0x30 .. 0x3F */







|







296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
** Return true if character 't' may be part of an FTS5 bareword, or false
** otherwise. Characters that may be part of barewords:
**
**   * All non-ASCII characters,
**   * The 52 upper and lower case ASCII characters, and
**   * The 10 integer ASCII characters.
**   * The underscore character "_" (0x5F).
**   * The unicode "substitute" character (0x1A).
*/
int sqlite3Fts5IsBareword(char t){
  u8 aBareword[128] = {
    0, 0, 0, 0, 0, 0, 0, 0,    0, 0, 0, 0, 0, 0, 0, 0,   /* 0x00 .. 0x0F */
    0, 0, 0, 0, 0, 0, 0, 0,    0, 0, 1, 0, 0, 0, 0, 0,   /* 0x10 .. 0x1F */
    0, 0, 0, 0, 0, 0, 0, 0,    0, 0, 0, 0, 0, 0, 0, 0,   /* 0x20 .. 0x2F */
    1, 1, 1, 1, 1, 1, 1, 1,    1, 1, 0, 0, 0, 0, 0, 0,   /* 0x30 .. 0x3F */

Changes to ext/fts5/fts5_expr.c.

2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
    }
  }
  pColset->nCol = iOut;
}

/*
** Recursively apply colset pColset to expression node pNode and all of
** its decendents. If (*ppFree) is not NULL, it contains a spare copy
** of pColset. This function may use the spare copy and set (*ppFree) to
** zero, or it may create copies of pColset using fts5CloneColset().
*/
static void fts5ParseSetColset(
  Fts5Parse *pParse, 
  Fts5ExprNode *pNode, 
  Fts5Colset *pColset,







|







2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
    }
  }
  pColset->nCol = iOut;
}

/*
** Recursively apply colset pColset to expression node pNode and all of
** its descendents. If (*ppFree) is not NULL, it contains a spare copy
** of pColset. This function may use the spare copy and set (*ppFree) to
** zero, or it may create copies of pColset using fts5CloneColset().
*/
static void fts5ParseSetColset(
  Fts5Parse *pParse, 
  Fts5ExprNode *pNode, 
  Fts5Colset *pColset,

Changes to ext/fts5/fts5_hash.c.

16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
..
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82

#include "fts5Int.h"

typedef struct Fts5HashEntry Fts5HashEntry;

/*
** This file contains the implementation of an in-memory hash table used
** to accumuluate "term -> doclist" content before it is flused to a level-0
** segment.
*/


struct Fts5Hash {
  int eDetail;                    /* Copy of Fts5Config.eDetail */
  int *pnByte;                    /* Pointer to bytes counter */
................................................................................
  u8 bContent;                    /* Set content-flag (detail=none mode) */
  i16 iCol;                       /* Column of last value written */
  int iPos;                       /* Position of last value written */
  i64 iRowid;                     /* Rowid of last value written */
};

/*
** Eqivalent to:
**
**   char *fts5EntryKey(Fts5HashEntry *pEntry){ return zKey; }
*/
#define fts5EntryKey(p) ( ((char *)(&(p)[1])) )


/*







|







 







|







16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
..
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82

#include "fts5Int.h"

typedef struct Fts5HashEntry Fts5HashEntry;

/*
** This file contains the implementation of an in-memory hash table used
** to accumulate "term -> doclist" content before it is flused to a level-0
** segment.
*/


struct Fts5Hash {
  int eDetail;                    /* Copy of Fts5Config.eDetail */
  int *pnByte;                    /* Pointer to bytes counter */
................................................................................
  u8 bContent;                    /* Set content-flag (detail=none mode) */
  i16 iCol;                       /* Column of last value written */
  int iPos;                       /* Position of last value written */
  i64 iRowid;                     /* Rowid of last value written */
};

/*
** Equivalent to:
**
**   char *fts5EntryKey(Fts5HashEntry *pEntry){ return zKey; }
*/
#define fts5EntryKey(p) ( ((char *)(&(p)[1])) )


/*

Changes to ext/fts5/fts5_index.c.

4134
4135
4136
4137
4138
4139
4140
4141
4142
4143
4144
4145
4146
4147
4148
....
4205
4206
4207
4208
4209
4210
4211
4212
4213
4214
4215
4216
4217
4218
4219
....
6166
6167
6168
6169
6170
6171
6172
6173
6174
6175
6176
6177
6178
6179
6180
  Fts5Buffer buf;
  memset(&buf, 0, sizeof(Fts5Buffer));
  for(i=0; i<pIter->nSeg && p->rc==SQLITE_OK; i++){
    Fts5SegIter *pSeg = &pIter->aSeg[i];
    if( pSeg->pSeg==0 ){
      /* no-op */
    }else if( pSeg->pLeaf==0 ){
      /* All keys from this input segment have been transfered to the output.
      ** Set both the first and last page-numbers to 0 to indicate that the
      ** segment is now empty. */
      pSeg->pSeg->pgnoLast = 0;
      pSeg->pSeg->pgnoFirst = 0;
    }else{
      int iOff = pSeg->iTermLeafOffset;     /* Offset on new first leaf page */
      i64 iLeafRowid;
................................................................................
}

/*
**
*/
static void fts5IndexMergeLevel(
  Fts5Index *p,                   /* FTS5 backend object */
  Fts5Structure **ppStruct,       /* IN/OUT: Stucture of index */
  int iLvl,                       /* Level to read input from */
  int *pnRem                      /* Write up to this many output leaves */
){
  Fts5Structure *pStruct = *ppStruct;
  Fts5StructureLevel *pLvl = &pStruct->aLevel[iLvl];
  Fts5StructureLevel *pLvlOut;
  Fts5Iter *pIter = 0;       /* Iterator to read input data */
................................................................................
  *pbDlidx = (int)(iRowid & 0x0001);
  iRowid >>= FTS5_DATA_DLI_B;

  *piSegid = (int)(iRowid & (((i64)1 << FTS5_DATA_ID_B) - 1));
}

static void fts5DebugRowid(int *pRc, Fts5Buffer *pBuf, i64 iKey){
  int iSegid, iHeight, iPgno, bDlidx;       /* Rowid compenents */
  fts5DecodeRowid(iKey, &iSegid, &bDlidx, &iHeight, &iPgno);

  if( iSegid==0 ){
    if( iKey==FTS5_AVERAGES_ROWID ){
      sqlite3Fts5BufferAppendPrintf(pRc, pBuf, "{averages} ");
    }else{
      sqlite3Fts5BufferAppendPrintf(pRc, pBuf, "{structure}");







|







 







|







 







|







4134
4135
4136
4137
4138
4139
4140
4141
4142
4143
4144
4145
4146
4147
4148
....
4205
4206
4207
4208
4209
4210
4211
4212
4213
4214
4215
4216
4217
4218
4219
....
6166
6167
6168
6169
6170
6171
6172
6173
6174
6175
6176
6177
6178
6179
6180
  Fts5Buffer buf;
  memset(&buf, 0, sizeof(Fts5Buffer));
  for(i=0; i<pIter->nSeg && p->rc==SQLITE_OK; i++){
    Fts5SegIter *pSeg = &pIter->aSeg[i];
    if( pSeg->pSeg==0 ){
      /* no-op */
    }else if( pSeg->pLeaf==0 ){
      /* All keys from this input segment have been transferred to the output.
      ** Set both the first and last page-numbers to 0 to indicate that the
      ** segment is now empty. */
      pSeg->pSeg->pgnoLast = 0;
      pSeg->pSeg->pgnoFirst = 0;
    }else{
      int iOff = pSeg->iTermLeafOffset;     /* Offset on new first leaf page */
      i64 iLeafRowid;
................................................................................
}

/*
**
*/
static void fts5IndexMergeLevel(
  Fts5Index *p,                   /* FTS5 backend object */
  Fts5Structure **ppStruct,       /* IN/OUT: Structure of index */
  int iLvl,                       /* Level to read input from */
  int *pnRem                      /* Write up to this many output leaves */
){
  Fts5Structure *pStruct = *ppStruct;
  Fts5StructureLevel *pLvl = &pStruct->aLevel[iLvl];
  Fts5StructureLevel *pLvlOut;
  Fts5Iter *pIter = 0;       /* Iterator to read input data */
................................................................................
  *pbDlidx = (int)(iRowid & 0x0001);
  iRowid >>= FTS5_DATA_DLI_B;

  *piSegid = (int)(iRowid & (((i64)1 << FTS5_DATA_ID_B) - 1));
}

static void fts5DebugRowid(int *pRc, Fts5Buffer *pBuf, i64 iKey){
  int iSegid, iHeight, iPgno, bDlidx;       /* Rowid components */
  fts5DecodeRowid(iKey, &iSegid, &bDlidx, &iHeight, &iPgno);

  if( iSegid==0 ){
    if( iKey==FTS5_AVERAGES_ROWID ){
      sqlite3Fts5BufferAppendPrintf(pRc, pBuf, "{averages} ");
    }else{
      sqlite3Fts5BufferAppendPrintf(pRc, pBuf, "{structure}");

Changes to ext/fts5/fts5_main.c.

1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
      eConflict = sqlite3_vtab_on_conflict(pConfig->db);
    }

    assert( eType0==SQLITE_INTEGER || eType0==SQLITE_NULL );
    assert( nArg!=1 || eType0==SQLITE_INTEGER );

    /* Filter out attempts to run UPDATE or DELETE on contentless tables.
    ** This is not suported.  */
    if( eType0==SQLITE_INTEGER && fts5IsContentless(pTab) ){
      pTab->p.base.zErrMsg = sqlite3_mprintf(
          "cannot %s contentless fts5 table: %s", 
          (nArg>1 ? "UPDATE" : "DELETE from"), pConfig->zName
      );
      rc = SQLITE_ERROR;
    }







|







1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
      eConflict = sqlite3_vtab_on_conflict(pConfig->db);
    }

    assert( eType0==SQLITE_INTEGER || eType0==SQLITE_NULL );
    assert( nArg!=1 || eType0==SQLITE_INTEGER );

    /* Filter out attempts to run UPDATE or DELETE on contentless tables.
    ** This is not supported.  */
    if( eType0==SQLITE_INTEGER && fts5IsContentless(pTab) ){
      pTab->p.base.zErrMsg = sqlite3_mprintf(
          "cannot %s contentless fts5 table: %s", 
          (nArg>1 ? "UPDATE" : "DELETE from"), pConfig->zName
      );
      rc = SQLITE_ERROR;
    }

Changes to ext/fts5/fts5_test_mi.c.

383
384
385
386
387
388
389
390
391
392
393
394
395
396
397

  if( rc==SQLITE_OK ){
    rc = fts5MatchinfoIter(pApi, pFts, p, fts5MatchinfoLocalCb);
  }
  if( rc!=SQLITE_OK ){
    sqlite3_result_error_code(pCtx, rc);
  }else{
    /* No errors has occured, so return a copy of the array of integers. */
    int nByte = p->nRet * sizeof(u32);
    sqlite3_result_blob(pCtx, (void*)p->aRet, nByte, SQLITE_TRANSIENT);
  }
}

int sqlite3Fts5TestRegisterMatchinfo(sqlite3 *db){
  int rc;                         /* Return code */







|







383
384
385
386
387
388
389
390
391
392
393
394
395
396
397

  if( rc==SQLITE_OK ){
    rc = fts5MatchinfoIter(pApi, pFts, p, fts5MatchinfoLocalCb);
  }
  if( rc!=SQLITE_OK ){
    sqlite3_result_error_code(pCtx, rc);
  }else{
    /* No errors has occurred, so return a copy of the array of integers. */
    int nByte = p->nRet * sizeof(u32);
    sqlite3_result_blob(pCtx, (void*)p->aRet, nByte, SQLITE_TRANSIENT);
  }
}

int sqlite3Fts5TestRegisterMatchinfo(sqlite3 *db){
  int rc;                         /* Return code */

Changes to ext/fts5/fts5_tokenize.c.

659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
  int bCons = bPrevCons;

  /* Scan for a vowel */
  for(i=0; i<nStem; i++){
    if( 0==(bCons = !fts5PorterIsVowel(zStem[i], bCons)) ) break;
  }

  /* Scan for a consonent */
  for(i++; i<nStem; i++){
    if( (bCons = !fts5PorterIsVowel(zStem[i], bCons)) ) return i+1;
  }
  return 0;
}

/* porter rule condition: (m > 0) */







|







659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
  int bCons = bPrevCons;

  /* Scan for a vowel */
  for(i=0; i<nStem; i++){
    if( 0==(bCons = !fts5PorterIsVowel(zStem[i], bCons)) ) break;
  }

  /* Scan for a consonant */
  for(i++; i<nStem; i++){
    if( (bCons = !fts5PorterIsVowel(zStem[i], bCons)) ) return i+1;
  }
  return 0;
}

/* porter rule condition: (m > 0) */

Changes to ext/fts5/fts5_unicode2.c.

21
22
23
24
25
26
27
28
29
30
31
32
33
34
35


/*
** If the argument is a codepoint corresponding to a lowercase letter
** in the ASCII range with a diacritic added, return the codepoint
** of the ASCII letter only. For example, if passed 235 - "LATIN
** SMALL LETTER E WITH DIAERESIS" - return 65 ("LATIN SMALL LETTER
** E"). The resuls of passing a codepoint that corresponds to an
** uppercase letter are undefined.
*/
static int fts5_remove_diacritic(int c, int bComplex){
  unsigned short aDia[] = {
        0,  1797,  1848,  1859,  1891,  1928,  1940,  1995, 
     2024,  2040,  2060,  2110,  2168,  2206,  2264,  2286, 
     2344,  2383,  2472,  2488,  2516,  2596,  2668,  2732, 







|







21
22
23
24
25
26
27
28
29
30
31
32
33
34
35


/*
** If the argument is a codepoint corresponding to a lowercase letter
** in the ASCII range with a diacritic added, return the codepoint
** of the ASCII letter only. For example, if passed 235 - "LATIN
** SMALL LETTER E WITH DIAERESIS" - return 65 ("LATIN SMALL LETTER
** E"). The results of passing a codepoint that corresponds to an
** uppercase letter are undefined.
*/
static int fts5_remove_diacritic(int c, int bComplex){
  unsigned short aDia[] = {
        0,  1797,  1848,  1859,  1891,  1928,  1940,  1995, 
     2024,  2040,  2060,  2110,  2168,  2206,  2264,  2286, 
     2344,  2383,  2472,  2488,  2516,  2596,  2668,  2732, 

Changes to ext/fts5/fts5_vocab.c.

58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
  sqlite3_stmt *pStmt;            /* Statement holding lock on pIndex */
  Fts5Table *pFts5;               /* Associated FTS5 table */

  int bEof;                       /* True if this cursor is at EOF */
  Fts5IndexIter *pIter;           /* Term/rowid iterator object */

  int nLeTerm;                    /* Size of zLeTerm in bytes */
  char *zLeTerm;                  /* (term <= $zLeTerm) paramater, or NULL */

  /* These are used by 'col' tables only */
  int iCol;
  i64 *aCnt;
  i64 *aDoc;

  /* Output values used by all tables. */







|







58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
  sqlite3_stmt *pStmt;            /* Statement holding lock on pIndex */
  Fts5Table *pFts5;               /* Associated FTS5 table */

  int bEof;                       /* True if this cursor is at EOF */
  Fts5IndexIter *pIter;           /* Term/rowid iterator object */

  int nLeTerm;                    /* Size of zLeTerm in bytes */
  char *zLeTerm;                  /* (term <= $zLeTerm) parameter, or NULL */

  /* These are used by 'col' tables only */
  int iCol;
  i64 *aCnt;
  i64 *aDoc;

  /* Output values used by all tables. */

Changes to ext/fts5/mkportersteps.tcl.

213
214
215
216
217
218
219
220
221
222
}
puts [string trim {
/* 
** GENERATED CODE ENDS HERE (mkportersteps.tcl)
***************************************************************************
**************************************************************************/
}]










<
<
<
213
214
215
216
217
218
219



}
puts [string trim {
/* 
** GENERATED CODE ENDS HERE (mkportersteps.tcl)
***************************************************************************
**************************************************************************/
}]



Changes to ext/fts5/test/fts5_common.tcl.

640
641
642
643
644
645
646
647
    }
  }
  sqlite3_fts5_create_tokenizer db tclnum tclnum_create
}
#
# End of tokenizer code.
#-------------------------------------------------------------------------








<
640
641
642
643
644
645
646

    }
  }
  sqlite3_fts5_create_tokenizer db tclnum tclnum_create
}
#
# End of tokenizer code.
#-------------------------------------------------------------------------

Changes to ext/fts5/test/fts5aux.test.

304
305
306
307
308
309
310
311
} {1 {unable to use function firstcol in the requested context}}

do_catchsql_test 10.1.4 {
  SELECT group_concat(firstcol(t1), '.') FROM t1 GROUP BY rowid
} {1 {unable to use function firstcol in the requested context}}

finish_test








<
304
305
306
307
308
309
310

} {1 {unable to use function firstcol in the requested context}}

do_catchsql_test 10.1.4 {
  SELECT group_concat(firstcol(t1), '.') FROM t1 GROUP BY rowid
} {1 {unable to use function firstcol in the requested context}}

finish_test

Changes to ext/fts5/test/fts5content.test.

290
291
292
293
294
295
296
297
  SELECT count(*) FROM t1;
} {1 {recursively defined fts5 content table}}
do_catchsql_test 7.2.5 { 
  SELECT * FROM t1('abc') ORDER BY rank;
} {1 {recursively defined fts5 content table}}

finish_test








<
290
291
292
293
294
295
296

  SELECT count(*) FROM t1;
} {1 {recursively defined fts5 content table}}
do_catchsql_test 7.2.5 { 
  SELECT * FROM t1('abc') ORDER BY rank;
} {1 {recursively defined fts5 content table}}

finish_test

Changes to ext/fts5/test/fts5corrupt3.test.

10322
10323
10324
10325
10326
10327
10328
10329

do_catchsql_test 69.2 {
  SELECT * FROM t1 WHERE a MATCH 'fx*'
} {1 {database disk image is malformed}}

sqlite3_fts5_may_be_corrupt 0
finish_test








<
10322
10323
10324
10325
10326
10327
10328


do_catchsql_test 69.2 {
  SELECT * FROM t1 WHERE a MATCH 'fx*'
} {1 {database disk image is malformed}}

sqlite3_fts5_may_be_corrupt 0
finish_test

Changes to ext/fts5/test/fts5corrupt4.test.

54
55
56
57
58
59
60
61
      execsql ROLLBACK
    }
  } {}
}

sqlite3_fts5_may_be_corrupt 0
finish_test








<
54
55
56
57
58
59
60

      execsql ROLLBACK
    }
  } {}
}

sqlite3_fts5_may_be_corrupt 0
finish_test

Changes to ext/fts5/test/fts5interrupt.test.

60
61
62
63
64
65
66
67
      }
      set {} {}
    } {}
  }
}

finish_test








<
60
61
62
63
64
65
66

      }
      set {} {}
    } {}
  }
}

finish_test

Changes to ext/fts5/test/fts5misc.test.

320
321
322
323
324
325
326
327
  SELECT * FROM ft NATURAL JOIN t1 WHERE ft MATCH('b')
} {b world {}}
do_execsql_test 12.3 {
  SELECT * FROM t2 JOIN ft USING (ft)
} {3 4 b b}

finish_test








<
320
321
322
323
324
325
326

  SELECT * FROM ft NATURAL JOIN t1 WHERE ft MATCH('b')
} {b world {}}
do_execsql_test 12.3 {
  SELECT * FROM t2 JOIN ft USING (ft)
} {3 4 b b}

finish_test

Changes to ext/fts5/test/fts5multi.test.

92
93
94
95
96
97
98
99
  SELECT rowid FROM t1 WHERE t1 MATCH '(NOT' AND t1 MATCH 'aa bb';
} {1 {fts5: syntax error near "NOT"}}
do_catchsql_test 2.1.2 {
  SELECT rowid FROM t1 WHERE t1 MATCH 'aa bb' AND t1 MATCH '(NOT';
} {1 {fts5: syntax error near "NOT"}}

finish_test








<
92
93
94
95
96
97
98

  SELECT rowid FROM t1 WHERE t1 MATCH '(NOT' AND t1 MATCH 'aa bb';
} {1 {fts5: syntax error near "NOT"}}
do_catchsql_test 2.1.2 {
  SELECT rowid FROM t1 WHERE t1 MATCH 'aa bb' AND t1 MATCH '(NOT';
} {1 {fts5: syntax error near "NOT"}}

finish_test

Changes to ext/fts5/test/fts5savepoint.test.

78
79
80
81
82
83
84
85
      INSERT INTO vt0 VALUES('x');
    COMMIT;
    INSERT INTO vt0(vt0) VALUES('integrity-check');
  }
}

finish_test








<
78
79
80
81
82
83
84

      INSERT INTO vt0 VALUES('x');
    COMMIT;
    INSERT INTO vt0(vt0) VALUES('integrity-check');
  }
}

finish_test

Changes to ext/fts5/test/fts5simple3.test.

77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
  INSERT INTO x3 VALUES('a b c');
  INSERT INTO x3 VALUES('c b a');
  INSERT INTO x3 VALUES('o t t');
  SELECT * FROM x3('x OR y OR z');
}

#-------------------------------------------------------------------------
# Test that a crash occuring when the second or subsequent tokens in a
# phrase matched zero rows has been fixed.
#
do_execsql_test 4.0 {
  CREATE VIRTUAL TABLE t1 USING fts5(x);
  INSERT INTO t1 VALUES('ab');
  INSERT INTO t1 VALUES('cd');
  INSERT INTO t1 VALUES('ab cd');







|







77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
  INSERT INTO x3 VALUES('a b c');
  INSERT INTO x3 VALUES('c b a');
  INSERT INTO x3 VALUES('o t t');
  SELECT * FROM x3('x OR y OR z');
}

#-------------------------------------------------------------------------
# Test that a crash occurring when the second or subsequent tokens in a
# phrase matched zero rows has been fixed.
#
do_execsql_test 4.0 {
  CREATE VIRTUAL TABLE t1 USING fts5(x);
  INSERT INTO t1 VALUES('ab');
  INSERT INTO t1 VALUES('cd');
  INSERT INTO t1 VALUES('ab cd');

Changes to ext/fts5/test/fts5umlaut.test.

58
59
60
61
62
63
64
65
    DELETE FROM t2;
    INSERT INTO t2(rowid, x) VALUES (1, $q);
    SELECT count(*) FROM t2('Ha Noi')
  } $res2
}

finish_test








<
58
59
60
61
62
63
64

    DELETE FROM t2;
    INSERT INTO t2(rowid, x) VALUES (1, $q);
    SELECT count(*) FROM t2('Ha Noi')
  } $res2
}

finish_test

Changes to ext/fts5/tool/fts5speed.tcl.

56
57
58
59
60
61
62
63
64
} else {
  puts "result:  [db eval $sql]"
}

for {set i 1} {$i < $nRepeat} {incr i} {
  db eval $sql
}









<
<
56
57
58
59
60
61
62


} else {
  puts "result:  [db eval $sql]"
}

for {set i 1} {$i < $nRepeat} {incr i} {
  db eval $sql
}


Changes to ext/fts5/tool/fts5txt2db.tcl.

220
221
222
223
224
225
226
227
228
229
    foreach c $cols s $A(colsize) {
      set R($c) [lrange $tokens $i [expr $i+$s-1]]
      incr i $s
    }
    db eval $sql
  }
if {$A(trans)} { db eval COMMIT }










<
<
<
220
221
222
223
224
225
226



    foreach c $cols s $A(colsize) {
      set R($c) [lrange $tokens $i [expr $i+$s-1]]
      incr i $s
    }
    db eval $sql
  }
if {$A(trans)} { db eval COMMIT }



Changes to ext/fts5/tool/loadfts5.tcl.

6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
...
160
161
162
163
164
165
166
167
168
169
170
171
172
  close $fd
  return $data
}

set ::nRow 0
set ::nRowPerDot 1000

proc load_hierachy {dir} {
  foreach f [glob -nocomplain -dir $dir *] {
    if {$::O(limit) && $::nRow>=$::O(limit)} break
    if {[file isdir $f]} {
      load_hierachy $f
    } else {
      db eval { INSERT INTO t1 VALUES($f, loadfile($f)) }
      incr ::nRow

      if {$::O(trans) && ($::nRow % $::O(trans))==0} {
        db eval { COMMIT }
        db eval { INSERT INTO t1(t1) VALUES('integrity-check') }
................................................................................
  }
  if {$O(crisismerge)>=0} {
    if {$O(vtab) == "fts5"} {
      db eval {INSERT INTO t1(t1, rank) VALUES('crisismerge', $O(crisismerge))}
    } else {
    }
  }
  load_hierachy [lindex $argv end]
db eval COMMIT
puts ""










|



|







 







|


<
<
<
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
...
160
161
162
163
164
165
166
167
168
169



  close $fd
  return $data
}

set ::nRow 0
set ::nRowPerDot 1000

proc load_hierarchy {dir} {
  foreach f [glob -nocomplain -dir $dir *] {
    if {$::O(limit) && $::nRow>=$::O(limit)} break
    if {[file isdir $f]} {
      load_hierarchy $f
    } else {
      db eval { INSERT INTO t1 VALUES($f, loadfile($f)) }
      incr ::nRow

      if {$::O(trans) && ($::nRow % $::O(trans))==0} {
        db eval { COMMIT }
        db eval { INSERT INTO t1(t1) VALUES('integrity-check') }
................................................................................
  }
  if {$O(crisismerge)>=0} {
    if {$O(vtab) == "fts5"} {
      db eval {INSERT INTO t1(t1, rank) VALUES('crisismerge', $O(crisismerge))}
    } else {
    }
  }
  load_hierarchy [lindex $argv end]
db eval COMMIT
puts ""



Changes to ext/fts5/tool/showfts5.tcl.

86
87
88
89
90
91
92
93
94
95
96
97

if {$O(segments)} {
  puts ""
  db eval "SELECT fts5_decode(rowid, block) AS d FROM ${tbl}_data WHERE id>10" {
    puts $d
  }
}












<
<
<
<
<
86
87
88
89
90
91
92






if {$O(segments)} {
  puts ""
  db eval "SELECT fts5_decode(rowid, block) AS d FROM ${tbl}_data WHERE id>10" {
    puts $d
  }
}





Changes to ext/icu/README.txt.

87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
          australian_penpal_name TEXT COLLATE australian,
          turkish_penpal_name    TEXT COLLATE turkish
        );
  
  1.4 SQL REGEXP Operator

    This extension provides an implementation of the SQL binary
    comparision operator "REGEXP", based on the regular expression functions
    provided by the ICU library. The syntax of the operator is as described
    in SQLite documentation:

        <string> REGEXP <re-pattern>

    This extension uses the ICU defaults for regular expression matching
    behavior. Specifically, this means that:







|







87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
          australian_penpal_name TEXT COLLATE australian,
          turkish_penpal_name    TEXT COLLATE turkish
        );
  
  1.4 SQL REGEXP Operator

    This extension provides an implementation of the SQL binary
    comparison operator "REGEXP", based on the regular expression functions
    provided by the ICU library. The syntax of the operator is as described
    in SQLite documentation:

        <string> REGEXP <re-pattern>

    This extension uses the ICU defaults for regular expression matching
    behavior. Specifically, this means that:

Changes to ext/lsm1/lsm-test/lsmtest5.c.

553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
      iPrng += sizeof(aRnd);
      snprintf(aValue, sizeof(aValue), "%d.%s", iThread, aRnd);
      nWrite += dbWriteOperation(&p->param, pDb, iKey, aValue, &rc);
    }
  }
  testClose(&pDb);

  /* If an error has occured, set the thread error code and the threadset 
  ** halt flag to tell the other test threads to halt. Otherwise, set the
  ** thread error code to 0 and post a message with the number of read
  ** and write operations completed.  */
  if( rc ){
    testThreadSetResult(pThreadSet, iThread, rc, 0);
    testThreadSetHalt(pThreadSet);
  }else{







|







553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
      iPrng += sizeof(aRnd);
      snprintf(aValue, sizeof(aValue), "%d.%s", iThread, aRnd);
      nWrite += dbWriteOperation(&p->param, pDb, iKey, aValue, &rc);
    }
  }
  testClose(&pDb);

  /* If an error has occurred, set the thread error code and the threadset 
  ** halt flag to tell the other test threads to halt. Otherwise, set the
  ** thread error code to 0 and post a message with the number of read
  ** and write operations completed.  */
  if( rc ){
    testThreadSetResult(pThreadSet, iThread, rc, 0);
    testThreadSetHalt(pThreadSet);
  }else{

Changes to ext/lsm1/lsmInt.h.

564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
  u32 iBlk;                       /* Block number */
  i64 iId;                        /* Largest snapshot id to use this block */
};

/*
** A snapshot of a database. A snapshot contains all the information required
** to read or write a database file on disk. See the description of struct
** Database below for futher details.
*/
struct Snapshot {
  Database *pDatabase;            /* Database this snapshot belongs to */
  u32 iCmpId;                     /* Id of compression scheme */
  Level *pLevel;                  /* Pointer to level 0 of snapshot (or NULL) */
  i64 iId;                        /* Snapshot id */
  i64 iLogOff;                    /* Log file offset */







|







564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
  u32 iBlk;                       /* Block number */
  i64 iId;                        /* Largest snapshot id to use this block */
};

/*
** A snapshot of a database. A snapshot contains all the information required
** to read or write a database file on disk. See the description of struct
** Database below for further details.
*/
struct Snapshot {
  Database *pDatabase;            /* Database this snapshot belongs to */
  u32 iCmpId;                     /* Id of compression scheme */
  Level *pLevel;                  /* Pointer to level 0 of snapshot (or NULL) */
  i64 iId;                        /* Snapshot id */
  i64 iLogOff;                    /* Log file offset */

Changes to ext/lsm1/lsm_ckpt.c.

165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
** The actual value used may be configured using LSM_CONFIG_MAX_FREELIST.
*/

/*
** The argument to this macro must be of type u32. On a little-endian
** architecture, it returns the u32 value that results from interpreting
** the 4 bytes as a big-endian value. On a big-endian architecture, it
** returns the value that would be produced by intepreting the 4 bytes
** of the input value as a little-endian integer.
*/
#define BYTESWAP32(x) ( \
   (((x)&0x000000FF)<<24) + (((x)&0x0000FF00)<<8)  \
 + (((x)&0x00FF0000)>>8)  + (((x)&0xFF000000)>>24) \
)








|







165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
** The actual value used may be configured using LSM_CONFIG_MAX_FREELIST.
*/

/*
** The argument to this macro must be of type u32. On a little-endian
** architecture, it returns the u32 value that results from interpreting
** the 4 bytes as a big-endian value. On a big-endian architecture, it
** returns the value that would be produced by interpreting the 4 bytes
** of the input value as a little-endian integer.
*/
#define BYTESWAP32(x) ( \
   (((x)&0x000000FF)<<24) + (((x)&0x0000FF00)<<8)  \
 + (((x)&0x00FF0000)>>8)  + (((x)&0xFF000000)>>24) \
)

Changes to ext/lsm1/lsm_file.c.

643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
...
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803

    /* Make a copy of the database and log file names. */
    memcpy(pFS->zDb, zDb, nDb+1);
    memcpy(pFS->zLog, zDb, nDb);
    memcpy(&pFS->zLog[nDb], "-log", 5);

    /* Allocate the hash-table here. At some point, it should be changed
    ** so that it can grow dynamicly. */
    pFS->nCacheMax = 2048*1024 / pFS->nPagesize;
    pFS->nHash = 4096;
    pFS->apHash = lsmMallocZeroRc(pDb->pEnv, sizeof(Page *) * pFS->nHash, &rc);

    /* Open the database file */
    pLsmFile = lsmDbRecycleFd(pDb);
    if( pLsmFile ){
................................................................................
** when the FileSystem object is destroyed, as this would cause any POSIX
** locks held by the other connections to be silently dropped (see "man close"
** for details). Instead, the file-descriptor is stored in a list by the
** lsm_shared.c module until it is either closed or reused.
**
** This function returns a pointer to an object that can be linked into
** the list described above. The returned object now 'owns' the database
** file descriptr, so that when the FileSystem object is destroyed, it
** will not be closed. 
**
** This function may be called at most once in the life-time of a 
** FileSystem object. The results of any operations involving the database 
** file descriptor are undefined once this function has been called.
**
** None of this is necessary on non-POSIX systems. But we do it anyway in







|







 







|







643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
...
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803

    /* Make a copy of the database and log file names. */
    memcpy(pFS->zDb, zDb, nDb+1);
    memcpy(pFS->zLog, zDb, nDb);
    memcpy(&pFS->zLog[nDb], "-log", 5);

    /* Allocate the hash-table here. At some point, it should be changed
    ** so that it can grow dynamically. */
    pFS->nCacheMax = 2048*1024 / pFS->nPagesize;
    pFS->nHash = 4096;
    pFS->apHash = lsmMallocZeroRc(pDb->pEnv, sizeof(Page *) * pFS->nHash, &rc);

    /* Open the database file */
    pLsmFile = lsmDbRecycleFd(pDb);
    if( pLsmFile ){
................................................................................
** when the FileSystem object is destroyed, as this would cause any POSIX
** locks held by the other connections to be silently dropped (see "man close"
** for details). Instead, the file-descriptor is stored in a list by the
** lsm_shared.c module until it is either closed or reused.
**
** This function returns a pointer to an object that can be linked into
** the list described above. The returned object now 'owns' the database
** file descriptor, so that when the FileSystem object is destroyed, it
** will not be closed. 
**
** This function may be called at most once in the life-time of a 
** FileSystem object. The results of any operations involving the database 
** file descriptor are undefined once this function has been called.
**
** None of this is necessary on non-POSIX systems. But we do it anyway in

Changes to ext/lsm1/lsm_log.c.

754
755
756
757
758
759
760
761
762
763
764
765
766
767
768

  pMark->iOff = pLog->iOff + pLog->buf.n;
  pMark->cksum0 = pLog->cksum0;
  pMark->cksum1 = pLog->cksum1;
}

/*
** Seek (rewind) back to the log file offset stored by an ealier call to
** lsmLogTell() in *pMark.
*/
void lsmLogSeek(
  lsm_db *pDb,                    /* Database handle */
  LogMark *pMark                  /* Object containing log offset to seek to */
){
  LogWriter *pLog;







|







754
755
756
757
758
759
760
761
762
763
764
765
766
767
768

  pMark->iOff = pLog->iOff + pLog->buf.n;
  pMark->cksum0 = pLog->cksum0;
  pMark->cksum1 = pLog->cksum1;
}

/*
** Seek (rewind) back to the log file offset stored by an earlier call to
** lsmLogTell() in *pMark.
*/
void lsmLogSeek(
  lsm_db *pDb,                    /* Database handle */
  LogMark *pMark                  /* Object containing log offset to seek to */
){
  LogWriter *pLog;

Changes to ext/lsm1/lsm_main.c.

767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
  }

  /* Allocate the multi-cursor. */
  if( rc==LSM_OK ){
    rc = lsmMCursorNew(pDb, &pCsr);
  }

  /* If an error has occured, set the output to NULL and delete any partially
  ** allocated cursor. If this means there are no open cursors, release the
  ** client snapshot.  */
  if( rc!=LSM_OK ){
    lsmMCursorClose(pCsr, 0);
    dbReleaseClientSnapshot(pDb);
  }








|







767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
  }

  /* Allocate the multi-cursor. */
  if( rc==LSM_OK ){
    rc = lsmMCursorNew(pDb, &pCsr);
  }

  /* If an error has occurred, set the output to NULL and delete any partially
  ** allocated cursor. If this means there are no open cursors, release the
  ** client snapshot.  */
  if( rc!=LSM_OK ){
    lsmMCursorClose(pCsr, 0);
    dbReleaseClientSnapshot(pDb);
  }

Changes to ext/lsm1/lsm_shared.c.

1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
  u32 nWrite = 0;                 /* Number of pages checkpointed */

  /* Attempt the checkpoint. If successful, nWrite is set to the number of
  ** pages written between this and the previous checkpoint.  */
  rc = lsmCheckpointWrite(pDb, &nWrite);

  /* If required, calculate the output variable (KB of data checkpointed). 
  ** Set it to zero if an error occured.  */
  if( pnKB ){
    int nKB = 0;
    if( rc==LSM_OK && nWrite ){
      nKB = (((i64)nWrite * lsmFsPageSize(pDb->pFS)) + 1023) / 1024;
    }
    *pnKB = nKB;
  }

  return rc;
}







|










1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
  u32 nWrite = 0;                 /* Number of pages checkpointed */

  /* Attempt the checkpoint. If successful, nWrite is set to the number of
  ** pages written between this and the previous checkpoint.  */
  rc = lsmCheckpointWrite(pDb, &nWrite);

  /* If required, calculate the output variable (KB of data checkpointed). 
  ** Set it to zero if an error occurred.  */
  if( pnKB ){
    int nKB = 0;
    if( rc==LSM_OK && nWrite ){
      nKB = (((i64)nWrite * lsmFsPageSize(pDb->pFS)) + 1023) / 1024;
    }
    *pnKB = nKB;
  }

  return rc;
}

Changes to ext/lsm1/lsm_sorted.c.

2648
2649
2650
2651
2652
2653
2654
2655
2656
2657
2658
2659
2660
2661
2662
....
3580
3581
3582
3583
3584
3585
3586
3587
3588
3589
3590
3591
3592
3593
3594
....
3980
3981
3982
3983
3984
3985
3986
3987
3988
3989
3990
3991
3992
3993
3994
}

int lsmSortedLoadFreelist(
  lsm_db *pDb,                    /* Database handle (must be worker) */
  void **ppVal,                   /* OUT: Blob containing LSM free-list */
  int *pnVal                      /* OUT: Size of *ppVal blob in bytes */
){
  MultiCursor *pCsr;              /* Cursor used to retreive free-list */
  int rc = LSM_OK;                /* Return Code */

  assert( pDb->pWorker );
  assert( *ppVal==0 && *pnVal==0 );

  pCsr = multiCursorNew(pDb, &rc);
  if( pCsr ){
................................................................................
  LsmPgno iKeyPg,
  void *pKey,
  int nKey
){
  Hierarchy *p = &pMW->hier;
  lsm_db *pDb = pMW->pDb;         /* Database handle */
  int rc = LSM_OK;                /* Return Code */
  int iLevel;                     /* Level of b-tree hierachy to write to */
  int nData;                      /* Size of aData[] in bytes */
  u8 *aData;                      /* Page data for level iLevel */
  int iOff;                       /* Offset on b-tree page to write record to */
  int nRec;                       /* Initial number of records on b-tree page */

  /* iKeyPg should be zero for an ordinary b-tree key, or non-zero for an
  ** indirect key. The flags byte for an indirect key is 0x00.  */
................................................................................
    nRec = pageGetNRec(aData, nData);
    iFPtr = (int)pageGetPtr(aData, nData);
    iRPtr = iPtr - iFPtr;
  }
     
  /* Figure out how much space is required by the new record. The space
  ** required is divided into two sections: the header and the body. The
  ** header consists of the intial varint fields. The body are the blobs 
  ** of data that correspond to the key and value data. The entire header 
  ** must be stored on the page. The body may overflow onto the next and
  ** subsequent pages.
  **
  ** The header space is:
  **
  **     1) record type - 1 byte.







|







 







|







 







|







2648
2649
2650
2651
2652
2653
2654
2655
2656
2657
2658
2659
2660
2661
2662
....
3580
3581
3582
3583
3584
3585
3586
3587
3588
3589
3590
3591
3592
3593
3594
....
3980
3981
3982
3983
3984
3985
3986
3987
3988
3989
3990
3991
3992
3993
3994
}

int lsmSortedLoadFreelist(
  lsm_db *pDb,                    /* Database handle (must be worker) */
  void **ppVal,                   /* OUT: Blob containing LSM free-list */
  int *pnVal                      /* OUT: Size of *ppVal blob in bytes */
){
  MultiCursor *pCsr;              /* Cursor used to retrieve free-list */
  int rc = LSM_OK;                /* Return Code */

  assert( pDb->pWorker );
  assert( *ppVal==0 && *pnVal==0 );

  pCsr = multiCursorNew(pDb, &rc);
  if( pCsr ){
................................................................................
  LsmPgno iKeyPg,
  void *pKey,
  int nKey
){
  Hierarchy *p = &pMW->hier;
  lsm_db *pDb = pMW->pDb;         /* Database handle */
  int rc = LSM_OK;                /* Return Code */
  int iLevel;                     /* Level of b-tree hierarchy to write to */
  int nData;                      /* Size of aData[] in bytes */
  u8 *aData;                      /* Page data for level iLevel */
  int iOff;                       /* Offset on b-tree page to write record to */
  int nRec;                       /* Initial number of records on b-tree page */

  /* iKeyPg should be zero for an ordinary b-tree key, or non-zero for an
  ** indirect key. The flags byte for an indirect key is 0x00.  */
................................................................................
    nRec = pageGetNRec(aData, nData);
    iFPtr = (int)pageGetPtr(aData, nData);
    iRPtr = iPtr - iFPtr;
  }
     
  /* Figure out how much space is required by the new record. The space
  ** required is divided into two sections: the header and the body. The
  ** header consists of the initial varint fields. The body are the blobs 
  ** of data that correspond to the key and value data. The entire header 
  ** must be stored on the page. The body may overflow onto the next and
  ** subsequent pages.
  **
  ** The header space is:
  **
  **     1) record type - 1 byte.

Changes to ext/lsm1/lsm_vtab.c.

78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
** is T/6.  Type value 0 means that the value is an integer whose actual
** values is T/6 and there is no content.  The type-value-0 integer format
** only works for integers in the range of 0 through 40.
**
** There is no content for NULL or type-0 integers.  For BLOB and TEXT
** values, the content is the blob data or the UTF-8 text data.  For
** non-negative integers X, the content is a variable-length integer X*2.
** For negative integers Y, the content is varaible-length integer (1-Y)*2+1.
** For FLOAT values, the content is the IEEE754 floating point value in
** native byte-order.  This means that FLOAT values will be corrupted when
** database file is moved between big-endian and little-endian machines.
*/
#include "sqlite3ext.h"
SQLITE_EXTENSION_INIT1
#include "lsm.h"







|







78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
** is T/6.  Type value 0 means that the value is an integer whose actual
** values is T/6 and there is no content.  The type-value-0 integer format
** only works for integers in the range of 0 through 40.
**
** There is no content for NULL or type-0 integers.  For BLOB and TEXT
** values, the content is the blob data or the UTF-8 text data.  For
** non-negative integers X, the content is a variable-length integer X*2.
** For negative integers Y, the content is variable-length integer (1-Y)*2+1.
** For FLOAT values, the content is the IEEE754 floating point value in
** native byte-order.  This means that FLOAT values will be corrupted when
** database file is moved between big-endian and little-endian machines.
*/
#include "sqlite3ext.h"
SQLITE_EXTENSION_INIT1
#include "lsm.h"

Changes to ext/misc/amatch.c.

480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
*/
struct amatch_rule {
  amatch_rule *pNext;      /* Next rule in order of increasing rCost */
  char *zFrom;             /* Transform from (a string from user input) */
  amatch_cost rCost;       /* Cost of this transformation */
  amatch_langid iLang;     /* The langauge to which this rule belongs */
  amatch_len nFrom, nTo;   /* Length of the zFrom and zTo strings */
  char zTo[4];             /* Tranform to V.W value (extra space appended) */
};

/* 
** A amatch virtual-table object 
*/
struct amatch_vtab {
  sqlite3_vtab base;         /* Base class - must be first */







|







480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
*/
struct amatch_rule {
  amatch_rule *pNext;      /* Next rule in order of increasing rCost */
  char *zFrom;             /* Transform from (a string from user input) */
  amatch_cost rCost;       /* Cost of this transformation */
  amatch_langid iLang;     /* The langauge to which this rule belongs */
  amatch_len nFrom, nTo;   /* Length of the zFrom and zTo strings */
  char zTo[4];             /* Transform to V.W value (extra space appended) */
};

/* 
** A amatch virtual-table object 
*/
struct amatch_vtab {
  sqlite3_vtab base;         /* Base class - must be first */

Changes to ext/misc/btreeinfo.c.

45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
..
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
**
** The sqlite_dbpage virtual table must be available for this virtual table
** to operate.
**
** USAGE EXAMPLES:
**
** Show the table btrees in a schema order with the tables with the most
** rows occuring first:
**
**      SELECT name, nEntry
**        FROM sqlite_btreeinfo
**       WHERE type='table'
**       ORDER BY nEntry DESC, name;
**
** Show the names of all WITHOUT ROWID tables: 
................................................................................
  int szPage;                     /* size of a btree page.  0 if unknown */
  char *zSchema;                  /* Schema being interrogated */
};

/* The sqlite_btreeinfo table */
struct BinfoTable {
  sqlite3_vtab base;              /* Base class.  Must be first */
  sqlite3 *db;                    /* The databse connection */
};

/*
** Connect to the sqlite_btreeinfo virtual table.
*/
static int binfoConnect(
  sqlite3 *db,







|







 







|







45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
..
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
**
** The sqlite_dbpage virtual table must be available for this virtual table
** to operate.
**
** USAGE EXAMPLES:
**
** Show the table btrees in a schema order with the tables with the most
** rows occurring first:
**
**      SELECT name, nEntry
**        FROM sqlite_btreeinfo
**       WHERE type='table'
**       ORDER BY nEntry DESC, name;
**
** Show the names of all WITHOUT ROWID tables: 
................................................................................
  int szPage;                     /* size of a btree page.  0 if unknown */
  char *zSchema;                  /* Schema being interrogated */
};

/* The sqlite_btreeinfo table */
struct BinfoTable {
  sqlite3_vtab base;              /* Base class.  Must be first */
  sqlite3 *db;                    /* The database connection */
};

/*
** Connect to the sqlite_btreeinfo virtual table.
*/
static int binfoConnect(
  sqlite3 *db,

Changes to ext/misc/cksumvfs.c.

43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
** to the sqlite3_load_extension() API call.  Then you invoke the
** sqlite3_load_extension() API and shutdown the dummy database
** connection.  All subsequent database connections that are opened
** will include this extension.  For example:
**
**     sqlite3 *db;
**     sqlite3_open(":memory:", &db);
**     sqlite3_load_extention(db, "./cksumvfs");
**     sqlite3_close(db);
**
** If this extension is compiled with -DSQLITE_CKSUMVFS_STATIC and
** statically linked against the application, initialize it using
** a single API call as follows:
**
**     sqlite3_cksumvfs_init();







|







43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
** to the sqlite3_load_extension() API call.  Then you invoke the
** sqlite3_load_extension() API and shutdown the dummy database
** connection.  All subsequent database connections that are opened
** will include this extension.  For example:
**
**     sqlite3 *db;
**     sqlite3_open(":memory:", &db);
**     sqlite3_load_extension(db, "./cksumvfs");
**     sqlite3_close(db);
**
** If this extension is compiled with -DSQLITE_CKSUMVFS_STATIC and
** statically linked against the application, initialize it using
** a single API call as follows:
**
**     sqlite3_cksumvfs_init();

Changes to ext/misc/csv.c.

310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
...
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
...
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
  char *zData;                    /* Raw CSV data in lieu of zFilename */
  long iStart;                    /* Offset to start of data in zFilename */
  int nCol;                       /* Number of columns in the CSV file */
  unsigned int tstFlags;          /* Bit values used for testing */
} CsvTable;

/* Allowed values for tstFlags */
#define CSVTEST_FIDX  0x0001      /* Pretend that constrained searchs cost less*/

/* A cursor for the CSV virtual table */
typedef struct CsvCursor {
  sqlite3_vtab_cursor base;       /* Base class.  Must be first */
  CsvReader rdr;                  /* The CsvReader object */
  char **azVal;                   /* Value of the current row */
  int *aLen;                      /* Length of each entry */
................................................................................
** Return 1 if the parameter is seen, or 0 if not.  1 is returned
** even if there is an error.  If an error occurs, then an error message
** is left in p->zErr.  If there are no errors, p->zErr[0]==0.
*/
static int csv_string_parameter(
  CsvReader *p,            /* Leave the error message here, if there is one */
  const char *zParam,      /* Parameter we are checking for */
  const char *zArg,        /* Raw text of the virtual table argment */
  char **pzVal             /* Write the dequoted string value here */
){
  const char *zValue;
  zValue = csv_parameter(zParam,(int)strlen(zParam),zArg);
  if( zValue==0 ) return 0;
  p->zErr[0] = 0;
  if( *pzVal ){
................................................................................
  sqlite3_index_info *pIdxInfo
){
  pIdxInfo->estimatedCost = 1000000;
#ifdef SQLITE_TEST
  if( (((CsvTable*)tab)->tstFlags & CSVTEST_FIDX)!=0 ){
    /* The usual (and sensible) case is to always do a full table scan.
    ** The code in this branch only runs when testflags=1.  This code
    ** generates an artifical and unrealistic plan which is useful
    ** for testing virtual table logic but is not helpful to real applications.
    **
    ** Any ==, LIKE, or GLOB constraint is marked as usable by the virtual
    ** table (even though it is not) and the cost of running the virtual table
    ** is reduced from 1 million to just 10.  The constraints are *not* marked
    ** as omittable, however, so the query planner should still generate a
    ** plan that gives a correct answer, even if they plan is not optimal.







|







 







|







 







|







310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
...
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
...
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
  char *zData;                    /* Raw CSV data in lieu of zFilename */
  long iStart;                    /* Offset to start of data in zFilename */
  int nCol;                       /* Number of columns in the CSV file */
  unsigned int tstFlags;          /* Bit values used for testing */
} CsvTable;

/* Allowed values for tstFlags */
#define CSVTEST_FIDX  0x0001      /* Pretend that constrained searches cost less*/

/* A cursor for the CSV virtual table */
typedef struct CsvCursor {
  sqlite3_vtab_cursor base;       /* Base class.  Must be first */
  CsvReader rdr;                  /* The CsvReader object */
  char **azVal;                   /* Value of the current row */
  int *aLen;                      /* Length of each entry */
................................................................................
** Return 1 if the parameter is seen, or 0 if not.  1 is returned
** even if there is an error.  If an error occurs, then an error message
** is left in p->zErr.  If there are no errors, p->zErr[0]==0.
*/
static int csv_string_parameter(
  CsvReader *p,            /* Leave the error message here, if there is one */
  const char *zParam,      /* Parameter we are checking for */
  const char *zArg,        /* Raw text of the virtual table argument */
  char **pzVal             /* Write the dequoted string value here */
){
  const char *zValue;
  zValue = csv_parameter(zParam,(int)strlen(zParam),zArg);
  if( zValue==0 ) return 0;
  p->zErr[0] = 0;
  if( *pzVal ){
................................................................................
  sqlite3_index_info *pIdxInfo
){
  pIdxInfo->estimatedCost = 1000000;
#ifdef SQLITE_TEST
  if( (((CsvTable*)tab)->tstFlags & CSVTEST_FIDX)!=0 ){
    /* The usual (and sensible) case is to always do a full table scan.
    ** The code in this branch only runs when testflags=1.  This code
    ** generates an artificial and unrealistic plan which is useful
    ** for testing virtual table logic but is not helpful to real applications.
    **
    ** Any ==, LIKE, or GLOB constraint is marked as usable by the virtual
    ** table (even though it is not) and the cost of running the virtual table
    ** is reduced from 1 million to just 10.  The constraints are *not* marked
    ** as omittable, however, so the query planner should still generate a
    ** plan that gives a correct answer, even if they plan is not optimal.

Changes to ext/misc/dbdump.c.

222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
  ** is tricky.  We never need to preserve a rowid for a WITHOUT ROWID table
  ** or a table with an INTEGER PRIMARY KEY.  We are unable to preserve
  ** rowids on tables where the rowid is inaccessible because there are other
  ** columns in the table named "rowid", "_rowid_", and "oid".
  */
  if( isIPK ){
    /* If a single PRIMARY KEY column with type INTEGER was seen, then it
    ** might be an alise for the ROWID.  But it might also be a WITHOUT ROWID
    ** table or a INTEGER PRIMARY KEY DESC column, neither of which are
    ** ROWID aliases.  To distinguish these cases, check to see if
    ** there is a "pk" entry in "PRAGMA index_list".  There will be
    ** no "pk" index if the PRIMARY KEY really is an alias for the ROWID.
    */
    zSql = sqlite3_mprintf("SELECT 1 FROM pragma_index_list(%Q)"
                           " WHERE origin='pk'", zTab);







|







222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
  ** is tricky.  We never need to preserve a rowid for a WITHOUT ROWID table
  ** or a table with an INTEGER PRIMARY KEY.  We are unable to preserve
  ** rowids on tables where the rowid is inaccessible because there are other
  ** columns in the table named "rowid", "_rowid_", and "oid".
  */
  if( isIPK ){
    /* If a single PRIMARY KEY column with type INTEGER was seen, then it
    ** might be an alias for the ROWID.  But it might also be a WITHOUT ROWID
    ** table or a INTEGER PRIMARY KEY DESC column, neither of which are
    ** ROWID aliases.  To distinguish these cases, check to see if
    ** there is a "pk" entry in "PRAGMA index_list".  There will be
    ** no "pk" index if the PRIMARY KEY really is an alias for the ROWID.
    */
    zSql = sqlite3_mprintf("SELECT 1 FROM pragma_index_list(%Q)"
                           " WHERE origin='pk'", zTab);

Changes to ext/misc/decimal.c.

463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
  pB->sign = !pB->sign;
  decimal_add(pA, pB);
  decimal_result(context, pA);
  decimal_free(pA);
  decimal_free(pB);
}

/* Aggregate funcion:   decimal_sum(X)
**
** Works like sum() except that it uses decimal arithmetic for unlimited
** precision.
*/
static void decimalSumStep(
  sqlite3_context *context,
  int argc,







|







463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
  pB->sign = !pB->sign;
  decimal_add(pA, pB);
  decimal_result(context, pA);
  decimal_free(pA);
  decimal_free(pB);
}

/* Aggregate function:   decimal_sum(X)
**
** Works like sum() except that it uses decimal arithmetic for unlimited
** precision.
*/
static void decimalSumStep(
  sqlite3_context *context,
  int argc,

Changes to ext/misc/fileio.c.

497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
    }
  }
}

/*
** SQL function:   lsmode(MODE)
**
** Given a numberic st_mode from stat(), convert it into a human-readable
** text string in the style of "ls -l".
*/
static void lsModeFunc(
  sqlite3_context *context,
  int argc,
  sqlite3_value **argv
){







|







497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
    }
  }
}

/*
** SQL function:   lsmode(MODE)
**
** Given a numeric st_mode from stat(), convert it into a human-readable
** text string in the style of "ls -l".
*/
static void lsModeFunc(
  sqlite3_context *context,
  int argc,
  sqlite3_value **argv
){

Changes to ext/misc/fuzzer.c.

93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
**   SELECT vocabulary.w FROM f, vocabulary
**    WHERE f.word MATCH $word
**      AND f.distance<=200
**      AND f.word=vocabulary.w
**    LIMIT 20
**
** The query above gives the 20 closest words to the $word being tested.
** (Note that for good performance, the vocubulary.w column should be
** indexed.)
**
** A similar query can be used to find all words in the dictionary that
** begin with some prefix $prefix:
**
**   SELECT vocabulary.w FROM f, vocabulary
**    WHERE f.word MATCH $prefix







|







93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
**   SELECT vocabulary.w FROM f, vocabulary
**    WHERE f.word MATCH $word
**      AND f.distance<=200
**      AND f.word=vocabulary.w
**    LIMIT 20
**
** The query above gives the 20 closest words to the $word being tested.
** (Note that for good performance, the vocabulary.w column should be
** indexed.)
**
** A similar query can be used to find all words in the dictionary that
** begin with some prefix $prefix:
**
**   SELECT vocabulary.w FROM f, vocabulary
**    WHERE f.word MATCH $prefix

Changes to ext/misc/ieee754.c.

69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
** point value (the value 47.49 is used in the example) do:
**
**    WITH c(n) AS (VALUES(47.49))
**          ---------------^^^^^---- Replace with whatever you want
**    SELECT decimal_mul(ieee754_mantissa(c.n),pow2.v)
**      FROM pow2, c WHERE pow2.x=ieee754_exponent(c.n);
**
** Here is a query to show various boundry values for the binary64
** number format:
**
**    WITH c(name,bin) AS (VALUES
**       ('minimum positive value',        x'0000000000000001'),
**       ('maximum subnormal value',       x'000fffffffffffff'),
**       ('mininum positive nornal value', x'0010000000000000'),
**       ('maximum value',                 x'7fefffffffffffff'))
**    SELECT c.name, decimal_mul(ieee754_mantissa(c.bin),pow2.v)
**      FROM pow2, c WHERE pow2.x=ieee754_exponent(c.bin);
**
*/
#include "sqlite3ext.h"
SQLITE_EXTENSION_INIT1







|





|







69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
** point value (the value 47.49 is used in the example) do:
**
**    WITH c(n) AS (VALUES(47.49))
**          ---------------^^^^^---- Replace with whatever you want
**    SELECT decimal_mul(ieee754_mantissa(c.n),pow2.v)
**      FROM pow2, c WHERE pow2.x=ieee754_exponent(c.n);
**
** Here is a query to show various boundary values for the binary64
** number format:
**
**    WITH c(name,bin) AS (VALUES
**       ('minimum positive value',        x'0000000000000001'),
**       ('maximum subnormal value',       x'000fffffffffffff'),
**       ('minimum positive normal value', x'0010000000000000'),
**       ('maximum value',                 x'7fefffffffffffff'))
**    SELECT c.name, decimal_mul(ieee754_mantissa(c.bin),pow2.v)
**      FROM pow2, c WHERE pow2.x=ieee754_exponent(c.bin);
**
*/
#include "sqlite3ext.h"
SQLITE_EXTENSION_INIT1

Changes to ext/misc/json1.c.

1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
  char c;
  JsonString *pStr;
  UNUSED_PARAM(argc);
  UNUSED_PARAM(argv);
  pStr = (JsonString*)sqlite3_aggregate_context(ctx, 0);
#ifdef NEVER
  /* pStr is always non-NULL since jsonArrayStep() or jsonObjectStep() will
  ** always have been called to initalize it */
  if( NEVER(!pStr) ) return;
#endif
  z = pStr->zBuf;
  for(i=1; (c = z[i])!=',' || inStr || nNest; i++){
    if( i>=pStr->nUsed ){
      pStr->nUsed = 1;
      return;







|







1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
  char c;
  JsonString *pStr;
  UNUSED_PARAM(argc);
  UNUSED_PARAM(argv);
  pStr = (JsonString*)sqlite3_aggregate_context(ctx, 0);
#ifdef NEVER
  /* pStr is always non-NULL since jsonArrayStep() or jsonObjectStep() will
  ** always have been called to initialize it */
  if( NEVER(!pStr) ) return;
#endif
  z = pStr->zBuf;
  for(i=1; (c = z[i])!=',' || inStr || nNest; i++){
    if( i>=pStr->nUsed ){
      pStr->nUsed = 1;
      return;

Changes to ext/misc/nextchar.c.

16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
** string A given the vocabulary in T.F.  If the W value exists and is a
** non-empty string, then it is an SQL expression that limits the entries
** in T.F that will be considered.  If C exists and is a non-empty string,
** then it is the name of the collating sequence to use for comparison.  If
** 
** Only the first three arguments are required.  If the C parameter is 
** omitted or is NULL or is an empty string, then the default collating 
** sequence of T.F is used for comparision.  If the W parameter is omitted
** or is NULL or is an empty string, then no filtering of the output is
** done.
**
** The T.F column should be indexed using collation C or else this routine
** will be quite slow.
**
** For example, suppose an application has a dictionary like this:







|







16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
** string A given the vocabulary in T.F.  If the W value exists and is a
** non-empty string, then it is an SQL expression that limits the entries
** in T.F that will be considered.  If C exists and is a non-empty string,
** then it is the name of the collating sequence to use for comparison.  If
** 
** Only the first three arguments are required.  If the C parameter is 
** omitted or is NULL or is an empty string, then the default collating 
** sequence of T.F is used for comparison.  If the W parameter is omitted
** or is NULL or is an empty string, then no filtering of the output is
** done.
**
** The T.F column should be indexed using collation C or else this routine
** will be quite slow.
**
** For example, suppose an application has a dictionary like this:

Changes to ext/misc/sha1.c.

15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
**
**     sha1(X)
**     sha1_query(Y)
**
** The sha1(X) function computes the SHA1 hash of the input X, or NULL if
** X is NULL.
**
** The sha1_query(Y) function evalutes all queries in the SQL statements of Y
** and returns a hash of their results.
*/
#include "sqlite3ext.h"
SQLITE_EXTENSION_INIT1
#include <assert.h>
#include <string.h>
#include <stdarg.h>







|







15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
**
**     sha1(X)
**     sha1_query(Y)
**
** The sha1(X) function computes the SHA1 hash of the input X, or NULL if
** X is NULL.
**
** The sha1_query(Y) function evaluates all queries in the SQL statements of Y
** and returns a hash of their results.
*/
#include "sqlite3ext.h"
SQLITE_EXTENSION_INIT1
#include <assert.h>
#include <string.h>
#include <stdarg.h>

Changes to ext/misc/shathree.c.

15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
...
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
**
**     sha3(X,SIZE)
**     sha3_query(Y,SIZE)
**
** The sha3(X) function computes the SHA3 hash of the input X, or NULL if
** X is NULL.
**
** The sha3_query(Y) function evalutes all queries in the SQL statements of Y
** and returns a hash of their results.
**
** The SIZE argument is optional.  If omitted, the SHA3-256 hash algorithm
** is used.  If SIZE is included it must be one of the integers 224, 256,
** 384, or 512, to determine SHA3 hash variant that is computed.
*/
#include "sqlite3ext.h"
................................................................................
** B means blobs of <size> bytes.  T means text rendered as <size>
** bytes of UTF-8.  The <n> and <size> values are expressed as an ASCII
** text integers.
**
** For each SQL statement in the X input, there is one S segment.  Each
** S segment is followed by zero or more R segments, one for each row in the
** result set.  After each R, there are one or more N, I, F, B, or T segments,
** one for each column in the result set.  Segments are concatentated directly
** with no delimiters of any kind.
*/
static void sha3QueryFunc(
  sqlite3_context *context,
  int argc,
  sqlite3_value **argv
){







|







 







|







15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
...
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
**
**     sha3(X,SIZE)
**     sha3_query(Y,SIZE)
**
** The sha3(X) function computes the SHA3 hash of the input X, or NULL if
** X is NULL.
**
** The sha3_query(Y) function evaluates all queries in the SQL statements of Y
** and returns a hash of their results.
**
** The SIZE argument is optional.  If omitted, the SHA3-256 hash algorithm
** is used.  If SIZE is included it must be one of the integers 224, 256,
** 384, or 512, to determine SHA3 hash variant that is computed.
*/
#include "sqlite3ext.h"
................................................................................
** B means blobs of <size> bytes.  T means text rendered as <size>
** bytes of UTF-8.  The <n> and <size> values are expressed as an ASCII
** text integers.
**
** For each SQL statement in the X input, there is one S segment.  Each
** S segment is followed by zero or more R segments, one for each row in the
** result set.  After each R, there are one or more N, I, F, B, or T segments,
** one for each column in the result set.  Segments are concatenated directly
** with no delimiters of any kind.
*/
static void sha3QueryFunc(
  sqlite3_context *context,
  int argc,
  sqlite3_value **argv
){

Changes to ext/misc/spellfix.c.

332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
...
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
....
2361
2362
2363
2364
2365
2366
2367
2368
2369
2370
2371
2372
2373
2374
2375
    return 40;
  }
  if( classFrom>=CCLASS_B && classFrom<=CCLASS_Y
      && classTo>=CCLASS_B && classTo<=CCLASS_Y ){
    /* Convert from one consonant to another, but in a different class */
    return 75;
  }
  /* Any other subsitution */
  return 100;
}

/*
** Given two strings zA and zB which are pure ASCII, return the cost
** of transforming zA into zB.  If zA ends with '*' assume that it is
** a prefix of zB and give only minimal penalty for extra characters
................................................................................
/*
** Return TRUE (non-zero) of the next FROM character and the next TO
** character are the same.
*/
static int matchFromTo(
  EditDist3FromString *pStr,  /* Left hand string */
  int n1,                     /* Index of comparison character on the left */
  const char *z2,             /* Right-handl comparison character */
  int n2                      /* Bytes remaining in z2[] */
){
  int b1 = pStr->a[n1].nByte;
  if( b1>n2 ) return 0;
  assert( b1>0 );
  if( pStr->z[n1]!=z2[0] ) return 0;
  if( strncmp(pStr->z+n1, z2, b1)!=0 ) return 0;
  return 1;
}

/*
** Delete an EditDist3FromString objecct
*/
static void editDist3FromStringDelete(EditDist3FromString *p){
  int i;
  if( p ){
    for(i=0; i<p->n; i++){
      sqlite3_free(p->a[i].apDel);
      sqlite3_free(p->a[i].apSubst);
................................................................................

/*
** A structure used to pass information from spellfix1FilterForMatch()
** into spellfix1RunQuery().
*/
typedef struct MatchQuery {
  spellfix1_cursor *pCur;          /* The cursor being queried */
  sqlite3_stmt *pStmt;             /* shadow table query statment */
  char zHash[SPELLFIX_MX_HASH];    /* The current phonehash for zPattern */
  const char *zPattern;            /* Transliterated input string */
  int nPattern;                    /* Length of zPattern */
  EditDist3FromString *pMatchStr3; /* Original unicode string */
  EditDist3Config *pConfig3;       /* Edit-distance cost coefficients */
  const EditDist3Lang *pLang;      /* The selected language coefficients */
  int iLang;                       /* The language id */







|







 







|











|







 







|







332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
...
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
....
2361
2362
2363
2364
2365
2366
2367
2368
2369
2370
2371
2372
2373
2374
2375
    return 40;
  }
  if( classFrom>=CCLASS_B && classFrom<=CCLASS_Y
      && classTo>=CCLASS_B && classTo<=CCLASS_Y ){
    /* Convert from one consonant to another, but in a different class */
    return 75;
  }
  /* Any other substitution */
  return 100;
}

/*
** Given two strings zA and zB which are pure ASCII, return the cost
** of transforming zA into zB.  If zA ends with '*' assume that it is
** a prefix of zB and give only minimal penalty for extra characters
................................................................................
/*
** Return TRUE (non-zero) of the next FROM character and the next TO
** character are the same.
*/
static int matchFromTo(
  EditDist3FromString *pStr,  /* Left hand string */
  int n1,                     /* Index of comparison character on the left */
  const char *z2,             /* Right-hand comparison character */
  int n2                      /* Bytes remaining in z2[] */
){
  int b1 = pStr->a[n1].nByte;
  if( b1>n2 ) return 0;
  assert( b1>0 );
  if( pStr->z[n1]!=z2[0] ) return 0;
  if( strncmp(pStr->z+n1, z2, b1)!=0 ) return 0;
  return 1;
}

/*
** Delete an EditDist3FromString object
*/
static void editDist3FromStringDelete(EditDist3FromString *p){
  int i;
  if( p ){
    for(i=0; i<p->n; i++){
      sqlite3_free(p->a[i].apDel);
      sqlite3_free(p->a[i].apSubst);
................................................................................

/*
** A structure used to pass information from spellfix1FilterForMatch()
** into spellfix1RunQuery().
*/
typedef struct MatchQuery {
  spellfix1_cursor *pCur;          /* The cursor being queried */
  sqlite3_stmt *pStmt;             /* shadow table query statement */
  char zHash[SPELLFIX_MX_HASH];    /* The current phonehash for zPattern */
  const char *zPattern;            /* Transliterated input string */
  int nPattern;                    /* Length of zPattern */
  EditDist3FromString *pMatchStr3; /* Original unicode string */
  EditDist3Config *pConfig3;       /* Edit-distance cost coefficients */
  const EditDist3Lang *pLang;      /* The selected language coefficients */
  int iLang;                       /* The language id */

Changes to ext/misc/totype.c.

342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
      result = (double)s;
    }
  }

  /* store the result */
  *pResult = result;

  /* return true if number and no extra non-whitespace chracters after */
  return z>=zEnd && nDigits>0 && eValid && nonNum==0;
}

/*
** tointeger(X):  If X is any value (integer, double, blob, or string) that
** can be losslessly converted into an integer, then make the conversion and
** return the result.  Otherwise, return NULL.







|







342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
      result = (double)s;
    }
  }

  /* store the result */
  *pResult = result;

  /* return true if number and no extra non-whitespace characters after */
  return z>=zEnd && nDigits>0 && eValid && nonNum==0;
}

/*
** tointeger(X):  If X is any value (integer, double, blob, or string) that
** can be losslessly converted into an integer, then make the conversion and
** return the result.  Otherwise, return NULL.

Changes to ext/misc/uint.c.

12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
**
** This SQLite extension implements the UINT collating sequence.
**
** UINT works like BINARY for text, except that embedded strings
** of digits compare in numeric order.
**
**     *   Leading zeros are handled properly, in the sense that
**         they do not mess of the maginitude comparison of embedded
**         strings of digits.  "x00123y" is equal to "x123y".
**
**     *   Only unsigned integers are recognized.  Plus and minus
**         signs are ignored.  Decimal points and exponential notation
**         are ignored.
**
**     *   Embedded integers can be of arbitrary length.  Comparison







|







12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
**
** This SQLite extension implements the UINT collating sequence.
**
** UINT works like BINARY for text, except that embedded strings
** of digits compare in numeric order.
**
**     *   Leading zeros are handled properly, in the sense that
**         they do not mess of the magnitude comparison of embedded
**         strings of digits.  "x00123y" is equal to "x123y".
**
**     *   Only unsigned integers are recognized.  Plus and minus
**         signs are ignored.  Decimal points and exponential notation
**         are ignored.
**
**     *   Embedded integers can be of arbitrary length.  Comparison

Changes to ext/misc/unionvtab.c.

34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
**
** UNIONVTAB
**
**   A "unionvtab" virtual table is created as follows:
**
**     CREATE VIRTUAL TABLE <name> USING unionvtab(<sql-statement>);
**
**   The implementation evalutes <sql statement> whenever a unionvtab virtual
**   table is created or opened. It should return one row for each source
**   database table. The four columns required of each row are:
**
**     1. The name of the database containing the table ("main" or "temp" or
**        the name of an attached database). Or NULL to indicate that all
**        databases should be searched for the table in the usual fashion.
**







|







34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
**
** UNIONVTAB
**
**   A "unionvtab" virtual table is created as follows:
**
**     CREATE VIRTUAL TABLE <name> USING unionvtab(<sql-statement>);
**
**   The implementation evaluates <sql statement> whenever a unionvtab virtual
**   table is created or opened. It should return one row for each source
**   database table. The four columns required of each row are:
**
**     1. The name of the database containing the table ("main" or "temp" or
**        the name of an attached database). Or NULL to indicate that all
**        databases should be searched for the table in the usual fashion.
**

Changes to ext/misc/vfslog.c.

19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
** most VFS calls to be written into a file on disk.
**
** Each database connection creates a separate log file in the same
** directory as the original database and named after the original
** database.  A unique suffix is added to avoid name collisions.  
** Separate log files are used so that concurrent processes do not
** try to write log operations to the same file at the same instant, 
** resulting in overwritten or comingled log text.
**
** Each individual log file records operations by a single database
** connection on both the original database and its associated rollback
** journal.
**
** The log files are in the comma-separated-value (CSV) format.  The
** log files can be imported into an SQLite database using the ".import"







|







19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
** most VFS calls to be written into a file on disk.
**
** Each database connection creates a separate log file in the same
** directory as the original database and named after the original
** database.  A unique suffix is added to avoid name collisions.  
** Separate log files are used so that concurrent processes do not
** try to write log operations to the same file at the same instant, 
** resulting in overwritten or commingled log text.
**
** Each individual log file records operations by a single database
** connection on both the original database and its associated rollback
** journal.
**
** The log files are in the comma-separated-value (CSV) format.  The
** log files can be imported into an SQLite database using the ".import"

Changes to ext/misc/vfsstat.c.

20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
/*
** This module contains code for a wrapper VFS that cause stats for
** most VFS calls to be recorded.
**
** To use this module, first compile it as a loadable extension.  See
** https://www.sqlite.org/loadext.html#build for compilations instructions.
**
** After compliing, load this extension, then open database connections to be
** measured.  Query usages status using the vfsstat virtual table:
**
**         SELECT * FROM vfsstat;
**
** Reset counters using UPDATE statements against vfsstat:
**
**         UPDATE vfsstat SET count=0;







|







20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
/*
** This module contains code for a wrapper VFS that cause stats for
** most VFS calls to be recorded.
**
** To use this module, first compile it as a loadable extension.  See
** https://www.sqlite.org/loadext.html#build for compilations instructions.
**
** After compiling, load this extension, then open database connections to be
** measured.  Query usages status using the vfsstat virtual table:
**
**         SELECT * FROM vfsstat;
**
** Reset counters using UPDATE statements against vfsstat:
**
**         UPDATE vfsstat SET count=0;

Changes to ext/misc/vtablog.c.

99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
/* Decode a parameter that requires a dequoted string.
**
** Return non-zero on an error.
*/
static int vtablog_string_parameter(
  char **pzErr,            /* Leave the error message here, if there is one */
  const char *zParam,      /* Parameter we are checking for */
  const char *zArg,        /* Raw text of the virtual table argment */
  char **pzVal             /* Write the dequoted string value here */
){
  const char *zValue;
  zValue = vtablog_parameter(zParam,(int)strlen(zParam),zArg);
  if( zValue==0 ) return 0;
  if( *pzVal ){
    *pzErr = sqlite3_mprintf("more than one '%s' parameter", zParam);







|







99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
/* Decode a parameter that requires a dequoted string.
**
** Return non-zero on an error.
*/
static int vtablog_string_parameter(
  char **pzErr,            /* Leave the error message here, if there is one */
  const char *zParam,      /* Parameter we are checking for */
  const char *zArg,        /* Raw text of the virtual table argument */
  char **pzVal             /* Write the dequoted string value here */
){
  const char *zValue;
  zValue = vtablog_parameter(zParam,(int)strlen(zParam),zArg);
  if( zValue==0 ) return 0;
  if( *pzVal ){
    *pzErr = sqlite3_mprintf("more than one '%s' parameter", zParam);

Changes to ext/misc/vtshim.c.

421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
  rc = pAux->pMod->xRollbackTo(pVtab->pChild, n);
  if( rc!=SQLITE_OK ){
    VTSHIM_COPY_ERRMSG();
  }
  return rc;
}

/* The destructor function for a disposible module */
static void vtshimAuxDestructor(void *pXAux){
  vtshim_aux *pAux = (vtshim_aux*)pXAux;
  assert( pAux->pAllVtab==0 );
  if( !pAux->bDisposed && pAux->xChildDestroy ){
    pAux->xChildDestroy(pAux->pChildAux);
    pAux->xChildDestroy = 0;
  }







|







421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
  rc = pAux->pMod->xRollbackTo(pVtab->pChild, n);
  if( rc!=SQLITE_OK ){
    VTSHIM_COPY_ERRMSG();
  }
  return rc;
}

/* The destructor function for a disposable module */
static void vtshimAuxDestructor(void *pXAux){
  vtshim_aux *pAux = (vtshim_aux*)pXAux;
  assert( pAux->pAllVtab==0 );
  if( !pAux->bDisposed && pAux->xChildDestroy ){
    pAux->xChildDestroy(pAux->pChildAux);
    pAux->xChildDestroy = 0;
  }

Changes to ext/misc/zipfile.c.

548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
  return ((u32)(aBuf[3]) << 24)
       + ((u32)(aBuf[2]) << 16)
       + ((u32)(aBuf[1]) <<  8)
       + ((u32)(aBuf[0]) <<  0);
}

/*
** Write a 16-bit little endiate integer into buffer aBuf.
*/
static void zipfilePutU16(u8 *aBuf, u16 val){
  aBuf[0] = val & 0xFF;
  aBuf[1] = (val>>8) & 0xFF;
}

/*
** Write a 32-bit little endiate integer into buffer aBuf.
*/
static void zipfilePutU32(u8 *aBuf, u32 val){
  aBuf[0] = val & 0xFF;
  aBuf[1] = (val>>8) & 0xFF;
  aBuf[2] = (val>>16) & 0xFF;
  aBuf[3] = (val>>24) & 0xFF;
}







|







|







548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
  return ((u32)(aBuf[3]) << 24)
       + ((u32)(aBuf[2]) << 16)
       + ((u32)(aBuf[1]) <<  8)
       + ((u32)(aBuf[0]) <<  0);
}

/*
** Write a 16-bit little endian integer into buffer aBuf.
*/
static void zipfilePutU16(u8 *aBuf, u16 val){
  aBuf[0] = val & 0xFF;
  aBuf[1] = (val>>8) & 0xFF;
}

/*
** Write a 32-bit little endian integer into buffer aBuf.
*/
static void zipfilePutU32(u8 *aBuf, u32 val){
  aBuf[0] = val & 0xFF;
  aBuf[1] = (val>>8) & 0xFF;
  aBuf[2] = (val>>16) & 0xFF;
  aBuf[3] = (val>>24) & 0xFF;
}

Changes to ext/rbu/rbu_common.tcl.

103
104
105
106
107
108
109
110
    rbu close
  }] {SQLITE_DONE}]

  uplevel [list do_execsql_test $tn.2 {
    PRAGMA integrity_check
  } ok]
}








<
103
104
105
106
107
108
109

    rbu close
  }] {SQLITE_DONE}]

  uplevel [list do_execsql_test $tn.2 {
    PRAGMA integrity_check
  } ok]
}

Changes to ext/rbu/rbuexpr.test.

86
87
88
89
90
91
92
93
  run_rbu test.db rbu.db
} {SQLITE_DONE}

sqlite3 db test.db
integrity_check 2.3

finish_test








<
86
87
88
89
90
91
92

  run_rbu test.db rbu.db
} {SQLITE_DONE}

sqlite3 db test.db
integrity_check 2.3

finish_test

Changes to ext/rbu/rbuvacuum4.test.

109
110
111
112
113
114
115
116
  INSERT INTO x VALUES('a'), ('b'), ('d');
  CREATE UNIQUE INDEX y ON x(a);
}

do_rbu_vacuum_test 5.1 1

finish_test








<
109
110
111
112
113
114
115

  INSERT INTO x VALUES('a'), ('b'), ('d');
  CREATE UNIQUE INDEX y ON x(a);
}

do_rbu_vacuum_test 5.1 1

finish_test

Changes to ext/rbu/sqlite3rbu.c.

1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
....
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
....
3845
3846
3847
3848
3849
3850
3851
3852
3853
3854
3855
3856
3857
3858
3859
....
4690
4691
4692
4693
4694
4695
4696
4697
4698
4699
4700
4701
4702
4703
4704
....
4986
4987
4988
4989
4990
4991
4992
4993
4994
4995
4996
4997
4998
4999
5000
  }
  va_end(ap);
  return zSql;
}

/*
** Argument zFmt is a sqlite3_mprintf() style format string. The trailing
** arguments are the usual subsitution values. This function performs
** the printf() style substitutions and executes the result as an SQL
** statement on the RBU handles database.
**
** If an error occurs, an error code and error message is stored in the
** RBU handle. If an error has already occurred when this function is
** called, it is a no-op.
*/
................................................................................
**
**   (index-field1, index-field2, ...) > (?, ?, ...)
**
** except that the "?" placeholders are replaced with literal values.
**
** If the expression cannot be created, NULL is returned. In this case,
** the caller has to use an OFFSET clause to extract only the required 
** rows from the sourct table, just as it does for an RBU update operation.
*/
char *rbuVacuumIndexStart(
  sqlite3rbu *p,                  /* RBU handle */
  RbuObjIter *pIter               /* RBU iterator object */
){
  char *zOrder = 0;
  char *zLhs = 0;
................................................................................
    p->zVfsName = 0;
  }
}

/*
** This user-defined SQL function is invoked with a single argument - the
** name of a table expected to appear in the target database. It returns
** the number of auxilliary indexes on the table.
*/
static void rbuIndexCntFunc(
  sqlite3_context *pCtx, 
  int nVal,
  sqlite3_value **apVal
){
  sqlite3rbu *p = (sqlite3rbu*)sqlite3_user_data(pCtx);
................................................................................
static int rbuVfsFileSize(sqlite3_file *pFile, sqlite_int64 *pSize){
  rbu_file *p = (rbu_file *)pFile;
  int rc;
  rc = p->pReal->pMethods->xFileSize(p->pReal, pSize);

  /* If this is an RBU vacuum operation and this is the target database,
  ** pretend that it has at least one page. Otherwise, SQLite will not
  ** check for the existance of a *-wal file. rbuVfsRead() contains 
  ** similar logic.  */
  if( rc==SQLITE_OK && *pSize==0 
   && p->pRbu && rbuIsVacuum(p->pRbu) 
   && (p->openFlags & SQLITE_OPEN_MAIN_DB)
  ){
    *pSize = 1024;
  }
................................................................................
      ** or xOpen() to operate on the *-wal file.  */
      pFd->zWal = sqlite3_filename_wal(zName);
    }
    else if( flags & SQLITE_OPEN_WAL ){
      rbu_file *pDb = rbuFindMaindb(pRbuVfs, zName, 0);
      if( pDb ){
        if( pDb->pRbu && pDb->pRbu->eStage==RBU_STAGE_OAL ){
          /* This call is to open a *-wal file. Intead, open the *-oal. This
          ** code ensures that the string passed to xOpen() is terminated by a
          ** pair of '\0' bytes in case the VFS attempts to extract a URI 
          ** parameter from it.  */
          const char *zBase = zName;
          size_t nCopy;
          char *zCopy;
          if( rbuIsVacuum(pDb->pRbu) ){







|







 







|







 







|







 







|







 







|







1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
....
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
....
3845
3846
3847
3848
3849
3850
3851
3852
3853
3854
3855
3856
3857
3858
3859
....
4690
4691
4692
4693
4694
4695
4696
4697
4698
4699
4700
4701
4702
4703
4704
....
4986
4987
4988
4989
4990
4991
4992
4993
4994
4995
4996
4997
4998
4999
5000
  }
  va_end(ap);
  return zSql;
}

/*
** Argument zFmt is a sqlite3_mprintf() style format string. The trailing
** arguments are the usual substitution values. This function performs
** the printf() style substitutions and executes the result as an SQL
** statement on the RBU handles database.
**
** If an error occurs, an error code and error message is stored in the
** RBU handle. If an error has already occurred when this function is
** called, it is a no-op.
*/
................................................................................
**
**   (index-field1, index-field2, ...) > (?, ?, ...)
**
** except that the "?" placeholders are replaced with literal values.
**
** If the expression cannot be created, NULL is returned. In this case,
** the caller has to use an OFFSET clause to extract only the required 
** rows from the source table, just as it does for an RBU update operation.
*/
char *rbuVacuumIndexStart(
  sqlite3rbu *p,                  /* RBU handle */
  RbuObjIter *pIter               /* RBU iterator object */
){
  char *zOrder = 0;
  char *zLhs = 0;
................................................................................
    p->zVfsName = 0;
  }
}

/*
** This user-defined SQL function is invoked with a single argument - the
** name of a table expected to appear in the target database. It returns
** the number of auxiliary indexes on the table.
*/
static void rbuIndexCntFunc(
  sqlite3_context *pCtx, 
  int nVal,
  sqlite3_value **apVal
){
  sqlite3rbu *p = (sqlite3rbu*)sqlite3_user_data(pCtx);
................................................................................
static int rbuVfsFileSize(sqlite3_file *pFile, sqlite_int64 *pSize){
  rbu_file *p = (rbu_file *)pFile;
  int rc;
  rc = p->pReal->pMethods->xFileSize(p->pReal, pSize);

  /* If this is an RBU vacuum operation and this is the target database,
  ** pretend that it has at least one page. Otherwise, SQLite will not
  ** check for the existence of a *-wal file. rbuVfsRead() contains 
  ** similar logic.  */
  if( rc==SQLITE_OK && *pSize==0 
   && p->pRbu && rbuIsVacuum(p->pRbu) 
   && (p->openFlags & SQLITE_OPEN_MAIN_DB)
  ){
    *pSize = 1024;
  }
................................................................................
      ** or xOpen() to operate on the *-wal file.  */
      pFd->zWal = sqlite3_filename_wal(zName);
    }
    else if( flags & SQLITE_OPEN_WAL ){
      rbu_file *pDb = rbuFindMaindb(pRbuVfs, zName, 0);
      if( pDb ){
        if( pDb->pRbu && pDb->pRbu->eStage==RBU_STAGE_OAL ){
          /* This call is to open a *-wal file. Instead, open the *-oal. This
          ** code ensures that the string passed to xOpen() is terminated by a
          ** pair of '\0' bytes in case the VFS attempts to extract a URI 
          ** parameter from it.  */
          const char *zBase = zName;
          size_t nCopy;
          char *zCopy;
          if( rbuIsVacuum(pDb->pRbu) ){

Changes to ext/rbu/sqlite3rbu.h.

438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
SQLITE_API int sqlite3rbu_savestate(sqlite3rbu *pRbu);

/*
** Close an RBU handle. 
**
** If the RBU update has been completely applied, mark the RBU database
** as fully applied. Otherwise, assuming no error has occurred, save the
** current state of the RBU update appliation to the RBU database.
**
** If an error has already occurred as part of an sqlite3rbu_step()
** or sqlite3rbu_open() call, or if one occurs within this function, an
** SQLite error code is returned. Additionally, if pzErrmsg is not NULL,
** *pzErrmsg may be set to point to a buffer containing a utf-8 formatted
** English language error message. It is the responsibility of the caller to
** eventually free any such buffer using sqlite3_free().







|







438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
SQLITE_API int sqlite3rbu_savestate(sqlite3rbu *pRbu);

/*
** Close an RBU handle. 
**
** If the RBU update has been completely applied, mark the RBU database
** as fully applied. Otherwise, assuming no error has occurred, save the
** current state of the RBU update application to the RBU database.
**
** If an error has already occurred as part of an sqlite3rbu_step()
** or sqlite3rbu_open() call, or if one occurs within this function, an
** SQLite error code is returned. Additionally, if pzErrmsg is not NULL,
** *pzErrmsg may be set to point to a buffer containing a utf-8 formatted
** English language error message. It is the responsibility of the caller to
** eventually free any such buffer using sqlite3_free().

Changes to ext/repair/checkindex.c.

106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
#define IIC_SCANNER_SQL   4
  rc = sqlite3_declare_vtab(db,
      "CREATE TABLE xyz("
      " errmsg TEXT,"            /* Error message or NULL if everything is ok */
      " current_key TEXT,"       /* SQLite quote() text of key values */
      " index_name HIDDEN,"      /* IN: name of the index being scanned */
      " after_key HIDDEN,"       /* IN: Start scanning after this key */
      " scanner_sql HIDDEN"      /* debuggingn info: SQL used for scanner */
      ")"
  );
  pRet = cidxMalloc(&rc, sizeof(CidxTable));
  if( pRet ){
    pRet->db = db;
  }








|







106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
#define IIC_SCANNER_SQL   4
  rc = sqlite3_declare_vtab(db,
      "CREATE TABLE xyz("
      " errmsg TEXT,"            /* Error message or NULL if everything is ok */
      " current_key TEXT,"       /* SQLite quote() text of key values */
      " index_name HIDDEN,"      /* IN: name of the index being scanned */
      " after_key HIDDEN,"       /* IN: Start scanning after this key */
      " scanner_sql HIDDEN"      /* debugging info: SQL used for scanner */
      ")"
  );
  pRet = cidxMalloc(&rc, sizeof(CidxTable));
  if( pRet ){
    pRet->db = db;
  }

Changes to ext/repair/sqlite3_checker.tcl.

209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
}
if {![file readable $root_filename]} {
  puts stderr "File is not readable: $root_filename"
  exit 1
}

if {[catch {sqlite3 db $file_to_analyze} res]} {
  puts stderr "Cannot open datababase $root_filename: $res"
  exit 1
}

if {$bFreelistCheck || $bAll} {
  puts -nonewline "freelist-check: "
  flush stdout
  db eval BEGIN







|







209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
}
if {![file readable $root_filename]} {
  puts stderr "File is not readable: $root_filename"
  exit 1
}

if {[catch {sqlite3 db $file_to_analyze} res]} {
  puts stderr "Cannot open database $root_filename: $res"
  exit 1
}

if {$bFreelistCheck || $bAll} {
  puts -nonewline "freelist-check: "
  flush stdout
  db eval BEGIN

Changes to ext/rtree/README.

102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
  and use it as a dynamically loadable SQLite extension. To do this
  using gcc on *nix:

    gcc -shared rtree.c -o libSqliteRtree.so

  You may need to add "-I" flags so that gcc can find sqlite3ext.h
  and sqlite3.h. The resulting shared lib, libSqliteRtree.so, may be
  loaded into sqlite in the same way as any other dynamicly loadable
  extension.


3. REFERENCES

  [1]  Atonin Guttman, "R-trees - A Dynamic Index Structure For Spatial 
       Searching", University of California Berkeley, 1984.

  [2]  Norbert Beckmann, Hans-Peter Kriegel, Ralf Schneider, Bernhard Seeger,
       "The R*-tree: An Efficient and Robust Access Method for Points and
       Rectangles", Universitaet Bremen, 1990.







|





|





102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
  and use it as a dynamically loadable SQLite extension. To do this
  using gcc on *nix:

    gcc -shared rtree.c -o libSqliteRtree.so

  You may need to add "-I" flags so that gcc can find sqlite3ext.h
  and sqlite3.h. The resulting shared lib, libSqliteRtree.so, may be
  loaded into sqlite in the same way as any other dynamically loadable
  extension.


3. REFERENCES

  [1]  Antonin Guttman, "R-trees - A Dynamic Index Structure For Spatial 
       Searching", University of California Berkeley, 1984.

  [2]  Norbert Beckmann, Hans-Peter Kriegel, Ralf Schneider, Bernhard Seeger,
       "The R*-tree: An Efficient and Robust Access Method for Points and
       Rectangles", Universitaet Bremen, 1990.

Changes to ext/rtree/geopoly.c.

757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
...
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
}


/*
** Determine if point (x0,y0) is beneath line segment (x1,y1)->(x2,y2).
** Returns:
**
**    +2  x0,y0 is on the line segement
**
**    +1  x0,y0 is beneath line segment
**
**    0   x0,y0 is not on or beneath the line segment or the line segment
**        is vertical and x0,y0 is not on the line segment
**
** The left-most coordinate min(x1,x2) is not considered to be part of
................................................................................
      sqlite3_result_int(context, x==2 ? 1 : x==4 ? 2 : 0);
    }
  }
  sqlite3_free(p1);
  sqlite3_free(p2);
}

/* Objects used by the overlap algorihm. */
typedef struct GeoEvent GeoEvent;
typedef struct GeoSegment GeoSegment;
typedef struct GeoOverlap GeoOverlap;
struct GeoEvent {
  double x;              /* X coordinate at which event occurs */
  int eType;             /* 0 for ADD, 1 for REMOVE */
  GeoSegment *pSeg;      /* The segment to be added or removed */







|







 







|







757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
...
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
}


/*
** Determine if point (x0,y0) is beneath line segment (x1,y1)->(x2,y2).
** Returns:
**
**    +2  x0,y0 is on the line segment
**
**    +1  x0,y0 is beneath line segment
**
**    0   x0,y0 is not on or beneath the line segment or the line segment
**        is vertical and x0,y0 is not on the line segment
**
** The left-most coordinate min(x1,x2) is not considered to be part of
................................................................................
      sqlite3_result_int(context, x==2 ? 1 : x==4 ? 2 : 0);
    }
  }
  sqlite3_free(p1);
  sqlite3_free(p2);
}

/* Objects used by the overlap algorithm. */
typedef struct GeoEvent GeoEvent;
typedef struct GeoSegment GeoSegment;
typedef struct GeoOverlap GeoOverlap;
struct GeoEvent {
  double x;              /* X coordinate at which event occurs */
  int eType;             /* 0 for ADD, 1 for REMOVE */
  GeoSegment *pSeg;      /* The segment to be added or removed */

Changes to ext/rtree/rtree.c.

324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
....
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
....
4388
4389
4390
4391
4392
4393
4394
4395
4396
4397
4398
4399
4400
4401
4402
#define RTREE_GE    0x44  /* D */
#define RTREE_GT    0x45  /* E */
#define RTREE_MATCH 0x46  /* F: Old-style sqlite3_rtree_geometry_callback() */
#define RTREE_QUERY 0x47  /* G: New-style sqlite3_rtree_query_callback() */

/* Special operators available only on cursors.  Needs to be consecutive
** with the normal values above, but must be less than RTREE_MATCH.  These
** are used in the cursor for contraints such as x=NULL (RTREE_FALSE) or
** x<'xyz' (RTREE_TRUE) */
#define RTREE_TRUE  0x3f  /* ? */
#define RTREE_FALSE 0x40  /* @ */

/* 
** An rtree structure node.
*/
................................................................................
  pIdxInfo->estimatedCost = (double)6.0 * (double)nRow;
  pIdxInfo->estimatedRows = nRow;

  return rc;
}

/*
** Return the N-dimensional volumn of the cell stored in *p.
*/
static RtreeDValue cellArea(Rtree *pRtree, RtreeCell *p){
  RtreeDValue area = (RtreeDValue)1;
  assert( pRtree->nDim>=1 && pRtree->nDim<=5 );
#ifndef SQLITE_RTREE_INT_ONLY
  if( pRtree->eCoordType==RTREE_COORD_REAL32 ){
    switch( pRtree->nDim ){
................................................................................
  return rc;
}

/*
** This routine deletes the RtreeGeomCallback object that was attached
** one of the SQL functions create by sqlite3_rtree_geometry_callback()
** or sqlite3_rtree_query_callback().  In other words, this routine is the
** destructor for an RtreeGeomCallback objecct.  This routine is called when
** the corresponding SQL function is deleted.
*/
static void rtreeFreeCallback(void *p){
  RtreeGeomCallback *pInfo = (RtreeGeomCallback*)p;
  if( pInfo->xDestructor ) pInfo->xDestructor(pInfo->pContext);
  sqlite3_free(p);
}







|







 







|







 







|







324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
....
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
....
4388
4389
4390
4391
4392
4393
4394
4395
4396
4397
4398
4399
4400
4401
4402
#define RTREE_GE    0x44  /* D */
#define RTREE_GT    0x45  /* E */
#define RTREE_MATCH 0x46  /* F: Old-style sqlite3_rtree_geometry_callback() */
#define RTREE_QUERY 0x47  /* G: New-style sqlite3_rtree_query_callback() */

/* Special operators available only on cursors.  Needs to be consecutive
** with the normal values above, but must be less than RTREE_MATCH.  These
** are used in the cursor for constraints such as x=NULL (RTREE_FALSE) or
** x<'xyz' (RTREE_TRUE) */
#define RTREE_TRUE  0x3f  /* ? */
#define RTREE_FALSE 0x40  /* @ */

/* 
** An rtree structure node.
*/
................................................................................
  pIdxInfo->estimatedCost = (double)6.0 * (double)nRow;
  pIdxInfo->estimatedRows = nRow;

  return rc;
}

/*
** Return the N-dimensional volume of the cell stored in *p.
*/
static RtreeDValue cellArea(Rtree *pRtree, RtreeCell *p){
  RtreeDValue area = (RtreeDValue)1;
  assert( pRtree->nDim>=1 && pRtree->nDim<=5 );
#ifndef SQLITE_RTREE_INT_ONLY
  if( pRtree->eCoordType==RTREE_COORD_REAL32 ){
    switch( pRtree->nDim ){
................................................................................
  return rc;
}

/*
** This routine deletes the RtreeGeomCallback object that was attached
** one of the SQL functions create by sqlite3_rtree_geometry_callback()
** or sqlite3_rtree_query_callback().  In other words, this routine is the
** destructor for an RtreeGeomCallback object.  This routine is called when
** the corresponding SQL function is deleted.
*/
static void rtreeFreeCallback(void *p){
  RtreeGeomCallback *pInfo = (RtreeGeomCallback*)p;
  if( pInfo->xDestructor ) pInfo->xDestructor(pInfo->pContext);
  sqlite3_free(p);
}

Changes to ext/rtree/rtree4.test.

144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
...
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
...
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
...
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
...
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
...
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
      lappend where mx$j>=$mn mn$j<=$mx
    }
    set where "WHERE [join $where { AND }]"
    do_test rtree4-$nDim.2.$i.3 {
      list $where [db eval "SELECT id FROM rx $where ORDER BY id"]
    } [list $where [db eval "SELECT id FROM bx $where ORDER BY id"]]

    # Do a contained-in query with surplus contraints at the beginning.
    # This should force a full-table scan on the rtree.
    #
    set where {}
    for {set j 0} {$j<$nDim} {incr j} {
      lappend where mn$j>-10000 mx$j<10000
    }
    for {set j 0} {$j<$nDim} {incr j} {
................................................................................
      lappend where mn$j>=$mn mx$j<=$mx
    }
    set where "WHERE [join $where { AND }]"
    do_test rtree4-$nDim.2.$i.3 {
      list $where [db eval "SELECT id FROM rx $where ORDER BY id"]
    } [list $where [db eval "SELECT id FROM bx $where ORDER BY id"]]

    # Do an overlaps query with surplus contraints at the beginning.
    # This should force a full-table scan on the rtree.
    #
    set where {}
    for {set j 0} {$j<$nDim} {incr j} {
      lappend where mn$j>=-10000 mx$j<=10000
    }
    for {set j 0} {$j<$nDim} {incr j} {
................................................................................
      lappend where mx$j>$mn mn$j<$mx
    }
    set where "WHERE [join $where { AND }]"
    do_test rtree4-$nDim.2.$i.4 {
      list $where [db eval "SELECT id FROM rx $where ORDER BY id"]
    } [list $where [db eval "SELECT id FROM bx $where ORDER BY id"]]

    # Do a contained-in query with surplus contraints at the end
    #
    set where {}
    for {set j 0} {$j<$nDim} {incr j} {
      set mn [rand 10000]
      set mx [expr {$mn+[randincr 500]}]
      lappend where mn$j>=$mn mx$j<$mx
    }
................................................................................
      lappend where mn$j>=-10000 mx$j<10000
    }
    set where "WHERE [join $where { AND }]"
    do_test rtree4-$nDim.2.$i.5 {
      list $where [db eval "SELECT id FROM rx $where ORDER BY id"]
    } [list $where [db eval "SELECT id FROM bx $where ORDER BY id"]]

    # Do an overlaps query with surplus contraints at the end
    #
    set where {}
    for {set j [expr {$nDim-1}]} {$j>=0} {incr j -1} {
      set mn [rand 10000]
      set mx [expr {$mn+[randincr 500]}]
      lappend where mx$j>$mn mn$j<=$mx
    }
................................................................................
      lappend where mx$j>-10000 mn$j<=10000
    }
    set where "WHERE [join $where { AND }]"
    do_test rtree4-$nDim.2.$i.6 {
      list $where [db eval "SELECT id FROM rx $where ORDER BY id"]
    } [list $where [db eval "SELECT id FROM bx $where ORDER BY id"]]

    # Do a contained-in query with surplus contraints where the 
    # constraints appear in a random order.
    #
    set where {}
    for {set j 0} {$j<$nDim} {incr j} {
      set mn1 [rand 10000]
      set mn2 [expr {$mn1+[randincr 100]}]
      set mx1 [expr {$mn2+[randincr 400]}]
................................................................................
      lappend where mn$j>=$mn1 mn$j>$mn2 mx$j<$mx1 mx$j<=$mx2
    }
    set where "WHERE [join [scramble $where] { AND }]"
    do_test rtree4-$nDim.2.$i.7 {
      list $where [db eval "SELECT id FROM rx $where ORDER BY id"]
    } [list $where [db eval "SELECT id FROM bx $where ORDER BY id"]]

    # Do an overlaps query with surplus contraints where the
    # constraints appear in a random order.
    #
    set where {}
    for {set j 0} {$j<$nDim} {incr j} {
      set mn1 [rand 10000]
      set mn2 [expr {$mn1+[randincr 100]}]
      set mx1 [expr {$mn2+[randincr 400]}]







|







 







|







 







|







 







|







 







|







 







|







144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
...
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
...
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
...
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
...
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
...
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
      lappend where mx$j>=$mn mn$j<=$mx
    }
    set where "WHERE [join $where { AND }]"
    do_test rtree4-$nDim.2.$i.3 {
      list $where [db eval "SELECT id FROM rx $where ORDER BY id"]
    } [list $where [db eval "SELECT id FROM bx $where ORDER BY id"]]

    # Do a contained-in query with surplus constraints at the beginning.
    # This should force a full-table scan on the rtree.
    #
    set where {}
    for {set j 0} {$j<$nDim} {incr j} {
      lappend where mn$j>-10000 mx$j<10000
    }
    for {set j 0} {$j<$nDim} {incr j} {
................................................................................
      lappend where mn$j>=$mn mx$j<=$mx
    }
    set where "WHERE [join $where { AND }]"
    do_test rtree4-$nDim.2.$i.3 {
      list $where [db eval "SELECT id FROM rx $where ORDER BY id"]
    } [list $where [db eval "SELECT id FROM bx $where ORDER BY id"]]

    # Do an overlaps query with surplus constraints at the beginning.
    # This should force a full-table scan on the rtree.
    #
    set where {}
    for {set j 0} {$j<$nDim} {incr j} {
      lappend where mn$j>=-10000 mx$j<=10000
    }
    for {set j 0} {$j<$nDim} {incr j} {
................................................................................
      lappend where mx$j>$mn mn$j<$mx
    }
    set where "WHERE [join $where { AND }]"
    do_test rtree4-$nDim.2.$i.4 {
      list $where [db eval "SELECT id FROM rx $where ORDER BY id"]
    } [list $where [db eval "SELECT id FROM bx $where ORDER BY id"]]

    # Do a contained-in query with surplus constraints at the end
    #
    set where {}
    for {set j 0} {$j<$nDim} {incr j} {
      set mn [rand 10000]
      set mx [expr {$mn+[randincr 500]}]
      lappend where mn$j>=$mn mx$j<$mx
    }
................................................................................
      lappend where mn$j>=-10000 mx$j<10000
    }
    set where "WHERE [join $where { AND }]"
    do_test rtree4-$nDim.2.$i.5 {
      list $where [db eval "SELECT id FROM rx $where ORDER BY id"]
    } [list $where [db eval "SELECT id FROM bx $where ORDER BY id"]]

    # Do an overlaps query with surplus constraints at the end
    #
    set where {}
    for {set j [expr {$nDim-1}]} {$j>=0} {incr j -1} {
      set mn [rand 10000]
      set mx [expr {$mn+[randincr 500]}]
      lappend where mx$j>$mn mn$j<=$mx
    }
................................................................................
      lappend where mx$j>-10000 mn$j<=10000
    }
    set where "WHERE [join $where { AND }]"
    do_test rtree4-$nDim.2.$i.6 {
      list $where [db eval "SELECT id FROM rx $where ORDER BY id"]
    } [list $where [db eval "SELECT id FROM bx $where ORDER BY id"]]

    # Do a contained-in query with surplus constraints where the 
    # constraints appear in a random order.
    #
    set where {}
    for {set j 0} {$j<$nDim} {incr j} {
      set mn1 [rand 10000]
      set mn2 [expr {$mn1+[randincr 100]}]
      set mx1 [expr {$mn2+[randincr 400]}]
................................................................................
      lappend where mn$j>=$mn1 mn$j>$mn2 mx$j<$mx1 mx$j<=$mx2
    }
    set where "WHERE [join [scramble $where] { AND }]"
    do_test rtree4-$nDim.2.$i.7 {
      list $where [db eval "SELECT id FROM rx $where ORDER BY id"]
    } [list $where [db eval "SELECT id FROM bx $where ORDER BY id"]]

    # Do an overlaps query with surplus constraints where the
    # constraints appear in a random order.
    #
    set where {}
    for {set j 0} {$j<$nDim} {incr j} {
      set mn1 [rand 10000]
      set mn2 [expr {$mn1+[randincr 100]}]
      set mx1 [expr {$mn2+[randincr 400]}]

Changes to ext/rtree/rtree_util.tcl.

190
191
192
193
194
195
196
197
  set d [rtree_depth $db $zTab]
  rtree_nodetreedump $db $zTab "" $d 1
}

proc do_rtree_integrity_test {tn tbl} {
  uplevel [list do_execsql_test $tn "SELECT rtreecheck('$tbl')" ok]
}








<
190
191
192
193
194
195
196

  set d [rtree_depth $db $zTab]
  rtree_nodetreedump $db $zTab "" $d 1
}

proc do_rtree_integrity_test {tn tbl} {
  uplevel [list do_execsql_test $tn "SELECT rtreecheck('$tbl')" ok]
}

Changes to ext/rtree/rtreecirc.test.

59
60
61
62
63
64
65
66
  do_execsql_test  1.1.$tn.1 $schema
  do_catchsql_test 1.1.$tn.2 $sql {1 {no such table: main.rt}}
  db close
}


finish_test








<
59
60
61
62
63
64
65

  do_execsql_test  1.1.$tn.1 $schema
  do_catchsql_test 1.1.$tn.2 $sql {1 {no such table: main.rt}}
  db close
}


finish_test

Changes to ext/session/changesetfuzz1.test.

77
78
79
80
81
82
83
84
      forcecopy input.patchset-0 input.patchset
    }
  }
} {}


finish_test








<
77
78
79
80
81
82
83

      forcecopy input.patchset-0 input.patchset
    }
  }
} {}


finish_test

Changes to ext/session/sessionwor.test.

93
94
95
96
97
98
99
100
    DELETE FROM t1;
  } {
    {DELETE t1 0 X. {i 1 t four} {}}
  }
}

finish_test








<
93
94
95
96
97
98
99

    DELETE FROM t1;
  } {
    {DELETE t1 0 X. {i 1 t four} {}}
  }
}

finish_test

Changes to ext/session/sqlite3session.c.

1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
....
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
....
2363
2364
2365
2366
2367
2368
2369
2370
2371
2372
2373
2374
2375
2376
2377
....
3779
3780
3781
3782
3783
3784
3785
3786
3787
3788
3789
3790
3791
3792
3793
....
3824
3825
3826
3827
3828
3829
3830
3831
3832
3833
3834
3835
3836
3837
3838
....
3843
3844
3845
3846
3847
3848
3849
3850
3851
3852
3853
3854
3855
3856
3857
  sqlite3_finalize(pStmt);
  return rc;
}

/*
** This function is only called from within a pre-update handler for a
** write to table pTab, part of session pSession. If this is the first
** write to this table, initalize the SessionTable.nCol, azCol[] and
** abPK[] arrays accordingly.
**
** If an error occurs, an error code is stored in sqlite3_session.rc and
** non-zero returned. Or, if no error occurs but the table has no primary
** key, sqlite3_session.rc is left set to SQLITE_OK and non-zero returned to
** indicate that updates on this table should be ignored. SessionTable.abPK 
** is set to NULL in this case.
................................................................................
    *zOut++ = '"';
    p->nBuf = (int)((u8 *)zOut - p->aBuf);
  }
}

/*
** This function is a no-op if *pRc is other than SQLITE_OK when it is
** called. Otherwse, it appends the serialized version of the value stored
** in column iCol of the row that SQL statement pStmt currently points
** to to the buffer.
*/
static void sessionAppendCol(
  SessionBuffer *p,               /* Buffer to append to */
  sqlite3_stmt *pStmt,            /* Handle pointing to row containing value */
  int iCol,                       /* Column to read value from */
................................................................................
  int (*xOutput)(void *pOut, const void *pData, int nData),
  void *pOut,                     /* First argument for xOutput */
  int *pnChangeset,               /* OUT: Size of buffer at *ppChangeset */
  void **ppChangeset              /* OUT: Buffer containing changeset */
){
  sqlite3 *db = pSession->db;     /* Source database handle */
  SessionTable *pTab;             /* Used to iterate through attached tables */
  SessionBuffer buf = {0,0,0};    /* Buffer in which to accumlate changeset */
  int rc;                         /* Return code */

  assert( xOutput==0 || (pnChangeset==0 && ppChangeset==0 ) );

  /* Zero the output variables in case an error occurs. If this session
  ** object is already in the error state (sqlite3_session.rc != SQLITE_OK),
  ** this call will be a no-op.  */
................................................................................
/*
** Iterator pIter must point to an SQLITE_INSERT entry. This function 
** transfers new.* values from the current iterator entry to statement
** pStmt. The table being inserted into has nCol columns.
**
** New.* value $i from the iterator is bound to variable ($i+1) of 
** statement pStmt. If parameter abPK is NULL, all values from 0 to (nCol-1)
** are transfered to the statement. Otherwise, if abPK is not NULL, it points
** to an array nCol elements in size. In this case only those values for 
** which abPK[$i] is true are read from the iterator and bound to the 
** statement.
**
** An SQLite error code is returned if an error occurs. Otherwise, SQLITE_OK.
*/
static int sessionBindRow(
................................................................................

/*
** SQL statement pSelect is as generated by the sessionSelectRow() function.
** This function binds the primary key values from the change that changeset
** iterator pIter points to to the SELECT and attempts to seek to the table
** entry. If a row is found, the SELECT statement left pointing at the row 
** and SQLITE_ROW is returned. Otherwise, if no row is found and no error
** has occured, the statement is reset and SQLITE_OK is returned. If an
** error occurs, the statement is reset and an SQLite error code is returned.
**
** If this function returns SQLITE_ROW, the caller must eventually reset() 
** statement pSelect. If any other value is returned, the statement does
** not require a reset().
**
** If the iterator currently points to an INSERT record, bind values from the
................................................................................
  sqlite3 *db,                    /* Database handle */
  sqlite3_changeset_iter *pIter,  /* Changeset iterator */
  u8 *abPK,                       /* Primary key flags array */
  sqlite3_stmt *pSelect           /* SELECT statement from sessionSelectRow() */
){
  int rc;                         /* Return code */
  int nCol;                       /* Number of columns in table */
  int op;                         /* Changset operation (SQLITE_UPDATE etc.) */
  const char *zDummy;             /* Unused */

  sqlite3changeset_op(pIter, &zDummy, &nCol, &op, 0);
  rc = sessionBindRow(pIter, 
      op==SQLITE_INSERT ? sqlite3changeset_new : sqlite3changeset_old,
      nCol, abPK, pSelect
  );







|







 







|







 







|







 







|







 







|







 







|







1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
....
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
....
2363
2364
2365
2366
2367
2368
2369
2370
2371
2372
2373
2374
2375
2376
2377
....
3779
3780
3781
3782
3783
3784
3785
3786
3787
3788
3789
3790
3791
3792
3793
....
3824
3825
3826
3827
3828
3829
3830
3831
3832
3833
3834
3835
3836
3837
3838
....
3843
3844
3845
3846
3847
3848
3849
3850
3851
3852
3853
3854
3855
3856
3857
  sqlite3_finalize(pStmt);
  return rc;
}

/*
** This function is only called from within a pre-update handler for a
** write to table pTab, part of session pSession. If this is the first
** write to this table, initialize the SessionTable.nCol, azCol[] and
** abPK[] arrays accordingly.
**
** If an error occurs, an error code is stored in sqlite3_session.rc and
** non-zero returned. Or, if no error occurs but the table has no primary
** key, sqlite3_session.rc is left set to SQLITE_OK and non-zero returned to
** indicate that updates on this table should be ignored. SessionTable.abPK 
** is set to NULL in this case.
................................................................................
    *zOut++ = '"';
    p->nBuf = (int)((u8 *)zOut - p->aBuf);
  }
}

/*
** This function is a no-op if *pRc is other than SQLITE_OK when it is
** called. Otherwise, it appends the serialized version of the value stored
** in column iCol of the row that SQL statement pStmt currently points
** to to the buffer.
*/
static void sessionAppendCol(
  SessionBuffer *p,               /* Buffer to append to */
  sqlite3_stmt *pStmt,            /* Handle pointing to row containing value */
  int iCol,                       /* Column to read value from */
................................................................................
  int (*xOutput)(void *pOut, const void *pData, int nData),
  void *pOut,                     /* First argument for xOutput */
  int *pnChangeset,               /* OUT: Size of buffer at *ppChangeset */
  void **ppChangeset              /* OUT: Buffer containing changeset */
){
  sqlite3 *db = pSession->db;     /* Source database handle */
  SessionTable *pTab;             /* Used to iterate through attached tables */
  SessionBuffer buf = {0,0,0};    /* Buffer in which to accumulate changeset */
  int rc;                         /* Return code */

  assert( xOutput==0 || (pnChangeset==0 && ppChangeset==0 ) );

  /* Zero the output variables in case an error occurs. If this session
  ** object is already in the error state (sqlite3_session.rc != SQLITE_OK),
  ** this call will be a no-op.  */
................................................................................
/*
** Iterator pIter must point to an SQLITE_INSERT entry. This function 
** transfers new.* values from the current iterator entry to statement
** pStmt. The table being inserted into has nCol columns.
**
** New.* value $i from the iterator is bound to variable ($i+1) of 
** statement pStmt. If parameter abPK is NULL, all values from 0 to (nCol-1)
** are transferred to the statement. Otherwise, if abPK is not NULL, it points
** to an array nCol elements in size. In this case only those values for 
** which abPK[$i] is true are read from the iterator and bound to the 
** statement.
**
** An SQLite error code is returned if an error occurs. Otherwise, SQLITE_OK.
*/
static int sessionBindRow(
................................................................................

/*
** SQL statement pSelect is as generated by the sessionSelectRow() function.
** This function binds the primary key values from the change that changeset
** iterator pIter points to to the SELECT and attempts to seek to the table
** entry. If a row is found, the SELECT statement left pointing at the row 
** and SQLITE_ROW is returned. Otherwise, if no row is found and no error
** has occurred, the statement is reset and SQLITE_OK is returned. If an
** error occurs, the statement is reset and an SQLite error code is returned.
**
** If this function returns SQLITE_ROW, the caller must eventually reset() 
** statement pSelect. If any other value is returned, the statement does
** not require a reset().
**
** If the iterator currently points to an INSERT record, bind values from the
................................................................................
  sqlite3 *db,                    /* Database handle */
  sqlite3_changeset_iter *pIter,  /* Changeset iterator */
  u8 *abPK,                       /* Primary key flags array */
  sqlite3_stmt *pSelect           /* SELECT statement from sessionSelectRow() */
){
  int rc;                         /* Return code */
  int nCol;                       /* Number of columns in table */
  int op;                         /* Changeset operation (SQLITE_UPDATE etc.) */
  const char *zDummy;             /* Unused */

  sqlite3changeset_op(pIter, &zDummy, &nCol, &op, 0);
  rc = sessionBindRow(pIter, 
      op==SQLITE_INSERT ? sqlite3changeset_new : sqlite3changeset_old,
      nCol, abPK, pSelect
  );

Changes to ext/userauth/sqlite3userauth.h.

31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
** appropriate username and password prior to enable read and write
** access to the database.
**
** Return SQLITE_OK on success or SQLITE_ERROR if the username/password
** combination is incorrect or unknown.
**
** If the SQLITE_USER table is not present in the database file, then
** this interface is a harmless no-op returnning SQLITE_OK.
*/
int sqlite3_user_authenticate(
  sqlite3 *db,           /* The database connection */
  const char *zUsername, /* Username */
  const char *aPW,       /* Password or credentials */
  int nPW                /* Number of bytes in aPW[] */
);







|







31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
** appropriate username and password prior to enable read and write
** access to the database.
**
** Return SQLITE_OK on success or SQLITE_ERROR if the username/password
** combination is incorrect or unknown.
**
** If the SQLITE_USER table is not present in the database file, then
** this interface is a harmless no-op returning SQLITE_OK.
*/
int sqlite3_user_authenticate(
  sqlite3 *db,           /* The database connection */
  const char *zUsername, /* Username */
  const char *aPW,       /* Password or credentials */
  int nPW                /* Number of bytes in aPW[] */
);

Changes to ext/userauth/user-auth.txt.

150
151
152
153
154
155
156
157
158
159
160
161
162
163
164

    sqlite_user.pw == sqlite_crypt(X, sqlite_user.pw)

To compute an appropriate sqlite_user.pw value from a new or modified
password X, sqlite_crypt(X,NULL) is run.  A new random salt is selected
when the second argument is NULL.

The built-in version of of sqlite_crypt() uses a simple Ceasar-cypher
which prevents passwords from being revealed by searching the raw database
for ASCII text, but is otherwise trivally broken.  For better password
security, the database should be encrypted using the SQLite Encryption
Extension or similar technology.  Or, the application can use the
sqlite3_create_function() interface to provide an alternative
implementation of sqlite_crypt() that computes a stronger password hash,
perhaps using a cryptographic hash function like SHA1.







|

|





150
151
152
153
154
155
156
157
158
159
160
161
162
163
164

    sqlite_user.pw == sqlite_crypt(X, sqlite_user.pw)

To compute an appropriate sqlite_user.pw value from a new or modified
password X, sqlite_crypt(X,NULL) is run.  A new random salt is selected
when the second argument is NULL.

The built-in version of of sqlite_crypt() uses a simple Caesar-cypher
which prevents passwords from being revealed by searching the raw database
for ASCII text, but is otherwise trivially broken.  For better password
security, the database should be encrypted using the SQLite Encryption
Extension or similar technology.  Or, the application can use the
sqlite3_create_function() interface to provide an alternative
implementation of sqlite_crypt() that computes a stronger password hash,
perhaps using a cryptographic hash function like SHA1.

Changes to ext/userauth/userauth.c.

184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
** appropriate username and password prior to enable read and write
** access to the database.
**
** Return SQLITE_OK on success or SQLITE_ERROR if the username/password
** combination is incorrect or unknown.
**
** If the SQLITE_USER table is not present in the database file, then
** this interface is a harmless no-op returnning SQLITE_OK.
*/
int sqlite3_user_authenticate(
  sqlite3 *db,           /* The database connection */
  const char *zUsername, /* Username */
  const char *zPW,       /* Password or credentials */
  int nPW                /* Number of bytes in aPW[] */
){







|







184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
** appropriate username and password prior to enable read and write
** access to the database.
**
** Return SQLITE_OK on success or SQLITE_ERROR if the username/password
** combination is incorrect or unknown.
**
** If the SQLITE_USER table is not present in the database file, then
** this interface is a harmless no-op returning SQLITE_OK.
*/
int sqlite3_user_authenticate(
  sqlite3 *db,           /* The database connection */
  const char *zUsername, /* Username */
  const char *zPW,       /* Password or credentials */
  int nPW                /* Number of bytes in aPW[] */
){

Changes to ltmain.sh.

2432
2433
2434
2435
2436
2437
2438
2439
2440
2441
2442
2443
2444
2445
2446
....
3536
3537
3538
3539
3540
3541
3542
3543
3544
3545
3546
3547
3548
3549
3550
....
8454
8455
8456
8457
8458
8459
8460
8461
  const char *name;
  void *address;
} lt_dlsymlist;
"
	  case $host in
	  *cygwin* | *mingw* | *cegcc* )
	    $ECHO >> "$output_objdir/$my_dlsyms" "\
/* DATA imports from DLLs on WIN32 con't be const, because
   runtime relocations are performed -- see ld's documentation
   on pseudo-relocs.  */"
	    lt_dlsym_const= ;;
	  *osf5*)
	    echo >> "$output_objdir/$my_dlsyms" "\
/* This system does not cope well with relocations in const data */"
	    lt_dlsym_const= ;;
................................................................................
            lt_fatal ("%s missing required argument", env_append_opt);
          continue;
        }
      if (strncmp (argv[i], ltwrapper_option_prefix, opt_prefix_len) == 0)
        {
          /* however, if there is an option in the LTWRAPPER_OPTION_PREFIX
             namespace, but it is not one of the ones we know about and
             have already dealt with, above (inluding dump-script), then
             report an error. Otherwise, targets might begin to believe
             they are allowed to use options in the LTWRAPPER_OPTION_PREFIX
             namespace. The first time any user complains about this, we'll
             need to make LTWRAPPER_OPTION_PREFIX a configure-time option
             or a configure.ac-settable value.
           */
          lt_fatal ("Unrecognized option in %s namespace: '%s'",
................................................................................
# ### END LIBTOOL TAG CONFIG: disable-static

# Local Variables:
# mode:shell-script
# sh-indentation:2
# End:
# vi:sw=2








|







 







|







 







<
2432
2433
2434
2435
2436
2437
2438
2439
2440
2441
2442
2443
2444
2445
2446
....
3536
3537
3538
3539
3540
3541
3542
3543
3544
3545
3546
3547
3548
3549
3550
....
8454
8455
8456
8457
8458
8459
8460

  const char *name;
  void *address;
} lt_dlsymlist;
"
	  case $host in
	  *cygwin* | *mingw* | *cegcc* )
	    $ECHO >> "$output_objdir/$my_dlsyms" "\
/* DATA imports from DLLs on WIN32 can't be const, because
   runtime relocations are performed -- see ld's documentation
   on pseudo-relocs.  */"
	    lt_dlsym_const= ;;
	  *osf5*)
	    echo >> "$output_objdir/$my_dlsyms" "\
/* This system does not cope well with relocations in const data */"
	    lt_dlsym_const= ;;
................................................................................
            lt_fatal ("%s missing required argument", env_append_opt);
          continue;
        }
      if (strncmp (argv[i], ltwrapper_option_prefix, opt_prefix_len) == 0)
        {
          /* however, if there is an option in the LTWRAPPER_OPTION_PREFIX
             namespace, but it is not one of the ones we know about and
             have already dealt with, above (including dump-script), then
             report an error. Otherwise, targets might begin to believe
             they are allowed to use options in the LTWRAPPER_OPTION_PREFIX
             namespace. The first time any user complains about this, we'll
             need to make LTWRAPPER_OPTION_PREFIX a configure-time option
             or a configure.ac-settable value.
           */
          lt_fatal ("Unrecognized option in %s namespace: '%s'",
................................................................................
# ### END LIBTOOL TAG CONFIG: disable-static

# Local Variables:
# mode:shell-script
# sh-indentation:2
# End:
# vi:sw=2

Changes to main.mk.

1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
# This target will fail if the SQLite amalgamation contains any exported
# symbols that do not begin with "sqlite3_". It is run as part of the
# releasetest.tcl script.
#
checksymbols: sqlite3.o
	nm -g --defined-only sqlite3.o | grep -v " sqlite3_" ; test $$? -ne 0

# Build the amalgamation-autoconf package.  The amalamgation-tarball target builds
# a tarball named for the version number.  Ex:  sqlite-autoconf-3110000.tar.gz.
# The snapshot-tarball target builds a tarball named by the SHA1 hash
#
amalgamation-tarball: sqlite3.c sqlite3rc.h
	TOP=$(TOP) sh $(TOP)/tool/mkautoconfamal.sh --normal

snapshot-tarball: sqlite3.c sqlite3rc.h







|







1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
# This target will fail if the SQLite amalgamation contains any exported
# symbols that do not begin with "sqlite3_". It is run as part of the
# releasetest.tcl script.
#
checksymbols: sqlite3.o
	nm -g --defined-only sqlite3.o | grep -v " sqlite3_" ; test $$? -ne 0

# Build the amalgamation-autoconf package.  The amalagmation-tarball target builds
# a tarball named for the version number.  Ex:  sqlite-autoconf-3110000.tar.gz.
# The snapshot-tarball target builds a tarball named by the SHA1 hash
#
amalgamation-tarball: sqlite3.c sqlite3rc.h
	TOP=$(TOP) sh $(TOP)/tool/mkautoconfamal.sh --normal

snapshot-tarball: sqlite3.c sqlite3rc.h

Changes to src/alter.c.

941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
  for(pp=&pCtx->pList; *pp!=pBest; pp=&(*pp)->pNext);
  *pp = pBest->pNext;

  return pBest;
}

/*
** An error occured while parsing or otherwise processing a database
** object (either pParse->pNewTable, pNewIndex or pNewTrigger) as part of an
** ALTER TABLE RENAME COLUMN program. The error message emitted by the
** sub-routine is currently stored in pParse->zErrMsg. This function
** adds context to the error message and then stores it in pCtx.
*/
static void renameColumnParseError(
  sqlite3_context *pCtx, 







|







941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
  for(pp=&pCtx->pList; *pp!=pBest; pp=&(*pp)->pNext);
  *pp = pBest->pNext;

  return pBest;
}

/*
** An error occurred while parsing or otherwise processing a database
** object (either pParse->pNewTable, pNewIndex or pNewTrigger) as part of an
** ALTER TABLE RENAME COLUMN program. The error message emitted by the
** sub-routine is currently stored in pParse->zErrMsg. This function
** adds context to the error message and then stores it in pCtx.
*/
static void renameColumnParseError(
  sqlite3_context *pCtx, 

Changes to src/auth.c.

201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
  const char *zArg1,
  const char *zArg2,
  const char *zArg3
){
  sqlite3 *db = pParse->db;
  int rc;

  /* Don't do any authorization checks if the database is initialising
  ** or if the parser is being invoked from within sqlite3_declare_vtab.
  */
  assert( !IN_RENAME_OBJECT || db->xAuth==0 );
  if( db->init.busy || IN_SPECIAL_PARSE ){
    return SQLITE_OK;
  }








|







201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
  const char *zArg1,
  const char *zArg2,
  const char *zArg3
){
  sqlite3 *db = pParse->db;
  int rc;

  /* Don't do any authorization checks if the database is initializing
  ** or if the parser is being invoked from within sqlite3_declare_vtab.
  */
  assert( !IN_RENAME_OBJECT || db->xAuth==0 );
  if( db->init.busy || IN_SPECIAL_PARSE ){
    return SQLITE_OK;
  }

Changes to src/bitvec.c.

184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
  if( p->iSize<=BITVEC_NBIT ){
    p->u.aBitmap[i/BITVEC_SZELEM] |= 1 << (i&(BITVEC_SZELEM-1));
    return SQLITE_OK;
  }
  h = BITVEC_HASH(i++);
  /* if there wasn't a hash collision, and this doesn't */
  /* completely fill the hash, then just add it without */
  /* worring about sub-dividing and re-hashing. */
  if( !p->u.aHash[h] ){
    if (p->nSet<(BITVEC_NINT-1)) {
      goto bitvec_set_end;
    } else {
      goto bitvec_set_rehash;
    }
  }







|







184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
  if( p->iSize<=BITVEC_NBIT ){
    p->u.aBitmap[i/BITVEC_SZELEM] |= 1 << (i&(BITVEC_SZELEM-1));
    return SQLITE_OK;
  }
  h = BITVEC_HASH(i++);
  /* if there wasn't a hash collision, and this doesn't */
  /* completely fill the hash, then just add it without */
  /* worrying about sub-dividing and re-hashing. */
  if( !p->u.aHash[h] ){
    if (p->nSet<(BITVEC_NINT-1)) {
      goto bitvec_set_end;
    } else {
      goto bitvec_set_rehash;
    }
  }

Changes to src/btmutex.c.

174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
** connection.  This is needed (for example) prior to parsing
** a statement since we will be comparing table and column names
** against all schemas and we do not want those schemas being
** reset out from under us.
**
** There is a corresponding leave-all procedures.
**
** Enter the mutexes in accending order by BtShared pointer address
** to avoid the possibility of deadlock when two threads with
** two or more btrees in common both try to lock all their btrees
** at the same instant.
*/
static void SQLITE_NOINLINE btreeEnterAll(sqlite3 *db){
  int i;
  int skipOk = 1;







|







174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
** connection.  This is needed (for example) prior to parsing
** a statement since we will be comparing table and column names
** against all schemas and we do not want those schemas being
** reset out from under us.
**
** There is a corresponding leave-all procedures.
**
** Enter the mutexes in ascending order by BtShared pointer address
** to avoid the possibility of deadlock when two threads with
** two or more btrees in common both try to lock all their btrees
** at the same instant.
*/
static void SQLITE_NOINLINE btreeEnterAll(sqlite3 *db){
  int i;
  int skipOk = 1;

Changes to src/btree.c.

480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
...
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
....
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
....
2244
2245
2246
2247
2248
2249
2250
2251
2252
2253
2254
2255
2256
2257
2258
....
2674
2675
2676
2677
2678
2679
2680
2681
2682
2683
2684
2685
2686
2687
2688
....
2906
2907
2908
2909
2910
2911
2912
2913
2914
2915
2916
2917
2918
2919
2920
....
4628
4629
4630
4631
4632
4633
4634
4635
4636
4637
4638
4639
4640
4641
4642
....
5437
5438
5439
5440
5441
5442
5443
5444
5445
5446
5447
5448
5449
5450
5451
....
7317
7318
7319
7320
7321
7322
7323
7324
7325
7326
7327
7328
7329
7330
7331
....
7908
7909
7910
7911
7912
7913
7914
7915
7916
7917
7918
7919
7920
7921
7922
....
8508
8509
8510
8511
8512
8513
8514
8515
8516
8517
8518
8519
8520
8521
8522
*/
#ifdef SQLITE_DEBUG
static int cursorHoldsMutex(BtCursor *p){
  return sqlite3_mutex_held(p->pBt->mutex);
}

/* Verify that the cursor and the BtShared agree about what is the current
** database connetion. This is important in shared-cache mode. If the database 
** connection pointers get out-of-sync, it is possible for routines like
** btreeInitPage() to reference an stale connection pointer that references a
** a connection that has already closed.  This routine is used inside assert()
** statements only and for the purpose of double-checking that the btree code
** does keep the database connection pointers up-to-date.
*/
static int cursorOwnsBtShared(BtCursor *p){
................................................................................

/*
** This routine restores a cursor back to its original position after it
** has been moved by some outside activity (such as a btree rebalance or
** a row having been deleted out from under the cursor).  
**
** On success, the *pDifferentRow parameter is false if the cursor is left
** pointing at exactly the same row.  *pDifferntRow is the row the cursor
** was pointing to has been deleted, forcing the cursor to point to some
** nearby row.
**
** This routine should only be called for a cursor that just returned
** TRUE from sqlite3BtreeCursorHasMoved().
*/
int sqlite3BtreeCursorRestore(BtCursor *pCur, int *pDifferentRow){
................................................................................
** The first byte of the new free block is pPage->aData[iStart]
** and the size of the block is iSize bytes.
**
** Adjacent freeblocks are coalesced.
**
** Even though the freeblock list was checked by btreeComputeFreeSpace(),
** that routine will not detect overlap between cells or freeblocks.  Nor
** does it detect cells or freeblocks that encrouch into the reserved bytes
** at the end of the page.  So do additional corruption checks inside this
** routine and return SQLITE_CORRUPT if any problems are found.
*/
static int freeSpace(MemPage *pPage, u16 iStart, u16 iSize){
  u16 iPtr;                             /* Address of ptr to next freeblock */
  u16 iFreeBlk;                         /* Address of the next freeblock */
  u8 hdr;                               /* Page header size.  0 or 100 */
................................................................................

/*
** Get an unused page.
**
** This works just like btreeGetPage() with the addition:
**
**   *  If the page is already in use for some other purpose, immediately
**      release it and return an SQLITE_CURRUPT error.
**   *  Make sure the isInit flag is clear
*/
static int btreeGetUnusedPage(
  BtShared *pBt,       /* The btree */
  Pgno pgno,           /* Number of the page to fetch */
  MemPage **ppPage,    /* Return the page in this parameter */
  int flags            /* PAGER_GET_NOCONTENT or PAGER_GET_READONLY */
................................................................................
    /* One of the uses of pBt->pTmpSpace is to format cells before
    ** inserting them into a leaf page (function fillInCell()). If
    ** a cell is less than 4 bytes in size, it is rounded up to 4 bytes
    ** by the various routines that manipulate binary cells. Which
    ** can mean that fillInCell() only initializes the first 2 or 3
    ** bytes of pTmpSpace, but that the first 4 bytes are copied from
    ** it into a database page. This is not actually a problem, but it
    ** does cause a valgrind error when the 1 or 2 bytes of unitialized 
    ** data is passed to system call write(). So to avoid this error,
    ** zero the first 4 bytes of temp space here.
    **
    ** Also:  Provide four bytes of initialized space before the
    ** beginning of pTmpSpace as an area available to prepend the
    ** left-child pointer to the beginning of a cell.
    */
................................................................................
  assert( sqlite3_mutex_held(p->pBt->mutex) );
  n = p->pBt->pageSize - p->pBt->usableSize;
  return n;
}

/*
** Return the number of bytes of space at the end of every page that
** are intentually left unused.  This is the "reserved" space that is
** sometimes used by extensions.
**
** The value returned is the larger of the current reserve size and
** the latest reserve size requested by SQLITE_FILECTRL_RESERVE_BYTES.
** The amount of reserve can only grow - never shrink.
*/
int sqlite3BtreeGetRequestedReserve(Btree *p){
................................................................................
** Return an upper bound on the size of any record for the table
** that the cursor is pointing into.
**
** This is an optimization.  Everything will still work if this
** routine always returns 2147483647 (which is the largest record
** that SQLite can handle) or more.  But returning a smaller value might
** prevent large memory allocations when trying to interpret a
** corrupt datrabase.
**
** The current implementation merely returns the size of the underlying
** database file.
*/
sqlite3_int64 sqlite3BtreeMaxRecordSize(BtCursor *pCur){
  assert( cursorHoldsMutex(pCur) );
  assert( pCur->eState==CURSOR_VALID );
................................................................................
      if( (pCur->curFlags & BTCF_AtLast)!=0 ){
        *pRes = -1;
        return SQLITE_OK;
      }
      /* If the requested key is one more than the previous key, then
      ** try to get there using sqlite3BtreeNext() rather than a full
      ** binary search.  This is an optimization only.  The correct answer
      ** is still obtained without this case, only a little more slowely */
      if( pCur->info.nKey+1==intKey ){
        *pRes = 0;
        rc = sqlite3BtreeNext(pCur, 0);
        if( rc==SQLITE_OK ){
          getCellInfo(pCur);
          if( pCur->info.nKey==intKey ){
            return SQLITE_OK;
................................................................................
    }
    pNew->nFree = pBt->usableSize - pNew->cellOffset - 2 - szCell;

    /* If this is an auto-vacuum database, update the pointer map
    ** with entries for the new page, and any pointer from the 
    ** cell on the page to an overflow page. If either of these
    ** operations fails, the return code is set, but the contents
    ** of the parent page are still manipulated by thh code below.
    ** That is Ok, at this point the parent page is guaranteed to
    ** be marked as dirty. Returning an error code will cause a
    ** rollback, undoing any changes made to the parent page.
    */
    if( ISAUTOVACUUM ){
      ptrmapPut(pBt, pgnoNew, PTRMAP_BTREE, pParent->pgno, &rc);
      if( szCell>pNew->minLocal ){
................................................................................
    szNew[i-1] = szLeft;
    if( cntNew[i-1] <= (i>1 ? cntNew[i-2] : 0) ){
      rc = SQLITE_CORRUPT_BKPT;
      goto balance_cleanup;
    }
  }

  /* Sanity check:  For a non-corrupt database file one of the follwing
  ** must be true:
  **    (1) We found one or more cells (cntNew[0])>0), or
  **    (2) pPage is a virtual root page.  A virtual root page is when
  **        the real root page is page 1 and we are the only child of
  **        that page.
  */
  assert( cntNew[0]>0 || (pParent->pgno==1 && pParent->nCell==0) || CORRUPT_DB);
................................................................................
  u8 *pDest,                /* Pointer to the place to start writing */
  const BtreePayload *pX,   /* Source of data to write */
  int iOffset,              /* Offset of first byte to write */
  int iAmt                  /* Number of bytes to be written */
){
  int nData = pX->nData - iOffset;
  if( nData<=0 ){
    /* Overwritting with zeros */
    int i;
    for(i=0; i<iAmt && pDest[i]==0; i++){}
    if( i<iAmt ){
      int rc = sqlite3PagerWrite(pPage->pDbPage);
      if( rc ) return rc;
      memset(pDest + i, 0, iAmt - i);
    }







|







 







|







 







|







 







|







 







|







 







|







 







|







 







|







 







|







 







|







 







|







480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
...
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
....
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
....
2244
2245
2246
2247
2248
2249
2250
2251
2252
2253
2254
2255
2256
2257
2258
....
2674
2675
2676
2677
2678
2679
2680
2681
2682
2683
2684
2685
2686
2687
2688
....
2906
2907
2908
2909
2910
2911
2912
2913
2914
2915
2916
2917
2918
2919
2920
....
4628
4629
4630
4631
4632
4633
4634
4635
4636
4637
4638
4639
4640
4641
4642
....
5437
5438
5439
5440
5441
5442
5443
5444
5445
5446
5447
5448
5449
5450
5451
....
7317
7318
7319
7320
7321
7322
7323
7324
7325
7326
7327
7328
7329
7330
7331
....
7908
7909
7910
7911
7912
7913
7914
7915
7916
7917
7918
7919
7920
7921
7922
....
8508
8509
8510
8511
8512
8513
8514
8515
8516
8517
8518
8519
8520
8521
8522
*/
#ifdef SQLITE_DEBUG
static int cursorHoldsMutex(BtCursor *p){
  return sqlite3_mutex_held(p->pBt->mutex);
}

/* Verify that the cursor and the BtShared agree about what is the current
** database connection. This is important in shared-cache mode. If the database 
** connection pointers get out-of-sync, it is possible for routines like
** btreeInitPage() to reference an stale connection pointer that references a
** a connection that has already closed.  This routine is used inside assert()
** statements only and for the purpose of double-checking that the btree code
** does keep the database connection pointers up-to-date.
*/
static int cursorOwnsBtShared(BtCursor *p){
................................................................................

/*
** This routine restores a cursor back to its original position after it
** has been moved by some outside activity (such as a btree rebalance or
** a row having been deleted out from under the cursor).  
**
** On success, the *pDifferentRow parameter is false if the cursor is left
** pointing at exactly the same row.  *pDifferentRow is the row the cursor
** was pointing to has been deleted, forcing the cursor to point to some
** nearby row.
**
** This routine should only be called for a cursor that just returned
** TRUE from sqlite3BtreeCursorHasMoved().
*/
int sqlite3BtreeCursorRestore(BtCursor *pCur, int *pDifferentRow){
................................................................................
** The first byte of the new free block is pPage->aData[iStart]
** and the size of the block is iSize bytes.
**
** Adjacent freeblocks are coalesced.
**
** Even though the freeblock list was checked by btreeComputeFreeSpace(),
** that routine will not detect overlap between cells or freeblocks.  Nor
** does it detect cells or freeblocks that encroach into the reserved bytes
** at the end of the page.  So do additional corruption checks inside this
** routine and return SQLITE_CORRUPT if any problems are found.
*/
static int freeSpace(MemPage *pPage, u16 iStart, u16 iSize){
  u16 iPtr;                             /* Address of ptr to next freeblock */
  u16 iFreeBlk;                         /* Address of the next freeblock */
  u8 hdr;                               /* Page header size.  0 or 100 */
................................................................................

/*
** Get an unused page.
**
** This works just like btreeGetPage() with the addition:
**
**   *  If the page is already in use for some other purpose, immediately
**      release it and return an SQLITE_CORRUPT error.
**   *  Make sure the isInit flag is clear
*/
static int btreeGetUnusedPage(
  BtShared *pBt,       /* The btree */
  Pgno pgno,           /* Number of the page to fetch */
  MemPage **ppPage,    /* Return the page in this parameter */
  int flags            /* PAGER_GET_NOCONTENT or PAGER_GET_READONLY */
................................................................................
    /* One of the uses of pBt->pTmpSpace is to format cells before
    ** inserting them into a leaf page (function fillInCell()). If
    ** a cell is less than 4 bytes in size, it is rounded up to 4 bytes
    ** by the various routines that manipulate binary cells. Which
    ** can mean that fillInCell() only initializes the first 2 or 3
    ** bytes of pTmpSpace, but that the first 4 bytes are copied from
    ** it into a database page. This is not actually a problem, but it
    ** does cause a valgrind error when the 1 or 2 bytes of uninitialized 
    ** data is passed to system call write(). So to avoid this error,
    ** zero the first 4 bytes of temp space here.
    **
    ** Also:  Provide four bytes of initialized space before the
    ** beginning of pTmpSpace as an area available to prepend the
    ** left-child pointer to the beginning of a cell.
    */
................................................................................
  assert( sqlite3_mutex_held(p->pBt->mutex) );
  n = p->pBt->pageSize - p->pBt->usableSize;
  return n;
}

/*
** Return the number of bytes of space at the end of every page that
** are intentionally left unused.  This is the "reserved" space that is
** sometimes used by extensions.
**
** The value returned is the larger of the current reserve size and
** the latest reserve size requested by SQLITE_FILECTRL_RESERVE_BYTES.
** The amount of reserve can only grow - never shrink.
*/
int sqlite3BtreeGetRequestedReserve(Btree *p){
................................................................................
** Return an upper bound on the size of any record for the table
** that the cursor is pointing into.
**
** This is an optimization.  Everything will still work if this
** routine always returns 2147483647 (which is the largest record
** that SQLite can handle) or more.  But returning a smaller value might
** prevent large memory allocations when trying to interpret a
** corrupt database.
**
** The current implementation merely returns the size of the underlying
** database file.
*/
sqlite3_int64 sqlite3BtreeMaxRecordSize(BtCursor *pCur){
  assert( cursorHoldsMutex(pCur) );
  assert( pCur->eState==CURSOR_VALID );
................................................................................
      if( (pCur->curFlags & BTCF_AtLast)!=0 ){
        *pRes = -1;
        return SQLITE_OK;
      }
      /* If the requested key is one more than the previous key, then
      ** try to get there using sqlite3BtreeNext() rather than a full
      ** binary search.  This is an optimization only.  The correct answer
      ** is still obtained without this case, only a little more slowly */
      if( pCur->info.nKey+1==intKey ){
        *pRes = 0;
        rc = sqlite3BtreeNext(pCur, 0);
        if( rc==SQLITE_OK ){
          getCellInfo(pCur);
          if( pCur->info.nKey==intKey ){
            return SQLITE_OK;
................................................................................
    }
    pNew->nFree = pBt->usableSize - pNew->cellOffset - 2 - szCell;

    /* If this is an auto-vacuum database, update the pointer map
    ** with entries for the new page, and any pointer from the 
    ** cell on the page to an overflow page. If either of these
    ** operations fails, the return code is set, but the contents
    ** of the parent page are still manipulated by the code below.
    ** That is Ok, at this point the parent page is guaranteed to
    ** be marked as dirty. Returning an error code will cause a
    ** rollback, undoing any changes made to the parent page.
    */
    if( ISAUTOVACUUM ){
      ptrmapPut(pBt, pgnoNew, PTRMAP_BTREE, pParent->pgno, &rc);
      if( szCell>pNew->minLocal ){
................................................................................
    szNew[i-1] = szLeft;
    if( cntNew[i-1] <= (i>1 ? cntNew[i-2] : 0) ){
      rc = SQLITE_CORRUPT_BKPT;
      goto balance_cleanup;
    }
  }

  /* Sanity check:  For a non-corrupt database file one of the following
  ** must be true:
  **    (1) We found one or more cells (cntNew[0])>0), or
  **    (2) pPage is a virtual root page.  A virtual root page is when
  **        the real root page is page 1 and we are the only child of
  **        that page.
  */
  assert( cntNew[0]>0 || (pParent->pgno==1 && pParent->nCell==0) || CORRUPT_DB);
................................................................................
  u8 *pDest,                /* Pointer to the place to start writing */
  const BtreePayload *pX,   /* Source of data to write */
  int iOffset,              /* Offset of first byte to write */
  int iAmt                  /* Number of bytes to be written */
){
  int nData = pX->nData - iOffset;
  if( nData<=0 ){
    /* Overwriting with zeros */
    int i;
    for(i=0; i<iAmt && pDest[i]==0; i++){}
    if( i<iAmt ){
      int rc = sqlite3PagerWrite(pPage->pDbPage);
      if( rc ) return rc;
      memset(pDest + i, 0, iAmt - i);
    }

Changes to src/btree.h.

171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
**
**     The design of the _RANGE hint is aid b-tree implementations that try
**     to prefetch content from remote machines - to provide those
**     implementations with limits on what needs to be prefetched and thereby
**     reduce network bandwidth.
**
** Note that BTREE_HINT_FLAGS with BTREE_BULKLOAD is the only hint used by
** standard SQLite.  The other hints are provided for extentions that use
** the SQLite parser and code generator but substitute their own storage
** engine.
*/
#define BTREE_HINT_RANGE 0       /* Range constraints on queries */

/*
** Values that may be OR'd together to form the argument to the







|







171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
**
**     The design of the _RANGE hint is aid b-tree implementations that try
**     to prefetch content from remote machines - to provide those
**     implementations with limits on what needs to be prefetched and thereby
**     reduce network bandwidth.
**
** Note that BTREE_HINT_FLAGS with BTREE_BULKLOAD is the only hint used by
** standard SQLite.  The other hints are provided for extensions that use
** the SQLite parser and code generator but substitute their own storage
** engine.
*/
#define BTREE_HINT_RANGE 0       /* Range constraints on queries */

/*
** Values that may be OR'd together to form the argument to the

Changes to src/btreeInt.h.

259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
#define PTF_LEAF      0x08

/*
** An instance of this object stores information about each a single database
** page that has been loaded into memory.  The information in this object
** is derived from the raw on-disk page content.
**
** As each database page is loaded into memory, the pager allocats an
** instance of this object and zeros the first 8 bytes.  (This is the
** "extra" information associated with each page of the pager.)
**
** Access to all fields of this structure is controlled by the mutex
** stored in MemPage.pBt->mutex.
*/
struct MemPage {







|







259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
#define PTF_LEAF      0x08

/*
** An instance of this object stores information about each a single database
** page that has been loaded into memory.  The information in this object
** is derived from the raw on-disk page content.
**
** As each database page is loaded into memory, the pager allocates an
** instance of this object and zeros the first 8 bytes.  (This is the
** "extra" information associated with each page of the pager.)
**
** Access to all fields of this structure is controlled by the mutex
** stored in MemPage.pBt->mutex.
*/
struct MemPage {

Changes to src/build.c.

1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
....
3537
3538
3539
3540
3541
3542
3543
3544
3545
3546
3547
3548
3549
3550
3551
** This routine is called by the parser while in the middle of
** parsing a CREATE TABLE statement.
*/
void sqlite3AddDefaultValue(
  Parse *pParse,           /* Parsing context */
  Expr *pExpr,             /* The parsed expression of the default value */
  const char *zStart,      /* Start of the default value text */
  const char *zEnd         /* First character past end of defaut value text */
){
  Table *p;
  Column *pCol;
  sqlite3 *db = pParse->db;
  p = pParse->pNewTable;
  if( p!=0 ){
    int isInit = db->init.busy && db->init.iDb!=1;
................................................................................
    iDb = sqlite3TwoPartName(pParse, pName1, pName2, &pName);
    if( iDb<0 ) goto exit_create_index;
    assert( pName && pName->z );

#ifndef SQLITE_OMIT_TEMPDB
    /* If the index name was unqualified, check if the table
    ** is a temp table. If so, set the database to 1. Do not do this
    ** if initialising a database schema.
    */
    if( !db->init.busy ){
      pTab = sqlite3SrcListLookup(pParse, pTblName);
      if( pName2->n==0 && pTab && pTab->pSchema==db->aDb[1].pSchema ){
        iDb = 1;
      }
    }







|







 







|







1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
....
3537
3538
3539
3540
3541
3542
3543
3544
3545
3546
3547
3548
3549
3550
3551
** This routine is called by the parser while in the middle of
** parsing a CREATE TABLE statement.
*/
void sqlite3AddDefaultValue(
  Parse *pParse,           /* Parsing context */
  Expr *pExpr,             /* The parsed expression of the default value */
  const char *zStart,      /* Start of the default value text */
  const char *zEnd         /* First character past end of default value text */
){
  Table *p;
  Column *pCol;
  sqlite3 *db = pParse->db;
  p = pParse->pNewTable;
  if( p!=0 ){
    int isInit = db->init.busy && db->init.iDb!=1;
................................................................................
    iDb = sqlite3TwoPartName(pParse, pName1, pName2, &pName);
    if( iDb<0 ) goto exit_create_index;
    assert( pName && pName->z );

#ifndef SQLITE_OMIT_TEMPDB
    /* If the index name was unqualified, check if the table
    ** is a temp table. If so, set the database to 1. Do not do this
    ** if initialiing a database schema.
    */
    if( !db->init.busy ){
      pTab = sqlite3SrcListLookup(pParse, pTblName);
      if( pName2->n==0 && pTab && pTab->pSchema==db->aDb[1].pSchema ){
        iDb = 1;
      }
    }

Changes to src/delete.c.

145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
  ExprList *pOrderBy,          /* The ORDER BY clause.  May be null */
  Expr *pLimit,                /* The LIMIT clause.  May be null */
  char *zStmtType              /* Either DELETE or UPDATE.  For err msgs. */
){
  sqlite3 *db = pParse->db;
  Expr *pLhs = NULL;           /* LHS of IN(SELECT...) operator */
  Expr *pInClause = NULL;      /* WHERE rowid IN ( select ) */
  ExprList *pEList = NULL;     /* Expression list contaning only pSelectRowid */
  SrcList *pSelectSrc = NULL;  /* SELECT rowid FROM x ... (dup of pSrc) */
  Select *pSelect = NULL;      /* Complete SELECT tree */
  Table *pTab;

  /* Check that there isn't an ORDER BY without a LIMIT clause.
  */
  if( pOrderBy && pLimit==0 ) {







|







145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
  ExprList *pOrderBy,          /* The ORDER BY clause.  May be null */
  Expr *pLimit,                /* The LIMIT clause.  May be null */
  char *zStmtType              /* Either DELETE or UPDATE.  For err msgs. */
){
  sqlite3 *db = pParse->db;
  Expr *pLhs = NULL;           /* LHS of IN(SELECT...) operator */
  Expr *pInClause = NULL;      /* WHERE rowid IN ( select ) */
  ExprList *pEList = NULL;     /* Expressions containing only pSelectRowid */
  SrcList *pSelectSrc = NULL;  /* SELECT rowid FROM x ... (dup of pSrc) */
  Select *pSelect = NULL;      /* Complete SELECT tree */
  Table *pTab;

  /* Check that there isn't an ORDER BY without a LIMIT clause.
  */
  if( pOrderBy && pLimit==0 ) {

Changes to src/expr.c.

216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
....
2451
2452
2453
2454
2455
2456
2457
2458
2459
2460
2461
2462
2463
2464
2465
....
2473
2474
2475
2476
2477
2478
2479
2480
2481
2482
2483
2484
2485
2486
2487
2488
2489
2490
2491
2492
....
2809
2810
2811
2812
2813
2814
2815
2816
2817
2818
2819
2820
2821
2822
2823
....
3624
3625
3626
3627
3628
3629
3630
3631
3632
3633
3634
3635
3636
3637
3638
....
3839
3840
3841
3842
3843
3844
3845
3846
3847
3848
3849
3850
3851
3852
3853
....
5423
5424
5425
5426
5427
5428
5429
5430
5431
5432
5433
5434
5435
5436
5437
  }
  return pColl;
}

/*
** Return the collation sequence for the expression pExpr. If
** there is no defined collating sequence, return a pointer to the
** defautl collation sequence.
**
** See also: sqlite3ExprCollSeq()
**
** The sqlite3ExprCollSeq() routine works the same except that it
** returns NULL if there is no defined collation.
*/
CollSeq *sqlite3ExprNNCollSeq(Parse *pParse, const Expr *pExpr){
................................................................................
**
** The returned value of this function indicates the b-tree type, as follows:
**
**   IN_INDEX_ROWID      - The cursor was opened on a database table.
**   IN_INDEX_INDEX_ASC  - The cursor was opened on an ascending index.
**   IN_INDEX_INDEX_DESC - The cursor was opened on a descending index.
**   IN_INDEX_EPH        - The cursor was opened on a specially created and
**                         populated epheremal table.
**   IN_INDEX_NOOP       - No cursor was allocated.  The IN operator must be
**                         implemented as a sequence of comparisons.
**
** An existing b-tree might be used if the RHS expression pX is a simple
** subquery such as:
**
**     SELECT <column1>, <column2>... FROM <table>
................................................................................
** IN_INDEX_MEMBERSHIP or IN_INDEX_LOOP but not both.  If inFlags contains
** IN_INDEX_MEMBERSHIP, then the generated table will be used for a fast
** membership test.  When the IN_INDEX_LOOP bit is set, the IN index will
** be used to loop over all values of the RHS of the IN operator.
**
** When IN_INDEX_LOOP is used (and the b-tree will be used to iterate
** through the set members) then the b-tree must not contain duplicates.
** An epheremal table will be created unless the selected columns are guaranteed
** to be unique - either because it is an INTEGER PRIMARY KEY or due to
** a UNIQUE constraint or index.
**
** When IN_INDEX_MEMBERSHIP is used (and the b-tree will be used 
** for fast set membership tests) then an epheremal table must 
** be used unless <columns> is a single INTEGER PRIMARY KEY column or an 
** index can be found with the specified <columns> as its left-most.
**
** If the IN_INDEX_NOOP_OK and IN_INDEX_MEMBERSHIP are both set and
** if the RHS of the IN operator is a list (not a subquery) then this
** routine might decide that creating an ephemeral b-tree for membership
** testing is too expensive and return IN_INDEX_NOOP.  In that case, the
................................................................................
** in the RHS of an IN operator.  The IN operator can be in either of two
** forms:
**
**     x IN (4,5,11)              -- IN operator with list on right-hand side
**     x IN (SELECT a FROM b)     -- IN operator with subquery on the right
**
** The pExpr parameter is the IN operator.  The cursor number for the
** constructed ephermeral table is returned.  The first time the ephemeral
** table is computed, the cursor number is also stored in pExpr->iTable,
** however the cursor number returned might not be the same, as it might
** have been duplicated using OP_OpenDup.
**
** If the LHS expression ("x" in the examples) is a column value, or
** the SELECT statement returns a column value, then the affinity of that
** column is used to build the index keys. If both 'x' and the
................................................................................
  p->op = TK_REGISTER;
  p->iTable = iReg;
  ExprClearProperty(p, EP_Skip);
}

/*
** Evaluate an expression (either a vector or a scalar expression) and store
** the result in continguous temporary registers.  Return the index of
** the first register used to store the result.
**
** If the returned result register is a temporary scalar, then also write
** that register number into *piFreeable.  If the returned result register
** is not a temporary or if the expression is a vector set *piFreeable
** to 0.
*/
................................................................................
    }
    case TK_COLUMN: {
      int iTab = pExpr->iTable;
      int iReg;
      if( ExprHasProperty(pExpr, EP_FixedCol) ){
        /* This COLUMN expression is really a constant due to WHERE clause
        ** constraints, and that constant is coded by the pExpr->pLeft
        ** expresssion.  However, make sure the constant has the correct
        ** datatype by applying the Affinity of the table column to the
        ** constant.
        */
        int aff;
        iReg = sqlite3ExprCodeTarget(pParse, pExpr->pLeft,target);
        if( pExpr->y.pTab ){
          aff = sqlite3TableColumnAffinity(pExpr->y.pTab, pExpr->iColumn);
................................................................................
**
**     pE1: x==5       pE2: x==5             Result: true
**     pE1: x>0        pE2: x==5             Result: false
**     pE1: x=21       pE2: x=21 OR y=43     Result: true
**     pE1: x!=123     pE2: x IS NOT NULL    Result: true
**     pE1: x!=?1      pE2: x IS NOT NULL    Result: true
**     pE1: x IS NULL  pE2: x IS NOT NULL    Result: false
**     pE1: x IS ?2    pE2: x IS NOT NULL    Reuslt: false
**
** When comparing TK_COLUMN nodes between pE1 and pE2, if pE2 has
** Expr.iTable<0 then assume a table number given by iTab.
**
** If pParse is not NULL, then the values of bound variables in pE1 are 
** compared against literal values in pE2 and pParse->pVdbe->expmask is
** modified to record which bound variables are referenced.  If pParse 







|







 







|







 







|




|







 







|







 







|







 







|







 







|







216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
....
2451
2452
2453
2454
2455
2456
2457
2458
2459
2460
2461
2462
2463
2464
2465
....
2473
2474
2475
2476
2477
2478
2479
2480
2481
2482
2483
2484
2485
2486
2487
2488
2489
2490
2491
2492
....
2809
2810
2811
2812
2813
2814
2815
2816
2817
2818
2819
2820
2821
2822
2823
....
3624
3625
3626
3627
3628
3629
3630
3631
3632
3633
3634
3635
3636
3637
3638
....
3839
3840
3841
3842
3843
3844
3845
3846
3847
3848
3849
3850
3851
3852
3853
....
5423
5424
5425
5426
5427
5428
5429
5430
5431
5432
5433
5434
5435
5436
5437
  }
  return pColl;
}

/*
** Return the collation sequence for the expression pExpr. If
** there is no defined collating sequence, return a pointer to the
** default collation sequence.
**
** See also: sqlite3ExprCollSeq()
**
** The sqlite3ExprCollSeq() routine works the same except that it
** returns NULL if there is no defined collation.
*/
CollSeq *sqlite3ExprNNCollSeq(Parse *pParse, const Expr *pExpr){
................................................................................
**
** The returned value of this function indicates the b-tree type, as follows:
**
**   IN_INDEX_ROWID      - The cursor was opened on a database table.
**   IN_INDEX_INDEX_ASC  - The cursor was opened on an ascending index.
**   IN_INDEX_INDEX_DESC - The cursor was opened on a descending index.
**   IN_INDEX_EPH        - The cursor was opened on a specially created and
**                         populated ephemeral table.
**   IN_INDEX_NOOP       - No cursor was allocated.  The IN operator must be
**                         implemented as a sequence of comparisons.
**
** An existing b-tree might be used if the RHS expression pX is a simple
** subquery such as:
**
**     SELECT <column1>, <column2>... FROM <table>
................................................................................
** IN_INDEX_MEMBERSHIP or IN_INDEX_LOOP but not both.  If inFlags contains
** IN_INDEX_MEMBERSHIP, then the generated table will be used for a fast
** membership test.  When the IN_INDEX_LOOP bit is set, the IN index will
** be used to loop over all values of the RHS of the IN operator.
**
** When IN_INDEX_LOOP is used (and the b-tree will be used to iterate
** through the set members) then the b-tree must not contain duplicates.
** An ephemeral table will be created unless the selected columns are guaranteed
** to be unique - either because it is an INTEGER PRIMARY KEY or due to
** a UNIQUE constraint or index.
**
** When IN_INDEX_MEMBERSHIP is used (and the b-tree will be used 
** for fast set membership tests) then an ephemeral table must 
** be used unless <columns> is a single INTEGER PRIMARY KEY column or an 
** index can be found with the specified <columns> as its left-most.
**
** If the IN_INDEX_NOOP_OK and IN_INDEX_MEMBERSHIP are both set and
** if the RHS of the IN operator is a list (not a subquery) then this
** routine might decide that creating an ephemeral b-tree for membership
** testing is too expensive and return IN_INDEX_NOOP.  In that case, the
................................................................................
** in the RHS of an IN operator.  The IN operator can be in either of two
** forms:
**
**     x IN (4,5,11)              -- IN operator with list on right-hand side
**     x IN (SELECT a FROM b)     -- IN operator with subquery on the right
**
** The pExpr parameter is the IN operator.  The cursor number for the
** constructed ephemeral table is returned.  The first time the ephemeral
** table is computed, the cursor number is also stored in pExpr->iTable,
** however the cursor number returned might not be the same, as it might
** have been duplicated using OP_OpenDup.
**
** If the LHS expression ("x" in the examples) is a column value, or
** the SELECT statement returns a column value, then the affinity of that
** column is used to build the index keys. If both 'x' and the
................................................................................
  p->op = TK_REGISTER;
  p->iTable = iReg;
  ExprClearProperty(p, EP_Skip);
}

/*
** Evaluate an expression (either a vector or a scalar expression) and store
** the result in contiguous temporary registers.  Return the index of
** the first register used to store the result.
**
** If the returned result register is a temporary scalar, then also write
** that register number into *piFreeable.  If the returned result register
** is not a temporary or if the expression is a vector set *piFreeable
** to 0.
*/
................................................................................
    }
    case TK_COLUMN: {
      int iTab = pExpr->iTable;
      int iReg;
      if( ExprHasProperty(pExpr, EP_FixedCol) ){
        /* This COLUMN expression is really a constant due to WHERE clause
        ** constraints, and that constant is coded by the pExpr->pLeft
        ** expression.  However, make sure the constant has the correct
        ** datatype by applying the Affinity of the table column to the
        ** constant.
        */
        int aff;
        iReg = sqlite3ExprCodeTarget(pParse, pExpr->pLeft,target);
        if( pExpr->y.pTab ){
          aff = sqlite3TableColumnAffinity(pExpr->y.pTab, pExpr->iColumn);
................................................................................
**
**     pE1: x==5       pE2: x==5             Result: true
**     pE1: x>0        pE2: x==5             Result: false
**     pE1: x=21       pE2: x=21 OR y=43     Result: true
**     pE1: x!=123     pE2: x IS NOT NULL    Result: true
**     pE1: x!=?1      pE2: x IS NOT NULL    Result: true
**     pE1: x IS NULL  pE2: x IS NOT NULL    Result: false
**     pE1: x IS ?2    pE2: x IS NOT NULL    Result: false
**
** When comparing TK_COLUMN nodes between pE1 and pE2, if pE2 has
** Expr.iTable<0 then assume a table number given by iTab.
**
** If pParse is not NULL, then the values of bound variables in pE1 are 
** compared against literal values in pE2 and pParse->pVdbe->expmask is
** modified to record which bound variables are referenced.  If pParse 

Changes to src/func.c.

609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
  u8 matchOne;          /* "?" or "_" */
  u8 matchSet;          /* "[" or 0 */
  u8 noCase;            /* true to ignore case differences */
};

/*
** For LIKE and GLOB matching on EBCDIC machines, assume that every
** character is exactly one byte in size.  Also, provde the Utf8Read()
** macro for fast reading of the next character in the common case where
** the next character is ASCII.
*/
#if defined(SQLITE_EBCDIC)
# define sqlite3Utf8Read(A)        (*((*A)++))
# define Utf8Read(A)               (*(A++))
#else







|







609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
  u8 matchOne;          /* "?" or "_" */
  u8 matchSet;          /* "[" or 0 */
  u8 noCase;            /* true to ignore case differences */
};

/*
** For LIKE and GLOB matching on EBCDIC machines, assume that every
** character is exactly one byte in size.  Also, provide the Utf8Read()
** macro for fast reading of the next character in the common case where
** the next character is ASCII.
*/
#if defined(SQLITE_EBCDIC)
# define sqlite3Utf8Read(A)        (*((*A)++))
# define Utf8Read(A)               (*(A++))
#else

Changes to src/insert.c.

1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
....
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
....
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
....
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
....
2714
2715
2716
2717
2718
2719
2720
2721
2722
2723
2724
2725
2726
2727
2728
*/
#define CKCNSTRNT_COLUMN   0x01    /* CHECK constraint uses a changing column */
#define CKCNSTRNT_ROWID    0x02    /* CHECK constraint references the ROWID */

/* This is the Walker callback from sqlite3ExprReferencesUpdatedColumn().
*  Set bit 0x01 of pWalker->eCode if pWalker->eCode to 0 and if this
** expression node references any of the
** columns that are being modifed by an UPDATE statement.
*/
static int checkConstraintExprNode(Walker *pWalker, Expr *pExpr){
  if( pExpr->op==TK_COLUMN ){
    assert( pExpr->iColumn>=0 || pExpr->iColumn==-1 );
    if( pExpr->iColumn>=0 ){
      if( pWalker->u.aiCol[pExpr->iColumn]>=0 ){
        pWalker->eCode |= CKCNSTRNT_COLUMN;
................................................................................
  u8 pkChng,           /* Non-zero if the rowid or PRIMARY KEY changed */
  u8 overrideError,    /* Override onError to this if not OE_Default */
  int ignoreDest,      /* Jump to this label on an OE_Ignore resolution */
  int *pbMayReplace,   /* OUT: Set to true if constraint may cause a replace */
  int *aiChng,         /* column i is unchanged if aiChng[i]<0 */
  Upsert *pUpsert      /* ON CONFLICT clauses, if any.  NULL otherwise */
){
  Vdbe *v;             /* VDBE under constrution */
  Index *pIdx;         /* Pointer to one of the indices */
  Index *pPk = 0;      /* The PRIMARY KEY index */
  sqlite3 *db;         /* Database connection */
  int i;               /* loop counter */
  int ix;              /* Index loop counter */
  int nCol;            /* Number of columns */
  int onError;         /* Conflict resolution strategy */
................................................................................

    /* If the response to a rowid conflict is REPLACE but the response
    ** to some other UNIQUE constraint is FAIL or IGNORE, then we need
    ** to defer the running of the rowid conflict checking until after
    ** the UNIQUE constraints have run.
    */
    if( onError==OE_Replace      /* IPK rule is REPLACE */
     && onError!=overrideError   /* Rules for other contraints are different */
     && pTab->pIndex             /* There exist other constraints */
    ){
      ipkTop = sqlite3VdbeAddOp0(v, OP_Goto)+1;
      VdbeComment((v, "defer IPK REPLACE until last"));
    }

    if( isUpdate ){
................................................................................
  ** index and making sure that duplicate entries do not already exist.
  ** Compute the revised record entries for indices as we go.
  **
  ** This loop also handles the case of the PRIMARY KEY index for a
  ** WITHOUT ROWID table.
  */
  for(ix=0, pIdx=pTab->pIndex; pIdx; pIdx=pIdx->pNext, ix++){
    int regIdx;          /* Range of registers hold conent for pIdx */
    int regR;            /* Range of registers holding conflicting PK */
    int iThisCur;        /* Cursor for this UNIQUE index */
    int addrUniqueOk;    /* Jump here if the UNIQUE constraint is satisfied */
    int addrConflictCk;  /* First opcode in the conflict check logic */

    if( aRegIdx[ix]==0 ) continue;  /* Skip indices that do not change */
    if( pUpIdx==pIdx ){
................................................................................
  }
#ifndef SQLITE_OMIT_CHECK
  if( pDest->pCheck && sqlite3ExprListCompare(pSrc->pCheck,pDest->pCheck,-1) ){
    return 0;   /* Tables have different CHECK constraints.  Ticket #2252 */
  }
#endif
#ifndef SQLITE_OMIT_FOREIGN_KEY
  /* Disallow the transfer optimization if the destination table constains
  ** any foreign key constraints.  This is more restrictive than necessary.
  ** But the main beneficiary of the transfer optimization is the VACUUM 
  ** command, and the VACUUM command disables foreign key constraints.  So
  ** the extra complication to make this rule less restrictive is probably
  ** not worth the effort.  Ticket [6284df89debdfa61db8073e062908af0c9b6118e]
  */
  if( (db->flags & SQLITE_ForeignKeys)!=0 && pDest->pFKey!=0 ){







|







 







|







 







|







 







|







 







|







1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
....
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
....
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
....
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
....
2714
2715
2716
2717
2718
2719
2720
2721
2722
2723
2724
2725
2726
2727
2728
*/
#define CKCNSTRNT_COLUMN   0x01    /* CHECK constraint uses a changing column */
#define CKCNSTRNT_ROWID    0x02    /* CHECK constraint references the ROWID */

/* This is the Walker callback from sqlite3ExprReferencesUpdatedColumn().
*  Set bit 0x01 of pWalker->eCode if pWalker->eCode to 0 and if this
** expression node references any of the
** columns that are being modified by an UPDATE statement.
*/
static int checkConstraintExprNode(Walker *pWalker, Expr *pExpr){
  if( pExpr->op==TK_COLUMN ){
    assert( pExpr->iColumn>=0 || pExpr->iColumn==-1 );
    if( pExpr->iColumn>=0 ){
      if( pWalker->u.aiCol[pExpr->iColumn]>=0 ){
        pWalker->eCode |= CKCNSTRNT_COLUMN;
................................................................................
  u8 pkChng,           /* Non-zero if the rowid or PRIMARY KEY changed */
  u8 overrideError,    /* Override onError to this if not OE_Default */
  int ignoreDest,      /* Jump to this label on an OE_Ignore resolution */
  int *pbMayReplace,   /* OUT: Set to true if constraint may cause a replace */
  int *aiChng,         /* column i is unchanged if aiChng[i]<0 */
  Upsert *pUpsert      /* ON CONFLICT clauses, if any.  NULL otherwise */
){
  Vdbe *v;             /* VDBE under construction */
  Index *pIdx;         /* Pointer to one of the indices */
  Index *pPk = 0;      /* The PRIMARY KEY index */
  sqlite3 *db;         /* Database connection */
  int i;               /* loop counter */
  int ix;              /* Index loop counter */
  int nCol;            /* Number of columns */
  int onError;         /* Conflict resolution strategy */
................................................................................

    /* If the response to a rowid conflict is REPLACE but the response
    ** to some other UNIQUE constraint is FAIL or IGNORE, then we need
    ** to defer the running of the rowid conflict checking until after
    ** the UNIQUE constraints have run.
    */
    if( onError==OE_Replace      /* IPK rule is REPLACE */
     && onError!=overrideError   /* Rules for other constraints are different */
     && pTab->pIndex             /* There exist other constraints */
    ){
      ipkTop = sqlite3VdbeAddOp0(v, OP_Goto)+1;
      VdbeComment((v, "defer IPK REPLACE until last"));
    }

    if( isUpdate ){
................................................................................
  ** index and making sure that duplicate entries do not already exist.
  ** Compute the revised record entries for indices as we go.
  **
  ** This loop also handles the case of the PRIMARY KEY index for a
  ** WITHOUT ROWID table.
  */
  for(ix=0, pIdx=pTab->pIndex; pIdx; pIdx=pIdx->pNext, ix++){
    int regIdx;          /* Range of registers hold content for pIdx */
    int regR;            /* Range of registers holding conflicting PK */
    int iThisCur;        /* Cursor for this UNIQUE index */
    int addrUniqueOk;    /* Jump here if the UNIQUE constraint is satisfied */
    int addrConflictCk;  /* First opcode in the conflict check logic */

    if( aRegIdx[ix]==0 ) continue;  /* Skip indices that do not change */
    if( pUpIdx==pIdx ){
................................................................................
  }
#ifndef SQLITE_OMIT_CHECK
  if( pDest->pCheck && sqlite3ExprListCompare(pSrc->pCheck,pDest->pCheck,-1) ){
    return 0;   /* Tables have different CHECK constraints.  Ticket #2252 */
  }
#endif
#ifndef SQLITE_OMIT_FOREIGN_KEY
  /* Disallow the transfer optimization if the destination table contains
  ** any foreign key constraints.  This is more restrictive than necessary.
  ** But the main beneficiary of the transfer optimization is the VACUUM 
  ** command, and the VACUUM command disables foreign key constraints.  So
  ** the extra complication to make this rule less restrictive is probably
  ** not worth the effort.  Ticket [6284df89debdfa61db8073e062908af0c9b6118e]
  */
  if( (db->flags & SQLITE_ForeignKeys)!=0 && pDest->pFKey!=0 ){

Changes to src/main.c.

3167
3168
3169
3170
3171
3172
3173
3174
3175
3176
3177
3178
3179
3180
3181
....
3680
3681
3682
3683
3684
3685
3686
3687
3688
3689
3690
3691
3692
3693
3694
....
4445
4446
4447
4448
4449
4450
4451
4452
4453
4454
4455
4456
4457
4458
4459
**     undefined               on                          on   
**         3                   on                          on
**         2                   on                         off
**         1                  off                          on
**         0                  off                         off
**
** Legacy behavior is 3 (double-quoted string literals are allowed anywhere)
** and so that is the default.  But developers are encouranged to use
** -DSQLITE_DQS=0 (best) or -DSQLITE_DQS=1 (second choice) if possible.
*/
#if !defined(SQLITE_DQS)
# define SQLITE_DQS 3
#endif
#if (SQLITE_DQS&1)==1
                 | SQLITE_DqsDML
................................................................................
  if( !pTab || pTab->pSelect ){
    pTab = 0;
    goto error_out;
  }

  /* Find the column for which info is requested */
  if( zColumnName==0 ){
    /* Query for existance of table only */
  }else{
    for(iCol=0; iCol<pTab->nCol; iCol++){
      pCol = &pTab->aCol[iCol];
      if( 0==sqlite3StrICmp(pCol->zName, zColumnName) ){
        break;
      }
    }
................................................................................

  sqlite3_mutex_leave(db->mutex);
#endif   /* SQLITE_OMIT_WAL */
  return rc;
}

/*
** Open a read-transaction on the snapshot idendified by pSnapshot.
*/
int sqlite3_snapshot_open(
  sqlite3 *db, 
  const char *zDb, 
  sqlite3_snapshot *pSnapshot
){
  int rc = SQLITE_ERROR;







|







 







|







 







|







3167
3168
3169
3170
3171
3172
3173
3174
3175
3176
3177
3178
3179
3180
3181
....
3680
3681
3682
3683
3684
3685
3686
3687
3688
3689
3690
3691
3692
3693
3694
....
4445
4446
4447
4448
4449
4450
4451
4452
4453
4454
4455
4456
4457
4458
4459
**     undefined               on                          on   
**         3                   on                          on
**         2                   on                         off
**         1                  off                          on
**         0                  off                         off
**
** Legacy behavior is 3 (double-quoted string literals are allowed anywhere)
** and so that is the default.  But developers are encouraged to use
** -DSQLITE_DQS=0 (best) or -DSQLITE_DQS=1 (second choice) if possible.
*/
#if !defined(SQLITE_DQS)
# define SQLITE_DQS 3
#endif
#if (SQLITE_DQS&1)==1
                 | SQLITE_DqsDML
................................................................................
  if( !pTab || pTab->pSelect ){
    pTab = 0;
    goto error_out;
  }

  /* Find the column for which info is requested */
  if( zColumnName==0 ){
    /* Query for existence of table only */
  }else{
    for(iCol=0; iCol<pTab->nCol; iCol++){
      pCol = &pTab->aCol[iCol];
      if( 0==sqlite3StrICmp(pCol->zName, zColumnName) ){
        break;
      }
    }
................................................................................

  sqlite3_mutex_leave(db->mutex);
#endif   /* SQLITE_OMIT_WAL */
  return rc;
}

/*
** Open a read-transaction on the snapshot identified by pSnapshot.
*/
int sqlite3_snapshot_open(
  sqlite3 *db, 
  const char *zDb, 
  sqlite3_snapshot *pSnapshot
){
  int rc = SQLITE_ERROR;

Changes to src/mem1.c.

152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
...
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
}

/*
** Like free() but works for allocations obtained from sqlite3MemMalloc()
** or sqlite3MemRealloc().
**
** For this low-level routine, we already know that pPrior!=0 since
** cases where pPrior==0 will have been intecepted and dealt with
** by higher-level routines.
*/
static void sqlite3MemFree(void *pPrior){
#ifdef SQLITE_MALLOCSIZE
  SQLITE_FREE(pPrior);
#else
  sqlite3_int64 *p = (sqlite3_int64*)pPrior;
................................................................................
#if defined(__APPLE__) && !defined(SQLITE_WITHOUT_ZONEMALLOC)
  int cpuCount;
  size_t len;
  if( _sqliteZone_ ){
    return SQLITE_OK;
  }
  len = sizeof(cpuCount);
  /* One usually wants to use hw.acctivecpu for MT decisions, but not here */
  sysctlbyname("hw.ncpu", &cpuCount, &len, NULL, 0);
  if( cpuCount>1 ){
    /* defer MT decisions to system malloc */
    _sqliteZone_ = malloc_default_zone();
  }else{
    /* only 1 core, use our own zone to contention over global locks, 
    ** e.g. we have our own dedicated locks */







|







 







|







152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
...
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
}

/*
** Like free() but works for allocations obtained from sqlite3MemMalloc()
** or sqlite3MemRealloc().
**
** For this low-level routine, we already know that pPrior!=0 since
** cases where pPrior==0 will have been intercepted and dealt with
** by higher-level routines.
*/
static void sqlite3MemFree(void *pPrior){
#ifdef SQLITE_MALLOCSIZE
  SQLITE_FREE(pPrior);
#else
  sqlite3_int64 *p = (sqlite3_int64*)pPrior;
................................................................................
#if defined(__APPLE__) && !defined(SQLITE_WITHOUT_ZONEMALLOC)
  int cpuCount;
  size_t len;
  if( _sqliteZone_ ){
    return SQLITE_OK;
  }
  len = sizeof(cpuCount);
  /* One usually wants to use hw.activecpu for MT decisions, but not here */
  sysctlbyname("hw.ncpu", &cpuCount, &len, NULL, 0);
  if( cpuCount>1 ){
    /* defer MT decisions to system malloc */
    _sqliteZone_ = malloc_default_zone();
  }else{
    /* only 1 core, use our own zone to contention over global locks, 
    ** e.g. we have our own dedicated locks */

Changes to src/mutex_unix.c.

22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
*/
#ifdef SQLITE_MUTEX_PTHREADS

#include <pthread.h>

/*
** The sqlite3_mutex.id, sqlite3_mutex.nRef, and sqlite3_mutex.owner fields
** are necessary under two condidtions:  (1) Debug builds and (2) using
** home-grown mutexes.  Encapsulate these conditions into a single #define.
*/
#if defined(SQLITE_DEBUG) || defined(SQLITE_HOMEGROWN_RECURSIVE_MUTEX)
# define SQLITE_MUTEX_NREF 1
#else
# define SQLITE_MUTEX_NREF 0
#endif







|







22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
*/
#ifdef SQLITE_MUTEX_PTHREADS

#include <pthread.h>

/*
** The sqlite3_mutex.id, sqlite3_mutex.nRef, and sqlite3_mutex.owner fields
** are necessary under two conditions:  (1) Debug builds and (2) using
** home-grown mutexes.  Encapsulate these conditions into a single #define.
*/
#if defined(SQLITE_DEBUG) || defined(SQLITE_HOMEGROWN_RECURSIVE_MUTEX)
# define SQLITE_MUTEX_NREF 1
#else
# define SQLITE_MUTEX_NREF 0
#endif

Changes to src/mutex_w32.c.

34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
/*
** Each recursive mutex is an instance of the following structure.
*/
struct sqlite3_mutex {
  CRITICAL_SECTION mutex;    /* Mutex controlling the lock */
  int id;                    /* Mutex type */
#ifdef SQLITE_DEBUG
  volatile int nRef;         /* Number of enterances */
  volatile DWORD owner;      /* Thread holding this mutex */
  volatile LONG trace;       /* True to trace changes */
#endif
};

/*
** These are the initializer values used when declaring a "static" mutex







|







34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
/*
** Each recursive mutex is an instance of the following structure.
*/
struct sqlite3_mutex {
  CRITICAL_SECTION mutex;    /* Mutex controlling the lock */
  int id;                    /* Mutex type */
#ifdef SQLITE_DEBUG
  volatile int nRef;         /* Number of recursive entrances */
  volatile DWORD owner;      /* Thread holding this mutex */
  volatile LONG trace;       /* True to trace changes */
#endif
};

/*
** These are the initializer values used when declaring a "static" mutex

Changes to src/os.c.

129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
#ifdef SQLITE_TEST
  if( op!=SQLITE_FCNTL_COMMIT_PHASETWO
   && op!=SQLITE_FCNTL_LOCK_TIMEOUT
  ){
    /* Faults are not injected into COMMIT_PHASETWO because, assuming SQLite
    ** is using a regular VFS, it is called after the corresponding
    ** transaction has been committed. Injecting a fault at this point
    ** confuses the test scripts - the COMMIT comand returns SQLITE_NOMEM
    ** but the transaction is committed anyway.
    **
    ** The core must call OsFileControl() though, not OsFileControlHint(),
    ** as if a custom VFS (e.g. zipvfs) returns an error here, it probably
    ** means the commit really has failed and an error should be returned
    ** to the user.  */
    DO_OS_MALLOC_TEST(id);







|







129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
#ifdef SQLITE_TEST
  if( op!=SQLITE_FCNTL_COMMIT_PHASETWO
   && op!=SQLITE_FCNTL_LOCK_TIMEOUT
  ){
    /* Faults are not injected into COMMIT_PHASETWO because, assuming SQLite
    ** is using a regular VFS, it is called after the corresponding
    ** transaction has been committed. Injecting a fault at this point
    ** confuses the test scripts - the COMMIT command returns SQLITE_NOMEM
    ** but the transaction is committed anyway.
    **
    ** The core must call OsFileControl() though, not OsFileControlHint(),
    ** as if a custom VFS (e.g. zipvfs) returns an error here, it probably
    ** means the commit really has failed and an error should be returned
    ** to the user.  */
    DO_OS_MALLOC_TEST(id);

Changes to src/os.h.

123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
** byte ranges are used for Unix.  This leaves open the possibility of having
** clients on win95, winNT, and unix all talking to the same shared file
** and all locking correctly.  To do so would require that samba (or whatever
** tool is being used for file sharing) implements locks correctly between
** windows and unix.  I'm guessing that isn't likely to happen, but by
** using the same locking range we are at least open to the possibility.
**
** Locking in windows is manditory.  For this reason, we cannot store
** actual data in the bytes used for locking.  The pager never allocates
** the pages involved in locking therefore.  SHARED_SIZE is selected so
** that all locks will fit on a single page even at the minimum page size.
** PENDING_BYTE defines the beginning of the locks.  By default PENDING_BYTE
** is set high so that we don't have to allocate an unused page except
** for very large databases.  But one should test the page skipping logic 
** by setting PENDING_BYTE low and running the entire regression suite.







|







123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
** byte ranges are used for Unix.  This leaves open the possibility of having
** clients on win95, winNT, and unix all talking to the same shared file
** and all locking correctly.  To do so would require that samba (or whatever
** tool is being used for file sharing) implements locks correctly between
** windows and unix.  I'm guessing that isn't likely to happen, but by
** using the same locking range we are at least open to the possibility.
**
** Locking in windows is mandatory.  For this reason, we cannot store
** actual data in the bytes used for locking.  The pager never allocates
** the pages involved in locking therefore.  SHARED_SIZE is selected so
** that all locks will fit on a single page even at the minimum page size.
** PENDING_BYTE defines the beginning of the locks.  By default PENDING_BYTE
** is set high so that we don't have to allocate an unused page except
** for very large databases.  But one should test the page skipping logic 
** by setting PENDING_BYTE low and running the entire regression suite.

Changes to src/os_unix.c.

1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
....
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
....
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
....
2929
2930
2931
2932
2933
2934
2935
2936
2937
2938
2939
2940
2941
2942
2943
....
3045
3046
3047
3048
3049
3050
3051
3052
3053
3054
3055
3056
3057
3058
3059
....
3346
3347
3348
3349
3350
3351
3352
3353
3354
3355
3356
3357
3358
3359
3360
....
3481
3482
3483
3484
3485
3486
3487
3488
3489
3490
3491
3492
3493
3494
3495
....
3603
3604
3605
3606
3607
3608
3609
3610
3611
3612
3613
3614
3615
3616
3617
....
7312
7313
7314
7315
7316
7317
7318
7319
7320
7321
7322
7323
7324
7325
7326
**
** Yet another problem:  LinuxThreads do not play well with posix locks.
**
** Many older versions of linux use the LinuxThreads library which is
** not posix compliant.  Under LinuxThreads, a lock created by thread
** A cannot be modified or overridden by a different thread B.
** Only thread A can modify the lock.  Locking behavior is correct
** if the appliation uses the newer Native Posix Thread Library (NPTL)
** on linux - with NPTL a lock created by thread A can override locks
** in thread B.  But there is no way to know at compile-time which
** threading library is being used.  So there is no way to know at
** compile-time whether or not thread A can override locks on thread B.
** One has to do a run-time check to discover the behavior of the
** current process.
**
................................................................................
  /* The following describes the implementation of the various locks and
  ** lock transitions in terms of the POSIX advisory shared and exclusive
  ** lock primitives (called read-locks and write-locks below, to avoid
  ** confusion with SQLite lock names). The algorithms are complicated
  ** slightly in order to be compatible with Windows95 systems simultaneously
  ** accessing the same database file, in case that is ever required.
  **
  ** Symbols defined in os.h indentify the 'pending byte' and the 'reserved
  ** byte', each single bytes at well known offsets, and the 'shared byte
  ** range', a range of 510 bytes at a well known offset.
  **
  ** To obtain a SHARED lock, a read-lock is obtained on the 'pending
  ** byte'.  If this is successful, 'shared byte range' is read-locked
  ** and the lock on the 'pending byte' released.  (Legacy note:  When
  ** SQLite was first developed, Windows95 systems were still very common,
  ** and Widnows95 lacks a shared-lock capability.  So on Windows95, a
  ** single randomly selected by from the 'shared byte range' is locked.
  ** Windows95 is now pretty much extinct, but this work-around for the
  ** lack of shared-locks on Windows95 lives on, for backwards
  ** compatibility.)
  **
  ** A process may only obtain a RESERVED lock after it has a SHARED lock.
  ** A RESERVED lock is implemented by grabbing a write-lock on the
................................................................................
    OSTRACE(("LOCK    %d %s ok (already held) (unix)\n", pFile->h,
            azFileLock(eFileLock)));
    return SQLITE_OK;
  }

  /* Make sure the locking sequence is correct.
  **  (1) We never move from unlocked to anything higher than shared lock.
  **  (2) SQLite never explicitly requests a pendig lock.
  **  (3) A shared lock is always held when a reserve lock is requested.
  */
  assert( pFile->eFileLock!=NO_LOCK || eFileLock==SHARED_LOCK );
  assert( eFileLock!=PENDING_LOCK );
  assert( eFileLock!=RESERVED_LOCK || pFile->eFileLock==SHARED_LOCK );

  /* This mutex is needed because pFile->pInode is shared across threads
................................................................................
    OSTRACE(("LOCK    %d %s ok (already held) (afp)\n", pFile->h,
           azFileLock(eFileLock)));
    return SQLITE_OK;
  }

  /* Make sure the locking sequence is correct
  **  (1) We never move from unlocked to anything higher than shared lock.
  **  (2) SQLite never explicitly requests a pendig lock.
  **  (3) A shared lock is always held when a reserve lock is requested.
  */
  assert( pFile->eFileLock!=NO_LOCK || eFileLock==SHARED_LOCK );
  assert( eFileLock!=PENDING_LOCK );
  assert( eFileLock!=RESERVED_LOCK || pFile->eFileLock==SHARED_LOCK );
  
  /* This mutex is needed because pFile->pInode is shared across threads
................................................................................
        
      /* Remove the shared lock before trying the range.  we'll need to 
      ** reestablish the shared lock if we can't get the  afpUnlock
      */
      if( !(failed = afpSetLock(context->dbPath, pFile, SHARED_FIRST +
                         pInode->sharedByte, 1, 0)) ){
        int failed2 = SQLITE_OK;
        /* now attemmpt to get the exclusive lock range */
        failed = afpSetLock(context->dbPath, pFile, SHARED_FIRST, 
                               SHARED_SIZE, 1);
        if( failed && (failed2 = afpSetLock(context->dbPath, pFile, 
                       SHARED_FIRST + pInode->sharedByte, 1, 1)) ){
          /* Can't reestablish the shared lock.  Sqlite can't deal, this is
          ** a critical I/O error
          */
................................................................................
  assert( pFile->pPreallocatedUnused==0
       || offset>=PENDING_BYTE+512
       || offset+amt<=PENDING_BYTE 
  );
#endif

#if SQLITE_MAX_MMAP_SIZE>0
  /* Deal with as much of this read request as possible by transfering
  ** data from the memory mapping using memcpy().  */
  if( offset<pFile->mmapSize ){
    if( offset+amt <= pFile->mmapSize ){
      memcpy(pBuf, &((u8 *)(pFile->pMapRegion))[offset], amt);
      return SQLITE_OK;
    }else{
      int nCopy = pFile->mmapSize - offset;
................................................................................
        pFile->transCntrChng = 1;  /* The transaction counter has changed */
      }
    }
  }
#endif

#if defined(SQLITE_MMAP_READWRITE) && SQLITE_MAX_MMAP_SIZE>0
  /* Deal with as much of this write request as possible by transfering
  ** data from the memory mapping using memcpy().  */
  if( offset<pFile->mmapSize ){
    if( offset+amt <= pFile->mmapSize ){
      memcpy(&((u8 *)(pFile->pMapRegion))[offset], pBuf, amt);
      return SQLITE_OK;
    }else{
      int nCopy = pFile->mmapSize - offset;
................................................................................
  if( fullSync ) sqlite3_fullsync_count++;
  sqlite3_sync_count++;
#endif

  /* If we compiled with the SQLITE_NO_SYNC flag, then syncing is a
  ** no-op.  But go ahead and call fstat() to validate the file
  ** descriptor as we need a method to provoke a failure during
  ** coverate testing.
  */
#ifdef SQLITE_NO_SYNC
  {
    struct stat buf;
    rc = osFstat(fd, &buf);
  }
#elif HAVE_FULLFSYNC
................................................................................
              rc = osFchmod(conchFile->h, cmode);
            }while( rc==(-1) && errno==EINTR );
            if( rc!=0 ){
              int code = errno;
              fprintf(stderr, "fchmod %o FAILED with %d %s\n",
                      cmode, code, strerror(code));
            } else {
              fprintf(stderr, "fchmod %o SUCCEDED\n",cmode);
            }
          }else{
            int code = errno;
            fprintf(stderr, "STAT FAILED[%d] with %d %s\n", 
                    err, code, strerror(code));
#endif
          }







|







 







|







|







 







|







 







|







 







|







 







|







 







|







 







|







 







|







1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
....
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
....
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
....
2929
2930
2931
2932
2933
2934
2935
2936
2937
2938
2939
2940
2941
2942
2943
....
3045
3046
3047
3048
3049
3050
3051
3052
3053
3054
3055
3056
3057
3058
3059
....
3346
3347
3348
3349
3350
3351
3352
3353
3354
3355
3356
3357
3358
3359
3360
....
3481
3482
3483
3484
3485
3486
3487
3488
3489
3490
3491
3492
3493
3494
3495
....
3603
3604
3605
3606
3607
3608
3609
3610
3611
3612
3613
3614
3615
3616
3617
....
7312
7313
7314
7315
7316
7317
7318
7319
7320
7321
7322
7323
7324
7325
7326
**
** Yet another problem:  LinuxThreads do not play well with posix locks.
**
** Many older versions of linux use the LinuxThreads library which is
** not posix compliant.  Under LinuxThreads, a lock created by thread
** A cannot be modified or overridden by a different thread B.
** Only thread A can modify the lock.  Locking behavior is correct
** if the application uses the newer Native Posix Thread Library (NPTL)
** on linux - with NPTL a lock created by thread A can override locks
** in thread B.  But there is no way to know at compile-time which
** threading library is being used.  So there is no way to know at
** compile-time whether or not thread A can override locks on thread B.
** One has to do a run-time check to discover the behavior of the
** current process.
**
................................................................................
  /* The following describes the implementation of the various locks and
  ** lock transitions in terms of the POSIX advisory shared and exclusive
  ** lock primitives (called read-locks and write-locks below, to avoid
  ** confusion with SQLite lock names). The algorithms are complicated
  ** slightly in order to be compatible with Windows95 systems simultaneously
  ** accessing the same database file, in case that is ever required.
  **
  ** Symbols defined in os.h identify the 'pending byte' and the 'reserved
  ** byte', each single bytes at well known offsets, and the 'shared byte
  ** range', a range of 510 bytes at a well known offset.
  **
  ** To obtain a SHARED lock, a read-lock is obtained on the 'pending
  ** byte'.  If this is successful, 'shared byte range' is read-locked
  ** and the lock on the 'pending byte' released.  (Legacy note:  When
  ** SQLite was first developed, Windows95 systems were still very common,
  ** and Windows95 lacks a shared-lock capability.  So on Windows95, a
  ** single randomly selected by from the 'shared byte range' is locked.
  ** Windows95 is now pretty much extinct, but this work-around for the
  ** lack of shared-locks on Windows95 lives on, for backwards
  ** compatibility.)
  **
  ** A process may only obtain a RESERVED lock after it has a SHARED lock.
  ** A RESERVED lock is implemented by grabbing a write-lock on the
................................................................................
    OSTRACE(("LOCK    %d %s ok (already held) (unix)\n", pFile->h,
            azFileLock(eFileLock)));
    return SQLITE_OK;
  }

  /* Make sure the locking sequence is correct.
  **  (1) We never move from unlocked to anything higher than shared lock.
  **  (2) SQLite never explicitly requests a pending lock.
  **  (3) A shared lock is always held when a reserve lock is requested.
  */
  assert( pFile->eFileLock!=NO_LOCK || eFileLock==SHARED_LOCK );
  assert( eFileLock!=PENDING_LOCK );
  assert( eFileLock!=RESERVED_LOCK || pFile->eFileLock==SHARED_LOCK );

  /* This mutex is needed because pFile->pInode is shared across threads
................................................................................
    OSTRACE(("LOCK    %d %s ok (already held) (afp)\n", pFile->h,
           azFileLock(eFileLock)));
    return SQLITE_OK;
  }

  /* Make sure the locking sequence is correct
  **  (1) We never move from unlocked to anything higher than shared lock.
  **  (2) SQLite never explicitly requests a pending lock.
  **  (3) A shared lock is always held when a reserve lock is requested.
  */
  assert( pFile->eFileLock!=NO_LOCK || eFileLock==SHARED_LOCK );
  assert( eFileLock!=PENDING_LOCK );
  assert( eFileLock!=RESERVED_LOCK || pFile->eFileLock==SHARED_LOCK );
  
  /* This mutex is needed because pFile->pInode is shared across threads
................................................................................
        
      /* Remove the shared lock before trying the range.  we'll need to 
      ** reestablish the shared lock if we can't get the  afpUnlock
      */
      if( !(failed = afpSetLock(context->dbPath, pFile, SHARED_FIRST +
                         pInode->sharedByte, 1, 0)) ){
        int failed2 = SQLITE_OK;
        /* now attempt to get the exclusive lock range */
        failed = afpSetLock(context->dbPath, pFile, SHARED_FIRST, 
                               SHARED_SIZE, 1);
        if( failed && (failed2 = afpSetLock(context->dbPath, pFile, 
                       SHARED_FIRST + pInode->sharedByte, 1, 1)) ){
          /* Can't reestablish the shared lock.  Sqlite can't deal, this is
          ** a critical I/O error
          */
................................................................................
  assert( pFile->pPreallocatedUnused==0
       || offset>=PENDING_BYTE+512
       || offset+amt<=PENDING_BYTE 
  );
#endif

#if SQLITE_MAX_MMAP_SIZE>0
  /* Deal with as much of this read request as possible by transferring
  ** data from the memory mapping using memcpy().  */
  if( offset<pFile->mmapSize ){
    if( offset+amt <= pFile->mmapSize ){
      memcpy(pBuf, &((u8 *)(pFile->pMapRegion))[offset], amt);
      return SQLITE_OK;
    }else{
      int nCopy = pFile->mmapSize - offset;
................................................................................
        pFile->transCntrChng = 1;  /* The transaction counter has changed */
      }
    }
  }
#endif

#if defined(SQLITE_MMAP_READWRITE) && SQLITE_MAX_MMAP_SIZE>0
  /* Deal with as much of this write request as possible by transferring
  ** data from the memory mapping using memcpy().  */
  if( offset<pFile->mmapSize ){
    if( offset+amt <= pFile->mmapSize ){
      memcpy(&((u8 *)(pFile->pMapRegion))[offset], pBuf, amt);
      return SQLITE_OK;
    }else{
      int nCopy = pFile->mmapSize - offset;
................................................................................
  if( fullSync ) sqlite3_fullsync_count++;
  sqlite3_sync_count++;
#endif

  /* If we compiled with the SQLITE_NO_SYNC flag, then syncing is a
  ** no-op.  But go ahead and call fstat() to validate the file
  ** descriptor as we need a method to provoke a failure during
  ** coverage testing.
  */
#ifdef SQLITE_NO_SYNC
  {
    struct stat buf;
    rc = osFstat(fd, &buf);
  }
#elif HAVE_FULLFSYNC
................................................................................
              rc = osFchmod(conchFile->h, cmode);
            }while( rc==(-1) && errno==EINTR );
            if( rc!=0 ){
              int code = errno;
              fprintf(stderr, "fchmod %o FAILED with %d %s\n",
                      cmode, code, strerror(code));
            } else {
              fprintf(stderr, "fchmod %o SUCCEEDED\n",cmode);
            }
          }else{
            int code = errno;
            fprintf(stderr, "STAT FAILED[%d] with %d %s\n", 
                    err, code, strerror(code));
#endif
          }

Changes to src/os_win.c.

2728
2729
2730
2731
2732
2733
2734
2735
2736
2737
2738
2739
2740
2741
2742
....
2806
2807
2808
2809
2810
2811
2812
2813
2814
2815
2816
2817
2818
2819
2820
....
2916
2917
2918
2919
2920
2921
2922
2923
2924
2925
2926
2927
2928
2929
2930
....
2975
2976
2977
2978
2979
2980
2981
2982
2983
2984
2985
2986
2987
2988
2989
  assert( offset>=0 );
  SimulateIOError(return SQLITE_IOERR_READ);
  OSTRACE(("READ pid=%lu, pFile=%p, file=%p, buffer=%p, amount=%d, "
           "offset=%lld, lock=%d\n", osGetCurrentProcessId(), pFile,
           pFile->h, pBuf, amt, offset, pFile->locktype));

#if SQLITE_MAX_MMAP_SIZE>0
  /* Deal with as much of this read request as possible by transfering
  ** data from the memory mapping using memcpy().  */
  if( offset<pFile->mmapSize ){
    if( offset+amt <= pFile->mmapSize ){
      memcpy(pBuf, &((u8 *)(pFile->pMapRegion))[offset], amt);
      OSTRACE(("READ-MMAP pid=%lu, pFile=%p, file=%p, rc=SQLITE_OK\n",
               osGetCurrentProcessId(), pFile, pFile->h));
      return SQLITE_OK;
................................................................................
  SimulateDiskfullError(return SQLITE_FULL);

  OSTRACE(("WRITE pid=%lu, pFile=%p, file=%p, buffer=%p, amount=%d, "
           "offset=%lld, lock=%d\n", osGetCurrentProcessId(), pFile,
           pFile->h, pBuf, amt, offset, pFile->locktype));

#if defined(SQLITE_MMAP_READWRITE) && SQLITE_MAX_MMAP_SIZE>0
  /* Deal with as much of this write request as possible by transfering
  ** data from the memory mapping using memcpy().  */
  if( offset<pFile->mmapSize ){
    if( offset+amt <= pFile->mmapSize ){
      memcpy(&((u8 *)(pFile->pMapRegion))[offset], pBuf, amt);
      OSTRACE(("WRITE-MMAP pid=%lu, pFile=%p, file=%p, rc=SQLITE_OK\n",
               osGetCurrentProcessId(), pFile, pFile->h));
      return SQLITE_OK;
................................................................................
    ** though some folks might complain that the file is bigger than it
    ** needs to be.
    **
    ** The only feasible work-around is to defer the truncation until after
    ** all references to memory-mapped content are closed.  That is doable,
    ** but involves adding a few branches in the common write code path which
    ** could slow down normal operations slightly.  Hence, we have decided for
    ** now to simply make trancations a no-op if there are pending reads.  We
    ** can maybe revisit this decision in the future.
    */
    return SQLITE_OK;
  }
#endif

  assert( pFile );
................................................................................
           osGetCurrentProcessId(), pFile, pFile->h, sqlite3ErrName(rc)));
  return rc;
}

#ifdef SQLITE_TEST
/*
** Count the number of fullsyncs and normal syncs.  This is used to test
** that syncs and fullsyncs are occuring at the right times.
*/
int sqlite3_sync_count = 0;
int sqlite3_fullsync_count = 0;
#endif

/*
** Make sure all writes to a particular file are committed to disk.







|







 







|







 







|







 







|







2728
2729
2730
2731
2732
2733
2734
2735
2736
2737
2738
2739
2740
2741
2742
....
2806
2807
2808
2809
2810
2811
2812
2813
2814
2815
2816
2817
2818
2819
2820
....
2916
2917
2918
2919
2920
2921
2922
2923
2924
2925
2926
2927
2928
2929
2930
....
2975
2976
2977
2978
2979
2980
2981
2982
2983
2984
2985
2986
2987
2988
2989
  assert( offset>=0 );
  SimulateIOError(return SQLITE_IOERR_READ);
  OSTRACE(("READ pid=%lu, pFile=%p, file=%p, buffer=%p, amount=%d, "
           "offset=%lld, lock=%d\n", osGetCurrentProcessId(), pFile,
           pFile->h, pBuf, amt, offset, pFile->locktype));

#if SQLITE_MAX_MMAP_SIZE>0
  /* Deal with as much of this read request as possible by transferring
  ** data from the memory mapping using memcpy().  */
  if( offset<pFile->mmapSize ){
    if( offset+amt <= pFile->mmapSize ){
      memcpy(pBuf, &((u8 *)(pFile->pMapRegion))[offset], amt);
      OSTRACE(("READ-MMAP pid=%lu, pFile=%p, file=%p, rc=SQLITE_OK\n",
               osGetCurrentProcessId(), pFile, pFile->h));
      return SQLITE_OK;
................................................................................
  SimulateDiskfullError(return SQLITE_FULL);

  OSTRACE(("WRITE pid=%lu, pFile=%p, file=%p, buffer=%p, amount=%d, "
           "offset=%lld, lock=%d\n", osGetCurrentProcessId(), pFile,
           pFile->h, pBuf, amt, offset, pFile->locktype));

#if defined(SQLITE_MMAP_READWRITE) && SQLITE_MAX_MMAP_SIZE>0
  /* Deal with as much of this write request as possible by transferring
  ** data from the memory mapping using memcpy().  */
  if( offset<pFile->mmapSize ){
    if( offset+amt <= pFile->mmapSize ){
      memcpy(&((u8 *)(pFile->pMapRegion))[offset], pBuf, amt);
      OSTRACE(("WRITE-MMAP pid=%lu, pFile=%p, file=%p, rc=SQLITE_OK\n",
               osGetCurrentProcessId(), pFile, pFile->h));
      return SQLITE_OK;
................................................................................
    ** though some folks might complain that the file is bigger than it
    ** needs to be.
    **
    ** The only feasible work-around is to defer the truncation until after
    ** all references to memory-mapped content are closed.  That is doable,
    ** but involves adding a few branches in the common write code path which
    ** could slow down normal operations slightly.  Hence, we have decided for
    ** now to simply make truncations a no-op if there are pending reads.  We
    ** can maybe revisit this decision in the future.
    */
    return SQLITE_OK;
  }
#endif

  assert( pFile );
................................................................................
           osGetCurrentProcessId(), pFile, pFile->h, sqlite3ErrName(rc)));
  return rc;
}

#ifdef SQLITE_TEST
/*
** Count the number of fullsyncs and normal syncs.  This is used to test
** that syncs and fullsyncs are occurring at the right times.
*/
int sqlite3_sync_count = 0;
int sqlite3_fullsync_count = 0;
#endif

/*
** Make sure all writes to a particular file are committed to disk.

Changes to src/pager.c.

30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
..
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
...
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
....
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
....
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
....
3116
3117
3118
3119
3120
3121
3122
3123
3124
3125
3126
3127
3128
3129
3130
....
3499
3500
3501
3502
3503
3504
3505
3506
3507
3508
3509
3510
3511
3512
3513
....
5005
5006
5007
5008
5009
5010
5011
5012
5013
5014
5015
5016
5017
5018
5019
....
7274
7275
7276
7277
7278
7279
7280
7281
7282
7283
7284
7285
7286
7287
7288
** journal_mode=MEMORY, or journal_mode=OFF.
**
** Within this comment block, a page is deemed to have been synced
** automatically as soon as it is written when PRAGMA synchronous=OFF.
** Otherwise, the page is not synced until the xSync method of the VFS
** is called successfully on the file containing the page.
**
** Definition:  A page of the database file is said to be "overwriteable" if
** one or more of the following are true about the page:
** 
**     (a)  The original content of the page as it was at the beginning of
**          the transaction has been written into the rollback journal and
**          synced.
** 
**     (b)  The page was a freelist leaf page at the start of the transaction.
................................................................................
** 
**     (c)  The page number is greater than the largest page that existed in
**          the database file at the start of the transaction.
** 
** (1) A page of the database file is never overwritten unless one of the
**     following are true:
** 
**     (a) The page and all other pages on the same sector are overwriteable.
** 
**     (b) The atomic page write optimization is enabled, and the entire
**         transaction other than the update of the transaction sequence
**         number consists of a single page change.
** 
** (2) The content of a page written into the rollback journal exactly matches
**     both the content in the database when the rollback journal was written
................................................................................
**    instead of READER following such an error.
**
**    Once it has entered the ERROR state, any attempt to use the pager
**    to read or write data returns an error. Eventually, once all 
**    outstanding transactions have been abandoned, the pager is able to
**    transition back to OPEN state, discarding the contents of the 
**    page-cache and any other in-memory state at the same time. Everything
**    is reloaded from disk (and, if necessary, hot-journal rollback peformed)
**    when a read-transaction is next opened on the pager (transitioning
**    the pager into READER state). At that point the system has recovered 
**    from the error.
**
**    Specifically, the pager jumps into the ERROR state if:
**
**      1. An error occurs while attempting a rollback. This happens in
................................................................................
   || (0 != (rc = sqlite3OsWrite(pPager->jfd, aJournalMagic, 8,
                                 iHdrOff+4+nSuper+8)))
  ){
    return rc;
  }
  pPager->journalOff += (nSuper+20);

  /* If the pager is in peristent-journal mode, then the physical 
  ** journal-file may extend past the end of the super-journal name
  ** and 8 bytes of magic data just written to the file. This is 
  ** dangerous because the code to rollback a hot-journal file
  ** will not be able to find the super-journal name to determine 
  ** whether or not the journal is hot. 
  **
  ** Easiest thing to do in this scenario is to truncate the journal 
................................................................................
  pPager->journalOff = 0;
  pPager->journalHdr = 0;
  pPager->setSuper = 0;
}

/*
** This function is called whenever an IOERR or FULL error that requires
** the pager to transition into the ERROR state may ahve occurred.
** The first argument is a pointer to the pager structure, the second 
** the error-code about to be returned by a pager API function. The 
** value returned is a copy of the second argument to this function. 
**
** If the second argument is SQLITE_FULL, SQLITE_IOERR or one of the
** IOERR sub-codes, the pager enters the ERROR state and the error code
** is stored in Pager.errCode. While the pager remains in the ERROR state,
................................................................................
  int rc;                         /* Return code */
  int nList;                      /* Number of pages in pList */
  PgHdr *p;                       /* For looping over pages */

  assert( pPager->pWal );
  assert( pList );
#ifdef SQLITE_DEBUG
  /* Verify that the page list is in accending order */
  for(p=pList; p && p->pDirty; p=p->pDirty){
    assert( p->pgno < p->pDirty->pgno );
  }
#endif

  assert( pList->pDirty==0 || isCommit );
  if( isCommit ){
................................................................................
**
**    OFF       sqlite3OsSync() is never called.  This is the default
**              for temporary and transient files.
**
**    NORMAL    The journal is synced once before writes begin on the
**              database.  This is normally adequate protection, but
**              it is theoretically possible, though very unlikely,
**              that an inopertune power failure could leave the journal
**              in a state which would cause damage to the database
**              when it is rolled back.
**
**    FULL      The journal is synced twice before writes begin on the
**              database (with some additional information - the nRec field
**              of the journal header - being written in between the two
**              syncs).  If we assume that writing a
................................................................................

  *ppPager = pPager;
  return SQLITE_OK;
}

/*
** Return the sqlite3_file for the main database given the name
** of the corresonding WAL or Journal name as passed into
** xOpen.
*/
sqlite3_file *sqlite3_database_file_object(const char *zName){
  Pager *pPager;
  while( zName[-1]!=0 || zName[-2]!=0 || zName[-3]!=0 || zName[-4]!=0 ){
    zName--;
  }
................................................................................

  if( eMode!=eOld ){

    /* Change the journal mode. */
    assert( pPager->eState!=PAGER_ERROR );
    pPager->journalMode = (u8)eMode;

    /* When transistioning from TRUNCATE or PERSIST to any other journal
    ** mode except WAL, unless the pager is in locking_mode=exclusive mode,
    ** delete the journal file.
    */
    assert( (PAGER_JOURNALMODE_TRUNCATE & 5)==1 );
    assert( (PAGER_JOURNALMODE_PERSIST & 5)==1 );
    assert( (PAGER_JOURNALMODE_DELETE & 5)==0 );
    assert( (PAGER_JOURNALMODE_MEMORY & 5)==4 );







|







 







|







 







|







 







|







 







|







 







|







 







|







 







|







 







|







30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
..
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
...
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
....
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
....
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
....
3116
3117
3118
3119
3120
3121
3122
3123
3124
3125
3126
3127
3128
3129
3130
....
3499
3500
3501
3502
3503
3504
3505
3506
3507
3508
3509
3510
3511
3512
3513
....
5005
5006
5007
5008
5009
5010
5011
5012
5013
5014
5015
5016
5017
5018
5019
....
7274
7275
7276
7277
7278
7279
7280
7281
7282
7283
7284
7285
7286
7287
7288
** journal_mode=MEMORY, or journal_mode=OFF.
**
** Within this comment block, a page is deemed to have been synced
** automatically as soon as it is written when PRAGMA synchronous=OFF.
** Otherwise, the page is not synced until the xSync method of the VFS
** is called successfully on the file containing the page.
**
** Definition:  A page of the database file is said to be "overwritable" if
** one or more of the following are true about the page:
** 
**     (a)  The original content of the page as it was at the beginning of
**          the transaction has been written into the rollback journal and
**          synced.
** 
**     (b)  The page was a freelist leaf page at the start of the transaction.
................................................................................
** 
**     (c)  The page number is greater than the largest page that existed in
**          the database file at the start of the transaction.
** 
** (1) A page of the database file is never overwritten unless one of the
**     following are true:
** 
**     (a) The page and all other pages on the same sector are overwritable.
** 
**     (b) The atomic page write optimization is enabled, and the entire
**         transaction other than the update of the transaction sequence
**         number consists of a single page change.
** 
** (2) The content of a page written into the rollback journal exactly matches
**     both the content in the database when the rollback journal was written
................................................................................
**    instead of READER following such an error.
**
**    Once it has entered the ERROR state, any attempt to use the pager
**    to read or write data returns an error. Eventually, once all 
**    outstanding transactions have been abandoned, the pager is able to
**    transition back to OPEN state, discarding the contents of the 
**    page-cache and any other in-memory state at the same time. Everything
**    is reloaded from disk (and, if necessary, hot-journal rollback performed)
**    when a read-transaction is next opened on the pager (transitioning
**    the pager into READER state). At that point the system has recovered 
**    from the error.
**
**    Specifically, the pager jumps into the ERROR state if:
**
**      1. An error occurs while attempting a rollback. This happens in
................................................................................
   || (0 != (rc = sqlite3OsWrite(pPager->jfd, aJournalMagic, 8,
                                 iHdrOff+4+nSuper+8)))
  ){
    return rc;
  }
  pPager->journalOff += (nSuper+20);

  /* If the pager is in persistent-journal mode, then the physical 
  ** journal-file may extend past the end of the super-journal name
  ** and 8 bytes of magic data just written to the file. This is 
  ** dangerous because the code to rollback a hot-journal file
  ** will not be able to find the super-journal name to determine 
  ** whether or not the journal is hot. 
  **
  ** Easiest thing to do in this scenario is to truncate the journal 
................................................................................
  pPager->journalOff = 0;
  pPager->journalHdr = 0;
  pPager->setSuper = 0;
}

/*
** This function is called whenever an IOERR or FULL error that requires
** the pager to transition into the ERROR state may have occurred.
** The first argument is a pointer to the pager structure, the second 
** the error-code about to be returned by a pager API function. The 
** value returned is a copy of the second argument to this function. 
**
** If the second argument is SQLITE_FULL, SQLITE_IOERR or one of the
** IOERR sub-codes, the pager enters the ERROR state and the error code
** is stored in Pager.errCode. While the pager remains in the ERROR state,
................................................................................
  int rc;                         /* Return code */
  int nList;                      /* Number of pages in pList */
  PgHdr *p;                       /* For looping over pages */

  assert( pPager->pWal );
  assert( pList );
#ifdef SQLITE_DEBUG
  /* Verify that the page list is in ascending order */
  for(p=pList; p && p->pDirty; p=p->pDirty){
    assert( p->pgno < p->pDirty->pgno );
  }
#endif

  assert( pList->pDirty==0 || isCommit );
  if( isCommit ){
................................................................................
**
**    OFF       sqlite3OsSync() is never called.  This is the default
**              for temporary and transient files.
**
**    NORMAL    The journal is synced once before writes begin on the
**              database.  This is normally adequate protection, but
**              it is theoretically possible, though very unlikely,
**              that an inopportune power failure could leave the journal
**              in a state which would cause damage to the database
**              when it is rolled back.
**
**    FULL      The journal is synced twice before writes begin on the
**              database (with some additional information - the nRec field
**              of the journal header - being written in between the two
**              syncs).  If we assume that writing a
................................................................................

  *ppPager = pPager;
  return SQLITE_OK;
}

/*
** Return the sqlite3_file for the main database given the name
** of the corresponding WAL or Journal name as passed into
** xOpen.
*/
sqlite3_file *sqlite3_database_file_object(const char *zName){
  Pager *pPager;
  while( zName[-1]!=0 || zName[-2]!=0 || zName[-3]!=0 || zName[-4]!=0 ){
    zName--;
  }
................................................................................

  if( eMode!=eOld ){

    /* Change the journal mode. */
    assert( pPager->eState!=PAGER_ERROR );
    pPager->journalMode = (u8)eMode;

    /* When transitioning from TRUNCATE or PERSIST to any other journal
    ** mode except WAL, unless the pager is in locking_mode=exclusive mode,
    ** delete the journal file.
    */
    assert( (PAGER_JOURNALMODE_TRUNCATE & 5)==1 );
    assert( (PAGER_JOURNALMODE_PERSIST & 5)==1 );
    assert( (PAGER_JOURNALMODE_DELETE & 5)==0 );
    assert( (PAGER_JOURNALMODE_MEMORY & 5)==4 );

Changes to src/parse.y.

245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
...
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
  GENERATED ALWAYS
%endif
  REINDEX RENAME CTIME_KW IF
  .
%wildcard ANY.

// Define operator precedence early so that this is the first occurrence
// of the operator tokens in the grammer.  Keeping the operators together
// causes them to be assigned integer values that are close together,
// which keeps parser tables smaller.
//
// The token values assigned to these symbols is determined by the order
// in which lemon first sees them.  It must be the case that ISNULL/NOTNULL,
// NE/EQ, GT/LE, and GE/LT are separated by only a single value.  See
// the sqlite3ExprIfFalse() routine for additional information on this
................................................................................
joinop(X) ::= JOIN_KW(A) JOIN.
                  {X = sqlite3JoinType(pParse,&A,0,0);  /*X-overwrites-A*/}
joinop(X) ::= JOIN_KW(A) nm(B) JOIN.
                  {X = sqlite3JoinType(pParse,&A,&B,0); /*X-overwrites-A*/}
joinop(X) ::= JOIN_KW(A) nm(B) nm(C) JOIN.
                  {X = sqlite3JoinType(pParse,&A,&B,&C);/*X-overwrites-A*/}

// There is a parsing abiguity in an upsert statement that uses a
// SELECT on the RHS of a the INSERT:
//
//      INSERT INTO tab SELECT * FROM aaa JOIN bbb ON CONFLICT ...
//                                        here ----^^
//
// When the ON token is encountered, the parser does not know if it is
// the beginning of an ON CONFLICT clause, or the beginning of an ON







|







 







|







245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
...
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
  GENERATED ALWAYS
%endif
  REINDEX RENAME CTIME_KW IF
  .
%wildcard ANY.

// Define operator precedence early so that this is the first occurrence
// of the operator tokens in the grammar.  Keeping the operators together
// causes them to be assigned integer values that are close together,
// which keeps parser tables smaller.
//
// The token values assigned to these symbols is determined by the order
// in which lemon first sees them.  It must be the case that ISNULL/NOTNULL,
// NE/EQ, GT/LE, and GE/LT are separated by only a single value.  See
// the sqlite3ExprIfFalse() routine for additional information on this
................................................................................
joinop(X) ::= JOIN_KW(A) JOIN.
                  {X = sqlite3JoinType(pParse,&A,0,0);  /*X-overwrites-A*/}
joinop(X) ::= JOIN_KW(A) nm(B) JOIN.
                  {X = sqlite3JoinType(pParse,&A,&B,0); /*X-overwrites-A*/}
joinop(X) ::= JOIN_KW(A) nm(B) nm(C) JOIN.
                  {X = sqlite3JoinType(pParse,&A,&B,&C);/*X-overwrites-A*/}

// There is a parsing ambiguity in an upsert statement that uses a
// SELECT on the RHS of a the INSERT:
//
//      INSERT INTO tab SELECT * FROM aaa JOIN bbb ON CONFLICT ...
//                                        here ----^^
//
// When the ON token is encountered, the parser does not know if it is
// the beginning of an ON CONFLICT clause, or the beginning of an ON

Changes to src/pcache.c.

239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
...
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
*/
static int numberOfCachePages(PCache *p){
  if( p->szCache>=0 ){
    /* IMPLEMENTATION-OF: R-42059-47211 If the argument N is positive then the
    ** suggested cache size is set to N. */
    return p->szCache;
  }else{
    /* IMPLEMANTATION-OF: R-59858-46238 If the argument N is negative, then the
    ** number of cache pages is adjusted to be a number of pages that would
    ** use approximately abs(N*1024) bytes of memory based on the current
    ** page size. */
    return (int)((-1024*(i64)p->szCache)/(p->szPage+p->szExtra));
  }
}

................................................................................
      }
    }
  }
  return result.pDirty;
}

/*
** Sort the list of pages in accending order by pgno.  Pages are
** connected by pDirty pointers.  The pDirtyPrev pointers are
** corrupted by this sort.
**
** Since there cannot be more than 2^31 distinct pages in a database,
** there cannot be more than 31 buckets required by the merge sorter.
** One extra bucket is added to catch overflow in case something
** ever changes to make the previous sentence incorrect.







|







 







|







239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
...
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
*/
static int numberOfCachePages(PCache *p){
  if( p->szCache>=0 ){
    /* IMPLEMENTATION-OF: R-42059-47211 If the argument N is positive then the
    ** suggested cache size is set to N. */
    return p->szCache;
  }else{
    /* IMPLEMENTATION-OF: R-59858-46238 If the argument N is negative, then the
    ** number of cache pages is adjusted to be a number of pages that would
    ** use approximately abs(N*1024) bytes of memory based on the current
    ** page size. */
    return (int)((-1024*(i64)p->szCache)/(p->szPage+p->szExtra));
  }
}

................................................................................
      }
    }
  }
  return result.pDirty;
}

/*
** Sort the list of pages in ascending order by pgno.  Pages are
** connected by pDirty pointers.  The pDirtyPrev pointers are
** corrupted by this sort.
**
** Since there cannot be more than 2^31 distinct pages in a database,
** there cannot be more than 31 buckets required by the merge sorter.
** One extra bucket is added to catch overflow in case something
** ever changes to make the previous sentence incorrect.

Changes to src/pcache1.c.

95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
**
** Note: Variables isBulkLocal and isAnchor were once type "u8". That works,
** but causes a 2-byte gap in the structure for most architectures (since 
** pointers must be either 4 or 8-byte aligned). As this structure is located
** in memory directly after the associated page data, if the database is
** corrupt, code at the b-tree layer may overread the page buffer and 
** read part of this structure before the corruption is detected. This
** can cause a valgrind error if the unitialized gap is accessed. Using u16
** ensures there is no such gap, and therefore no bytes of unitialized memory
** in the structure.
*/
struct PgHdr1 {
  sqlite3_pcache_page page;      /* Base class. Must be first. pBuf & pExtra */
  unsigned int iKey;             /* Key value (page number) */
  u16 isBulkLocal;               /* This page from bulk local storage */
  u16 isAnchor;                  /* This is the PGroup.lru element */







|
|







95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
**
** Note: Variables isBulkLocal and isAnchor were once type "u8". That works,
** but causes a 2-byte gap in the structure for most architectures (since 
** pointers must be either 4 or 8-byte aligned). As this structure is located
** in memory directly after the associated page data, if the database is
** corrupt, code at the b-tree layer may overread the page buffer and 
** read part of this structure before the corruption is detected. This
** can cause a valgrind error if the uninitialized gap is accessed. Using u16
** ensures there is no such gap, and therefore no bytes of uninitialized memory
** in the structure.
*/
struct PgHdr1 {
  sqlite3_pcache_page page;      /* Base class. Must be first. pBuf & pExtra */
  unsigned int iKey;             /* Key value (page number) */
  u16 isBulkLocal;               /* This page from bulk local storage */
  u16 isAnchor;                  /* This is the PGroup.lru element */

Changes to src/pragma.c.

837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
....
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
  /*
  **  PRAGMA [schema.]cache_spill
  **  PRAGMA cache_spill=BOOLEAN
  **  PRAGMA [schema.]cache_spill=N
  **
  ** The first form reports the current local setting for the
  ** page cache spill size. The second form turns cache spill on
  ** or off.  When turnning cache spill on, the size is set to the
  ** current cache_size.  The third form sets a spill size that
  ** may be different form the cache size.
  ** If N is positive then that is the
  ** number of pages in the cache.  If N is negative, then the
  ** number of pages is adjusted so that the cache uses -N kibibytes
  ** of memory.
  **
................................................................................
  **    PRAGMA quick_check(N)
  **
  ** Verify the integrity of the database.
  **
  ** The "quick_check" is reduced version of 
  ** integrity_check designed to detect most database corruption
  ** without the overhead of cross-checking indexes.  Quick_check
  ** is linear time wherease integrity_check is O(NlogN).
  **
  ** The maximum nubmer of errors is 100 by default.  A different default
  ** can be specified using a numeric parameter N.
  **
  ** Or, the parameter N can be the name of a table.  In that case, only
  ** the one table named is verified.  The freelist is only verified if
  ** the named table is "sqlite_schema" (or one of its aliases).
  **
  ** All schemas are checked by default.  To check just a single







|







 







|

|







837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
....
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
  /*
  **  PRAGMA [schema.]cache_spill
  **  PRAGMA cache_spill=BOOLEAN
  **  PRAGMA [schema.]cache_spill=N
  **
  ** The first form reports the current local setting for the
  ** page cache spill size. The second form turns cache spill on
  ** or off.  When turning cache spill on, the size is set to the
  ** current cache_size.  The third form sets a spill size that
  ** may be different form the cache size.
  ** If N is positive then that is the
  ** number of pages in the cache.  If N is negative, then the
  ** number of pages is adjusted so that the cache uses -N kibibytes
  ** of memory.
  **
................................................................................
  **    PRAGMA quick_check(N)
  **
  ** Verify the integrity of the database.
  **
  ** The "quick_check" is reduced version of 
  ** integrity_check designed to detect most database corruption
  ** without the overhead of cross-checking indexes.  Quick_check
  ** is linear time whereas integrity_check is O(NlogN).
  **
  ** The maximum number of errors is 100 by default.  A different default
  ** can be specified using a numeric parameter N.
  **
  ** Or, the parameter N can be the name of a table.  In that case, only
  ** the one table named is verified.  The freelist is only verified if
  ** the named table is "sqlite_schema" (or one of its aliases).
  **
  ** All schemas are checked by default.  To check just a single

Changes to src/printf.c.

13
14
15
16
17
18
19
20
21
22
23
24
25
26
27

/*
** Conversion types fall into various categories as defined by the
** following enumeration.
*/
#define etRADIX       0 /* non-decimal integer types.  %x %o */
#define etFLOAT       1 /* Floating point.  %f */
#define etEXP         2 /* Exponentional notation. %e and %E */
#define etGENERIC     3 /* Floating or exponential, depending on exponent. %g */
#define etSIZE        4 /* Return number of characters processed so far. %n */
#define etSTRING      5 /* Strings. %s */
#define etDYNSTRING   6 /* Dynamically allocated strings. %z */
#define etPERCENT     7 /* Percent symbol. %% */
#define etCHARX       8 /* Characters. %c */
/* The rest are extensions, not normally found in printf() */







|







13
14
15
16
17
18
19
20
21
22
23
24
25
26
27

/*
** Conversion types fall into various categories as defined by the
** following enumeration.
*/
#define etRADIX       0 /* non-decimal integer types.  %x %o */
#define etFLOAT       1 /* Floating point.  %f */
#define etEXP         2 /* Exponental notation. %e and %E */
#define etGENERIC     3 /* Floating or exponential, depending on exponent. %g */
#define etSIZE        4 /* Return number of characters processed so far. %n */
#define etSTRING      5 /* Strings. %s */
#define etDYNSTRING   6 /* Dynamically allocated strings. %z */
#define etPERCENT     7 /* Percent symbol. %% */
#define etCHARX       8 /* Characters. %c */
/* The rest are extensions, not normally found in printf() */

Changes to src/resolve.c.

908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
....
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
....
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
....
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
        if( (pDef->funcFlags & SQLITE_FUNC_INTERNAL)!=0
         && pParse->nested==0
         && (pParse->db->mDbFlags & DBFLAG_InternalFunc)==0
        ){
          /* Internal-use-only functions are disallowed unless the
          ** SQL is being compiled using sqlite3NestedParse() or
          ** the SQLITE_TESTCTRL_INTERNAL_FUNCTIONS test-control has be
          ** used to activate internal functionsn for testing purposes */
          no_such_func = 1;
          pDef = 0;
        }else
        if( (pDef->funcFlags & (SQLITE_FUNC_DIRECT|SQLITE_FUNC_UNSAFE))!=0
         && !IN_RENAME_OBJECT
        ){
          sqlite3ExprFunctionUsable(pParse, pExpr, pDef);
................................................................................
          ** is deleted.
          **
          ** Or, if this is running as part of an ALTER TABLE operation,
          ** resolve the symbols in the actual expression, not a duplicate.
          ** And, if one of the comparisons is successful, leave the expression
          ** as is instead of transforming it to an integer as in the usual
          ** case. This allows the code in alter.c to modify column
          ** refererences within the ORDER BY expression as required.  */
          if( IN_RENAME_OBJECT ){
            pDup = pE;
          }else{
            pDup = sqlite3ExprDup(db, pE, 0);
          }
          if( !db->mallocFailed ){
            assert(pDup);
................................................................................
    /* Otherwise, treat the ORDER BY term as an ordinary expression */
    pItem->u.x.iOrderByCol = 0;
    if( sqlite3ResolveExprNames(pNC, pE) ){
      return 1;
    }
    for(j=0; j<pSelect->pEList->nExpr; j++){
      if( sqlite3ExprCompare(0, pE, pSelect->pEList->a[j].pExpr, -1)==0 ){
        /* Since this expresion is being changed into a reference
        ** to an identical expression in the result set, remove all Window
        ** objects belonging to the expression from the Select.pWin list. */
        windowRemoveExprFromSelect(pSelect, pE);
        pItem->u.x.iOrderByCol = j+1;
      }
    }
  }
................................................................................
  }
  pNC->ncFlags |= savedHasAgg;
  return WRC_Continue;
}

/*
** Resolve all names in all expressions of a SELECT and in all
** decendents of the SELECT, including compounds off of p->pPrior,
** subqueries in expressions, and subqueries used as FROM clause
** terms.
**
** See sqlite3ResolveExprNames() for a description of the kinds of
** transformations that occur.
**
** All SELECT statements should have been expanded using







|







 







|







 







|







 







|







908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
....
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
....
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
....
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
        if( (pDef->funcFlags & SQLITE_FUNC_INTERNAL)!=0
         && pParse->nested==0
         && (pParse->db->mDbFlags & DBFLAG_InternalFunc)==0
        ){
          /* Internal-use-only functions are disallowed unless the
          ** SQL is being compiled using sqlite3NestedParse() or
          ** the SQLITE_TESTCTRL_INTERNAL_FUNCTIONS test-control has be
          ** used to activate internal functions for testing purposes */
          no_such_func = 1;
          pDef = 0;
        }else
        if( (pDef->funcFlags & (SQLITE_FUNC_DIRECT|SQLITE_FUNC_UNSAFE))!=0
         && !IN_RENAME_OBJECT
        ){
          sqlite3ExprFunctionUsable(pParse, pExpr, pDef);
................................................................................
          ** is deleted.
          **
          ** Or, if this is running as part of an ALTER TABLE operation,
          ** resolve the symbols in the actual expression, not a duplicate.
          ** And, if one of the comparisons is successful, leave the expression
          ** as is instead of transforming it to an integer as in the usual
          ** case. This allows the code in alter.c to modify column
          ** references within the ORDER BY expression as required.  */
          if( IN_RENAME_OBJECT ){
            pDup = pE;
          }else{
            pDup = sqlite3ExprDup(db, pE, 0);
          }
          if( !db->mallocFailed ){
            assert(pDup);
................................................................................
    /* Otherwise, treat the ORDER BY term as an ordinary expression */
    pItem->u.x.iOrderByCol = 0;
    if( sqlite3ResolveExprNames(pNC, pE) ){
      return 1;
    }
    for(j=0; j<pSelect->pEList->nExpr; j++){
      if( sqlite3ExprCompare(0, pE, pSelect->pEList->a[j].pExpr, -1)==0 ){
        /* Since this expression is being changed into a reference
        ** to an identical expression in the result set, remove all Window
        ** objects belonging to the expression from the Select.pWin list. */
        windowRemoveExprFromSelect(pSelect, pE);
        pItem->u.x.iOrderByCol = j+1;
      }
    }
  }
................................................................................
  }
  pNC->ncFlags |= savedHasAgg;
  return WRC_Continue;
}

/*
** Resolve all names in all expressions of a SELECT and in all
** descendents of the SELECT, including compounds off of p->pPrior,
** subqueries in expressions, and subqueries used as FROM clause
** terms.
**
** See sqlite3ResolveExprNames() for a description of the kinds of
** transformations that occur.
**
** All SELECT statements should have been expanded using

Changes to src/rowset.c.

38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
** allocated in chunks so most INSERTs do no allocation.  There is an 
** upper bound on the size of allocated memory.  No memory is freed
** until DESTROY.
**
** The TEST primitive includes a "batch" number.  The TEST primitive
** will only see elements that were inserted before the last change
** in the batch number.  In other words, if an INSERT occurs between
** two TESTs where the TESTs have the same batch nubmer, then the
** value added by the INSERT will not be visible to the second TEST.
** The initial batch number is zero, so if the very first TEST contains
** a non-zero batch number, it will see all prior INSERTs.
**
** No INSERTs may occurs after a SMALLEST.  An assertion will fail if
** that is attempted.
**







|







38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
** allocated in chunks so most INSERTs do no allocation.  There is an 
** upper bound on the size of allocated memory.  No memory is freed
** until DESTROY.
**
** The TEST primitive includes a "batch" number.  The TEST primitive
** will only see elements that were inserted before the last change
** in the batch number.  In other words, if an INSERT occurs between
** two TESTs where the TESTs have the same batch number, then the
** value added by the INSERT will not be visible to the second TEST.
** The initial batch number is zero, so if the very first TEST contains
** a non-zero batch number, it will see all prior INSERTs.
**
** No INSERTs may occurs after a SMALLEST.  An assertion will fail if
** that is attempted.
**

Changes to src/select.c.

607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
....
2343
2344
2345
2346
2347
2348
2349
2350
2351
2352
2353
2354
2355
2356
2357
....
2487
2488
2489
2490
2491
2492
2493
2494
2495
2496
2497
2498
2499
2500
2501
....
2911
2912
2913
2914
2915
2916
2917
2918
2919
2920
2921
2922
2923
2924
2925
....
4115
4116
4117
4118
4119
4120
4121
4122
4123
4124
4125
4126
4127
4128
4129
....
4174
4175
4176
4177
4178
4179
4180
4181
4182
4183
4184
4185
4186
4187
4188
....
5895
5896
5897
5898
5899
5900
5901
5902
5903
5904
5905
5906
5907
5908
5909
....
6478
6479
6480
6481
6482
6483
6484
6485
6486
6487
6488
6489
6490
6491
6492
  **       by a prior OP_MakeRecord.  In this case nData==1 and regData
  **       will be completely unrelated to regOrigData.
  **   (2) All output columns are included in the sort record.  In that
  **       case regData==regOrigData.
  **   (3) Some output columns are omitted from the sort record due to
  **       the SQLITE_ENABLE_SORTER_REFERENCE optimization, or due to the
  **       SQLITE_ECEL_OMITREF optimization, or due to the 
  **       SortCtx.pDeferredRowLoad optimiation.  In any of these cases
  **       regOrigData is 0 to prevent this routine from trying to copy
  **       values that might not yet exist.
  */
  assert( nData==1 || regData==regOrigData || regOrigData==0 );

  if( nPrefixReg ){
    assert( nPrefixReg==nExpr+bSeq );
................................................................................
  int addrTop;                  /* Top of the loop */
  int addrCont, addrBreak;      /* CONTINUE and BREAK addresses */
  int iCurrent = 0;             /* The Current table */
  int regCurrent;               /* Register holding Current table */
  int iQueue;                   /* The Queue table */
  int iDistinct = 0;            /* To ensure unique results if UNION */
  int eDest = SRT_Fifo;         /* How to write to Queue */
  SelectDest destQueue;         /* SelectDest targetting the Queue table */
  int i;                        /* Loop counter */
  int rc;                       /* Result code */
  ExprList *pOrderBy;           /* The ORDER BY clause */
  Expr *pLimit;                 /* Saved LIMIT and OFFSET */
  int regLimit, regOffset;      /* Registers used by LIMIT and OFFSET */

#ifndef SQLITE_OMIT_WINDOWFUNC
................................................................................
**   (1) There is no LIMIT or OFFSET or else there is a LIMIT of exactly 1
**   (2) All terms are UNION ALL
**   (3) There is no ORDER BY clause
**
** The "LIMIT of exactly 1" case of condition (1) comes about when a VALUES
** clause occurs within scalar expression (ex: "SELECT (VALUES(1),(2),(3))").
** The sqlite3CodeSubselect will have added the LIMIT 1 clause in tht case.
** Since the limit is exactly 1, we only need to evalutes the left-most VALUES.
*/
static int multiSelectValues(
  Parse *pParse,        /* Parsing context */
  Select *p,            /* The right-most of SELECTs to be coded */
  SelectDest *pDest     /* What to do with query results */
){
  int nRow = 1;
................................................................................
    sqlite3ErrorMsg(pParse, "SELECTs to the left and right of %s"
      " do not have the same number of result columns", selectOpName(p->op));
  }
}

/*
** Code an output subroutine for a coroutine implementation of a
** SELECT statment.
**
** The data to be output is contained in pIn->iSdst.  There are
** pIn->nSdst columns to be output.  pDest is where the output should
** be sent.
**
** regReturn is the number of the register holding the subroutine
** return address.
................................................................................
    ** "a" we substitute "x*3" and every place we see "b" we substitute "y+10".
    */
    if( pSub->pOrderBy && (pParent->selFlags & SF_NoopOrderBy)==0 ){
      /* At this point, any non-zero iOrderByCol values indicate that the
      ** ORDER BY column expression is identical to the iOrderByCol'th
      ** expression returned by SELECT statement pSub. Since these values
      ** do not necessarily correspond to columns in SELECT statement pParent,
      ** zero them before transfering the ORDER BY clause.
      **
      ** Not doing this may cause an error if a subsequent call to this
      ** function attempts to flatten a compound sub-query into pParent
      ** (the only way this can happen is if the compound sub-query is
      ** currently part of pSub->pSrc). See ticket [d11a6e908f].  */
      ExprList *pOrderBy = pSub->pOrderBy;
      for(i=0; i<pOrderBy->nExpr; i++){
................................................................................
    /* Recompute the SrcList_item.colUsed masks for the flattened
    ** tables. */
    for(i=0; i<nSubSrc; i++){
      recomputeColumnsUsed(pParent, &pSrc->a[i+iFrom]);
    }
  }

  /* Finially, delete what is left of the subquery and return
  ** success.
  */
  sqlite3AggInfoPersistWalkerInit(&w, pParse);
  sqlite3WalkSelect(&w,pSub1);
  sqlite3SelectDelete(db, pSub1);

#if SELECTTRACE_ENABLED
................................................................................
    ){
      SELECTTRACE(0x100,pParse,p,
                ("LEFT-JOIN simplifies to JOIN on term %d\n",i));
      pItem->fg.jointype &= ~(JT_LEFT|JT_OUTER);
      unsetJoinExpr(p->pWhere, pItem->iCursor);
    }

    /* No futher action if this term of the FROM clause is no a subquery */
    if( pSub==0 ) continue;

    /* Catch mismatch in the declared columns of a view and the number of
    ** columns in the SELECT on the RHS */
    if( pTab->nCol!=pSub->pEList->nExpr ){
      sqlite3ErrorMsg(pParse, "expected %d columns for '%s' but got %d",
                      pTab->nCol, pTab->zName, pSub->pEList->nExpr);
................................................................................


    /* Processing for aggregates with GROUP BY is very different and
    ** much more complex than aggregates without a GROUP BY.
    */
    if( pGroupBy ){
      KeyInfo *pKeyInfo;  /* Keying information for the group by clause */
      int addr1;          /* A-vs-B comparision jump */
      int addrOutputRow;  /* Start of subroutine that outputs a result row */
      int regOutputRow;   /* Return address register for output subroutine */
      int addrSetAbort;   /* Set the abort flag and return */
      int addrTopOfLoop;  /* Top of the input loop */
      int addrSortingIdx; /* The OP_OpenEphemeral for the sorting index */
      int addrReset;      /* Subroutine for resetting the accumulator */
      int regReset;       /* Return address register for reset subroutine */







|







 







|







 







|







 







|







 







|







 







|







 







|







 







|







607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
....
2343
2344
2345
2346
2347
2348
2349
2350
2351
2352
2353
2354
2355
2356
2357
....
2487
2488
2489
2490
2491
2492
2493
2494
2495
2496
2497
2498
2499
2500
2501
....
2911
2912
2913
2914
2915
2916
2917
2918
2919
2920
2921
2922
2923
2924
2925
....
4115
4116
4117
4118
4119
4120
4121
4122
4123
4124
4125
4126
4127
4128
4129
....
4174
4175
4176
4177
4178
4179
4180
4181
4182
4183
4184
4185
4186
4187
4188
....
5895
5896
5897
5898
5899
5900
5901
5902
5903
5904
5905
5906
5907
5908
5909
....
6478
6479
6480
6481
6482
6483
6484
6485
6486
6487
6488
6489
6490
6491
6492
  **       by a prior OP_MakeRecord.  In this case nData==1 and regData
  **       will be completely unrelated to regOrigData.
  **   (2) All output columns are included in the sort record.  In that
  **       case regData==regOrigData.
  **   (3) Some output columns are omitted from the sort record due to
  **       the SQLITE_ENABLE_SORTER_REFERENCE optimization, or due to the
  **       SQLITE_ECEL_OMITREF optimization, or due to the 
  **       SortCtx.pDeferredRowLoad optimization.  In any of these cases
  **       regOrigData is 0 to prevent this routine from trying to copy
  **       values that might not yet exist.
  */
  assert( nData==1 || regData==regOrigData || regOrigData==0 );

  if( nPrefixReg ){
    assert( nPrefixReg==nExpr+bSeq );
................................................................................
  int addrTop;                  /* Top of the loop */
  int addrCont, addrBreak;      /* CONTINUE and BREAK addresses */
  int iCurrent = 0;             /* The Current table */
  int regCurrent;               /* Register holding Current table */
  int iQueue;                   /* The Queue table */
  int iDistinct = 0;            /* To ensure unique results if UNION */
  int eDest = SRT_Fifo;         /* How to write to Queue */
  SelectDest destQueue;         /* SelectDest targeting the Queue table */
  int i;                        /* Loop counter */
  int rc;                       /* Result code */
  ExprList *pOrderBy;           /* The ORDER BY clause */
  Expr *pLimit;                 /* Saved LIMIT and OFFSET */
  int regLimit, regOffset;      /* Registers used by LIMIT and OFFSET */

#ifndef SQLITE_OMIT_WINDOWFUNC
................................................................................
**   (1) There is no LIMIT or OFFSET or else there is a LIMIT of exactly 1
**   (2) All terms are UNION ALL
**   (3) There is no ORDER BY clause
**
** The "LIMIT of exactly 1" case of condition (1) comes about when a VALUES
** clause occurs within scalar expression (ex: "SELECT (VALUES(1),(2),(3))").
** The sqlite3CodeSubselect will have added the LIMIT 1 clause in tht case.
** Since the limit is exactly 1, we only need to evaluate the left-most VALUES.
*/
static int multiSelectValues(
  Parse *pParse,        /* Parsing context */
  Select *p,            /* The right-most of SELECTs to be coded */
  SelectDest *pDest     /* What to do with query results */
){
  int nRow = 1;
................................................................................
    sqlite3ErrorMsg(pParse, "SELECTs to the left and right of %s"
      " do not have the same number of result columns", selectOpName(p->op));
  }
}

/*
** Code an output subroutine for a coroutine implementation of a
** SELECT statement.
**
** The data to be output is contained in pIn->iSdst.  There are
** pIn->nSdst columns to be output.  pDest is where the output should
** be sent.
**
** regReturn is the number of the register holding the subroutine
** return address.
................................................................................
    ** "a" we substitute "x*3" and every place we see "b" we substitute "y+10".
    */
    if( pSub->pOrderBy && (pParent->selFlags & SF_NoopOrderBy)==0 ){
      /* At this point, any non-zero iOrderByCol values indicate that the
      ** ORDER BY column expression is identical to the iOrderByCol'th
      ** expression returned by SELECT statement pSub. Since these values
      ** do not necessarily correspond to columns in SELECT statement pParent,
      ** zero them before transferring the ORDER BY clause.
      **
      ** Not doing this may cause an error if a subsequent call to this
      ** function attempts to flatten a compound sub-query into pParent
      ** (the only way this can happen is if the compound sub-query is
      ** currently part of pSub->pSrc). See ticket [d11a6e908f].  */
      ExprList *pOrderBy = pSub->pOrderBy;
      for(i=0; i<pOrderBy->nExpr; i++){
................................................................................
    /* Recompute the SrcList_item.colUsed masks for the flattened
    ** tables. */
    for(i=0; i<nSubSrc; i++){
      recomputeColumnsUsed(pParent, &pSrc->a[i+iFrom]);
    }
  }

  /* Finally, delete what is left of the subquery and return
  ** success.
  */
  sqlite3AggInfoPersistWalkerInit(&w, pParse);
  sqlite3WalkSelect(&w,pSub1);
  sqlite3SelectDelete(db, pSub1);

#if SELECTTRACE_ENABLED
................................................................................
    ){
      SELECTTRACE(0x100,pParse,p,
                ("LEFT-JOIN simplifies to JOIN on term %d\n",i));
      pItem->fg.jointype &= ~(JT_LEFT|JT_OUTER);
      unsetJoinExpr(p->pWhere, pItem->iCursor);
    }

    /* No further action if this term of the FROM clause is no a subquery */
    if( pSub==0 ) continue;

    /* Catch mismatch in the declared columns of a view and the number of
    ** columns in the SELECT on the RHS */
    if( pTab->nCol!=pSub->pEList->nExpr ){
      sqlite3ErrorMsg(pParse, "expected %d columns for '%s' but got %d",
                      pTab->nCol, pTab->zName, pSub->pEList->nExpr);
................................................................................


    /* Processing for aggregates with GROUP BY is very different and
    ** much more complex than aggregates without a GROUP BY.
    */
    if( pGroupBy ){
      KeyInfo *pKeyInfo;  /* Keying information for the group by clause */
      int addr1;          /* A-vs-B comparison jump */
      int addrOutputRow;  /* Start of subroutine that outputs a result row */
      int regOutputRow;   /* Return address register for output subroutine */
      int addrSetAbort;   /* Set the abort flag and return */
      int addrTopOfLoop;  /* Top of the input loop */
      int addrSortingIdx; /* The OP_OpenEphemeral for the sorting index */
      int addrReset;      /* Subroutine for resetting the accumulator */
      int regReset;       /* Return address register for reset subroutine */

Changes to src/shell.c.in.

1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
....
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
....
3430
3431
3432
3433
3434
3435
3436
3437
3438
3439
3440
3441
3442
3443
3444
....
3481
3482
3483
3484
3485
3486
3487
3488
3489
3490
3491
3492
3493
3494
3495
....
3616
3617
3618
3619
3620
3621
3622
3623
3624
3625
3626
3627
3628
3629
3630
....
3827
3828
3829
3830
3831
3832
3833
3834
3835
3836
3837
3838
3839
3840
3841
....
3953
3954
3955
3956
3957
3958
3959
3960
3961
3962
3963
3964
3965
3966
3967
....
4631
4632
4633
4634
4635
4636
4637
4638
4639
4640
4641
4642
4643
4644
4645
.....
10819
10820
10821
10822
10823
10824
10825
10826
10827
10828
10829
10830
10831
10832
10833
** State information about the database connection is contained in an
** instance of the following structure.
*/
typedef struct ShellState ShellState;
struct ShellState {
  sqlite3 *db;           /* The database */
  u8 autoExplain;        /* Automatically turn on .explain mode */
  u8 autoEQP;            /* Run EXPLAIN QUERY PLAN prior to seach SQL stmt */
  u8 autoEQPtest;        /* autoEQP is in test mode */
  u8 autoEQPtrace;       /* autoEQP is in trace mode */
  u8 statsOn;            /* True to display memory stats before each finalize */
  u8 scanstatsOn;        /* True to display scan stats before each finalize */
  u8 openMode;           /* SHELL_OPEN_NORMAL, _APPENDVFS, or _ZIPFILE */
  u8 doXdgOpen;          /* Invoke start/open/xdg-open in output_reset() */
  u8 nEqpLevel;          /* Depth of the EQP output graph */
  u8 eTraceType;         /* SHELL_TRACE_* value for type of trace */
  unsigned mEqpLines;    /* Mask of veritical lines in the EQP output graph */
  int outCount;          /* Revert to stdout when reaching zero */
  int cnt;               /* Number of records displayed so far */
  int lineno;            /* Line number of last line read from in */
  int openFlags;         /* Additional flags to open.  (SQLITE_OPEN_NOFOLLOW) */
  FILE *in;              /* Read commands from this stream */
  FILE *out;             /* Write results here */
  FILE *traceOut;        /* Output for sqlite3_trace() */
................................................................................
    raw_printf(out, "'");
  }
  setTextMode(out, 1);
}

/*
** Output the given string as a quoted string using SQL quoting conventions.
** Additionallly , escape the "\n" and "\r" characters so that they do not
** get corrupted by end-of-line translation facilities in some operating
** systems.
**
** This is like output_quoted_string() but with the addition of the \r\n
** escape mechanism.
*/
static void output_quoted_escaped_string(FILE *out, const char *z){
................................................................................
        while( IsSpace(zSql[0]) ) zSql++;
        continue;
      }
      zStmtSql = sqlite3_sql(pStmt);
      if( zStmtSql==0 ) zStmtSql = "";
      while( IsSpace(zStmtSql[0]) ) zStmtSql++;

      /* save off the prepared statment handle and reset row count */
      if( pArg ){
        pArg->pStmt = pStmt;
        pArg->cnt = 0;
      }

      /* echo the sql statement if echo on */
      if( pArg && ShellHasFlag(pArg, SHFLG_Echo) ){
................................................................................
            explain_data_delete(pArg);
          }
          sqlite3_finalize(pExplain);
          sqlite3_free(zEQP);
        }
        if( pArg->autoEQP>=AUTOEQP_trigger && triggerEQP==0 ){
          sqlite3_db_config(db, SQLITE_DBCONFIG_TRIGGER_EQP, 0, 0);
          /* Reprepare pStmt before reactiving trace modes */
          sqlite3_finalize(pStmt);
          sqlite3_prepare_v2(db, zSql, -1, &pStmt, 0);
          if( pArg ) pArg->pStmt = pStmt;
        }
        restore_debug_trace_modes();
      }

................................................................................
  ** is tricky.  We never need to preserve a rowid for a WITHOUT ROWID table
  ** or a table with an INTEGER PRIMARY KEY.  We are unable to preserve
  ** rowids on tables where the rowid is inaccessible because there are other
  ** columns in the table named "rowid", "_rowid_", and "oid".
  */
  if( preserveRowid && isIPK ){
    /* If a single PRIMARY KEY column with type INTEGER was seen, then it
    ** might be an alise for the ROWID.  But it might also be a WITHOUT ROWID
    ** table or a INTEGER PRIMARY KEY DESC column, neither of which are
    ** ROWID aliases.  To distinguish these cases, check to see if
    ** there is a "pk" entry in "PRAGMA index_list".  There will be
    ** no "pk" index if the PRIMARY KEY really is an alias for the ROWID.
    */
    zSql = sqlite3_mprintf("SELECT 1 FROM pragma_index_list(%Q)"
                           " WHERE origin='pk'", zTab);
................................................................................
  return rc;
}

/*
** Text of help messages.
**
** The help text for each individual command begins with a line that starts
** with ".".  Subsequent lines are supplimental information.
**
** There must be two or more spaces between the end of the command and the
** start of the description of what that command does.
*/
static const char *(azHelp[]) = {
#if defined(SQLITE_HAVE_ZLIB) && !defined(SQLITE_OMIT_VIRTUALTABLE)
  ".archive ...             Manage SQL archives",
................................................................................
#ifdef SQLITE_DEBUG
  ".oom ?--repeat M? ?N?    Simulate an OOM error on the N-th allocation",
#endif 
  ".open ?OPTIONS? ?FILE?   Close existing database and reopen FILE",
  "     Options:",
  "        --append        Use appendvfs to append database to the end of FILE",
#ifdef SQLITE_ENABLE_DESERIALIZE
  "        --deserialize   Load into memory useing sqlite3_deserialize()",
  "        --hexdb         Load the output of \"dbtotxt\" as an in-memory db",
  "        --maxsize N     Maximum size for --hexdb or --deserialized database",
#endif
  "        --new           Initialize FILE to an empty database",
  "        --nofollow      Do not follow symbolic links",
  "        --readonly      Open FILE readonly",
  "        --zip           FILE is a ZIP archive",
................................................................................
      }
    }
#endif
  }
}

/*
** Attempt to close the databaes connection.  Report errors.
*/
void close_db(sqlite3 *db){
  int rc = sqlite3_close(db);
  if( rc ){
    utf8_printf(stderr, "Error: sqlite3_close() returns %d: %s\n",
        rc, sqlite3_errmsg(db));
  } 
................................................................................
  for(i=1; i<argc; i++){
    char *z;
    z = argv[i];
    if( z[0]!='-' ){
      if( data.zDbFilename==0 ){
        data.zDbFilename = z;
      }else{
        /* Excesss arguments are interpreted as SQL (or dot-commands) and
        ** mean that nothing is read from stdin */
        readStdin = 0;
        nCmd++;
        azCmd = realloc(azCmd, sizeof(azCmd[0])*nCmd);
        if( azCmd==0 ) shell_out_of_memory();
        azCmd[nCmd-1] = z;
      }







|








|







 







|







 







|







 







|







 







|







 







|







 







|







 







|







 







|







1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
....
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
....
3430
3431
3432
3433
3434
3435
3436
3437
3438
3439
3440
3441
3442
3443
3444
....
3481
3482
3483
3484
3485
3486
3487
3488
3489
3490
3491
3492
3493
3494
3495
....
3616
3617
3618
3619
3620
3621
3622
3623
3624
3625
3626
3627
3628
3629
3630
....
3827
3828
3829
3830
3831
3832
3833
3834
3835
3836
3837
3838
3839
3840
3841
....
3953
3954
3955
3956
3957
3958
3959
3960
3961
3962
3963
3964
3965
3966
3967
....
4631
4632
4633
4634
4635
4636
4637
4638
4639
4640
4641
4642
4643
4644
4645
.....
10819
10820
10821
10822
10823
10824
10825
10826
10827
10828
10829
10830
10831
10832
10833
** State information about the database connection is contained in an
** instance of the following structure.
*/
typedef struct ShellState ShellState;
struct ShellState {
  sqlite3 *db;           /* The database */
  u8 autoExplain;        /* Automatically turn on .explain mode */
  u8 autoEQP;            /* Run EXPLAIN QUERY PLAN prior to each SQL stmt */
  u8 autoEQPtest;        /* autoEQP is in test mode */
  u8 autoEQPtrace;       /* autoEQP is in trace mode */
  u8 statsOn;            /* True to display memory stats before each finalize */
  u8 scanstatsOn;        /* True to display scan stats before each finalize */
  u8 openMode;           /* SHELL_OPEN_NORMAL, _APPENDVFS, or _ZIPFILE */
  u8 doXdgOpen;          /* Invoke start/open/xdg-open in output_reset() */
  u8 nEqpLevel;          /* Depth of the EQP output graph */
  u8 eTraceType;         /* SHELL_TRACE_* value for type of trace */
  unsigned mEqpLines;    /* Mask of vertical lines in the EQP output graph */
  int outCount;          /* Revert to stdout when reaching zero */
  int cnt;               /* Number of records displayed so far */
  int lineno;            /* Line number of last line read from in */
  int openFlags;         /* Additional flags to open.  (SQLITE_OPEN_NOFOLLOW) */
  FILE *in;              /* Read commands from this stream */
  FILE *out;             /* Write results here */
  FILE *traceOut;        /* Output for sqlite3_trace() */
................................................................................
    raw_printf(out, "'");
  }
  setTextMode(out, 1);
}

/*
** Output the given string as a quoted string using SQL quoting conventions.
** Additionally , escape the "\n" and "\r" characters so that they do not
** get corrupted by end-of-line translation facilities in some operating
** systems.
**
** This is like output_quoted_string() but with the addition of the \r\n
** escape mechanism.
*/
static void output_quoted_escaped_string(FILE *out, const char *z){
................................................................................
        while( IsSpace(zSql[0]) ) zSql++;
        continue;
      }
      zStmtSql = sqlite3_sql(pStmt);
      if( zStmtSql==0 ) zStmtSql = "";
      while( IsSpace(zStmtSql[0]) ) zStmtSql++;

      /* save off the prepared statement handle and reset row count */
      if( pArg ){
        pArg->pStmt = pStmt;
        pArg->cnt = 0;
      }

      /* echo the sql statement if echo on */
      if( pArg && ShellHasFlag(pArg, SHFLG_Echo) ){
................................................................................
            explain_data_delete(pArg);
          }
          sqlite3_finalize(pExplain);
          sqlite3_free(zEQP);
        }
        if( pArg->autoEQP>=AUTOEQP_trigger && triggerEQP==0 ){
          sqlite3_db_config(db, SQLITE_DBCONFIG_TRIGGER_EQP, 0, 0);
          /* Reprepare pStmt before reactivating trace modes */
          sqlite3_finalize(pStmt);
          sqlite3_prepare_v2(db, zSql, -1, &pStmt, 0);
          if( pArg ) pArg->pStmt = pStmt;
        }
        restore_debug_trace_modes();
      }

................................................................................
  ** is tricky.  We never need to preserve a rowid for a WITHOUT ROWID table
  ** or a table with an INTEGER PRIMARY KEY.  We are unable to preserve
  ** rowids on tables where the rowid is inaccessible because there are other
  ** columns in the table named "rowid", "_rowid_", and "oid".
  */
  if( preserveRowid && isIPK ){
    /* If a single PRIMARY KEY column with type INTEGER was seen, then it
    ** might be an alias for the ROWID.  But it might also be a WITHOUT ROWID
    ** table or a INTEGER PRIMARY KEY DESC column, neither of which are
    ** ROWID aliases.  To distinguish these cases, check to see if
    ** there is a "pk" entry in "PRAGMA index_list".  There will be
    ** no "pk" index if the PRIMARY KEY really is an alias for the ROWID.
    */
    zSql = sqlite3_mprintf("SELECT 1 FROM pragma_index_list(%Q)"
                           " WHERE origin='pk'", zTab);
................................................................................
  return rc;
}

/*
** Text of help messages.
**
** The help text for each individual command begins with a line that starts
** with ".".  Subsequent lines are supplemental information.
**
** There must be two or more spaces between the end of the command and the
** start of the description of what that command does.
*/
static const char *(azHelp[]) = {
#if defined(SQLITE_HAVE_ZLIB) && !defined(SQLITE_OMIT_VIRTUALTABLE)
  ".archive ...             Manage SQL archives",
................................................................................
#ifdef SQLITE_DEBUG
  ".oom ?--repeat M? ?N?    Simulate an OOM error on the N-th allocation",
#endif 
  ".open ?OPTIONS? ?FILE?   Close existing database and reopen FILE",
  "     Options:",
  "        --append        Use appendvfs to append database to the end of FILE",
#ifdef SQLITE_ENABLE_DESERIALIZE
  "        --deserialize   Load into memory using sqlite3_deserialize()",
  "        --hexdb         Load the output of \"dbtotxt\" as an in-memory db",
  "        --maxsize N     Maximum size for --hexdb or --deserialized database",
#endif
  "        --new           Initialize FILE to an empty database",
  "        --nofollow      Do not follow symbolic links",
  "        --readonly      Open FILE readonly",
  "        --zip           FILE is a ZIP archive",
................................................................................
      }
    }
#endif
  }
}

/*
** Attempt to close the database connection.  Report errors.
*/
void close_db(sqlite3 *db){
  int rc = sqlite3_close(db);
  if( rc ){
    utf8_printf(stderr, "Error: sqlite3_close() returns %d: %s\n",
        rc, sqlite3_errmsg(db));
  } 
................................................................................
  for(i=1; i<argc; i++){
    char *z;
    z = argv[i];
    if( z[0]!='-' ){
      if( data.zDbFilename==0 ){
        data.zDbFilename = z;
      }else{
        /* Excess arguments are interpreted as SQL (or dot-commands) and
        ** mean that nothing is read from stdin */
        readStdin = 0;
        nCmd++;
        azCmd = realloc(azCmd, sizeof(azCmd[0])*nCmd);
        if( azCmd==0 ) shell_out_of_memory();
        azCmd[nCmd-1] = z;
      }

Changes to src/sqlite.h.in.

2318
2319
2320
2321
2322
2323
2324
2325
2326
2327
2328
2329
2330
2331
2332
....
3543
3544
3545
3546
3547
3548
3549
3550
3551
3552
3553
3554
3555
3556
3557
....
3656
3657
3658
3659
3660
3661
3662
3663
3664
3665
3666
3667
3668
3669
3670
....
9151
9152
9153
9154
9155
9156
9157
9158
9159
9160
9161
9162
9163
9164
9165
9166
9167
9168
9169
9170
9171
9172
9173
** practical use, but is provided so that SQLite can continue to claim the
** ability to generate new database files that are compatible with  version
** 3.0.0.
** <p>Note that when the SQLITE_DBCONFIG_LEGACY_FILE_FORMAT setting is on,
** the [VACUUM] command will fail with an obscure error when attempting to
** process a table with generated columns and a descending index.  This is
** not considered a bug since SQLite versions 3.3.0 and earlier do not support
** either generated columns or decending indexes.
** </dd>
** </dl>
*/
#define SQLITE_DBCONFIG_MAINDBNAME            1000 /* const char* */
#define SQLITE_DBCONFIG_LOOKASIDE             1001 /* void* int int */
#define SQLITE_DBCONFIG_ENABLE_FKEY           1002 /* int int* */
#define SQLITE_DBCONFIG_ENABLE_TRIGGER        1003 /* int int* */
................................................................................
** that check if a database file was a URI that contained a specific query 
** parameter, and if so obtains the value of that query parameter.
**
** The first parameter to these interfaces (hereafter referred to
** as F) must be one of:
** <ul>
** <li> A database filename pointer created by the SQLite core and
** passed into the xOpen() method of a VFS implemention, or
** <li> A filename obtained from [sqlite3_db_filename()], or
** <li> A new filename constructed using [sqlite3_create_filename()].
** </ul>
** If the F parameter is not one of the above, then the behavior is
** undefined and probably undesirable.  Older versions of SQLite were
** more tolerant of invalid F parameters than newer versions.
**
................................................................................
** behavior.
*/
sqlite3_file *sqlite3_database_file_object(const char*);

/*
** CAPI3REF: Create and Destroy VFS Filenames
**
** These interfces are provided for use by [VFS shim] implementations and
** are not useful outside of that context.
**
** The sqlite3_create_filename(D,J,W,N,P) allocates memory to hold a version of
** database filename D with corresponding journal file J and WAL file W and
** with N URI parameters key/values pairs in the array P.  The result from
** sqlite3_create_filename(D,J,W,N,P) is a pointer to a database filename that
** is safe to pass to routines like:
................................................................................
** SQLITE_CONSTRAINT, in which case SQLite falls back to OR ABORT 
** constraint handling.
** </dd>
**
** [[SQLITE_VTAB_DIRECTONLY]]<dt>SQLITE_VTAB_DIRECTONLY</dt>
** <dd>Calls of the form
** [sqlite3_vtab_config](db,SQLITE_VTAB_DIRECTONLY) from within the
** the [xConnect] or [xCreate] methods of a [virtual table] implmentation
** prohibits that virtual table from being used from within triggers and
** views.
** </dd>
**
** [[SQLITE_VTAB_INNOCUOUS]]<dt>SQLITE_VTAB_INNOCUOUS</dt>
** <dd>Calls of the form
** [sqlite3_vtab_config](db,SQLITE_VTAB_INNOCUOUS) from within the
** the [xConnect] or [xCreate] methods of a [virtual table] implmentation
** identify that virtual table as being safe to use from within triggers
** and views.  Conceptually, the SQLITE_VTAB_INNOCUOUS tag means that the
** virtual table can do no serious harm even if it is controlled by a
** malicious hacker.  Developers should avoid setting the SQLITE_VTAB_INNOCUOUS
** flag unless absolutely necessary.
** </dd>
** </dl>







|







 







|







 







|







 







|







|







2318
2319
2320
2321
2322
2323
2324
2325
2326
2327
2328
2329
2330
2331
2332
....
3543
3544
3545
3546
3547
3548
3549
3550
3551
3552
3553
3554
3555
3556
3557
....
3656
3657
3658
3659
3660
3661
3662
3663
3664
3665
3666
3667
3668
3669
3670
....
9151
9152
9153
9154
9155
9156
9157
9158
9159
9160
9161
9162
9163
9164
9165
9166
9167
9168
9169
9170
9171
9172
9173
** practical use, but is provided so that SQLite can continue to claim the
** ability to generate new database files that are compatible with  version
** 3.0.0.
** <p>Note that when the SQLITE_DBCONFIG_LEGACY_FILE_FORMAT setting is on,
** the [VACUUM] command will fail with an obscure error when attempting to
** process a table with generated columns and a descending index.  This is
** not considered a bug since SQLite versions 3.3.0 and earlier do not support
** either generated columns or descending indexes.
** </dd>
** </dl>
*/
#define SQLITE_DBCONFIG_MAINDBNAME            1000 /* const char* */
#define SQLITE_DBCONFIG_LOOKASIDE             1001 /* void* int int */
#define SQLITE_DBCONFIG_ENABLE_FKEY           1002 /* int int* */
#define SQLITE_DBCONFIG_ENABLE_TRIGGER        1003 /* int int* */
................................................................................
** that check if a database file was a URI that contained a specific query 
** parameter, and if so obtains the value of that query parameter.
**
** The first parameter to these interfaces (hereafter referred to
** as F) must be one of:
** <ul>
** <li> A database filename pointer created by the SQLite core and
** passed into the xOpen() method of a VFS implementation, or
** <li> A filename obtained from [sqlite3_db_filename()], or
** <li> A new filename constructed using [sqlite3_create_filename()].
** </ul>
** If the F parameter is not one of the above, then the behavior is
** undefined and probably undesirable.  Older versions of SQLite were
** more tolerant of invalid F parameters than newer versions.
**
................................................................................
** behavior.
*/
sqlite3_file *sqlite3_database_file_object(const char*);

/*
** CAPI3REF: Create and Destroy VFS Filenames
**
** These interfaces are provided for use by [VFS shim] implementations and
** are not useful outside of that context.
**
** The sqlite3_create_filename(D,J,W,N,P) allocates memory to hold a version of
** database filename D with corresponding journal file J and WAL file W and
** with N URI parameters key/values pairs in the array P.  The result from
** sqlite3_create_filename(D,J,W,N,P) is a pointer to a database filename that
** is safe to pass to routines like:
................................................................................
** SQLITE_CONSTRAINT, in which case SQLite falls back to OR ABORT 
** constraint handling.
** </dd>
**
** [[SQLITE_VTAB_DIRECTONLY]]<dt>SQLITE_VTAB_DIRECTONLY</dt>
** <dd>Calls of the form
** [sqlite3_vtab_config](db,SQLITE_VTAB_DIRECTONLY) from within the
** the [xConnect] or [xCreate] methods of a [virtual table] implementation
** prohibits that virtual table from being used from within triggers and
** views.
** </dd>
**
** [[SQLITE_VTAB_INNOCUOUS]]<dt>SQLITE_VTAB_INNOCUOUS</dt>
** <dd>Calls of the form
** [sqlite3_vtab_config](db,SQLITE_VTAB_INNOCUOUS) from within the
** the [xConnect] or [xCreate] methods of a [virtual table] implementation
** identify that virtual table as being safe to use from within triggers
** and views.  Conceptually, the SQLITE_VTAB_INNOCUOUS tag means that the
** virtual table can do no serious harm even if it is controlled by a
** malicious hacker.  Developers should avoid setting the SQLITE_VTAB_INNOCUOUS
** flag unless absolutely necessary.
** </dd>
** </dl>

Changes to src/sqliteInt.h.

661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
....
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
....
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
....
2660
2661
2662
2663
2664
2665
2666
2667
2668
2669
2670
2671
2672
2673
2674

/*
** The default initial allocation for the pagecache when using separate
** pagecaches for each database connection.  A positive number is the
** number of pages.  A negative number N translations means that a buffer
** of -1024*N bytes is allocated and used for as many pages as it will hold.
**
** The default value of "20" was choosen to minimize the run-time of the
** speedtest1 test program with options: --shrink-memory --reprepare
*/
#ifndef SQLITE_DEFAULT_PCACHE_INITSZ
# define SQLITE_DEFAULT_PCACHE_INITSZ 20
#endif

/*
................................................................................
  Schema *pSchema;     /* Pointer to database schema (possibly shared) */
};

/*
** An instance of the following structure stores a database schema.
**
** Most Schema objects are associated with a Btree.  The exception is
** the Schema for the TEMP databaes (sqlite3.aDb[1]) which is free-standing.
** In shared cache mode, a single Schema object can be shared by multiple
** Btrees that refer to the same underlying BtShared object.
**
** Schema objects are automatically deallocated when the last Btree that
** references them is destroyed.   The TEMP Schema is manually freed by
** sqlite3_close().
*
................................................................................
  u16 szTrue;             /* True value of sz, even if disabled */
  u8 bMalloced;           /* True if pStart obtained from sqlite3_malloc() */
  u32 nSlot;              /* Number of lookaside slots allocated */
  u32 anStat[3];          /* 0: hits.  1: size misses.  2: full misses */
  LookasideSlot *pInit;   /* List of buffers not previously used */
  LookasideSlot *pFree;   /* List of available buffers */
#ifndef SQLITE_OMIT_TWOSIZE_LOOKASIDE
  LookasideSlot *pSmallInit; /* List of small buffers not prediously used */
  LookasideSlot *pSmallFree; /* List of available small buffers */
  void *pMiddle;          /* First byte past end of full-size buffers and
                          ** the first byte of LOOKASIDE_SMALL buffers */
#endif /* SQLITE_OMIT_TWOSIZE_LOOKASIDE */
  void *pStart;           /* First byte of available memory space */
  void *pEnd;             /* First byte past end of available space */
};
................................................................................
#if SQLITE_MAX_EXPR_DEPTH>0
  int nHeight;           /* Height of the tree headed by this node */
#endif
  int iTable;            /* TK_COLUMN: cursor number of table holding column
                         ** TK_REGISTER: register number
                         ** TK_TRIGGER: 1 -> new, 0 -> old
                         ** EP_Unlikely:  134217728 times likelihood
                         ** TK_IN: ephemerial table holding RHS
                         ** TK_SELECT_COLUMN: Number of columns on the LHS
                         ** TK_SELECT: 1st register of result vector */
  ynVar iColumn;         /* TK_COLUMN: column index.  -1 for rowid.
                         ** TK_VARIABLE: variable number (always >= 1).
                         ** TK_SELECT_COLUMN: column of the result vector */
  i16 iAgg;              /* Which entry in pAggInfo->aCol[] or ->aFunc[] */
  i16 iRightJoinTable;   /* If EP_FromJoin, the right table of the join */







|







 







|







 







|







 







|







661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
....
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
....
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
....
2660
2661
2662
2663
2664
2665
2666
2667
2668
2669
2670
2671
2672
2673
2674

/*
** The default initial allocation for the pagecache when using separate
** pagecaches for each database connection.  A positive number is the
** number of pages.  A negative number N translations means that a buffer
** of -1024*N bytes is allocated and used for as many pages as it will hold.
**
** The default value of "20" was chosen to minimize the run-time of the
** speedtest1 test program with options: --shrink-memory --reprepare
*/
#ifndef SQLITE_DEFAULT_PCACHE_INITSZ
# define SQLITE_DEFAULT_PCACHE_INITSZ 20
#endif

/*
................................................................................
  Schema *pSchema;     /* Pointer to database schema (possibly shared) */
};

/*
** An instance of the following structure stores a database schema.
**
** Most Schema objects are associated with a Btree.  The exception is
** the Schema for the TEMP database (sqlite3.aDb[1]) which is free-standing.
** In shared cache mode, a single Schema object can be shared by multiple
** Btrees that refer to the same underlying BtShared object.
**
** Schema objects are automatically deallocated when the last Btree that
** references them is destroyed.   The TEMP Schema is manually freed by
** sqlite3_close().
*
................................................................................
  u16 szTrue;             /* True value of sz, even if disabled */
  u8 bMalloced;           /* True if pStart obtained from sqlite3_malloc() */
  u32 nSlot;              /* Number of lookaside slots allocated */
  u32 anStat[3];          /* 0: hits.  1: size misses.  2: full misses */
  LookasideSlot *pInit;   /* List of buffers not previously used */
  LookasideSlot *pFree;   /* List of available buffers */
#ifndef SQLITE_OMIT_TWOSIZE_LOOKASIDE
  LookasideSlot *pSmallInit; /* List of small buffers not previously used */
  LookasideSlot *pSmallFree; /* List of available small buffers */
  void *pMiddle;          /* First byte past end of full-size buffers and
                          ** the first byte of LOOKASIDE_SMALL buffers */
#endif /* SQLITE_OMIT_TWOSIZE_LOOKASIDE */
  void *pStart;           /* First byte of available memory space */
  void *pEnd;             /* First byte past end of available space */
};
................................................................................
#if SQLITE_MAX_EXPR_DEPTH>0
  int nHeight;           /* Height of the tree headed by this node */
#endif
  int iTable;            /* TK_COLUMN: cursor number of table holding column
                         ** TK_REGISTER: register number
                         ** TK_TRIGGER: 1 -> new, 0 -> old
                         ** EP_Unlikely:  134217728 times likelihood
                         ** TK_IN: ephemeral table holding RHS
                         ** TK_SELECT_COLUMN: Number of columns on the LHS
                         ** TK_SELECT: 1st register of result vector */
  ynVar iColumn;         /* TK_COLUMN: column index.  -1 for rowid.
                         ** TK_VARIABLE: variable number (always >= 1).
                         ** TK_SELECT_COLUMN: column of the result vector */
  i16 iAgg;              /* Which entry in pAggInfo->aCol[] or ->aFunc[] */
  i16 iRightJoinTable;   /* If EP_FromJoin, the right table of the join */

Changes to src/sqliteLimit.h.

72
73
74
75
76
77
78
79
80
81
82
83
84
85
86

/*
** The maximum number of terms in a compound SELECT statement.
** The code generator for compound SELECT statements does one
** level of recursion for each term.  A stack overflow can result
** if the number of terms is too large.  In practice, most SQL
** never has more than 3 or 4 terms.  Use a value of 0 to disable
** any limit on the number of terms in a compount SELECT.
*/
#ifndef SQLITE_MAX_COMPOUND_SELECT
# define SQLITE_MAX_COMPOUND_SELECT 500
#endif

/*
** The maximum number of opcodes in a VDBE program.







|







72
73
74
75
76
77
78
79
80
81
82
83
84
85
86

/*
** The maximum number of terms in a compound SELECT statement.
** The code generator for compound SELECT statements does one
** level of recursion for each term.  A stack overflow can result
** if the number of terms is too large.  In practice, most SQL
** never has more than 3 or 4 terms.  Use a value of 0 to disable
** any limit on the number of terms in a compound SELECT.
*/
#ifndef SQLITE_MAX_COMPOUND_SELECT
# define SQLITE_MAX_COMPOUND_SELECT 500
#endif

/*
** The maximum number of opcodes in a VDBE program.

Changes to src/tclsqlite.c.

1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
....
3382
3383
3384
3385
3386
3387
3388
3389
3390
3391
3392
3393
3394
3395
3396
  }

  return Tcl_NewStringObj((char*)sqlite3_column_text(pStmt, iCol), -1);
}

/*
** If using Tcl version 8.6 or greater, use the NR functions to avoid
** recursive evalution of scripts by the [db eval] and [db trans]
** commands. Even if the headers used while compiling the extension
** are 8.6 or newer, the code still tests the Tcl version at runtime.
** This allows stubs-enabled builds to be used with older Tcl libraries.
*/
#if TCL_MAJOR_VERSION>8 || (TCL_MAJOR_VERSION==8 && TCL_MINOR_VERSION>=6)
# define SQLITE_TCL_NRE 1
static int DbUseNre(void){
................................................................................
  }

  /*    $db transaction [-deferred|-immediate|-exclusive] SCRIPT
  **
  ** Start a new transaction (if we are not already in the midst of a
  ** transaction) and execute the TCL script SCRIPT.  After SCRIPT
  ** completes, either commit the transaction or roll it back if SCRIPT
  ** throws an exception.  Or if no new transation was started, do nothing.
  ** pass the exception on up the stack.
  **
  ** This command was inspired by Dave Thomas's talk on Ruby at the
  ** 2005 O'Reilly Open Source Convention (OSCON).
  */
  case DB_TRANSACTION: {
    Tcl_Obj *pScript;







|







 







|







1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
....
3382
3383
3384
3385
3386
3387
3388
3389
3390
3391
3392
3393
3394
3395
3396
  }

  return Tcl_NewStringObj((char*)sqlite3_column_text(pStmt, iCol), -1);
}

/*
** If using Tcl version 8.6 or greater, use the NR functions to avoid
** recursive evaluation of scripts by the [db eval] and [db trans]
** commands. Even if the headers used while compiling the extension
** are 8.6 or newer, the code still tests the Tcl version at runtime.
** This allows stubs-enabled builds to be used with older Tcl libraries.
*/
#if TCL_MAJOR_VERSION>8 || (TCL_MAJOR_VERSION==8 && TCL_MINOR_VERSION>=6)
# define SQLITE_TCL_NRE 1
static int DbUseNre(void){
................................................................................
  }

  /*    $db transaction [-deferred|-immediate|-exclusive] SCRIPT
  **
  ** Start a new transaction (if we are not already in the midst of a
  ** transaction) and execute the TCL script SCRIPT.  After SCRIPT
  ** completes, either commit the transaction or roll it back if SCRIPT
  ** throws an exception.  Or if no new transaction was started, do nothing.
  ** pass the exception on up the stack.
  **
  ** This command was inspired by Dave Thomas's talk on Ruby at the
  ** 2005 O'Reilly Open Source Convention (OSCON).
  */
  case DB_TRANSACTION: {
    Tcl_Obj *pScript;

Changes to src/test1.c.

2636
2637
2638
2639
2640
2641
2642
2643
2644
2645
2646
2647
2648
2649
2650
....
5644
5645
5646
5647
5648
5649
5650
5651
5652
5653
5654
5655
5656
5657
5658
....
6970
6971
6972
6973
6974
6975
6976
6977
6978
6979
6980
6981
6982
6983
6984
  sqlite3_close(db);
  return TCL_OK;
}

/*
** Usage:  sqlite3_next_stmt  DB  STMT
**
** Return the next statment in sequence after STMT.
*/
static int SQLITE_TCLAPI test_next_stmt(
  void * clientData,
  Tcl_Interp *interp,
  int objc,
  Tcl_Obj *CONST objv[]
){
................................................................................
  assert( pMain==0 || pMain==sqlite3_vfs_find(0) );

  /* We can find a VFS by its name */
  assert( sqlite3_vfs_find("__one")==&one );
  assert( sqlite3_vfs_find("__two")==&two );

  /* Calling sqlite_vfs_register with non-zero second parameter changes the
  ** default VFS, even if the 1st parameter is an existig VFS that is
  ** previously registered as the non-default.
  */
  sqlite3_vfs_register(&one, 1);
  assert( sqlite3_vfs_find("__one")==&one );
  assert( sqlite3_vfs_find("__two")==&two );
  assert( sqlite3_vfs_find(0)==&one );
  sqlite3_vfs_register(&two, 1);
................................................................................
}
#endif

#if SQLITE_OS_WIN
/*
**      lock_win32_file FILENAME DELAY1 DELAY2
**
** Get an exclusive manditory lock on file for DELAY2 milliseconds.
** Wait DELAY1 milliseconds before acquiring the lock.
*/
static int SQLITE_TCLAPI win32_file_lock(
  void * clientData,
  Tcl_Interp *interp,
  int objc,
  Tcl_Obj *CONST objv[]







|







 







|







 







|







2636
2637
2638
2639
2640
2641
2642
2643
2644
2645
2646
2647
2648
2649
2650
....
5644
5645
5646
5647
5648
5649
5650
5651
5652
5653
5654
5655
5656
5657
5658
....
6970
6971
6972
6973
6974
6975
6976
6977
6978
6979
6980
6981
6982
6983
6984
  sqlite3_close(db);
  return TCL_OK;
}

/*
** Usage:  sqlite3_next_stmt  DB  STMT
**
** Return the next statement in sequence after STMT.
*/
static int SQLITE_TCLAPI test_next_stmt(
  void * clientData,
  Tcl_Interp *interp,
  int objc,
  Tcl_Obj *CONST objv[]
){
................................................................................
  assert( pMain==0 || pMain==sqlite3_vfs_find(0) );

  /* We can find a VFS by its name */
  assert( sqlite3_vfs_find("__one")==&one );
  assert( sqlite3_vfs_find("__two")==&two );

  /* Calling sqlite_vfs_register with non-zero second parameter changes the
  ** default VFS, even if the 1st parameter is an existing VFS that is
  ** previously registered as the non-default.
  */
  sqlite3_vfs_register(&one, 1);
  assert( sqlite3_vfs_find("__one")==&one );
  assert( sqlite3_vfs_find("__two")==&two );
  assert( sqlite3_vfs_find(0)==&one );
  sqlite3_vfs_register(&two, 1);
................................................................................
}
#endif

#if SQLITE_OS_WIN
/*
**      lock_win32_file FILENAME DELAY1 DELAY2
**
** Get an exclusive mandatory lock on file for DELAY2 milliseconds.
** Wait DELAY1 milliseconds before acquiring the lock.
*/
static int SQLITE_TCLAPI win32_file_lock(
  void * clientData,
  Tcl_Interp *interp,
  int objc,
  Tcl_Obj *CONST objv[]

Changes to src/test_init.c.

34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
#endif

static struct Wrapped {
  sqlite3_pcache_methods2 pcache;
  sqlite3_mem_methods     mem;
  sqlite3_mutex_methods   mutex;

  int mem_init;                /* True if mem subsystem is initalized */
  int mem_fail;                /* True to fail mem subsystem inialization */
  int mutex_init;              /* True if mutex subsystem is initalized */
  int mutex_fail;              /* True to fail mutex subsystem inialization */
  int pcache_init;             /* True if pcache subsystem is initalized */
  int pcache_fail;             /* True to fail pcache subsystem inialization */
} wrapped;

static int wrMemInit(void *pAppData){
  int rc;
  if( wrapped.mem_fail ){
    rc = SQLITE_ERROR;
  }else{







|
|
|
|
|
|







34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
#endif

static struct Wrapped {
  sqlite3_pcache_methods2 pcache;
  sqlite3_mem_methods     mem;
  sqlite3_mutex_methods   mutex;

  int mem_init;                /* True if mem subsystem is initialized */
  int mem_fail;                /* True to fail mem subsystem initalization */
  int mutex_init;              /* True if mutex subsystem is initialized */
  int mutex_fail;              /* True to fail mutex subsystem initalization */
  int pcache_init;             /* True if pcache subsystem is initialized */
  int pcache_fail;             /* True to fail pcache subsystem initalization */
} wrapped;

static int wrMemInit(void *pAppData){
  int rc;
  if( wrapped.mem_fail ){
    rc = SQLITE_ERROR;
  }else{

Changes to src/test_intarray.h.

19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
**
** The intarray virtual table is designed to facilitate using an
** array of integers as the right-hand side of an IN operator.  So
** instead of doing a prepared statement like this:
**
**     SELECT * FROM table WHERE x IN (?,?,?,...,?);
**
** And then binding indivdual integers to each of ? slots, a C-language
** application can create an intarray object (named "ex1" in the following
** example), prepare a statement like this:
**
**     SELECT * FROM table WHERE x IN ex1;
**
** Then bind an ordinary C/C++ array of integer values to the ex1 object
** to run the statement.







|







19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
**
** The intarray virtual table is designed to facilitate using an
** array of integers as the right-hand side of an IN operator.  So
** instead of doing a prepared statement like this:
**
**     SELECT * FROM table WHERE x IN (?,?,?,...,?);
**
** And then binding individual integers to each of ? slots, a C-language
** application can create an intarray object (named "ex1" in the following
** example), prepare a statement like this:
**
**     SELECT * FROM table WHERE x IN ex1;
**
** Then bind an ordinary C/C++ array of integer values to the ex1 object
** to run the statement.

Changes to src/test_multiplex.c.

347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
...
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
    }
  }
  return pSubOpen;
}

/*
** Return the size, in bytes, of chunk number iChunk.  If that chunk
** does not exist, then return 0.  This function does not distingish between
** non-existant files and zero-length files.
*/
static sqlite3_int64 multiplexSubSize(
  multiplexGroup *pGroup,    /* The multiplexor group */
  int iChunk,                /* Which chunk to open.  0==original file */
  int *rc                    /* Result code in and out */
){
................................................................................
          /* If the first overflow file exists and if the size of the main file
          ** is different from the chunk size, that means the chunk size is set
          ** set incorrectly.  So fix it.
          **
          ** Or, if the first overflow file does not exist and the main file is
          ** larger than the chunk size, that means the chunk size is too small.
          ** But we have no way of determining the intended chunk size, so 
          ** just disable the multiplexor all togethre.
          */
          rc = pOrigVfs->xAccess(pOrigVfs, pGroup->aReal[1].z,
              SQLITE_ACCESS_EXISTS, &bExists);
          bExists = multiplexSubSize(pGroup, 1, &rc)>0;
          if( rc==SQLITE_OK && bExists && sz64==(sz64&0xffff0000) && sz64>0
              && sz64!=pGroup->szChunk ){
            pGroup->szChunk = (int)sz64;







|







 







|







347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
...
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
    }
  }
  return pSubOpen;
}

/*
** Return the size, in bytes, of chunk number iChunk.  If that chunk
** does not exist, then return 0.  This function does not distinguish between
** non-existant files and zero-length files.
*/
static sqlite3_int64 multiplexSubSize(
  multiplexGroup *pGroup,    /* The multiplexor group */
  int iChunk,                /* Which chunk to open.  0==original file */
  int *rc                    /* Result code in and out */
){
................................................................................
          /* If the first overflow file exists and if the size of the main file
          ** is different from the chunk size, that means the chunk size is set
          ** set incorrectly.  So fix it.
          **
          ** Or, if the first overflow file does not exist and the main file is
          ** larger than the chunk size, that means the chunk size is too small.
          ** But we have no way of determining the intended chunk size, so 
          ** just disable the multiplexor all together.
          */
          rc = pOrigVfs->xAccess(pOrigVfs, pGroup->aReal[1].z,
              SQLITE_ACCESS_EXISTS, &bExists);
          bExists = multiplexSubSize(pGroup, 1, &rc)>0;
          if( rc==SQLITE_OK && bExists && sz64==(sz64&0xffff0000) && sz64>0
              && sz64!=pGroup->szChunk ){
            pGroup->szChunk = (int)sz64;

Changes to src/test_mutex.c.

41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
  sqlite3_mutex *pReal;
  int eType;
};

/* State variables */
static struct test_mutex_globals {
  int isInstalled;           /* True if installed */
  int disableInit;           /* True to cause sqlite3_initalize() to fail */
  int disableTry;            /* True to force sqlite3_mutex_try() to fail */
  int isInit;                /* True if initialized */
  sqlite3_mutex_methods m;   /* Interface to "real" mutex system */
  int aCounter[MAX_MUTEXES]; /* Number of grabs of each type of mutex */
  sqlite3_mutex aStatic[STATIC_MUTEXES]; /* The static mutexes */
} g = {0};








|







41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
  sqlite3_mutex *pReal;
  int eType;
};

/* State variables */
static struct test_mutex_globals {
  int isInstalled;           /* True if installed */
  int disableInit;           /* True to cause sqlite3_initialize() to fail */
  int disableTry;            /* True to force sqlite3_mutex_try() to fail */
  int isInit;                /* True if initialized */
  sqlite3_mutex_methods m;   /* Interface to "real" mutex system */
  int aCounter[MAX_MUTEXES]; /* Number of grabs of each type of mutex */
  sqlite3_mutex aStatic[STATIC_MUTEXES]; /* The static mutexes */
} g = {0};

Changes to src/test_pcache.c.

32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
...
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
*/
typedef struct testpcacheGlobalType testpcacheGlobalType;
struct testpcacheGlobalType {
  void *pDummy;             /* Dummy allocation to simulate failures */
  int nInstance;            /* Number of current instances */
  unsigned discardChance;   /* Chance of discarding on an unpin (0-100) */
  unsigned prngSeed;        /* Seed for the PRNG */
  unsigned highStress;      /* Call xStress agressively */
};
static testpcacheGlobalType testpcacheGlobal;

/*
** Initializer.
**
** Verify that the initializer is only called when the system is
................................................................................
** page.  0 means never discard (unless the discard flag is set).
** 100 means always discard.
*/
void installTestPCache(
  int installFlag,            /* True to install.  False to uninstall. */
  unsigned discardChance,     /* 0-100.  Chance to discard on unpin */
  unsigned prngSeed,          /* Seed for the PRNG */
  unsigned highStress         /* Call xStress agressively */
){
  static const sqlite3_pcache_methods2 testPcache = {
    1,
    (void*)&testpcacheGlobal,
    testpcacheInit,
    testpcacheShutdown,
    testpcacheCreate,







|







 







|







32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
...
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
*/
typedef struct testpcacheGlobalType testpcacheGlobalType;
struct testpcacheGlobalType {
  void *pDummy;             /* Dummy allocation to simulate failures */
  int nInstance;            /* Number of current instances */
  unsigned discardChance;   /* Chance of discarding on an unpin (0-100) */
  unsigned prngSeed;        /* Seed for the PRNG */
  unsigned highStress;      /* Call xStress aggressively */
};
static testpcacheGlobalType testpcacheGlobal;

/*
** Initializer.
**
** Verify that the initializer is only called when the system is
................................................................................
** page.  0 means never discard (unless the discard flag is set).
** 100 means always discard.
*/
void installTestPCache(
  int installFlag,            /* True to install.  False to uninstall. */
  unsigned discardChance,     /* 0-100.  Chance to discard on unpin */
  unsigned prngSeed,          /* Seed for the PRNG */
  unsigned highStress         /* Call xStress aggressively */
){
  static const sqlite3_pcache_methods2 testPcache = {
    1,
    (void*)&testpcacheGlobal,
    testpcacheInit,
    testpcacheShutdown,
    testpcacheCreate,

Changes to src/test_quota.c.

995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
....
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
** Write content into a quota_FILE.  Invoke the quota callback and block
** the write if we exceed quota.
*/
size_t sqlite3_quota_fwrite(
  const void *pBuf,      /* Take content to write from here */
  size_t size,           /* Size of each element */
  size_t nmemb,          /* Number of elements */
  quota_FILE *p          /* Write to this quota_FILE objecct */
){
  sqlite3_int64 iOfst;
  sqlite3_int64 iEnd;
  sqlite3_int64 szNew;
  quotaFile *pFile;
  size_t rc;

................................................................................
static int SQLITE_TCLAPI test_quota_glob(
  void * clientData,
  Tcl_Interp *interp,
  int objc,
  Tcl_Obj *CONST objv[]
){
  const char *zPattern;          /* The glob pattern */
  const char *zText;             /* Text to compare agains the pattern */
  int rc;
  if( objc!=3 ){
    Tcl_WrongNumArgs(interp, 1, objv, "PATTERN TEXT");
    return TCL_ERROR;
  }
  zPattern = Tcl_GetString(objv[1]);
  zText = Tcl_GetString(objv[2]);







|







 







|







995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
....
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
** Write content into a quota_FILE.  Invoke the quota callback and block
** the write if we exceed quota.
*/
size_t sqlite3_quota_fwrite(
  const void *pBuf,      /* Take content to write from here */
  size_t size,           /* Size of each element */
  size_t nmemb,          /* Number of elements */
  quota_FILE *p          /* Write to this quota_FILE object */
){
  sqlite3_int64 iOfst;
  sqlite3_int64 iEnd;
  sqlite3_int64 szNew;
  quotaFile *pFile;
  size_t rc;

................................................................................
static int SQLITE_TCLAPI test_quota_glob(
  void * clientData,
  Tcl_Interp *interp,
  int objc,
  Tcl_Obj *CONST objv[]
){
  const char *zPattern;          /* The glob pattern */
  const char *zText;             /* Text to compare against the pattern */
  int rc;
  if( objc!=3 ){
    Tcl_WrongNumArgs(interp, 1, objv, "PATTERN TEXT");
    return TCL_ERROR;
  }
  zPattern = Tcl_GetString(objv[1]);
  zText = Tcl_GetString(objv[2]);

Changes to src/test_server.c.

145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
...
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
**** Restrictions Associated With SQLITE_ENABLE_MEMORY_MANAGEMENT ****
**
** If you compile with SQLITE_ENABLE_MEMORY_MANAGEMENT defined, then
** SQLite includes code that tracks how much memory is being used by
** each thread.  These memory counts can become confused if memory
** is allocated by one thread and then freed by another.  For that
** reason, when SQLITE_ENABLE_MEMORY_MANAGEMENT is used, all operations
** that might allocate or free memory should be performanced in the same
** thread that originally created the database connection.  In that case,
** many of the operations that are listed above as safe to be performed
** in separate threads would need to be sent over to the server to be
** done there.  If SQLITE_ENABLE_MEMORY_MANAGEMENT is defined, then
** the following functions can be used safely from different threads
** without messing up the allocation counts:
**
................................................................................
  }
  pthread_mutex_unlock(&g.serverMutex);
  return 0;
}

/*
** Start a server thread if one is not already running.  If there
** is aleady a server thread running, the new thread will quickly
** die and this routine is effectively a no-op.
*/
void sqlite3_server_start(void){
  pthread_t x;
  int rc;
  g.serverHalt = 0;
  rc = pthread_create(&x, 0, sqlite3_server, 0);







|







 







|







145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
...
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
**** Restrictions Associated With SQLITE_ENABLE_MEMORY_MANAGEMENT ****
**
** If you compile with SQLITE_ENABLE_MEMORY_MANAGEMENT defined, then
** SQLite includes code that tracks how much memory is being used by
** each thread.  These memory counts can become confused if memory
** is allocated by one thread and then freed by another.  For that
** reason, when SQLITE_ENABLE_MEMORY_MANAGEMENT is used, all operations
** that might allocate or free memory should be performed in the same
** thread that originally created the database connection.  In that case,
** many of the operations that are listed above as safe to be performed
** in separate threads would need to be sent over to the server to be
** done there.  If SQLITE_ENABLE_MEMORY_MANAGEMENT is defined, then
** the following functions can be used safely from different threads
** without messing up the allocation counts:
**
................................................................................
  }
  pthread_mutex_unlock(&g.serverMutex);
  return 0;
}

/*
** Start a server thread if one is not already running.  If there
** is already a server thread running, the new thread will quickly
** die and this routine is effectively a no-op.
*/
void sqlite3_server_start(void){
  pthread_t x;
  int rc;
  g.serverHalt = 0;
  rc = pthread_create(&x, 0, sqlite3_server, 0);

Changes to src/test_superlock.c.

87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
** of the file fd. If the lock cannot be obtained immediately, invoke
** the busy-handler until either it is obtained or the busy-handler
** callback returns 0.
*/
static int superlockShmLock(
  sqlite3_file *fd,               /* Database file handle */
  int idx,                        /* Offset of shm-lock to obtain */
  int nByte,                      /* Number of consective bytes to lock */
  SuperlockBusy *pBusy            /* Busy-handler wrapper object */
){
  int rc;
  int (*xShmLock)(sqlite3_file*, int, int, int) = fd->pMethods->xShmLock;
  do {
    rc = xShmLock(fd, idx, nByte, SQLITE_SHM_LOCK|SQLITE_SHM_EXCLUSIVE);
  }while( rc==SQLITE_BUSY && superlockBusyHandler((void *)pBusy, 0) );







|







87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
** of the file fd. If the lock cannot be obtained immediately, invoke
** the busy-handler until either it is obtained or the busy-handler
** callback returns 0.
*/
static int superlockShmLock(
  sqlite3_file *fd,               /* Database file handle */
  int idx,                        /* Offset of shm-lock to obtain */
  int nByte,                      /* Number of consecutive bytes to lock */
  SuperlockBusy *pBusy            /* Busy-handler wrapper object */
){
  int rc;
  int (*xShmLock)(sqlite3_file*, int, int, int) = fd->pMethods->xShmLock;
  do {
    rc = xShmLock(fd, idx, nByte, SQLITE_SHM_LOCK|SQLITE_SHM_EXCLUSIVE);
  }while( rc==SQLITE_BUSY && superlockBusyHandler((void *)pBusy, 0) );

Changes to src/test_windirent.h.

136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
*/

#ifndef is_filtered
#  define is_filtered(a) ((((a).attrib)&_A_HIDDEN) || (((a).attrib)&_A_SYSTEM))
#endif

/*
** Provide the function prototype for the POSIX compatiable getenv()
** function.  This function is not thread-safe.
*/

extern const char *windirent_getenv(const char *name);

/*
** Finally, we can provide the function prototypes for the opendir(),







|







136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
*/

#ifndef is_filtered
#  define is_filtered(a) ((((a).attrib)&_A_HIDDEN) || (((a).attrib)&_A_SYSTEM))
#endif

/*
** Provide the function prototype for the POSIX compatible getenv()
** function.  This function is not thread-safe.
*/

extern const char *windirent_getenv(const char *name);

/*
** Finally, we can provide the function prototypes for the opendir(),

Changes to src/update.c.

1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287

  if( eOnePass==ONEPASS_OFF ){
    /* End the virtual table scan */
    if( pSrc->nSrc==1 ){
      sqlite3WhereEnd(pWInfo);
    }

    /* Begin scannning through the ephemeral table. */
    addr = sqlite3VdbeAddOp1(v, OP_Rewind, ephemTab); VdbeCoverage(v);

    /* Extract arguments from the current row of the ephemeral table and 
    ** invoke the VUpdate method.  */
    for(i=0; i<nArg; i++){
      sqlite3VdbeAddOp3(v, OP_Column, ephemTab, i, regArg+i);
    }







|







1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287

  if( eOnePass==ONEPASS_OFF ){
    /* End the virtual table scan */
    if( pSrc->nSrc==1 ){
      sqlite3WhereEnd(pWInfo);
    }

    /* Begin scanning through the ephemeral table. */
    addr = sqlite3VdbeAddOp1(v, OP_Rewind, ephemTab); VdbeCoverage(v);

    /* Extract arguments from the current row of the ephemeral table and 
    ** invoke the VUpdate method.  */
    for(i=0; i<nArg; i++){
      sqlite3VdbeAddOp3(v, OP_Column, ephemTab, i, regArg+i);
    }

Changes to src/util.c.

30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
...
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
...
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
  dummy += (unsigned)x;
}
#endif

/*
** Calls to sqlite3FaultSim() are used to simulate a failure during testing,
** or to bypass normal error detection during testing in order to let 
** execute proceed futher downstream.
**
** In deployment, sqlite3FaultSim() *always* return SQLITE_OK (0).  The
** sqlite3FaultSim() function only returns non-zero during testing.
**
** During testing, if the test harness has set a fault-sim callback using
** a call to sqlite3_test_control(SQLITE_TESTCTRL_FAULT_INSTALL), then
** each call to sqlite3FaultSim() is relayed to that application-supplied
................................................................................
      }
    }
  }

  /* store the result */
  *pResult = result;

  /* return true if number and no extra non-whitespace chracters after */
  if( z==zEnd && nDigit>0 && eValid && eType>0 ){
    return eType;
  }else if( eType>=2 && (eType==3 || eValid) && nDigit>0 ){
    return -1;
  }else{
    return 0;
  }
................................................................................
** The caller must ensure that zOut[] is at least 21 bytes in size.
*/
void sqlite3Int64ToText(i64 v, char *zOut){
  int i;
  u64 x;
  char zTemp[22];
  if( v<0 ){
    x = (v==SMALLEST_INT64) ? ((u64)1)<<63 : -v;
  }else{
    x = v;
  }
  i = sizeof(zTemp)-2;
  zTemp[sizeof(zTemp)-1] = 0;
  do{
    zTemp[i--] = (x%10) + '0';







|







 







|







 







|







30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
...
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
...
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
  dummy += (unsigned)x;
}
#endif

/*
** Calls to sqlite3FaultSim() are used to simulate a failure during testing,
** or to bypass normal error detection during testing in order to let 
** execute proceed further downstream.
**
** In deployment, sqlite3FaultSim() *always* return SQLITE_OK (0).  The
** sqlite3FaultSim() function only returns non-zero during testing.
**
** During testing, if the test harness has set a fault-sim callback using
** a call to sqlite3_test_control(SQLITE_TESTCTRL_FAULT_INSTALL), then
** each call to sqlite3FaultSim() is relayed to that application-supplied
................................................................................
      }
    }
  }

  /* store the result */
  *pResult = result;

  /* return true if number and no extra non-whitespace characters after */
  if( z==zEnd && nDigit>0 && eValid && eType>0 ){
    return eType;
  }else if( eType>=2 && (eType==3 || eValid) && nDigit>0 ){
    return -1;
  }else{
    return 0;
  }
................................................................................
** The caller must ensure that zOut[] is at least 21 bytes in size.
*/
void sqlite3Int64ToText(i64 v, char *zOut){
  int i;
  u64 x;
  char zTemp[22];
  if( v<0 ){
    x = (v==SMALLEST_INT64) ? ((u64)1)<<63 : (u64)-v;
  }else{
    x = v;
  }
  i = sizeof(zTemp)-2;
  zTemp[sizeof(zTemp)-1] = 0;
  do{
    zTemp[i--] = (x%10) + '0';

Changes to src/vdbe.c.

863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
....
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
....
3937
3938
3939
3940
3941
3942
3943
3944
3945
3946
3947
3948
3949
3950
3951
....
5063
5064
5065
5066
5067
5068
5069
5070
5071
5072
5073
5074
5075
5076
5077
....
6808
6809
6810
6811
6812
6813
6814
6815
6816
6817
6818
6819
6820
6821
6822
....
7660
7661
7662
7663
7664
7665
7666
7667
7668
7669
7670
7671
7672
7673
7674
....
7678
7679
7680
7681
7682
7683
7684
7685
7686
7687
7688
7689
7690
7691
7692
....
7959
7960
7961
7962
7963
7964
7965
7966
7967
7968
7969
7970
7971
7972
7973
** is sometimes set to 1 instead of 0 as a hint to the command-line shell
** that this Goto is the bottom of a loop and that the lines from P2 down
** to the current line should be indented for EXPLAIN output.
*/
case OP_Goto: {             /* jump */

#ifdef SQLITE_DEBUG
  /* In debuggging mode, when the p5 flags is set on an OP_Goto, that
  ** means we should really jump back to the preceeding OP_ReleaseReg
  ** instruction. */
  if( pOp->p5 ){
    assert( pOp->p2 < (int)(pOp - aOp) );
    assert( pOp->p2 > 1 );
    pOp = &aOp[pOp->p2 - 2];
    assert( pOp[1].opcode==OP_ReleaseReg );
................................................................................
** VDBE, but do not rollback the transaction. 
**
** If P4 is not null then it is an error message string.
**
** P5 is a value between 0 and 4, inclusive, that modifies the P4 string.
**
**    0:  (no change)
**    1:  NOT NULL contraint failed: P4
**    2:  UNIQUE constraint failed: P4
**    3:  CHECK constraint failed: P4
**    4:  FOREIGN KEY constraint failed: P4
**
** If P5 is not zero and P4 is NULL, then everything after the ":" is
** omitted.
**
................................................................................
      SQLITE_OPEN_EXCLUSIVE |
      SQLITE_OPEN_DELETEONCLOSE |
      SQLITE_OPEN_TRANSIENT_DB;
  assert( pOp->p1>=0 );
  assert( pOp->p2>=0 );
  pCx = p->apCsr[pOp->p1];
  if( pCx && pCx->pBtx ){
    /* If the ephermeral table is already open, erase all existing content
    ** so that the table is empty again, rather than creating a new table. */
    assert( pCx->isEphemeral );
    pCx->seqCount = 0;
    pCx->cacheStatus = CACHE_STALE;
    rc = sqlite3BtreeClearTable(pCx->pBtx, pCx->pgnoRoot, 0);
  }else{
    pCx = allocateCursor(p, pOp->p1, pOp->p2, -1, CURTYPE_BTREE);
................................................................................
        zDb, pTab, pC->movetoTarget,
        pOp->p3
    );
  }
  if( opflags & OPFLAG_ISNOOP ) break;
#endif
 
  /* Only flags that can be set are SAVEPOISTION and AUXDELETE */ 
  assert( (pOp->p5 & ~(OPFLAG_SAVEPOSITION|OPFLAG_AUXDELETE))==0 );
  assert( OPFLAG_SAVEPOSITION==BTREE_SAVEPOSITION );
  assert( OPFLAG_AUXDELETE==BTREE_AUXDELETE );

#ifdef SQLITE_DEBUG
  if( p->pFrame==0 ){
    if( pC->isEphemeral==0
................................................................................
    pMem->uTemp = 0x1122e0e3;
  }
#endif

  /* If this function is inside of a trigger, the register array in aMem[]
  ** might change from one evaluation to the next.  The next block of code
  ** checks to see if the register array has changed, and if so it
  ** reinitializes the relavant parts of the sqlite3_context object */
  if( pCtx->pMem != pMem ){
    pCtx->pMem = pMem;
    for(i=pCtx->argc-1; i>=0; i--) pCtx->argv[i] = &aMem[pOp->p2+i];
  }

#ifdef SQLITE_DEBUG
  for(i=0; i<pCtx->argc; i++){
................................................................................
** whether meta data associated with a user function argument using the
** sqlite3_set_auxdata() API may be safely retained until the next
** invocation of this opcode.
**
** This opcode works exactly like OP_Function.  The only difference is in
** its name.  This opcode is used in places where the function must be
** purely non-deterministic.  Some built-in date/time functions can be
** either determinitic of non-deterministic, depending on their arguments.
** When those function are used in a non-deterministic way, they will check
** to see if they were called using OP_PureFunc instead of OP_Function, and
** if they were, they throw an error.
**
** See also: AggStep, AggFinal, Function
*/
case OP_PureFunc:              /* group */
................................................................................

  assert( pOp->p4type==P4_FUNCCTX );
  pCtx = pOp->p4.pCtx;

  /* If this function is inside of a trigger, the register array in aMem[]
  ** might change from one evaluation to the next.  The next block of code
  ** checks to see if the register array has changed, and if so it
  ** reinitializes the relavant parts of the sqlite3_context object */
  pOut = &aMem[pOp->p3];
  if( pCtx->pOut != pOut ){
    pCtx->pVdbe = p;
    pCtx->pOut = pOut;
    for(i=pCtx->argc-1; i>=0; i--) pCtx->argv[i] = &aMem[pOp->p2+i];
  }
  assert( pCtx->pVdbe==p );
................................................................................
        registerTrace(pOrigOp->p2, &aMem[pOrigOp->p2]);
      }
      if( opProperty & OPFLG_OUT3 ){
        registerTrace(pOrigOp->p3, &aMem[pOrigOp->p3]);
      }
      if( opProperty==0xff ){
        /* Never happens.  This code exists to avoid a harmless linkage
        ** warning aboud sqlite3VdbeRegisterDump() being defined but not
        ** used. */
        sqlite3VdbeRegisterDump(p);
      }
    }
#endif  /* SQLITE_DEBUG */
#endif  /* NDEBUG */
  }  /* The end of the for(;;) loop the loops through opcodes */







|







 







|







 







|







 







|







 







|







 







|







 







|







 







|







863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
....
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
....
3937
3938
3939
3940
3941
3942
3943
3944
3945
3946
3947
3948
3949
3950
3951
....
5063
5064
5065
5066
5067
5068
5069
5070
5071
5072
5073
5074
5075
5076
5077
....
6808
6809
6810
6811
6812
6813
6814
6815
6816
6817
6818
6819
6820
6821
6822
....
7660
7661
7662
7663
7664
7665
7666
7667
7668
7669
7670
7671
7672
7673
7674
....
7678
7679
7680
7681
7682
7683
7684
7685
7686
7687
7688
7689
7690
7691
7692
....
7959
7960
7961
7962
7963
7964
7965
7966
7967
7968
7969
7970
7971
7972
7973
** is sometimes set to 1 instead of 0 as a hint to the command-line shell
** that this Goto is the bottom of a loop and that the lines from P2 down
** to the current line should be indented for EXPLAIN output.
*/
case OP_Goto: {             /* jump */

#ifdef SQLITE_DEBUG
  /* In debugging mode, when the p5 flags is set on an OP_Goto, that
  ** means we should really jump back to the preceeding OP_ReleaseReg
  ** instruction. */
  if( pOp->p5 ){
    assert( pOp->p2 < (int)(pOp - aOp) );
    assert( pOp->p2 > 1 );
    pOp = &aOp[pOp->p2 - 2];
    assert( pOp[1].opcode==OP_ReleaseReg );
................................................................................
** VDBE, but do not rollback the transaction. 
**
** If P4 is not null then it is an error message string.
**
** P5 is a value between 0 and 4, inclusive, that modifies the P4 string.
**
**    0:  (no change)
**    1:  NOT NULL constraint failed: P4
**    2:  UNIQUE constraint failed: P4
**    3:  CHECK constraint failed: P4
**    4:  FOREIGN KEY constraint failed: P4
**
** If P5 is not zero and P4 is NULL, then everything after the ":" is
** omitted.
**
................................................................................
      SQLITE_OPEN_EXCLUSIVE |
      SQLITE_OPEN_DELETEONCLOSE |
      SQLITE_OPEN_TRANSIENT_DB;
  assert( pOp->p1>=0 );
  assert( pOp->p2>=0 );
  pCx = p->apCsr[pOp->p1];
  if( pCx && pCx->pBtx ){
    /* If the ephemeral table is already open, erase all existing content
    ** so that the table is empty again, rather than creating a new table. */
    assert( pCx->isEphemeral );
    pCx->seqCount = 0;
    pCx->cacheStatus = CACHE_STALE;
    rc = sqlite3BtreeClearTable(pCx->pBtx, pCx->pgnoRoot, 0);
  }else{
    pCx = allocateCursor(p, pOp->p1, pOp->p2, -1, CURTYPE_BTREE);
................................................................................
        zDb, pTab, pC->movetoTarget,
        pOp->p3
    );
  }
  if( opflags & OPFLAG_ISNOOP ) break;
#endif
 
  /* Only flags that can be set are SAVEPOSITION and AUXDELETE */ 
  assert( (pOp->p5 & ~(OPFLAG_SAVEPOSITION|OPFLAG_AUXDELETE))==0 );
  assert( OPFLAG_SAVEPOSITION==BTREE_SAVEPOSITION );
  assert( OPFLAG_AUXDELETE==BTREE_AUXDELETE );

#ifdef SQLITE_DEBUG
  if( p->pFrame==0 ){
    if( pC->isEphemeral==0
................................................................................
    pMem->uTemp = 0x1122e0e3;
  }
#endif

  /* If this function is inside of a trigger, the register array in aMem[]
  ** might change from one evaluation to the next.  The next block of code
  ** checks to see if the register array has changed, and if so it
  ** reinitializes the relevant parts of the sqlite3_context object */
  if( pCtx->pMem != pMem ){
    pCtx->pMem = pMem;
    for(i=pCtx->argc-1; i>=0; i--) pCtx->argv[i] = &aMem[pOp->p2+i];
  }

#ifdef SQLITE_DEBUG
  for(i=0; i<pCtx->argc; i++){
................................................................................
** whether meta data associated with a user function argument using the
** sqlite3_set_auxdata() API may be safely retained until the next
** invocation of this opcode.
**
** This opcode works exactly like OP_Function.  The only difference is in
** its name.  This opcode is used in places where the function must be
** purely non-deterministic.  Some built-in date/time functions can be
** either deterministic or non-deterministic, depending on their arguments.
** When those function are used in a non-deterministic way, they will check
** to see if they were called using OP_PureFunc instead of OP_Function, and
** if they were, they throw an error.
**
** See also: AggStep, AggFinal, Function
*/
case OP_PureFunc:              /* group */
................................................................................

  assert( pOp->p4type==P4_FUNCCTX );
  pCtx = pOp->p4.pCtx;

  /* If this function is inside of a trigger, the register array in aMem[]
  ** might change from one evaluation to the next.  The next block of code
  ** checks to see if the register array has changed, and if so it
  ** reinitializes the relevant parts of the sqlite3_context object */
  pOut = &aMem[pOp->p3];
  if( pCtx->pOut != pOut ){
    pCtx->pVdbe = p;
    pCtx->pOut = pOut;
    for(i=pCtx->argc-1; i>=0; i--) pCtx->argv[i] = &aMem[pOp->p2+i];
  }
  assert( pCtx->pVdbe==p );
................................................................................
        registerTrace(pOrigOp->p2, &aMem[pOrigOp->p2]);
      }
      if( opProperty & OPFLG_OUT3 ){
        registerTrace(pOrigOp->p3, &aMem[pOrigOp->p3]);
      }
      if( opProperty==0xff ){
        /* Never happens.  This code exists to avoid a harmless linkage
        ** warning about sqlite3VdbeRegisterDump() being defined but not
        ** used. */
        sqlite3VdbeRegisterDump(p);
      }
    }
#endif  /* SQLITE_DEBUG */
#endif  /* NDEBUG */
  }  /* The end of the for(;;) loop the loops through opcodes */

Changes to src/vdbe.h.

317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
...
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
# define VdbeModuleComment(X)
#endif

/*
** The VdbeCoverage macros are used to set a coverage testing point
** for VDBE branch instructions.  The coverage testing points are line
** numbers in the sqlite3.c source file.  VDBE branch coverage testing
** only works with an amalagmation build.  That's ok since a VDBE branch
** coverage build designed for testing the test suite only.  No application
** should ever ship with VDBE branch coverage measuring turned on.
**
**    VdbeCoverage(v)                  // Mark the previously coded instruction
**                                     // as a branch
**
**    VdbeCoverageIf(v, conditional)   // Mark previous if conditional true
................................................................................
**    VdbeCoverageNeverTaken(v)        // Previous branch is never taken
**
**    VdbeCoverageNeverNull(v)         // Previous three-way branch is only
**                                     // taken on the first two ways.  The
**                                     // NULL option is not possible
**
**    VdbeCoverageEqNe(v)              // Previous OP_Jump is only interested
**                                     // in distingishing equal and not-equal.
**
** Every VDBE branch operation must be tagged with one of the macros above.
** If not, then when "make test" is run with -DSQLITE_VDBE_COVERAGE and
** -DSQLITE_DEBUG then an ALWAYS() will fail in the vdbeTakeBranch()
** routine in vdbe.c, alerting the developer to the missed tag.
**
** During testing, the test application will invoke







|







 







|







317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
...
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
# define VdbeModuleComment(X)
#endif

/*
** The VdbeCoverage macros are used to set a coverage testing point
** for VDBE branch instructions.  The coverage testing points are line
** numbers in the sqlite3.c source file.  VDBE branch coverage testing
** only works with an amalgamation build.  That's ok since a VDBE branch
** coverage build designed for testing the test suite only.  No application
** should ever ship with VDBE branch coverage measuring turned on.
**
**    VdbeCoverage(v)                  // Mark the previously coded instruction
**                                     // as a branch
**
**    VdbeCoverageIf(v, conditional)   // Mark previous if conditional true
................................................................................
**    VdbeCoverageNeverTaken(v)        // Previous branch is never taken
**
**    VdbeCoverageNeverNull(v)         // Previous three-way branch is only
**                                     // taken on the first two ways.  The
**                                     // NULL option is not possible
**
**    VdbeCoverageEqNe(v)              // Previous OP_Jump is only interested
**                                     // in distinguishing equal and not-equal.
**
** Every VDBE branch operation must be tagged with one of the macros above.
** If not, then when "make test" is run with -DSQLITE_VDBE_COVERAGE and
** -DSQLITE_DEBUG then an ALWAYS() will fail in the vdbeTakeBranch()
** routine in vdbe.c, alerting the developer to the missed tag.
**
** During testing, the test application will invoke

Changes to src/vdbeapi.c.

1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
....
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
**     sqlite3_column_int()
**     sqlite3_column_int64()
**     sqlite3_column_text()
**     sqlite3_column_text16()
**     sqlite3_column_real()
**     sqlite3_column_bytes()
**     sqlite3_column_bytes16()
**     sqiite3_column_blob()
*/
static void columnMallocFailure(sqlite3_stmt *pStmt)
{
  /* If malloc() failed during an encoding conversion within an
  ** sqlite3_column_XXX API, then set the return code of the statement to
  ** SQLITE_NOMEM. The next call to _step() (if any) will return SQLITE_ERROR
  ** and _finalize() will return NOMEM.
................................................................................
/******************************* sqlite3_bind_  ***************************
** 
** Routines used to attach values to wildcards in a compiled SQL statement.
*/
/*
** Unbind the value bound to variable i in virtual machine p. This is the 
** the same as binding a NULL value to the column. If the "i" parameter is
** out of range, then SQLITE_RANGE is returned. Othewise SQLITE_OK.
**
** A successful evaluation of this routine acquires the mutex on p.
** the mutex is released if any kind of error occurs.
**
** The error code stored in database p->db is overwritten with the return
** value in any case.
*/







|







 







|







1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
....
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
**     sqlite3_column_int()
**     sqlite3_column_int64()
**     sqlite3_column_text()
**     sqlite3_column_text16()
**     sqlite3_column_real()
**     sqlite3_column_bytes()
**     sqlite3_column_bytes16()
**     sqlite3_column_blob()
*/
static void columnMallocFailure(sqlite3_stmt *pStmt)
{
  /* If malloc() failed during an encoding conversion within an
  ** sqlite3_column_XXX API, then set the return code of the statement to
  ** SQLITE_NOMEM. The next call to _step() (if any) will return SQLITE_ERROR
  ** and _finalize() will return NOMEM.
................................................................................
/******************************* sqlite3_bind_  ***************************
** 
** Routines used to attach values to wildcards in a compiled SQL statement.
*/
/*
** Unbind the value bound to variable i in virtual machine p. This is the 
** the same as binding a NULL value to the column. If the "i" parameter is
** out of range, then SQLITE_RANGE is returned. Otherwise SQLITE_OK.
**
** A successful evaluation of this routine acquires the mutex on p.
** the mutex is released if any kind of error occurs.
**
** The error code stored in database p->db is overwritten with the return
** value in any case.
*/

Changes to src/vdbeaux.c.

290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
....
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
....
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
....
4545
4546
4547
4548
4549
4550
4551
4552
4553
4554
4555
4556
4557
4558
4559
....
5014
5015
5016
5017
5018
5019
5020
5021
5022
5023
5024
5025
5026
5027
5028
int sqlite3VdbeLoadString(Vdbe *p, int iDest, const char *zStr){
  return sqlite3VdbeAddOp4(p, OP_String8, 0, iDest, 0, zStr, 0);
}

/*
** Generate code that initializes multiple registers to string or integer
** constants.  The registers begin with iDest and increase consecutively.
** One register is initialized for each characgter in zTypes[].  For each
** "s" character in zTypes[], the register is a string if the argument is
** not NULL, or OP_Null if the value is a null pointer.  For each "i" character
** in zTypes[], the register is initialized to an integer.
**
** If the input string does not end with "X" then an OP_ResultRow instruction
** is generated for the values inserted.
*/
................................................................................
    sqlite3VdbeChangeP2(p, addr, p->nOp);
  }
}


/*
** If the input FuncDef structure is ephemeral, then free it.  If
** the FuncDef is not ephermal, then do nothing.
*/
static void freeEphemeralFunction(sqlite3 *db, FuncDef *pDef){
  if( (pDef->funcFlags & SQLITE_FUNC_EPHEM)!=0 ){
    sqlite3DbFreeNN(db, pDef);
  }
}

................................................................................
  ** the sum of the number of rows in all trigger subprograms encountered
  ** so far.  The nRow value will increase as new trigger subprograms are
  ** encountered, but p->pc will eventually catch up to nRow.
  */
  nRow = p->nOp;
  if( pSub!=0 ){
    if( pSub->flags&MEM_Blob ){
      /* pSub is initiallly NULL.  It is initialized to a BLOB by
      ** the P4_SUBPROGRAM processing logic below */
      nSub = pSub->n/sizeof(Vdbe*);
      apSub = (SubProgram **)pSub->z;
    }
    for(i=0; i<nSub; i++){
      nRow += apSub[i]->nOp;
    }
................................................................................
    /* RHS is real */
    else if( pRhs->flags & MEM_Real ){
      serial_type = aKey1[idx1];
      if( serial_type>=10 ){
        /* Serial types 12 or greater are strings and blobs (greater than
        ** numbers). Types 10 and 11 are currently "reserved for future 
        ** use", so it doesn't really matter what the results of comparing
        ** them to numberic values are.  */
        rc = +1;
      }else if( serial_type==0 ){
        rc = -1;
      }else{
        sqlite3VdbeSerialGet(&aKey1[d1], serial_type, &mem1);
        if( serial_type==7 ){
          if( mem1.u.r<pRhs->u.r ){
................................................................................
void sqlite3VdbeSetChanges(sqlite3 *db, int nChange){
  assert( sqlite3_mutex_held(db->mutex) );
  db->nChange = nChange;
  db->nTotalChange += nChange;
}

/*
** Set a flag in the vdbe to update the change counter when it is finalised
** or reset.
*/
void sqlite3VdbeCountChanges(Vdbe *v){
  v->changeCntOn = 1;
}

/*







|







 







|







 







|







 







|







 







|







290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
....
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
....
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
....
4545
4546
4547
4548
4549
4550
4551
4552
4553
4554
4555
4556
4557
4558
4559
....
5014
5015
5016
5017
5018
5019
5020
5021
5022
5023
5024
5025
5026
5027
5028
int sqlite3VdbeLoadString(Vdbe *p, int iDest, const char *zStr){
  return sqlite3VdbeAddOp4(p, OP_String8, 0, iDest, 0, zStr, 0);
}

/*
** Generate code that initializes multiple registers to string or integer
** constants.  The registers begin with iDest and increase consecutively.
** One register is initialized for each character in zTypes[].  For each
** "s" character in zTypes[], the register is a string if the argument is
** not NULL, or OP_Null if the value is a null pointer.  For each "i" character
** in zTypes[], the register is initialized to an integer.
**
** If the input string does not end with "X" then an OP_ResultRow instruction
** is generated for the values inserted.
*/
................................................................................
    sqlite3VdbeChangeP2(p, addr, p->nOp);
  }
}


/*
** If the input FuncDef structure is ephemeral, then free it.  If
** the FuncDef is not ephemeral, then do nothing.
*/
static void freeEphemeralFunction(sqlite3 *db, FuncDef *pDef){
  if( (pDef->funcFlags & SQLITE_FUNC_EPHEM)!=0 ){
    sqlite3DbFreeNN(db, pDef);
  }
}

................................................................................
  ** the sum of the number of rows in all trigger subprograms encountered
  ** so far.  The nRow value will increase as new trigger subprograms are
  ** encountered, but p->pc will eventually catch up to nRow.
  */
  nRow = p->nOp;
  if( pSub!=0 ){
    if( pSub->flags&MEM_Blob ){
      /* pSub is initially NULL.  It is initialized to a BLOB by
      ** the P4_SUBPROGRAM processing logic below */
      nSub = pSub->n/sizeof(Vdbe*);
      apSub = (SubProgram **)pSub->z;
    }
    for(i=0; i<nSub; i++){
      nRow += apSub[i]->nOp;
    }
................................................................................
    /* RHS is real */
    else if( pRhs->flags & MEM_Real ){
      serial_type = aKey1[idx1];
      if( serial_type>=10 ){
        /* Serial types 12 or greater are strings and blobs (greater than
        ** numbers). Types 10 and 11 are currently "reserved for future 
        ** use", so it doesn't really matter what the results of comparing
        ** them to numeric values are.  */
        rc = +1;
      }else if( serial_type==0 ){
        rc = -1;
      }else{
        sqlite3VdbeSerialGet(&aKey1[d1], serial_type, &mem1);
        if( serial_type==7 ){
          if( mem1.u.r<pRhs->u.r ){
................................................................................
void sqlite3VdbeSetChanges(sqlite3 *db, int nChange){
  assert( sqlite3_mutex_held(db->mutex) );
  db->nChange = nChange;
  db->nTotalChange += nChange;
}

/*
** Set a flag in the vdbe to update the change counter when it is finalized
** or reset.
*/
void sqlite3VdbeCountChanges(Vdbe *v){
  v->changeCntOn = 1;
}

/*

Changes to src/vdbesort.c.

81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
...
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
...
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
....
2188
2189
2190
2191
2192
2193
2194
2195
2196
2197
2198
2199
2200
2201
2202
** A PMA created at this point is known as a "level-0 PMA". Higher levels
** of PMAs may be created by merging existing PMAs together - for example
** merging two or more level-0 PMAs together creates a level-1 PMA.
**
** The threshold for the amount of main memory to use before flushing 
** records to a PMA is roughly the same as the limit configured for the
** page-cache of the main database. Specifically, the threshold is set to 
** the value returned by "PRAGMA main.page_size" multipled by 
** that returned by "PRAGMA main.cache_size", in bytes.
**
** If the sorter is running in single-threaded mode, then all PMAs generated
** are appended to a single temporary file. Or, if the sorter is running in
** multi-threaded mode then up to (N+1) temporary files may be opened, where
** N is the configured number of worker threads. In this case, instead of
** sorting the records and writing the PMA to a temporary file itself, the
................................................................................
**
** When Rewind() is called, any data remaining in memory is flushed to a 
** final PMA. So at this point the data is stored in some number of sorted
** PMAs within temporary files on disk.
**
** If there are fewer than SORTER_MAX_MERGE_COUNT PMAs in total and the
** sorter is running in single-threaded mode, then these PMAs are merged
** incrementally as keys are retreived from the sorter by the VDBE.  The
** MergeEngine object, described in further detail below, performs this
** merge.
**
** Or, if running in multi-threaded mode, then a background thread is
** launched to merge the existing PMAs. Once the background thread has
** merged T bytes of data into a single sorted PMA, the main thread 
** begins reading keys from that PMA while the background thread proceeds
................................................................................
** other way. VdbeSorter.nTask is set to the number of worker threads allowed
** (see SQLITE_CONFIG_WORKER_THREADS) plus one (the main thread).  Thus for
** single-threaded operation, there is exactly one instance of this object
** and for multi-threaded operation there are two or more instances.
**
** Essentially, this structure contains all those fields of the VdbeSorter
** structure for which each thread requires a separate instance. For example,
** each thread requries its own UnpackedRecord object to unpack records in
** as part of comparison operations.
**
** Before a background thread is launched, variable bDone is set to 0. Then, 
** right before it exits, the thread itself sets bDone to 1. This is used for 
** two purposes:
**
**   1. When flushing the contents of memory to a level-0 PMA on disk, to
................................................................................
  sqlite3 *db = pTask->pSorter->db;

  /* eMode is always INCRINIT_NORMAL in single-threaded mode */
  assert( SQLITE_MAX_WORKER_THREADS>0 || eMode==INCRINIT_NORMAL );

  rc = vdbeMergeEngineInit(pTask, pIncr->pMerger, eMode);

  /* Set up the required files for pIncr. A multi-theaded IncrMerge object
  ** requires two temp files to itself, whereas a single-threaded object
  ** only requires a region of pTask->file2. */
  if( rc==SQLITE_OK ){
    int mxSz = pIncr->mxSz;
#if SQLITE_MAX_WORKER_THREADS>0
    if( pIncr->bUseThread ){
      rc = vdbeSorterOpenTempFile(db, mxSz, &pIncr->aFile[0].pFd);







|







 







|







 







|







 







|







81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
...
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
...
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
....
2188
2189
2190
2191
2192
2193
2194
2195
2196
2197
2198
2199
2200
2201
2202
** A PMA created at this point is known as a "level-0 PMA". Higher levels
** of PMAs may be created by merging existing PMAs together - for example
** merging two or more level-0 PMAs together creates a level-1 PMA.
**
** The threshold for the amount of main memory to use before flushing 
** records to a PMA is roughly the same as the limit configured for the
** page-cache of the main database. Specifically, the threshold is set to 
** the value returned by "PRAGMA main.page_size" multiplied by 
** that returned by "PRAGMA main.cache_size", in bytes.
**
** If the sorter is running in single-threaded mode, then all PMAs generated
** are appended to a single temporary file. Or, if the sorter is running in
** multi-threaded mode then up to (N+1) temporary files may be opened, where
** N is the configured number of worker threads. In this case, instead of
** sorting the records and writing the PMA to a temporary file itself, the
................................................................................
**
** When Rewind() is called, any data remaining in memory is flushed to a 
** final PMA. So at this point the data is stored in some number of sorted
** PMAs within temporary files on disk.
**
** If there are fewer than SORTER_MAX_MERGE_COUNT PMAs in total and the
** sorter is running in single-threaded mode, then these PMAs are merged
** incrementally as keys are retrieved from the sorter by the VDBE.  The
** MergeEngine object, described in further detail below, performs this
** merge.
**
** Or, if running in multi-threaded mode, then a background thread is
** launched to merge the existing PMAs. Once the background thread has
** merged T bytes of data into a single sorted PMA, the main thread 
** begins reading keys from that PMA while the background thread proceeds
................................................................................
** other way. VdbeSorter.nTask is set to the number of worker threads allowed
** (see SQLITE_CONFIG_WORKER_THREADS) plus one (the main thread).  Thus for
** single-threaded operation, there is exactly one instance of this object
** and for multi-threaded operation there are two or more instances.
**
** Essentially, this structure contains all those fields of the VdbeSorter
** structure for which each thread requires a separate instance. For example,
** each thread requires its own UnpackedRecord object to unpack records in
** as part of comparison operations.
**
** Before a background thread is launched, variable bDone is set to 0. Then, 
** right before it exits, the thread itself sets bDone to 1. This is used for 
** two purposes:
**
**   1. When flushing the contents of memory to a level-0 PMA on disk, to
................................................................................
  sqlite3 *db = pTask->pSorter->db;

  /* eMode is always INCRINIT_NORMAL in single-threaded mode */
  assert( SQLITE_MAX_WORKER_THREADS>0 || eMode==INCRINIT_NORMAL );

  rc = vdbeMergeEngineInit(pTask, pIncr->pMerger, eMode);

  /* Set up the required files for pIncr. A multi-threaded IncrMerge object
  ** requires two temp files to itself, whereas a single-threaded object
  ** only requires a region of pTask->file2. */
  if( rc==SQLITE_OK ){
    int mxSz = pIncr->mxSz;
#if SQLITE_MAX_WORKER_THREADS>0
    if( pIncr->bUseThread ){
      rc = vdbeSorterOpenTempFile(db, mxSz, &pIncr->aFile[0].pFd);

Changes to src/vtab.c.

1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
** Check to see if virtual table module pMod can be have an eponymous
** virtual table instance.  If it can, create one if one does not already
** exist. Return non-zero if the eponymous virtual table instance exists
** when this routine returns, and return zero if it does not exist.
**
** An eponymous virtual table instance is one that is named after its
** module, and more importantly, does not require a CREATE VIRTUAL TABLE
** statement in order to come into existance.  Eponymous virtual table
** instances always exist.  They cannot be DROP-ed.
**
** Any virtual table module for which xConnect and xCreate are the same
** method can have an eponymous virtual table instance.
*/
int sqlite3VtabEponymousTableInit(Parse *pParse, Module *pMod){
  const sqlite3_module *pModule = pMod->pModule;







|







1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
** Check to see if virtual table module pMod can be have an eponymous
** virtual table instance.  If it can, create one if one does not already
** exist. Return non-zero if the eponymous virtual table instance exists
** when this routine returns, and return zero if it does not exist.
**
** An eponymous virtual table instance is one that is named after its
** module, and more importantly, does not require a CREATE VIRTUAL TABLE
** statement in order to come into existence.  Eponymous virtual table
** instances always exist.  They cannot be DROP-ed.
**
** Any virtual table module for which xConnect and xCreate are the same
** method can have an eponymous virtual table instance.
*/
int sqlite3VtabEponymousTableInit(Parse *pParse, Module *pMod){
  const sqlite3_module *pModule = pMod->pModule;

Changes to src/wal.c.

277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
...
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
...
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
....
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
....
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
....
2195
2196
2197
2198
2199
2200
2201
2202
2203
2204
2205
2206
2207
2208
2209
/*
** Index numbers for various locking bytes.   WAL_NREADER is the number
** of available reader locks and should be at least 3.  The default
** is SQLITE_SHM_NLOCK==8 and  WAL_NREADER==5.
**
** Technically, the various VFSes are free to implement these locks however
** they see fit.  However, compatibility is encouraged so that VFSes can
** interoperate.  The standard implemention used on both unix and windows
** is for the index number to indicate a byte offset into the
** WalCkptInfo.aLock[] array in the wal-index header.  In other words, all
** locks are on the shm file.  The WALINDEX_LOCK_OFFSET constant (which
** should be 120) is the location in the shm file for the first locking
** byte.
*/
#define WAL_WRITE_LOCK         0
................................................................................
** database "backfilling".)  The nBackfill number is never greater than
** WalIndexHdr.mxFrame.  nBackfill can only be increased by threads
** holding the WAL_CKPT_LOCK lock (which includes a recovery thread).
** However, a WAL_WRITE_LOCK thread can move the value of nBackfill from
** mxFrame back to zero when the WAL is reset.
**
** nBackfillAttempted is the largest value of nBackfill that a checkpoint
** has attempted to achieve.  Normally nBackfill==nBackfillAtempted, however
** the nBackfillAttempted is set before any backfilling is done and the
** nBackfill is only set after all backfilling completes.  So if a checkpoint
** crashes, nBackfillAttempted might be larger than nBackfill.  The
** WalIndexHdr.mxFrame must never be less than nBackfillAttempted.
**
** The aLock[] field is a set of bytes used for locking.  These bytes should
** never be read or written.
................................................................................
  /* A frame is only valid if the salt values in the frame-header
  ** match the salt values in the wal-header. 
  */
  if( memcmp(&pWal->hdr.aSalt, &aFrame[8], 8)!=0 ){
    return 0;
  }

  /* A frame is only valid if the page number is creater than zero.
  */
  pgno = sqlite3Get4byte(&aFrame[0]);
  if( pgno==0 ){
    return 0;
  }

  /* A frame is only valid if a checksum of the WAL header,
  ** all prior frams, the first 16 bytes of this frame-header, 
  ** and the frame-data matches the checksum in the last 8 
  ** bytes of this frame-header.
  */
  nativeCksum = (pWal->hdr.bigEndCksum==SQLITE_BIGENDIAN);
  walChecksumBytes(nativeCksum, aFrame, 8, aCksum, aCksum);
  walChecksumBytes(nativeCksum, aData, pWal->szPage, aCksum, aCksum);
  if( aCksum[0]!=sqlite3Get4byte(&aFrame[16]) 
................................................................................
    *ppWal = pRet;
    WALTRACE(("WAL%d: opened\n", pRet));
  }
  return rc;
}

/*
** Change the size to which the WAL file is trucated on each reset.
*/
void sqlite3WalLimit(Wal *pWal, i64 iLimit){
  if( pWal ) pWal->mxWalSize = iLimit;
}

/*
** Find the smallest page number out of all pages held in the WAL that
................................................................................
** Free an iterator allocated by walIteratorInit().
*/
static void walIteratorFree(WalIterator *p){
  sqlite3_free(p);
}

/*
** Construct a WalInterator object that can be used to loop over all 
** pages in the WAL following frame nBackfill in ascending order. Frames
** nBackfill or earlier may be included - excluding them is an optimization
** only. The caller must hold the checkpoint lock.
**
** On success, make *pp point to the newly allocated WalInterator object
** return SQLITE_OK. Otherwise, return an error code. If this routine
** returns an error, the value of *pp is undefined.
**
** The calling routine should invoke walIteratorFree() to destroy the
** WalIterator object when it has finished with it.
*/
static int walIteratorInit(Wal *pWal, u32 nBackfill, WalIterator **pp){
................................................................................
  ** There are two copies of the header at the beginning of the wal-index.
  ** When reading, read [0] first then [1].  Writes are in the reverse order.
  ** Memory barriers are used to prevent the compiler or the hardware from
  ** reordering the reads and writes.  TSAN and similar tools can sometimes
  ** give false-positive warnings about these accesses because the tools do not
  ** account for the double-read and the memory barrier. The use of mutexes
  ** here would be problematic as the memory being accessed is potentially
  ** shared among multiple processes and not all mutex implementions work
  ** reliably in that environment.
  */
  aHdr = walIndexHdr(pWal);
  memcpy(&h1, (void *)&aHdr[0], sizeof(h1)); /* Possible TSAN false-positive */
  walShmBarrier(pWal);
  memcpy(&h2, (void *)&aHdr[1], sizeof(h2));








|







 







|







 







|







|







 







|







 







|




|







 







|







277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
...
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
...
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
....
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
....
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
....
2195
2196
2197
2198
2199
2200
2201
2202
2203
2204
2205
2206
2207
2208
2209
/*
** Index numbers for various locking bytes.   WAL_NREADER is the number
** of available reader locks and should be at least 3.  The default
** is SQLITE_SHM_NLOCK==8 and  WAL_NREADER==5.
**
** Technically, the various VFSes are free to implement these locks however
** they see fit.  However, compatibility is encouraged so that VFSes can
** interoperate.  The standard implementation used on both unix and windows
** is for the index number to indicate a byte offset into the
** WalCkptInfo.aLock[] array in the wal-index header.  In other words, all
** locks are on the shm file.  The WALINDEX_LOCK_OFFSET constant (which
** should be 120) is the location in the shm file for the first locking
** byte.
*/
#define WAL_WRITE_LOCK         0
................................................................................
** database "backfilling".)  The nBackfill number is never greater than
** WalIndexHdr.mxFrame.  nBackfill can only be increased by threads
** holding the WAL_CKPT_LOCK lock (which includes a recovery thread).
** However, a WAL_WRITE_LOCK thread can move the value of nBackfill from
** mxFrame back to zero when the WAL is reset.
**
** nBackfillAttempted is the largest value of nBackfill that a checkpoint
** has attempted to achieve.  Normally nBackfill==nBackfillAttempted, however
** the nBackfillAttempted is set before any backfilling is done and the
** nBackfill is only set after all backfilling completes.  So if a checkpoint
** crashes, nBackfillAttempted might be larger than nBackfill.  The
** WalIndexHdr.mxFrame must never be less than nBackfillAttempted.
**
** The aLock[] field is a set of bytes used for locking.  These bytes should
** never be read or written.
................................................................................
  /* A frame is only valid if the salt values in the frame-header
  ** match the salt values in the wal-header. 
  */
  if( memcmp(&pWal->hdr.aSalt, &aFrame[8], 8)!=0 ){
    return 0;
  }

  /* A frame is only valid if the page number is greater than zero.
  */
  pgno = sqlite3Get4byte(&aFrame[0]);
  if( pgno==0 ){
    return 0;
  }

  /* A frame is only valid if a checksum of the WAL header,
  ** all prior frames, the first 16 bytes of this frame-header, 
  ** and the frame-data matches the checksum in the last 8 
  ** bytes of this frame-header.
  */
  nativeCksum = (pWal->hdr.bigEndCksum==SQLITE_BIGENDIAN);
  walChecksumBytes(nativeCksum, aFrame, 8, aCksum, aCksum);
  walChecksumBytes(nativeCksum, aData, pWal->szPage, aCksum, aCksum);
  if( aCksum[0]!=sqlite3Get4byte(&aFrame[16]) 
................................................................................
    *ppWal = pRet;
    WALTRACE(("WAL%d: opened\n", pRet));
  }
  return rc;
}

/*
** Change the size to which the WAL file is truncated on each reset.
*/
void sqlite3WalLimit(Wal *pWal, i64 iLimit){
  if( pWal ) pWal->mxWalSize = iLimit;
}

/*
** Find the smallest page number out of all pages held in the WAL that
................................................................................
** Free an iterator allocated by walIteratorInit().
*/
static void walIteratorFree(WalIterator *p){
  sqlite3_free(p);
}

/*
** Construct a WalIterator object that can be used to loop over all 
** pages in the WAL following frame nBackfill in ascending order. Frames
** nBackfill or earlier may be included - excluding them is an optimization
** only. The caller must hold the checkpoint lock.
**
** On success, make *pp point to the newly allocated WalIterator object
** return SQLITE_OK. Otherwise, return an error code. If this routine
** returns an error, the value of *pp is undefined.
**
** The calling routine should invoke walIteratorFree() to destroy the
** WalIterator object when it has finished with it.
*/
static int walIteratorInit(Wal *pWal, u32 nBackfill, WalIterator **pp){
................................................................................
  ** There are two copies of the header at the beginning of the wal-index.
  ** When reading, read [0] first then [1].  Writes are in the reverse order.
  ** Memory barriers are used to prevent the compiler or the hardware from
  ** reordering the reads and writes.  TSAN and similar tools can sometimes
  ** give false-positive warnings about these accesses because the tools do not
  ** account for the double-read and the memory barrier. The use of mutexes
  ** here would be problematic as the memory being accessed is potentially
  ** shared among multiple processes and not all mutex implementations work
  ** reliably in that environment.
  */
  aHdr = walIndexHdr(pWal);
  memcpy(&h1, (void *)&aHdr[0], sizeof(h1)); /* Possible TSAN false-positive */
  walShmBarrier(pWal);
  memcpy(&h2, (void *)&aHdr[1], sizeof(h2));

Changes to src/walker.c.

205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
    }
    p = p->pPrior;
  }while( p!=0 );
  return WRC_Continue;
}

/* Increase the walkerDepth when entering a subquery, and
** descrease when leaving the subquery.
*/
int sqlite3WalkerDepthIncrease(Walker *pWalker, Select *pSelect){
  UNUSED_PARAMETER(pSelect);
  pWalker->walkerDepth++;
  return WRC_Continue;
}
void sqlite3WalkerDepthDecrease(Walker *pWalker, Select *pSelect){







|







205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
    }
    p = p->pPrior;
  }while( p!=0 );
  return WRC_Continue;
}

/* Increase the walkerDepth when entering a subquery, and
** decrease when leaving the subquery.
*/
int sqlite3WalkerDepthIncrease(Walker *pWalker, Select *pSelect){
  UNUSED_PARAMETER(pSelect);
  pWalker->walkerDepth++;
  return WRC_Continue;
}
void sqlite3WalkerDepthDecrease(Walker *pWalker, Select *pSelect){

Changes to src/where.c.

1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
....
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076
2077
2078
2079
....
2210
2211
2212
2213
2214
2215
2216
2217
2218
2219
2220
2221
2222
2223
2224
....
2387
2388
2389
2390
2391
2392
2393
2394
2395
2396
2397
2398
2399
2400
2401
....
2595
2596
2597
2598
2599
2600
2601
2602
2603
2604
2605
2606
2607
2608
2609
....
2891
2892
2893
2894
2895
2896
2897
2898
2899
2900
2901
2902
2903
2904
2905
....
4207
4208
4209
4210
4211
4212
4213
4214
4215
4216
4217
4218
4219
4220
4221
....
4653
4654
4655
4656
4657
4658
4659
4660
4661
4662
4663
4664
4665
4666
4667
....
4806
4807
4808
4809
4810
4811
4812
4813
4814
4815
4816
4817
4818
4819
4820
**
**   CREATE INDEX i1 ON t1(a, b, c);
**   SELECT * FROM t1 WHERE a=? AND c BETWEEN ? AND ?;
**
** Value pLoop->nOut is currently set to the estimated number of rows 
** visited for scanning (a=? AND b=?). This function reduces that estimate 
** by some factor to account for the (c BETWEEN ? AND ?) expression based
** on the stat4 data for the index. this scan will be peformed multiple 
** times (once for each (a,b) combination that matches a=?) is dealt with 
** by the caller.
**
** It does this by scanning through all stat4 samples, comparing values
** extracted from pLower and pUpper with the corresponding column in each
** sample. If L and U are the number of samples found to be less than or
** equal to the values extracted from pLower and pUpper respectively, and
................................................................................
                 || p->rSetup==pTemplate->rSetup );

    /* whereLoopAddBtree() always generates and inserts the automatic index
    ** case first.  Hence compatible candidate WhereLoops never have a larger
    ** rSetup. Call this SETUP-INVARIANT */
    assert( p->rSetup>=pTemplate->rSetup );

    /* Any loop using an appliation-defined index (or PRIMARY KEY or
    ** UNIQUE constraint) with one or more == constraints is better
    ** than an automatic index. Unless it is a skip-scan. */
    if( (p->wsFlags & WHERE_AUTO_INDEX)!=0
     && (pTemplate->nSkip)==0
     && (pTemplate->wsFlags & WHERE_INDEXED)!=0
     && (pTemplate->wsFlags & WHERE_COLUMN_EQ)!=0
     && (p->prereq & pTemplate->prereq)==pTemplate->prereq
................................................................................
    *ppPrev = p = sqlite3DbMallocRawNN(db, sizeof(WhereLoop));
    if( p==0 ) return SQLITE_NOMEM_BKPT;
    whereLoopInit(p);
    p->pNextLoop = 0;
  }else{
    /* We will be overwriting WhereLoop p[].  But before we do, first
    ** go through the rest of the list and delete any other entries besides
    ** p[] that are also supplated by pTemplate */
    WhereLoop **ppTail = &p->pNextLoop;
    WhereLoop *pToDel;
    while( *ppTail ){
      ppTail = whereLoopFindLesser(ppTail, pTemplate);
      if( ppTail==0 ) break;
      pToDel = *ppTail;
      if( pToDel==0 ) break;
................................................................................
    if( pColl==0 ) break;
    if( sqlite3StrICmp(pColl->zName, pIdx->azColl[i+nEq]) ) break;
  }
  return i;
}

/*
** Adjust the cost C by the costMult facter T.  This only occurs if
** compiled with -DSQLITE_ENABLE_COSTMULT
*/
#ifdef SQLITE_ENABLE_COSTMULT
# define ApplyCostMultiplier(C,T)  C += T
#else
# define ApplyCostMultiplier(C,T)
#endif
................................................................................
      pNew->wsFlags |= WHERE_COLUMN_RANGE|WHERE_BTM_LIMIT;
      pNew->u.btree.nBtm = whereRangeVectorLen(
          pParse, pSrc->iCursor, pProbe, saved_nEq, pTerm
      );
      pBtm = pTerm;
      pTop = 0;
      if( pTerm->wtFlags & TERM_LIKEOPT ){
        /* Range contraints that come from the LIKE optimization are
        ** always used in pairs. */
        pTop = &pTerm[1];
        assert( (pTop-(pTerm->pWC->a))<pTerm->pWC->nTerm );
        assert( pTop->wtFlags & TERM_LIKEOPT );
        assert( pTop->eOperator==WO_LT );
        if( whereLoopResize(db, pNew, pNew->nLTerm+1) ) break; /* OOM */
        pNew->aLTerm[pNew->nLTerm++] = pTop;
................................................................................
** log(nRow) factor is omitted from a non-covering index scan in order to
** bias the scoring in favor of using an index, since the worst-case
** performance of using an index is far better than the worst-case performance
** of a full table scan.
*/
static int whereLoopAddBtree(
  WhereLoopBuilder *pBuilder, /* WHERE clause information */
  Bitmask mPrereq             /* Extra prerequesites for using this table */
){
  WhereInfo *pWInfo;          /* WHERE analysis context */
  Index *pProbe;              /* An index we are evaluating */
  Index sPk;                  /* A fake index object for the primary key */
  LogEst aiRowEstPk[2];       /* The aiRowLogEst[] value for the sPk index */
  i16 aiColumnPk = -1;        /* The aColumn[] value for the sPk index */
  SrcList *pTabList;          /* The FROM clause */
................................................................................
        if( isOrdered>=0 && isOrdered<nOrderBy ){
          if( aSortCost[isOrdered]==0 ){
            aSortCost[isOrdered] = whereSortingCost(
                pWInfo, nRowEst, nOrderBy, isOrdered
            );
          }
          /* TUNING:  Add a small extra penalty (5) to sorting as an
          ** extra encouragment to the query planner to select a plan
          ** where the rows emerge in the correct order without any sorting
          ** required. */
          rCost = sqlite3LogEstAdd(rUnsorted, aSortCost[isOrdered]) + 5;

          WHERETRACE(0x002,
              ("---- sort cost=%-3d (%d/%d) increases cost %3d to %-3d\n",
               aSortCost[isOrdered], (nOrderBy-isOrdered), nOrderBy, 
................................................................................
** terms of the WHERE clause that use only terms in that loop and outer
** loops are evaluated and if false a jump is made around all subsequent
** inner loops (or around the "..." if the test occurs within the inner-
** most loop)
**
** OUTER JOINS
**
** An outer join of tables t1 and t2 is conceptally coded as follows:
**
**    foreach row1 in t1 do
**      flag = 0
**      foreach row2 in t2 do
**        start:
**          ...
**          flag = 1
................................................................................
    }
    ExplainQueryPlan((pParse, 0, "SCAN CONSTANT ROW"));
  }else{
    /* Assign a bit from the bitmask to every term in the FROM clause.
    **
    ** The N-th term of the FROM clause is assigned a bitmask of 1<<N.
    **
    ** The rule of the previous sentence ensures thta if X is the bitmask for
    ** a table T, then X-1 is the bitmask for all other tables to the left of T.
    ** Knowing the bitmask for all tables to the left of a left join is
    ** important.  Ticket #3015.
    **
    ** Note that bitmasks are created for all pTabList->nSrc tables in
    ** pTabList, not just the first nTabList tables.  nTabList is normally
    ** equal to pTabList->nSrc but might be shortened to 1 if the







|







 







|







 







|







 







|







 







|







 







|







 







|







 







|







 







|







1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
....
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076
2077
2078
2079
....
2210
2211
2212
2213
2214
2215
2216
2217
2218
2219
2220
2221
2222
2223
2224
....
2387
2388
2389
2390
2391
2392
2393
2394
2395
2396
2397
2398
2399
2400
2401
....
2595
2596
2597
2598
2599
2600
2601
2602
2603
2604
2605
2606
2607
2608
2609
....
2891
2892
2893
2894
2895
2896
2897
2898
2899
2900
2901
2902
2903
2904
2905
....
4207
4208
4209
4210
4211
4212
4213
4214
4215
4216
4217
4218
4219
4220
4221
....
4653
4654
4655
4656
4657
4658
4659
4660
4661
4662
4663
4664
4665
4666
4667
....
4806
4807
4808
4809
4810
4811
4812
4813
4814
4815
4816
4817
4818
4819
4820
**
**   CREATE INDEX i1 ON t1(a, b, c);
**   SELECT * FROM t1 WHERE a=? AND c BETWEEN ? AND ?;
**
** Value pLoop->nOut is currently set to the estimated number of rows 
** visited for scanning (a=? AND b=?). This function reduces that estimate 
** by some factor to account for the (c BETWEEN ? AND ?) expression based
** on the stat4 data for the index. this scan will be performed multiple 
** times (once for each (a,b) combination that matches a=?) is dealt with 
** by the caller.
**
** It does this by scanning through all stat4 samples, comparing values
** extracted from pLower and pUpper with the corresponding column in each
** sample. If L and U are the number of samples found to be less than or
** equal to the values extracted from pLower and pUpper respectively, and
................................................................................
                 || p->rSetup==pTemplate->rSetup );

    /* whereLoopAddBtree() always generates and inserts the automatic index
    ** case first.  Hence compatible candidate WhereLoops never have a larger
    ** rSetup. Call this SETUP-INVARIANT */
    assert( p->rSetup>=pTemplate->rSetup );

    /* Any loop using an application-defined index (or PRIMARY KEY or
    ** UNIQUE constraint) with one or more == constraints is better
    ** than an automatic index. Unless it is a skip-scan. */
    if( (p->wsFlags & WHERE_AUTO_INDEX)!=0
     && (pTemplate->nSkip)==0
     && (pTemplate->wsFlags & WHERE_INDEXED)!=0
     && (pTemplate->wsFlags & WHERE_COLUMN_EQ)!=0
     && (p->prereq & pTemplate->prereq)==pTemplate->prereq
................................................................................
    *ppPrev = p = sqlite3DbMallocRawNN(db, sizeof(WhereLoop));
    if( p==0 ) return SQLITE_NOMEM_BKPT;
    whereLoopInit(p);
    p->pNextLoop = 0;
  }else{
    /* We will be overwriting WhereLoop p[].  But before we do, first
    ** go through the rest of the list and delete any other entries besides
    ** p[] that are also to be replaced by pTemplate */
    WhereLoop **ppTail = &p->pNextLoop;
    WhereLoop *pToDel;
    while( *ppTail ){
      ppTail = whereLoopFindLesser(ppTail, pTemplate);
      if( ppTail==0 ) break;
      pToDel = *ppTail;
      if( pToDel==0 ) break;
................................................................................
    if( pColl==0 ) break;
    if( sqlite3StrICmp(pColl->zName, pIdx->azColl[i+nEq]) ) break;
  }
  return i;
}

/*
** Adjust the cost C by the costMult factor T.  This only occurs if
** compiled with -DSQLITE_ENABLE_COSTMULT
*/
#ifdef SQLITE_ENABLE_COSTMULT
# define ApplyCostMultiplier(C,T)  C += T
#else
# define ApplyCostMultiplier(C,T)
#endif
................................................................................
      pNew->wsFlags |= WHERE_COLUMN_RANGE|WHERE_BTM_LIMIT;
      pNew->u.btree.nBtm = whereRangeVectorLen(
          pParse, pSrc->iCursor, pProbe, saved_nEq, pTerm
      );
      pBtm = pTerm;
      pTop = 0;
      if( pTerm->wtFlags & TERM_LIKEOPT ){
        /* Range constraints that come from the LIKE optimization are
        ** always used in pairs. */
        pTop = &pTerm[1];
        assert( (pTop-(pTerm->pWC->a))<pTerm->pWC->nTerm );
        assert( pTop->wtFlags & TERM_LIKEOPT );
        assert( pTop->eOperator==WO_LT );
        if( whereLoopResize(db, pNew, pNew->nLTerm+1) ) break; /* OOM */
        pNew->aLTerm[pNew->nLTerm++] = pTop;
................................................................................
** log(nRow) factor is omitted from a non-covering index scan in order to
** bias the scoring in favor of using an index, since the worst-case
** performance of using an index is far better than the worst-case performance
** of a full table scan.
*/
static int whereLoopAddBtree(
  WhereLoopBuilder *pBuilder, /* WHERE clause information */
  Bitmask mPrereq             /* Extra prerequisites for using this table */
){
  WhereInfo *pWInfo;          /* WHERE analysis context */
  Index *pProbe;              /* An index we are evaluating */
  Index sPk;                  /* A fake index object for the primary key */
  LogEst aiRowEstPk[2];       /* The aiRowLogEst[] value for the sPk index */
  i16 aiColumnPk = -1;        /* The aColumn[] value for the sPk index */
  SrcList *pTabList;          /* The FROM clause */
................................................................................
        if( isOrdered>=0 && isOrdered<nOrderBy ){
          if( aSortCost[isOrdered]==0 ){
            aSortCost[isOrdered] = whereSortingCost(
                pWInfo, nRowEst, nOrderBy, isOrdered
            );
          }
          /* TUNING:  Add a small extra penalty (5) to sorting as an
          ** extra encouragement to the query planner to select a plan
          ** where the rows emerge in the correct order without any sorting
          ** required. */
          rCost = sqlite3LogEstAdd(rUnsorted, aSortCost[isOrdered]) + 5;

          WHERETRACE(0x002,
              ("---- sort cost=%-3d (%d/%d) increases cost %3d to %-3d\n",
               aSortCost[isOrdered], (nOrderBy-isOrdered), nOrderBy, 
................................................................................
** terms of the WHERE clause that use only terms in that loop and outer
** loops are evaluated and if false a jump is made around all subsequent
** inner loops (or around the "..." if the test occurs within the inner-
** most loop)
**
** OUTER JOINS
**
** An outer join of tables t1 and t2 is conceptually coded as follows:
**
**    foreach row1 in t1 do
**      flag = 0
**      foreach row2 in t2 do
**        start:
**          ...
**          flag = 1
................................................................................
    }
    ExplainQueryPlan((pParse, 0, "SCAN CONSTANT ROW"));
  }else{
    /* Assign a bit from the bitmask to every term in the FROM clause.
    **
    ** The N-th term of the FROM clause is assigned a bitmask of 1<<N.
    **
    ** The rule of the previous sentence ensures that if X is the bitmask for
    ** a table T, then X-1 is the bitmask for all other tables to the left of T.
    ** Knowing the bitmask for all tables to the left of a left join is
    ** important.  Ticket #3015.
    **
    ** Note that bitmasks are created for all pTabList->nSrc tables in
    ** pTabList, not just the first nTabList tables.  nTabList is normally
    ** equal to pTabList->nSrc but might be shortened to 1 if the

Changes to src/whereInt.h.

412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
  int nRecValid;            /* Number of valid fields currently in pRec */
#endif
  unsigned char bldFlags1;  /* First set of SQLITE_BLDF_* flags */
  unsigned char bldFlags2;  /* Second set of SQLITE_BLDF_* flags */
  unsigned int iPlanLimit;  /* Search limiter */
};

/* Allowed values for WhereLoopBuider.bldFlags */
#define SQLITE_BLDF1_INDEXED  0x0001   /* An index is used */
#define SQLITE_BLDF1_UNIQUE   0x0002   /* All keys of a UNIQUE index used */

#define SQLITE_BLDF2_2NDPASS  0x0004   /* Second builder pass needed */

/* The WhereLoopBuilder.iPlanLimit is used to limit the number of
** index+constraint combinations the query planner will consider for a







|







412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
  int nRecValid;            /* Number of valid fields currently in pRec */
#endif
  unsigned char bldFlags1;  /* First set of SQLITE_BLDF_* flags */
  unsigned char bldFlags2;  /* Second set of SQLITE_BLDF_* flags */
  unsigned int iPlanLimit;  /* Search limiter */
};

/* Allowed values for WhereLoopBuilder.bldFlags */
#define SQLITE_BLDF1_INDEXED  0x0001   /* An index is used */
#define SQLITE_BLDF1_UNIQUE   0x0002   /* All keys of a UNIQUE index used */

#define SQLITE_BLDF2_2NDPASS  0x0004   /* Second builder pass needed */

/* The WhereLoopBuilder.iPlanLimit is used to limit the number of
** index+constraint combinations the query planner will consider for a

Changes to src/wherecode.c.

872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
....
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
**
**   1) allocate a register and code an OP_Column instruction to read 
**      the specified column into the new register, and
**
**   2) transform the expression node to a TK_REGISTER node that reads 
**      from the newly populated register.
**
** Also, if the node is a TK_COLUMN that does access the table idenified
** by pCCurHint.iTabCur, and an index is being used (which we will
** know because CCurHint.pIdx!=0) then transform the TK_COLUMN into
** an access of the index rather than the original table.
*/
static int codeCursorHintFixExpr(Walker *pWalker, Expr *pExpr){
  int rc = WRC_Continue;
  struct CCurHint *pHint = pWalker->u.pCCurHint;
................................................................................
           /* TK_GT */  OP_SeekGT,
           /* TK_LE */  OP_SeekLE,
           /* TK_LT */  OP_SeekLT,
           /* TK_GE */  OP_SeekGE
      };
      assert( TK_LE==TK_GT+1 );      /* Make sure the ordering.. */
      assert( TK_LT==TK_GT+2 );      /*  ... of the TK_xx values... */
      assert( TK_GE==TK_GT+3 );      /*  ... is correcct. */

      assert( (pStart->wtFlags & TERM_VNULL)==0 );
      testcase( pStart->wtFlags & TERM_VIRTUAL );
      pX = pStart->pExpr;
      assert( pX!=0 );
      testcase( pStart->leftCursor!=iCur ); /* transitive constraints */
      if( sqlite3ExprIsVector(pX->pRight) ){







|







 







|







872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
....
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
**
**   1) allocate a register and code an OP_Column instruction to read 
**      the specified column into the new register, and
**
**   2) transform the expression node to a TK_REGISTER node that reads 
**      from the newly populated register.
**
** Also, if the node is a TK_COLUMN that does access the table identified
** by pCCurHint.iTabCur, and an index is being used (which we will
** know because CCurHint.pIdx!=0) then transform the TK_COLUMN into
** an access of the index rather than the original table.
*/
static int codeCursorHintFixExpr(Walker *pWalker, Expr *pExpr){
  int rc = WRC_Continue;
  struct CCurHint *pHint = pWalker->u.pCCurHint;
................................................................................
           /* TK_GT */  OP_SeekGT,
           /* TK_LE */  OP_SeekLE,
           /* TK_LT */  OP_SeekLT,
           /* TK_GE */  OP_SeekGE
      };
      assert( TK_LE==TK_GT+1 );      /* Make sure the ordering.. */
      assert( TK_LT==TK_GT+2 );      /*  ... of the TK_xx values... */
      assert( TK_GE==TK_GT+3 );      /*  ... is correct. */

      assert( (pStart->wtFlags & TERM_VNULL)==0 );
      testcase( pStart->wtFlags & TERM_VIRTUAL );
      pX = pStart->pExpr;
      assert( pX!=0 );
      testcase( pStart->leftCursor!=iCur ); /* transitive constraints */
      if( sqlite3ExprIsVector(pX->pRight) ){

Changes to src/whereexpr.c.

160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
...
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801

#ifndef SQLITE_OMIT_LIKE_OPTIMIZATION
/*
** Check to see if the given expression is a LIKE or GLOB operator that
** can be optimized using inequality constraints.  Return TRUE if it is
** so and false if not.
**
** In order for the operator to be optimizible, the RHS must be a string
** literal that does not begin with a wildcard.  The LHS must be a column
** that may only be NULL, a string, or a BLOB, never a number. (This means
** that virtual tables cannot participate in the LIKE optimization.)  The
** collating sequence for the column on the LHS must be appropriate for
** the operator.
*/
static int isLikeOrGlob(
................................................................................
          assert( j==1 );
          continue;
        }
        if( (chngToIN & sqlite3WhereGetMask(&pWInfo->sMaskSet,
                                            pOrTerm->leftCursor))==0 ){
          /* This term must be of the form t1.a==t2.b where t2 is in the
          ** chngToIN set but t1 is not.  This term will be either preceded
          ** or follwed by an inverted copy (t2.b==t1.a).  Skip this term 
          ** and use its inversion. */
          testcase( pOrTerm->wtFlags & TERM_COPIED );
          testcase( pOrTerm->wtFlags & TERM_VIRTUAL );
          assert( pOrTerm->wtFlags & (TERM_COPIED|TERM_VIRTUAL) );
          continue;
        }
        iColumn = pOrTerm->u.leftColumn;







|







 







|







160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
...
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801

#ifndef SQLITE_OMIT_LIKE_OPTIMIZATION
/*
** Check to see if the given expression is a LIKE or GLOB operator that
** can be optimized using inequality constraints.  Return TRUE if it is
** so and false if not.
**
** In order for the operator to be optimizable, the RHS must be a string
** literal that does not begin with a wildcard.  The LHS must be a column
** that may only be NULL, a string, or a BLOB, never a number. (This means
** that virtual tables cannot participate in the LIKE optimization.)  The
** collating sequence for the column on the LHS must be appropriate for
** the operator.
*/
static int isLikeOrGlob(
................................................................................
          assert( j==1 );
          continue;
        }
        if( (chngToIN & sqlite3WhereGetMask(&pWInfo->sMaskSet,
                                            pOrTerm->leftCursor))==0 ){
          /* This term must be of the form t1.a==t2.b where t2 is in the
          ** chngToIN set but t1 is not.  This term will be either preceded
          ** or followed by an inverted copy (t2.b==t1.a).  Skip this term 
          ** and use its inversion. */
          testcase( pOrTerm->wtFlags & TERM_COPIED );
          testcase( pOrTerm->wtFlags & TERM_VIRTUAL );
          assert( pOrTerm->wtFlags & (TERM_COPIED|TERM_VIRTUAL) );
          continue;
        }
        iColumn = pOrTerm->u.leftColumn;

Changes to src/window.c.

113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
....
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
....
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
....
2516
2517
2518
2519
2520
2521
2522
2523
2524
2525
2526
2527
2528
2529
2530
**     lag(expr [, offset [, default]])
**     first_value(expr)
**     last_value(expr)
**     nth_value(expr, N)
**   
**   These are the same built-in window functions supported by Postgres. 
**   Although the behaviour of aggregate window functions (functions that
**   can be used as either aggregates or window funtions) allows them to
**   be implemented using an API, built-in window functions are much more
**   esoteric. Additionally, some window functions (e.g. nth_value()) 
**   may only be implemented by caching the entire partition in memory.
**   As such, some built-in window functions use the same API as aggregate
**   window functions and some are implemented directly using VDBE 
**   instructions. Additionally, for those functions that use the API, the
**   window frame is sometimes modified before the SELECT statement is
................................................................................
    sqlite3ExprListDelete(pParse->db, pPartition);
    sqlite3ExprListDelete(pParse->db, pOrderBy);
  }
  return pWin;
}

/*
** Window *pWin has just been created from a WINDOW clause. Tokne pBase
** is the base window. Earlier windows from the same WINDOW clause are
** stored in the linked list starting at pWin->pNextWin. This function
** either updates *pWin according to the base specification, or else
** leaves an error in pParse.
*/
void sqlite3WindowChain(Parse *pParse, Window *pWin, Window *pList){
  if( pWin->zBase ){
................................................................................
**        callbacks have been invoked in it.
**
** start,current,end
**   Consider a window-frame similar to the following:
**
**     (ORDER BY a, b GROUPS BETWEEN 2 PRECEDING AND 2 FOLLOWING)
**
**   The windows functions implmentation caches the input rows in a temp
**   table, sorted by "a, b" (it actually populates the cache lazily, and
**   aggressively removes rows once they are no longer required, but that's
**   a mere detail). It keeps three cursors open on the temp table. One
**   (current) that points to the next row to return to the query engine
**   once its window function values have been calculated. Another (end)
**   points to the next row to call the xStep() method of each window function
**   on (so that it is 2 groups ahead of current). And a third (start) that
................................................................................
**     }
**     while( !eof csrCurrent ){
**       RETURN_ROW
**     }
**
** For the most part, the patterns above are adapted to support UNBOUNDED by
** assuming that it is equivalent to "infinity PRECEDING/FOLLOWING" and
** CURRENT ROW by assuming that it is equivilent to "0 PRECEDING/FOLLOWING".
** This is optimized of course - branches that will never be taken and
** conditions that are always true are omitted from the VM code. The only
** exceptional case is:
**
**   ROWS BETWEEN <expr1> FOLLOWING AND UNBOUNDED FOLLOWING
**
**     ... loop started by sqlite3WhereBegin() ...







|







 







|







 







|







 







|







113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
....
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
....
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
....
2516
2517
2518
2519
2520
2521
2522
2523
2524
2525
2526
2527
2528
2529
2530
**     lag(expr [, offset [, default]])
**     first_value(expr)
**     last_value(expr)
**     nth_value(expr, N)
**   
**   These are the same built-in window functions supported by Postgres. 
**   Although the behaviour of aggregate window functions (functions that
**   can be used as either aggregates or window functions) allows them to
**   be implemented using an API, built-in window functions are much more
**   esoteric. Additionally, some window functions (e.g. nth_value()) 
**   may only be implemented by caching the entire partition in memory.
**   As such, some built-in window functions use the same API as aggregate
**   window functions and some are implemented directly using VDBE 
**   instructions. Additionally, for those functions that use the API, the
**   window frame is sometimes modified before the SELECT statement is
................................................................................
    sqlite3ExprListDelete(pParse->db, pPartition);
    sqlite3ExprListDelete(pParse->db, pOrderBy);
  }
  return pWin;
}

/*
** Window *pWin has just been created from a WINDOW clause. Token pBase
** is the base window. Earlier windows from the same WINDOW clause are
** stored in the linked list starting at pWin->pNextWin. This function
** either updates *pWin according to the base specification, or else
** leaves an error in pParse.
*/
void sqlite3WindowChain(Parse *pParse, Window *pWin, Window *pList){
  if( pWin->zBase ){
................................................................................
**        callbacks have been invoked in it.
**
** start,current,end
**   Consider a window-frame similar to the following:
**
**     (ORDER BY a, b GROUPS BETWEEN 2 PRECEDING AND 2 FOLLOWING)
**
**   The windows functions implementation caches the input rows in a temp
**   table, sorted by "a, b" (it actually populates the cache lazily, and
**   aggressively removes rows once they are no longer required, but that's
**   a mere detail). It keeps three cursors open on the temp table. One
**   (current) that points to the next row to return to the query engine
**   once its window function values have been calculated. Another (end)
**   points to the next row to call the xStep() method of each window function
**   on (so that it is 2 groups ahead of current). And a third (start) that
................................................................................
**     }
**     while( !eof csrCurrent ){
**       RETURN_ROW
**     }
**
** For the most part, the patterns above are adapted to support UNBOUNDED by
** assuming that it is equivalent to "infinity PRECEDING/FOLLOWING" and
** CURRENT ROW by assuming that it is equivalent to "0 PRECEDING/FOLLOWING".
** This is optimized of course - branches that will never be taken and
** conditions that are always true are omitted from the VM code. The only
** exceptional case is:
**
**   ROWS BETWEEN <expr1> FOLLOWING AND UNBOUNDED FOLLOWING
**
**     ... loop started by sqlite3WhereBegin() ...

Changes to test/alter2.test.

371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
...
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
  execsql {
    SELECT a, typeof(a), b, typeof(b), c, typeof(c) FROM t1 LIMIT 1;
  }
} {1 integer -123 integer 5 text}

#-----------------------------------------------------------------------
# Test that UPDATE trigger tables work with default values, and that when
# a row is updated the default values are correctly transfered to the 
# new row.
# 
ifcapable trigger {
db function set_val {set ::val}
  do_test alter2-8.1 {
    execsql {
      CREATE TRIGGER trig1 BEFORE UPDATE ON t1 BEGIN
................................................................................
  do_test alter2-8.3 {
    set ::val
  } {-123 integer 5 text -123 integer 10 text}
}

#-----------------------------------------------------------------------
# Test that DELETE trigger tables work with default values, and that when
# a row is updated the default values are correctly transfered to the 
# new row.
# 
ifcapable trigger {
  do_test alter2-9.1 {
    execsql {
      CREATE TRIGGER trig2 BEFORE DELETE ON t1 BEGIN
      SELECT set_val(







|







 







|







371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
...
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
  execsql {
    SELECT a, typeof(a), b, typeof(b), c, typeof(c) FROM t1 LIMIT 1;
  }
} {1 integer -123 integer 5 text}

#-----------------------------------------------------------------------
# Test that UPDATE trigger tables work with default values, and that when
# a row is updated the default values are correctly transferred to the 
# new row.
# 
ifcapable trigger {
db function set_val {set ::val}
  do_test alter2-8.1 {
    execsql {
      CREATE TRIGGER trig1 BEFORE UPDATE ON t1 BEGIN
................................................................................
  do_test alter2-8.3 {
    set ::val
  } {-123 integer 5 text -123 integer 10 text}
}

#-----------------------------------------------------------------------
# Test that DELETE trigger tables work with default values, and that when
# a row is updated the default values are correctly transferred to the 
# new row.
# 
ifcapable trigger {
  do_test alter2-9.1 {
    execsql {
      CREATE TRIGGER trig2 BEFORE DELETE ON t1 BEGIN
      SELECT set_val(

Changes to test/alter3.test.

99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
    CREATE TABLE t3(a, b, UNIQUE(a, b));
    ALTER TABLE t3 ADD COLUMN c VARCHAR(10, 20);
    SELECT sql FROM sqlite_master WHERE tbl_name = 't3' AND type = 'table';
  }
} {{CREATE TABLE t3(a, b, c VARCHAR(10, 20), UNIQUE(a, b))}}
do_test alter3-1.99 {
  catchsql {
    # May not exist if foriegn-keys are omitted at compile time.
    DROP TABLE t2; 
  }
  execsql {
    DROP TABLE abc; 
    DROP TABLE t1; 
    DROP TABLE t3; 
  }







|







99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
    CREATE TABLE t3(a, b, UNIQUE(a, b));
    ALTER TABLE t3 ADD COLUMN c VARCHAR(10, 20);
    SELECT sql FROM sqlite_master WHERE tbl_name = 't3' AND type = 'table';
  }
} {{CREATE TABLE t3(a, b, c VARCHAR(10, 20), UNIQUE(a, b))}}
do_test alter3-1.99 {
  catchsql {
    # May not exist if foreign-keys are omitted at compile time.
    DROP TABLE t2; 
  }
  execsql {
    DROP TABLE abc; 
    DROP TABLE t1; 
    DROP TABLE t3; 
  }

Changes to test/alter4.test.

106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
    ALTER TABLE t3 ADD COLUMN c VARCHAR(10, 20);
    SELECT sql FROM sqlite_temp_master
     WHERE tbl_name = 't3' AND type = 'table';
  }
} {{CREATE TABLE t3(a, b, c VARCHAR(10, 20), UNIQUE(a, b))}}
do_test alter4-1.99 {
  catchsql {
    # May not exist if foriegn-keys are omitted at compile time.
    DROP TABLE t2; 
  }
  execsql {
    DROP TABLE abc; 
    DROP TABLE t1; 
    DROP TABLE t3; 
  }







|







106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
    ALTER TABLE t3 ADD COLUMN c VARCHAR(10, 20);
    SELECT sql FROM sqlite_temp_master
     WHERE tbl_name = 't3' AND type = 'table';
  }
} {{CREATE TABLE t3(a, b, c VARCHAR(10, 20), UNIQUE(a, b))}}
do_test alter4-1.99 {
  catchsql {
    # May not exist if foreign-keys are omitted at compile time.
    DROP TABLE t2; 
  }
  execsql {
    DROP TABLE abc; 
    DROP TABLE t1; 
    DROP TABLE t3; 
  }

Changes to test/altertab3.test.

597
598
599
600
601
602
603
604
  END;
}
#do_execsql_test 25.2 {
#  ALTER TABLE t2 RENAME COLUMN a TO aaa;
#}

finish_test








<
597
598
599
600
601
602
603

  END;
}
#do_execsql_test 25.2 {
#  ALTER TABLE t2 RENAME COLUMN a TO aaa;
#}

finish_test

Changes to test/analyze3.test.

34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
#               in the same way as constants when planning queries that
#               use LIKE expressions in the WHERE clause.
#
# analyze3-3.*: Test that binding to a variable does not invalidate the 
#               query plan when there is no way in which replanning the
#               query may produce a superior outcome.
#
# analyze3-4.*: Test that SQL or authorization callback errors occuring
#               within sqlite3Reprepare() are handled correctly.
#
# analyze3-5.*: Check that the query plans of applicable statements are
#               invalidated if the values of SQL parameter are modified
#               using the clear_bindings() or transfer_bindings() APIs.
# 
# analyze3-6.*: Test that the problem fixed by commit [127a5b776d] is fixed.







|







34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
#               in the same way as constants when planning queries that
#               use LIKE expressions in the WHERE clause.
#
# analyze3-3.*: Test that binding to a variable does not invalidate the 
#               query plan when there is no way in which replanning the
#               query may produce a superior outcome.
#
# analyze3-4.*: Test that SQL or authorization callback errors occurring
#               within sqlite3Reprepare() are handled correctly.
#
# analyze3-5.*: Check that the query plans of applicable statements are
#               invalidated if the values of SQL parameter are modified
#               using the clear_bindings() or transfer_bindings() APIs.
# 
# analyze3-6.*: Test that the problem fixed by commit [127a5b776d] is fixed.

Changes to test/attach4.test.

131
132
133
134
135
136
137
138
}

do_execsql_test 2.2 {
  DROP TRIGGER tr1;
}

finish_test








<
131
132
133
134
135
136
137

}

do_execsql_test 2.2 {
  DROP TRIGGER tr1;
}

finish_test

Changes to test/autoindex1.test.

205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
  |--SEARCH TABLE t501 USING INTEGER PRIMARY KEY (rowid=?)
  `--CORRELATED LIST SUBQUERY xxxxxx
     `--SCAN TABLE t502
}

# The following code checks a performance regression reported on the
# mailing list on 2010-10-19.  The problem is that the nRowEst field
# of ephermeral tables was not being initialized correctly and so no
# automatic index was being created for the emphemeral table when it was
# used as part of a join.
#
do_execsql_test autoindex1-600 {
  CREATE TABLE flock_owner(
    owner_rec_id INTEGER CONSTRAINT flock_owner_key PRIMARY KEY,
    flock_no VARCHAR(6) NOT NULL REFERENCES flock (flock_no),







|







205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
  |--SEARCH TABLE t501 USING INTEGER PRIMARY KEY (rowid=?)
  `--CORRELATED LIST SUBQUERY xxxxxx
     `--SCAN TABLE t502
}

# The following code checks a performance regression reported on the
# mailing list on 2010-10-19.  The problem is that the nRowEst field
# of ephemeral tables was not being initialized correctly and so no
# automatic index was being created for the emphemeral table when it was
# used as part of a join.
#
do_execsql_test autoindex1-600 {
  CREATE TABLE flock_owner(
    owner_rec_id INTEGER CONSTRAINT flock_owner_key PRIMARY KEY,
    flock_no VARCHAR(6) NOT NULL REFERENCES flock (flock_no),

Changes to test/bestindex4.test.

25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
# Virtual table callback for a virtual table named $tbl.
#
# The table created is:
#
#      "CREATE TABLE t1 (id, host, class)"
#
# The virtual table supports == operators on a subset of its columns. The
# exact subset depends on the value of bitmask paramater $param.
#
#   0x01 - == on "id" supported
#   0x02 - == on "host" supported
#   0x04 - == on "class" supported
#
# $param also supports the following bits:
#







|







25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
# Virtual table callback for a virtual table named $tbl.
#
# The table created is:
#
#      "CREATE TABLE t1 (id, host, class)"
#
# The virtual table supports == operators on a subset of its columns. The
# exact subset depends on the value of bitmask parameter $param.
#
#   0x01 - == on "id" supported
#   0x02 - == on "host" supported
#   0x04 - == on "class" supported
#
# $param also supports the following bits:
#

Changes to test/bestindex7.test.

72
73
74
75
76
77
78
79
do_execsql_test 1.9 { select * from vt1 WHERE a=1 OR a=0} {0}
do_execsql_test 1.10 { select * from vt1 WHERE a IN (2) } {}
do_execsql_test 1.10 { select * from vt1 WHERE a IN (0,1,2,3) } {0}
do_execsql_test 1.11 { select * from vt1 WHERE a IN (0, NULL) } {0}
do_execsql_test 1.12 { select * from vt1 WHERE a IN (NULL) } {}

finish_test








<
72
73
74
75
76
77
78

do_execsql_test 1.9 { select * from vt1 WHERE a=1 OR a=0} {0}
do_execsql_test 1.10 { select * from vt1 WHERE a IN (2) } {}
do_execsql_test 1.10 { select * from vt1 WHERE a IN (0,1,2,3) } {0}
do_execsql_test 1.11 { select * from vt1 WHERE a IN (0, NULL) } {0}
do_execsql_test 1.12 { select * from vt1 WHERE a IN (NULL) } {}

finish_test

Changes to test/bitvec.test.

145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
          1 25 121 125
          1 50 121 125
          2 25 121 125
          0
  }
} 0

# This procedure runs sqlite3BitvecBuiltinTest with argments "n" and
# "program".  But it also causes a malloc error to occur after the
# "failcnt"-th malloc.  The result should be "0" if no malloc failure
# occurs or "-1" if there is a malloc failure.
#
proc bitvec_malloc_test {label failcnt n program} {
  do_test $label [subst {
    sqlite3_memdebug_fail $failcnt







|







145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
          1 25 121 125
          1 50 121 125
          2 25 121 125
          0
  }
} 0

# This procedure runs sqlite3BitvecBuiltinTest with arguments "n" and
# "program".  But it also causes a malloc error to occur after the
# "failcnt"-th malloc.  The result should be "0" if no malloc failure
# occurs or "-1" if there is a malloc failure.
#
proc bitvec_malloc_test {label failcnt n program} {
  do_test $label [subst {
    sqlite3_memdebug_fail $failcnt

Changes to test/busy2.test.

125
126
127
128
129
130
131
132

  code1 {
    db1.2 close
  }
}

finish_test








<
125
126
127
128
129
130
131


  code1 {
    db1.2 close
  }
}

finish_test

Changes to test/collate1.test.

19
20
21
22
23
24
25
26
27
28
29
30
31
32
33

#
# Tests are roughly organised as follows:
#
# collate1-1.* - Single-field ORDER BY with an explicit COLLATE clause.
# collate1-2.* - Multi-field ORDER BY with an explicit COLLATE clause.
# collate1-3.* - ORDER BY using a default collation type. Also that an 
#                explict collate type overrides a default collate type.
# collate1-4.* - ORDER BY using a data type.
#

#
# Collation type 'HEX'. If an argument can be interpreted as a hexadecimal
# number, then it is converted to one before the comparison is performed. 
# Numbers are less than other strings. If neither argument is a number, 







|







19
20
21
22
23
24
25
26
27
28
29
30
31
32
33

#
# Tests are roughly organised as follows:
#
# collate1-1.* - Single-field ORDER BY with an explicit COLLATE clause.
# collate1-2.* - Multi-field ORDER BY with an explicit COLLATE clause.
# collate1-3.* - ORDER BY using a default collation type. Also that an 
#                explicit collate type overrides a default collate type.
# collate1-4.* - ORDER BY using a data type.
#

#
# Collation type 'HEX'. If an argument can be interpreted as a hexadecimal
# number, then it is converted to one before the comparison is performed. 
# Numbers are less than other strings. If neither argument is a number, 

Changes to test/collateB.test.

58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
  SELECT *,'|' FROM t1, t2, t3 WHERE b=x2 AND a=x1 AND 1=a;
} {1 11 1 11 |}
do_execsql_test collateB-1.17 {
  SELECT *,'|' FROM t1, t2, t3 WHERE b=x2 AND a=x1 AND 1=a;
} {1 11 1 11 |}

#-------------------------------------------------------------------------
# Test an assert() failure that was occuring if an index were created
# on a column explicitly declared "COLLATE binary".
reset_db
do_execsql_test 2.1 {
  CREATE TABLE t4(a COLLATE binary);
  CREATE INDEX i4 ON t4(a);
  INSERT INTO t4 VALUES('one'), ('two'), ('three');
  VACUUM;
} 

integrity_check 2.2

finish_test







|












58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
  SELECT *,'|' FROM t1, t2, t3 WHERE b=x2 AND a=x1 AND 1=a;
} {1 11 1 11 |}
do_execsql_test collateB-1.17 {
  SELECT *,'|' FROM t1, t2, t3 WHERE b=x2 AND a=x1 AND 1=a;
} {1 11 1 11 |}

#-------------------------------------------------------------------------
# Test an assert() failure that was occurring if an index were created
# on a column explicitly declared "COLLATE binary".
reset_db
do_execsql_test 2.1 {
  CREATE TABLE t4(a COLLATE binary);
  CREATE INDEX i4 ON t4(a);
  INSERT INTO t4 VALUES('one'), ('two'), ('three');
  VACUUM;
} 

integrity_check 2.2

finish_test

Changes to test/colmeta.test.

93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
  set tstbody [concat sqlite3_table_column_metadata $::DB $params] 
  do_test colmeta-$tn.2 {
    list [catch $tstbody msg] [set msg]
  } $results
}

# Calling sqlite3_table_column_metadata with a NULL column name merely
# checks for the existance of the table.
#
do_test colmeta-300 {
  catch {sqlite3_table_column_metadata $::DB main xyzzy} res
} {1}
do_test colmeta-301 {
  catch {sqlite3_table_column_metadata $::DB main abc} res
} {0}


finish_test







|










93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
  set tstbody [concat sqlite3_table_column_metadata $::DB $params] 
  do_test colmeta-$tn.2 {
    list [catch $tstbody msg] [set msg]
  } $results
}

# Calling sqlite3_table_column_metadata with a NULL column name merely
# checks for the existence of the table.
#
do_test colmeta-300 {
  catch {sqlite3_table_column_metadata $::DB main xyzzy} res
} {1}
do_test colmeta-301 {
  catch {sqlite3_table_column_metadata $::DB main abc} res
} {0}


finish_test

Changes to test/corruptL.test.

1287
1288
1289
1290
1291
1292
1293
1294
1295

do_catchsql_test 16.1 {
  PRAGMA writable_schema = ON;
  INSERT INTO t1(rowid, w, x, y, z) VALUES(5, 10, 11, 10, NULL);
} {1 {database disk image is malformed}}

finish_test









<
<
1287
1288
1289
1290
1291
1292
1293



do_catchsql_test 16.1 {
  PRAGMA writable_schema = ON;
  INSERT INTO t1(rowid, w, x, y, z) VALUES(5, 10, 11, 10, NULL);
} {1 {database disk image is malformed}}

finish_test


Changes to test/dbfuzz.c.

364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
    free(pVFile->a);
    pVFile->a = 0;
    return SQLITE_OK;
  }
  return SQLITE_IOERR_DELETE;
}

/* Check for the existance of a file
*/
static int inmemAccess(
  sqlite3_vfs *pVfs,
  const char *zFilename,
  int flags,
  int *pResOut
){







|







364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
    free(pVFile->a);
    pVFile->a = 0;
    return SQLITE_OK;
  }
  return SQLITE_IOERR_DELETE;
}

/* Check for the existence of a file
*/
static int inmemAccess(
  sqlite3_vfs *pVfs,
  const char *zFilename,
  int flags,
  int *pResOut
){

Changes to test/e_createtable.test.

684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
# EVIDENCE-OF: R-16667-09772 A table created using CREATE TABLE AS has
# no PRIMARY KEY and no constraints of any kind. The default value of
# each column is NULL. The default collation sequence for each column of
# the new table is BINARY.
#
#   The following tests create tables based on SELECT statements that read
#   from tables that have primary keys, constraints and explicit default 
#   collation sequences. None of this is transfered to the definition of
#   the new table as stored in the sqlite_master table.
#
#   Tests 2.3.2.* show that the default value of each column is NULL.
#
do_createtable_tests 2.3.1 -query {
  SELECT sql FROM sqlite_master ORDER BY rowid DESC LIMIT 1
} {







|







684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
# EVIDENCE-OF: R-16667-09772 A table created using CREATE TABLE AS has
# no PRIMARY KEY and no constraints of any kind. The default value of
# each column is NULL. The default collation sequence for each column of
# the new table is BINARY.
#
#   The following tests create tables based on SELECT statements that read
#   from tables that have primary keys, constraints and explicit default 
#   collation sequences. None of this is transferred to the definition of
#   the new table as stored in the sqlite_master table.
#
#   Tests 2.3.2.* show that the default value of each column is NULL.
#
do_createtable_tests 2.3.1 -query {
  SELECT sql FROM sqlite_master ORDER BY rowid DESC LIMIT 1
} {

Changes to test/enc.test.

6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
#    May you do good and not evil.
#    May you find forgiveness for yourself and forgive others.
#    May you share freely, never taking more than you give.
#
#***********************************************************************
# This file implements regression tests for SQLite library.  The focus of
# this file is testing the SQLite routines used for converting between the
# various suported unicode encodings (UTF-8, UTF-16, UTF-16le and
# UTF-16be).
#
# $Id: enc.test,v 1.7 2007/05/23 16:23:09 danielk1977 Exp $

set testdir [file dirname $argv0]
source $testdir/tester.tcl








|







6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
#    May you do good and not evil.
#    May you find forgiveness for yourself and forgive others.
#    May you share freely, never taking more than you give.
#
#***********************************************************************
# This file implements regression tests for SQLite library.  The focus of
# this file is testing the SQLite routines used for converting between the
# various supported unicode encodings (UTF-8, UTF-16, UTF-16le and
# UTF-16be).
#
# $Id: enc.test,v 1.7 2007/05/23 16:23:09 danielk1977 Exp $

set testdir [file dirname $argv0]
source $testdir/tester.tcl

Changes to test/enc2.test.

6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
..
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
#    May you do good and not evil.
#    May you find forgiveness for yourself and forgive others.
#    May you share freely, never taking more than you give.
#
#***********************************************************************
# This file implements regression tests for SQLite library.  The focus of
# this file is testing the SQLite routines used for converting between the
# various suported unicode encodings (UTF-8, UTF-16, UTF-16le and
# UTF-16be).
#

set testdir [file dirname $argv0]
source $testdir/tester.tcl

# If UTF16 support is disabled, ignore the tests in this file
................................................................................
# three times. Each time the file 'test.db' contains a database
# with the following contents:
set dbcontents {
  CREATE TABLE t1(a PRIMARY KEY, b, c);
  INSERT INTO t1 VALUES('one', 'I', 1);
}
# This proc tests that we can open and manipulate the test.db 
# database, and that it is possible to retreive values in
# various text encodings.
#
proc run_test_script {t enc} {

# Open the database and pull out a (the) row.
do_test $t.1 {
  sqlite3 db test.db; set DB [sqlite3_connection_pointer db]







|







 







|







6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
..
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
#    May you do good and not evil.
#    May you find forgiveness for yourself and forgive others.
#    May you share freely, never taking more than you give.
#
#***********************************************************************
# This file implements regression tests for SQLite library.  The focus of
# this file is testing the SQLite routines used for converting between the
# various supported unicode encodings (UTF-8, UTF-16, UTF-16le and
# UTF-16be).
#

set testdir [file dirname $argv0]
source $testdir/tester.tcl

# If UTF16 support is disabled, ignore the tests in this file
................................................................................
# three times. Each time the file 'test.db' contains a database
# with the following contents:
set dbcontents {
  CREATE TABLE t1(a PRIMARY KEY, b, c);
  INSERT INTO t1 VALUES('one', 'I', 1);
}
# This proc tests that we can open and manipulate the test.db 
# database, and that it is possible to retrieve values in
# various text encodings.
#
proc run_test_script {t enc} {

# Open the database and pull out a (the) row.
do_test $t.1 {
  sqlite3 db test.db; set DB [sqlite3_connection_pointer db]

Changes to test/enc4.test.

6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
#    May you do good and not evil.
#    May you find forgiveness for yourself and forgive others.
#    May you share freely, never taking more than you give.
#
#***********************************************************************
# This file implements regression tests for SQLite library.  The focus of
# this file is testing the SQLite routines used for converting between the
# various suported unicode encodings (UTF-8, UTF-16, UTF-16le and
# UTF-16be).
#
# $Id: enc4.test,v 1.0 2010/09/29 08:29:32 shaneh Exp $

set testdir [file dirname $argv0]
source $testdir/tester.tcl








|







6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
#    May you do good and not evil.
#    May you find forgiveness for yourself and forgive others.
#    May you share freely, never taking more than you give.
#
#***********************************************************************
# This file implements regression tests for SQLite library.  The focus of
# this file is testing the SQLite routines used for converting between the
# various supported unicode encodings (UTF-8, UTF-16, UTF-16le and
# UTF-16be).
#
# $Id: enc4.test,v 1.0 2010/09/29 08:29:32 shaneh Exp $

set testdir [file dirname $argv0]
source $testdir/tester.tcl

Changes to test/filter2.tcl.

124
125
126
127
128
129
130
131
132
    sum(b) FILTER (WHERE a%5=2),
    sum(b) FILTER (WHERE a%5=3),
    sum(b) FILTER (WHERE a%5=4)
  FROM t1 GROUP BY (a/5) ORDER BY 1;
}

finish_test









<
<
124
125
126
127
128
129
130


    sum(b) FILTER (WHERE a%5=2),
    sum(b) FILTER (WHERE a%5=3),
    sum(b) FILTER (WHERE a%5=4)
  FROM t1 GROUP BY (a/5) ORDER BY 1;
}

finish_test


Changes to test/fts3_common.tcl.

413
414
415
416
417
418
419
420

      }
    }
  }
}

##########################################################################








<
413
414
415
416
417
418
419


      }
    }
  }
}

##########################################################################

Changes to test/fts3corrupt6.test.

52
53
54
55
56
57
58
59
60

do_execsql_test 1.3 {
  SELECT 42+matchinfo(t1,'yxyyxy')  FROM t1 WHERE t1 MATCH  x'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';
} {42}

set sqlite_fts3_enable_parentheses $saved_sqlite_fts3_enable_parentheses
finish_test









<
<
52
53
54
55
56
57
58



do_execsql_test 1.3 {
  SELECT 42+matchinfo(t1,'yxyyxy')  FROM t1 WHERE t1 MATCH  x'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';
} {42}

set sqlite_fts3_enable_parentheses $saved_sqlite_fts3_enable_parentheses
finish_test


Changes to test/fts3cov.test.

21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
...
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
set DO_MALLOC_TEST 0
set testprefix fts3cov

#--------------------------------------------------------------------------
# When it first needs to read a block from the %_segments table, the FTS3 
# module compiles an SQL statement for that purpose. The statement is 
# stored and reused each subsequent time a block is read. This test case 
# tests the effects of an OOM error occuring while compiling the statement.
#
# Similarly, when FTS3 first needs to scan through a set of segment leaves
# to find a set of documents that matches a term, it allocates a string
# containing the text of the required SQL, and compiles one or more 
# statements to traverse the leaves. This test case tests that OOM errors
# that occur while allocating this string and statement are handled correctly
# also.
................................................................................
} {1 D E F 5 A B C}

#-------------------------------------------------------------------------
# If a set of documents are modified within a transaction, the 
# pending-terms table must be flushed each time a document with a docid
# less than or equal to the previous docid is modified. 
#
# This test checks the effects of an OOM error occuring when the 
# pending-terms table is flushed for this reason as part of a DELETE 
# statement.
#
do_malloc_test fts3cov-8 -sqlprep {
  BEGIN;
    CREATE VIRTUAL TABLE t8 USING fts3;
    INSERT INTO t8 VALUES('the output of each batch run');







|







 







|







21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
...
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
set DO_MALLOC_TEST 0
set testprefix fts3cov

#--------------------------------------------------------------------------
# When it first needs to read a block from the %_segments table, the FTS3 
# module compiles an SQL statement for that purpose. The statement is 
# stored and reused each subsequent time a block is read. This test case 
# tests the effects of an OOM error occurring while compiling the statement.
#
# Similarly, when FTS3 first needs to scan through a set of segment leaves
# to find a set of documents that matches a term, it allocates a string
# containing the text of the required SQL, and compiles one or more 
# statements to traverse the leaves. This test case tests that OOM errors
# that occur while allocating this string and statement are handled correctly
# also.
................................................................................
} {1 D E F 5 A B C}

#-------------------------------------------------------------------------
# If a set of documents are modified within a transaction, the 
# pending-terms table must be flushed each time a document with a docid
# less than or equal to the previous docid is modified. 
#
# This test checks the effects of an OOM error occurring when the 
# pending-terms table is flushed for this reason as part of a DELETE 
# statement.
#
do_malloc_test fts3cov-8 -sqlprep {
  BEGIN;
    CREATE VIRTUAL TABLE t8 USING fts3;
    INSERT INTO t8 VALUES('the output of each batch run');

Changes to test/fts3expr2.test.

42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
#
#   In step (2), 4 different fts3 expressions are created from each 
#   expression tree by varying the following boolean properties:
#
#     * Whether or not superflous parenthesis are included. i.e. if
#       "a OR b AND (c OR d)" or "a OR (b AND (c OR d))" is generated.
#
#     * Whether or not explict AND operators are used. i.e. if
#     "a OR b AND c" or "a OR b c" is generated.
#

set sqlite_fts3_enable_parentheses 1

proc strip_phrase_data {L} {
  if {[lindex $L 0] eq "PHRASE"} {







|







42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
#
#   In step (2), 4 different fts3 expressions are created from each 
#   expression tree by varying the following boolean properties:
#
#     * Whether or not superflous parenthesis are included. i.e. if
#       "a OR b AND (c OR d)" or "a OR (b AND (c OR d))" is generated.
#
#     * Whether or not explicit AND operators are used. i.e. if
#     "a OR b AND c" or "a OR b c" is generated.
#

set sqlite_fts3_enable_parentheses 1

proc strip_phrase_data {L} {
  if {[lindex $L 0] eq "PHRASE"} {

Changes to test/fts3snippet2.test.

53
54
55
56
57
58
59
60
do_execsql_test 2.2 {
  SELECT snippet(t0)  FROM t0 WHERE t0 MATCH 
  '(def AND (one NEAR abc)) OR one'
} {<b>one</b>}

set sqlite_fts3_enable_parentheses 0
finish_test








<
53
54
55
56
57
58
59

do_execsql_test 2.2 {
  SELECT snippet(t0)  FROM t0 WHERE t0 MATCH 
  '(def AND (one NEAR abc)) OR one'
} {<b>one</b>}

set sqlite_fts3_enable_parentheses 0
finish_test

Changes to test/fts4upfrom.test.

133
134
135
136
137
138
139
140
    12 b apple blueberry
    13 c banana clementine
    14 d cherry dewberry
  }
}

finish_test








<
133
134
135
136
137
138
139

    12 b apple blueberry
    13 c banana clementine
    14 d cherry dewberry
  }
}

finish_test

Changes to test/fuzzcheck.c.

1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
....
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
    free(pVFile->a);
    pVFile->a = 0;
    return SQLITE_OK;
  }
  return SQLITE_IOERR_DELETE;
}

/* Check for the existance of a file
*/
static int inmemAccess(
  sqlite3_vfs *pVfs,
  const char *zFilename,
  int flags,
  int *pResOut
){
................................................................................
  int timeoutTest = 0;         /* undocumented --timeout-test flag */
  int runFlags = 0;            /* Flags sent to runSql() */
  char *zMsg = 0;              /* Add this message */
  int nSrcDb = 0;              /* Number of source databases */
  char **azSrcDb = 0;          /* Array of source database names */
  int iSrcDb;                  /* Loop over all source databases */
  int nTest = 0;               /* Total number of tests performed */
  char *zDbName = "";          /* Appreviated name of a source database */
  const char *zFailCode = 0;   /* Value of the TEST_FAILURE env variable */
  int cellSzCkFlag = 0;        /* --cell-size-check */
  int sqlFuzz = 0;             /* True for SQL fuzz. False for DB fuzz */
  int iTimeout = 120;          /* Default 120-second timeout */
  int nMem = 0;                /* Memory limit override */
  int nMemThisDb = 0;          /* Memory limit set by the CONFIG table */
  char *zExpDb = 0;            /* Write Databases to files in this directory */







|







 







|







1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
....
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
    free(pVFile->a);
    pVFile->a = 0;
    return SQLITE_OK;
  }
  return SQLITE_IOERR_DELETE;
}

/* Check for the existence of a file
*/
static int inmemAccess(
  sqlite3_vfs *pVfs,
  const char *zFilename,
  int flags,
  int *pResOut
){
................................................................................
  int timeoutTest = 0;         /* undocumented --timeout-test flag */
  int runFlags = 0;            /* Flags sent to runSql() */
  char *zMsg = 0;              /* Add this message */
  int nSrcDb = 0;              /* Number of source databases */
  char **azSrcDb = 0;          /* Array of source database names */
  int iSrcDb;                  /* Loop over all source databases */
  int nTest = 0;               /* Total number of tests performed */
  char *zDbName = "";          /* Abbreviated name of a source database */
  const char *zFailCode = 0;   /* Value of the TEST_FAILURE env variable */
  int cellSzCkFlag = 0;        /* --cell-size-check */
  int sqlFuzz = 0;             /* True for SQL fuzz. False for DB fuzz */
  int iTimeout = 120;          /* Default 120-second timeout */
  int nMem = 0;                /* Memory limit override */
  int nMemThisDb = 0;          /* Memory limit set by the CONFIG table */
  char *zExpDb = 0;            /* Write Databases to files in this directory */

Changes to test/hook.test.

79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
do_test hook-3.7 {
  set ::commit_cnt
} {1 2 2 3 3 4 4 5 5 6 6 7}
do_test hook-3.8 {
  execsql {SELECT * FROM t2}
} {1 2 2 3 3 4 4 5 5 6}

# Test turnning off the commit hook
#
do_test hook-3.9 {
  db commit_hook {}
  set ::commit_cnt {}
  execsql {
    INSERT INTO t2 VALUES(7,8);
  }







|







79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
do_test hook-3.7 {
  set ::commit_cnt
} {1 2 2 3 3 4 4 5 5 6 6 7}
do_test hook-3.8 {
  execsql {SELECT * FROM t2}
} {1 2 2 3 3 4 4 5 5 6}

# Test turning off the commit hook
#
do_test hook-3.9 {
  db commit_hook {}
  set ::commit_cnt {}
  execsql {
    INSERT INTO t2 VALUES(7,8);
  }

Changes to test/in.test.

637
638
639
640
641
642
643
644
645
646
647
648
649
650
651


do_test in-13.X {
  db nullvalue ""
} {}

# At one point the following was causing valgrind to report a "jump
# depends on unitialized location" problem.
#
do_execsql_test in-14.0 {
  CREATE TABLE c1(a);
  INSERT INTO c1 VALUES(1), (2), (4), (3);
}
do_execsql_test in-14.1 {
  SELECT * FROM c1 WHERE a IN (SELECT a FROM c1) ORDER BY 1







|







637
638
639
640
641
642
643
644
645
646
647
648
649
650
651


do_test in-13.X {
  db nullvalue ""
} {}

# At one point the following was causing valgrind to report a "jump
# depends on uninitialized location" problem.
#
do_execsql_test in-14.0 {
  CREATE TABLE c1(a);
  INSERT INTO c1 VALUES(1), (2), (4), (3);
}
do_execsql_test in-14.1 {
  SELECT * FROM c1 WHERE a IN (SELECT a FROM c1) ORDER BY 1

Changes to test/indexfault.test.

167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
...
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
#        second temporary file.
#   3.4: As 7.3, but with a low (50KB) soft-heap-limit.
#
#   3.5: After a certain amount of data has been read from the main database
#        file (and written into the temporary b-tree), sqlite3_release_memory()
#        is called to free as much memory as possible. This causes the temp
#        b-tree to be flushed to disk. So that before its contents can be 
#        transfered to a PMA they must be read back from disk - creating extra
#        opportunities for IO errors.
#
install_custom_faultsim

# Set up a table to build indexes on. Save the setup using the 
# [faultsim_save_and_close] mechanism.
# 
................................................................................

uninstall_custom_faultsim

#-------------------------------------------------------------------------
# Test 4: After a certain amount of data has been read from the main database
# file (and written into the temporary b-tree), sqlite3_release_memory() is
# called to free as much memory as possible. This causes the temp b-tree to be
# flushed to disk. So that before its contents can be transfered to a PMA they
# must be read back from disk - creating extra opportunities for IO errors.
# 
install_custom_faultsim

catch { db close }
forcedelete test.db
sqlite3 db test.db







|







 







|







167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
...
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
#        second temporary file.
#   3.4: As 7.3, but with a low (50KB) soft-heap-limit.
#
#   3.5: After a certain amount of data has been read from the main database
#        file (and written into the temporary b-tree), sqlite3_release_memory()
#        is called to free as much memory as possible. This causes the temp
#        b-tree to be flushed to disk. So that before its contents can be 
#        transferred to a PMA they must be read back from disk - creating extra
#        opportunities for IO errors.
#
install_custom_faultsim

# Set up a table to build indexes on. Save the setup using the 
# [faultsim_save_and_close] mechanism.
# 
................................................................................

uninstall_custom_faultsim

#-------------------------------------------------------------------------
# Test 4: After a certain amount of data has been read from the main database
# file (and written into the temporary b-tree), sqlite3_release_memory() is
# called to free as much memory as possible. This causes the temp b-tree to be
# flushed to disk. So that before its contents can be transferred to a PMA they
# must be read back from disk - creating extra opportunities for IO errors.
# 
install_custom_faultsim

catch { db close }
forcedelete test.db
sqlite3 db test.db

Changes to test/insert.test.

325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
      DROP TABLE t1;
    }
  } {}
}

# Test that the special optimization for queries of the form 
# "SELECT max(x) FROM tbl" where there is an index on tbl(x) works with 
# INSERT statments.
do_test insert-7.1 {
  execsql {
    CREATE TABLE t1(a);
    INSERT INTO t1 VALUES(1);
    INSERT INTO t1 VALUES(2);
    CREATE INDEX i1 ON t1(a);
  }







|







325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
      DROP TABLE t1;
    }
  } {}
}

# Test that the special optimization for queries of the form 
# "SELECT max(x) FROM tbl" where there is an index on tbl(x) works with 
# INSERT statements.
do_test insert-7.1 {
  execsql {
    CREATE TABLE t1(a);
    INSERT INTO t1 VALUES(1);
    INSERT INTO t1 VALUES(2);
    CREATE INDEX i1 ON t1(a);
  }

Changes to test/insert4.test.

129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
# not, as appropriate.
#
#     xfer_check TESTID  XFER-USED   INIT-DATA   DEST-SCHEMA   SRC-SCHEMA 
#
# The TESTID argument is the symbolic name for this test.  The XFER-USED
# argument is true if the transfer optimization should be employed and
# false if not.  INIT-DATA is a single row of data that is to be 
# transfered.  DEST-SCHEMA and SRC-SCHEMA are table declarations for
# the destination and source tables.
#
proc xfer_check {testid xferused initdata destschema srcschema} {
  execsql "CREATE TABLE dest($destschema)"
  execsql "CREATE TABLE src($srcschema)"
  execsql "INSERT INTO src VALUES([join $initdata ,])"
  set ::sqlite3_xferopt_count 0







|







129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
# not, as appropriate.
#
#     xfer_check TESTID  XFER-USED   INIT-DATA   DEST-SCHEMA   SRC-SCHEMA 
#
# The TESTID argument is the symbolic name for this test.  The XFER-USED
# argument is true if the transfer optimization should be employed and
# false if not.  INIT-DATA is a single row of data that is to be 
# transferred.  DEST-SCHEMA and SRC-SCHEMA are table declarations for
# the destination and source tables.
#
proc xfer_check {testid xferused initdata destschema srcschema} {
  execsql "CREATE TABLE dest($destschema)"
  execsql "CREATE TABLE src($srcschema)"
  execsql "INSERT INTO src VALUES([join $initdata ,])"
  set ::sqlite3_xferopt_count 0

Changes to test/join.test.

1022
1023
1024
1025
1026
1027
1028
1029
} {}

do_execsql_test join-24.2 {
  SELECT * FROM t2 LEFT JOIN t1 ON a=0 WHERE (x='x' OR x IS NULL);
} {1 {} {}}

finish_test








<
1022
1023
1024
1025
1026
1027
1028

} {}

do_execsql_test join-24.2 {
  SELECT * FROM t2 LEFT JOIN t1 ON a=0 WHERE (x='x' OR x IS NULL);
} {1 {} {}}

finish_test

Changes to test/journal1.test.

16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
# $Id: journal1.test,v 1.2 2005/03/20 22:54:56 drh Exp $


set testdir [file dirname $argv0]
source $testdir/tester.tcl

# These tests will not work on windows because windows uses
# manditory file locking which breaks the copy_file command.
#
# Or with atomic_batch_write systems, as journal files are
# not created.
#
if {$tcl_platform(platform)=="windows"
 || [atomic_batch_write test.db]
} {







|







16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
# $Id: journal1.test,v 1.2 2005/03/20 22:54:56 drh Exp $


set testdir [file dirname $argv0]
source $testdir/tester.tcl

# These tests will not work on windows because windows uses
# mandatory file locking which breaks the copy_file command.
#
# Or with atomic_batch_write systems, as journal files are
# not created.
#
if {$tcl_platform(platform)=="windows"
 || [atomic_batch_write test.db]
} {

Changes to test/json101.test.

719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
do_execsql_test json-11.3 {
  /* Too deep by one { */
  SELECT json_valid(replace(printf('%.2001c0%.2001c','[','}'),'[','{"a":'));
  /* } */
} {0}

# 2017-10-27.  Demonstrate the ability to access an element from
# a json structure even though the element name constains a "."
# character, by quoting the element name in the path.
#
do_execsql_test json-12.100 {
  CREATE TABLE t12(x);
  INSERT INTO t12(x) VALUES(
    '{"settings":
        {"layer2":







|







719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
do_execsql_test json-11.3 {
  /* Too deep by one { */
  SELECT json_valid(replace(printf('%.2001c0%.2001c','[','}'),'[','{"a":'));
  /* } */
} {0}

# 2017-10-27.  Demonstrate the ability to access an element from
# a json structure even though the element name contains a "."
# character, by quoting the element name in the path.
#
do_execsql_test json-12.100 {
  CREATE TABLE t12(x);
  INSERT INTO t12(x) VALUES(
    '{"settings":
        {"layer2":

Changes to test/kvtest.c.

20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
** HOW TO COMPILE:
**
** (1) Gather this source file and a recent SQLite3 amalgamation with its
**     header into the working directory.  You should have:
**
**          kvtest.c       >--- this file
**          sqlite3.c      \___ SQLite
**          sqlite3.h      /    amlagamation & header
**
** (2) Run you compiler against the two C source code files.
**
**    (a) On linux or mac:
**
**        OPTS="-DSQLITE_THREADSAFE=0 -DSQLITE_OMIT_LOAD_EXTENSION"
**        gcc -Os -I. $OPTS kvtest.c sqlite3.c -o kvtest







|







20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
** HOW TO COMPILE:
**
** (1) Gather this source file and a recent SQLite3 amalgamation with its
**     header into the working directory.  You should have:
**
**          kvtest.c       >--- this file
**          sqlite3.c      \___ SQLite
**          sqlite3.h      /    amalagmation & header
**
** (2) Run you compiler against the two C source code files.
**
**    (a) On linux or mac:
**
**        OPTS="-DSQLITE_THREADSAFE=0 -DSQLITE_OMIT_LOAD_EXTENSION"
**        gcc -Os -I. $OPTS kvtest.c sqlite3.c -o kvtest

Changes to test/lock.test.

173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
#
do_test lock-2.2 {
  catchsql {SELECT * FROM t2} db2
} {0 {9 8}}

# If the other thread (the one that does not hold the transaction with
# a RESERVED lock) tries to get a RESERVED lock, we do get a busy callback
# as long as we were not orginally holding a READ lock.
#
do_test lock-2.3.1 {
  proc callback {count} {
    set ::callback_value $count
    break
  }
  set ::callback_value {}







|







173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
#
do_test lock-2.2 {
  catchsql {SELECT * FROM t2} db2
} {0 {9 8}}

# If the other thread (the one that does not hold the transaction with
# a RESERVED lock) tries to get a RESERVED lock, we do get a busy callback
# as long as we were not originally holding a READ lock.
#
do_test lock-2.3.1 {
  proc callback {count} {
    set ::callback_value $count
    break
  }
  set ::callback_value {}

Changes to test/lock2.test.

24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
# lock2-1.1: Connect a second process to the database.
# lock2-1.2: Establish a RESERVED lock with this process.
# lock2-1.3: Get a SHARED lock with the second process.
# lock2-1.4: Try for a RESERVED lock with process 2. This fails.
# lock2-1.5: Try to upgrade the first process to EXCLUSIVE, this fails so
#            it gets PENDING.
# lock2-1.6: Release the SHARED lock held by the second process. 
# lock2-1.7: Attempt to reaquire a SHARED lock with the second process.
#            this fails due to the PENDING lock.
# lock2-1.8: Ensure the first process can now upgrade to EXCLUSIVE.
#
do_test lock2-1.1 {
  set ::tf1 [launch_testfixture]
  testfixture $::tf1 {
    sqlite3 db test.db -key xyzzy







|







24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
# lock2-1.1: Connect a second process to the database.
# lock2-1.2: Establish a RESERVED lock with this process.
# lock2-1.3: Get a SHARED lock with the second process.
# lock2-1.4: Try for a RESERVED lock with process 2. This fails.
# lock2-1.5: Try to upgrade the first process to EXCLUSIVE, this fails so
#            it gets PENDING.
# lock2-1.6: Release the SHARED lock held by the second process. 
# lock2-1.7: Attempt to reacquire a SHARED lock with the second process.
#            this fails due to the PENDING lock.
# lock2-1.8: Ensure the first process can now upgrade to EXCLUSIVE.
#
do_test lock2-1.1 {
  set ::tf1 [launch_testfixture]
  testfixture $::tf1 {
    sqlite3 db test.db -key xyzzy

Changes to test/lock_common.tcl.

199
200
201
202
203
204
205
206
        append script "\n"
      }
    }
  }]
  close $fd
  set main_loop_written 1
}








<
199
200
201
202
203
204
205

        append script "\n"
      }
    }
  }]
  close $fd
  set main_loop_written 1
}

Changes to test/malloc_common.tcl.

367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
    # injected into the system this trial.
    #
    eval $O(-injectstart) $iFail
    set rc [catch $O(-body) res]
    set nfail [eval $O(-injectstop)]

    # Run the -test script. If it throws no error, consider this trial
    # sucessful. If it does throw an error, cause a [do_test] test to
    # fail (and print out the unexpected exception thrown by the -test
    # script at the same time).
    #
    set rc [catch [list faultsim_test_proc $rc $res $nfail] res]
    if {$rc == 0} {set res ok}
    do_test $testname.$iFail [list list $rc $res] {0 ok}








|







367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
    # injected into the system this trial.
    #
    eval $O(-injectstart) $iFail
    set rc [catch $O(-body) res]
    set nfail [eval $O(-injectstop)]

    # Run the -test script. If it throws no error, consider this trial
    # successful. If it does throw an error, cause a [do_test] test to
    # fail (and print out the unexpected exception thrown by the -test
    # script at the same time).
    #
    set rc [catch [list faultsim_test_proc $rc $res $nfail] res]
    if {$rc == 0} {set res ok}
    do_test $testname.$iFail [list list $rc $res] {0 ok}

Changes to test/malloctraceviewer.tcl.

246
247
248
249
250
251
252
253
  }
}

open_database
bind $O(tree) <<TreeviewSelect>> [list populate_text_widget mddb]

populate_tree_widget mddb [mddb one {SELECT zTest FROM malloc LIMIT 1}]








<
246
247
248
249
250
251
252

  }
}

open_database
bind $O(tree) <<TreeviewSelect>> [list populate_text_widget mddb]

populate_tree_widget mddb [mddb one {SELECT zTest FROM malloc LIMIT 1}]

Changes to test/misc1.test.

341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
  set rc [catch {db2 eval {SELECT count(*) FROM t1}} msg]
  db2 close
  lappend rc $msg
} {0 3}

# Make sure string comparisons really do compare strings in format4+.
# Similar tests in the format3.test file show that for format3 and earlier
# all comparisions where numeric if either operand looked like a number.
#
do_test misc1-12.1 {
  execsql {SELECT '0'=='0.0'}
} {0}
do_test misc1-12.2 {
  execsql {SELECT '0'==0.0}
} {0}







|







341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
  set rc [catch {db2 eval {SELECT count(*) FROM t1}} msg]
  db2 close
  lappend rc $msg
} {0 3}

# Make sure string comparisons really do compare strings in format4+.
# Similar tests in the format3.test file show that for format3 and earlier
# all comparisons where numeric if either operand looked like a number.
#
do_test misc1-12.1 {
  execsql {SELECT '0'=='0.0'}
} {0}
do_test misc1-12.2 {
  execsql {SELECT '0'==0.0}
} {0}

Changes to test/misc4.test.

216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
do_catchsql_test misc4-7.1 {
  CREATE TABLE t7(x);
  PRAGMA writable_schema=ON;
  UPDATE sqlite_master SET sql='CREATE TABLE [M%s%s%s%s%s%s%s%s%s%s%s%s%s';
  VACUUM;
} {1 {unrecognized token: "[M%s%s%s%s%s%s%s%s%s%s%s%s%s"}}

# 2015-05-18.  Use of ephermeral Mem content after the cursor that holds
# the canonical content has moved on.
#
do_execsql_test misc4-7.2 {
  CREATE TABLE t0(a,b);
  INSERT INTO t0 VALUES(1,0),(2,0);
  UPDATE t0 SET b=9 WHERE a AND (SELECT a FROM t0 WHERE a);
  SELECT * FROM t0 ORDER BY +a;
} {1 9 2 9}

finish_test







|










216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
do_catchsql_test misc4-7.1 {
  CREATE TABLE t7(x);
  PRAGMA writable_schema=ON;
  UPDATE sqlite_master SET sql='CREATE TABLE [M%s%s%s%s%s%s%s%s%s%s%s%s%s';
  VACUUM;
} {1 {unrecognized token: "[M%s%s%s%s%s%s%s%s%s%s%s%s%s"}}

# 2015-05-18.  Use of ephemeral Mem content after the cursor that holds
# the canonical content has moved on.
#
do_execsql_test misc4-7.2 {
  CREATE TABLE t0(a,b);
  INSERT INTO t0 VALUES(1,0),(2,0);
  UPDATE t0 SET b=9 WHERE a AND (SELECT a FROM t0 WHERE a);
  SELECT * FROM t0 ORDER BY +a;
} {1 9 2 9}

finish_test

Changes to test/pg_common.tcl.

168
169
170
171
172
173
174
175
  puts $::fd finish_test
  close $::fd
}

proc ifcapable {arg} {
   puts $::fd "ifcapable $arg { finish_test ; return }"
}








<
168
169
170
171
172
173
174

  puts $::fd finish_test
  close $::fd
}

proc ifcapable {arg} {
   puts $::fd "ifcapable $arg { finish_test ; return }"
}

Changes to test/releasetest.tcl.

736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
  set cflags   [join $cflags " "]
  set opts     [join $opts " "]

  lappend alltests [list \
      $title $dir $configOpts $testtarget $makeOpts $cflags $opts]
}

# The following procedure returns the "configure" command to be exectued for
# the current platform, which may be Windows (via MinGW, etc).
#
proc configureCommand {opts} {
  if {$::MSVC} return [list]; # This is not needed for MSVC.
  set result [list trace_cmd exec]
  if {$::tcl_platform(platform)=="windows"} {
    lappend result sh







|







736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
  set cflags   [join $cflags " "]
  set opts     [join $opts " "]

  lappend alltests [list \
      $title $dir $configOpts $testtarget $makeOpts $cflags $opts]
}

# The following procedure returns the "configure" command to be executed for
# the current platform, which may be Windows (via MinGW, etc).
#
proc configureCommand {opts} {
  if {$::MSVC} return [list]; # This is not needed for MSVC.
  set result [list trace_cmd exec]
  if {$::tcl_platform(platform)=="windows"} {
    lappend result sh

Changes to test/releasetest_data.tcl.

601
602
603
604
605
606
607
608
609
} elseif {[string match ${cmd}* platforms] && $n==0} {
  main_platforms
} elseif {[string match ${cmd}* tests]} {
  main_tests {*}[lrange $argv 1 end]
} else {
  usage
}









<
<
601
602
603
604
605
606
607


} elseif {[string match ${cmd}* platforms] && $n==0} {
  main_platforms
} elseif {[string match ${cmd}* tests]} {
  main_tests {*}[lrange $argv 1 end]
} else {
  usage
}


Changes to test/rollback2.test.

38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
#
# where SWITCHES are:
#
#   -setup      SQL script to open transaction and begin writing.
#   -select     SELECT to execute after -setup script
#   -result     Expected result of -select statement
#   -rollback   Use this SQL command ("ROLLBACK" or "ROLLBACK TO ...") to
#               rollback the transaction in the middle of the -select statment
#               execution.
#
proc do_rollback_test {tn args} {
  set A(-setup)    ""
  set A(-select)   ""
  set A(-result)   ""
  set A(-rollback) ROLLBACK







|







38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
#
# where SWITCHES are:
#
#   -setup      SQL script to open transaction and begin writing.
#   -select     SELECT to execute after -setup script
#   -result     Expected result of -select statement
#   -rollback   Use this SQL command ("ROLLBACK" or "ROLLBACK TO ...") to
#               rollback the transaction in the middle of the -select statement
#               execution.
#
proc do_rollback_test {tn args} {
  set A(-setup)    ""
  set A(-select)   ""
  set A(-result)   ""
  set A(-rollback) ROLLBACK

Changes to test/rowid.test.

685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
  execsql {SELECT rowid, a FROM t5 WHERE rowid<='abc' ORDER BY 1 DESC}
} {8 8 7 7 6 6 5 5 4 4 3 3 2 2 1 1}

# Test the automatic generation of rowids when the table already contains
# a rowid with the maximum value.
#
# Once the maximum rowid is taken, rowids are normally chosen at
# random.  By by reseting the random number generator, we can cause
# the rowid guessing loop to collide with prior rowids, and test the
# loop out to its limit of 100 iterations.  After 100 collisions, the
# rowid guesser gives up and reports SQLITE_FULL.
#
do_test rowid-12.1 {
  execsql {
    CREATE TABLE t7(x INTEGER PRIMARY KEY, y);







|







685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
  execsql {SELECT rowid, a FROM t5 WHERE rowid<='abc' ORDER BY 1 DESC}
} {8 8 7 7 6 6 5 5 4 4 3 3 2 2 1 1}

# Test the automatic generation of rowids when the table already contains
# a rowid with the maximum value.
#
# Once the maximum rowid is taken, rowids are normally chosen at
# random.  By by resetting the random number generator, we can cause
# the rowid guessing loop to collide with prior rowids, and test the
# loop out to its limit of 100 iterations.  After 100 collisions, the
# rowid guesser gives up and reports SQLITE_FULL.
#
do_test rowid-12.1 {
  execsql {
    CREATE TABLE t7(x INTEGER PRIMARY KEY, y);

Changes to test/savepoint.test.

14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
set testdir [file dirname $argv0]
source $testdir/tester.tcl
source $testdir/lock_common.tcl
source $testdir/malloc_common.tcl

#----------------------------------------------------------------------
# The following tests - savepoint-1.* - test that the SAVEPOINT, RELEASE
# and ROLLBACK TO comands are correctly parsed, and that the auto-commit
# flag is correctly set and unset as a result.
#
do_test savepoint-1.1 {
  wal_set_journal_mode
  execsql {
    SAVEPOINT sp1;
    RELEASE sp1;







|







14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
set testdir [file dirname $argv0]
source $testdir/tester.tcl
source $testdir/lock_common.tcl
source $testdir/malloc_common.tcl

#----------------------------------------------------------------------
# The following tests - savepoint-1.* - test that the SAVEPOINT, RELEASE
# and ROLLBACK TO commands are correctly parsed, and that the auto-commit
# flag is correctly set and unset as a result.
#
do_test savepoint-1.1 {
  wal_set_journal_mode
  execsql {
    SAVEPOINT sp1;
    RELEASE sp1;

Changes to test/sessionfuzz.c.

13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
** This file implements a program used for fuzz-testing the session
** module.
**
** Usage:
**
**      sessionfuzz setup         -- Generate starter test cases
**      sessionfuzz run FILE ...  -- Run a test fuzz on FILE
**      sesssiofuzz run SQLAR ... -- Run all test cases in the SQL Archive
**
** Compiling:
**
**    (1) Have a version of SQLite that supports SQLITE_ENABLE_MEMDB
**        in the local directory.
**    (2) Run:
**







|







13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
** This file implements a program used for fuzz-testing the session
** module.
**
** Usage:
**
**      sessionfuzz setup         -- Generate starter test cases
**      sessionfuzz run FILE ...  -- Run a test fuzz on FILE
**      sessionfuzz run SQLAR ... -- Run all test cases in the SQL Archive
**
** Compiling:
**
**    (1) Have a version of SQLite that supports SQLITE_ENABLE_MEMDB
**        in the local directory.
**    (2) Run:
**

Changes to test/shared.test.

310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
do_test shared-$av.4.1.3 {
  execsql {ATTACH 'test.db' AS test} db2
  set sqlite_open_file_count
  expr $sqlite_open_file_count-($extrafds_postlock*2)
} {2}

# Sanity check: Create a table in ./test.db via handle db, and test that handle
# db2 can "see" the new table immediately. A handle using a seperate pager
# cache would have to reload the database schema before this were possible.
#
do_test shared-$av.4.2.1 {
  execsql {
    CREATE TABLE abc(a, b, c);
    CREATE TABLE def(d, e, f);
    INSERT INTO abc VALUES('i', 'ii', 'iii');







|







310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
do_test shared-$av.4.1.3 {
  execsql {ATTACH 'test.db' AS test} db2
  set sqlite_open_file_count
  expr $sqlite_open_file_count-($extrafds_postlock*2)
} {2}

# Sanity check: Create a table in ./test.db via handle db, and test that handle
# db2 can "see" the new table immediately. A handle using a separate pager
# cache would have to reload the database schema before this were possible.
#
do_test shared-$av.4.2.1 {
  execsql {
    CREATE TABLE abc(a, b, c);
    CREATE TABLE def(d, e, f);
    INSERT INTO abc VALUES('i', 'ii', 'iii');

Changes to test/shell3.test.

12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
..
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
# The focus of this file is testing the CLI shell tool.
#
# $Id: shell2.test,v 1.7 2009/07/17 16:54:48 shaneh Exp $
#

# Test plan:
#
#   shell3-1.*: Basic tests for running SQL statments from command line.
#   shell3-2.*: Basic tests for running SQL file from command line.
#
set testdir [file dirname $argv0]
source $testdir/tester.tcl
set CLI [test_find_cli]
db close
forcedelete test.db test.db-journal test.db-wal
................................................................................
#
if {$::tcl_platform(platform)=="windows"} {
  finish_test
  return
}

#----------------------------------------------------------------------------
#   shell3-1.*: Basic tests for running SQL statments from command line.
#

# Run SQL statement from command line
do_test shell3-1.1 {
  forcedelete foo.db
  set rc [ catchcmd "foo.db \"CREATE TABLE t1(a);\"" ]
  set fexist [file exist foo.db]







|







 







|







12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
..
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
# The focus of this file is testing the CLI shell tool.
#
# $Id: shell2.test,v 1.7 2009/07/17 16:54:48 shaneh Exp $
#

# Test plan:
#
#   shell3-1.*: Basic tests for running SQL statements from command line.
#   shell3-2.*: Basic tests for running SQL file from command line.
#
set testdir [file dirname $argv0]
source $testdir/tester.tcl
set CLI [test_find_cli]
db close
forcedelete test.db test.db-journal test.db-wal
................................................................................
#
if {$::tcl_platform(platform)=="windows"} {
  finish_test
  return
}

#----------------------------------------------------------------------------
#   shell3-1.*: Basic tests for running SQL statements from command line.
#

# Run SQL statement from command line
do_test shell3-1.1 {
  forcedelete foo.db
  set rc [ catchcmd "foo.db \"CREATE TABLE t1(a);\"" ]
  set fexist [file exist foo.db]

Changes to test/speed1.test.

27
28
29
30
31
32
33
34
35
36
37
38
39
40
41

set sqlout [open speed1.txt w]
proc tracesql {sql} {
  puts $::sqlout $sql\;
}
#db trace tracesql

# The number_name procedure below converts its argment (an integer)
# into a string which is the English-language name for that number.
#
# Example:
#
#     puts [number_name 123]   ->  "one hundred twenty three"
#
set ones {zero one two three four five six seven eight nine







|







27
28
29
30
31
32
33
34
35
36
37
38
39
40
41

set sqlout [open speed1.txt w]
proc tracesql {sql} {
  puts $::sqlout $sql\;
}
#db trace tracesql

# The number_name procedure below converts its argument (an integer)
# into a string which is the English-language name for that number.
#
# Example:
#
#     puts [number_name 123]   ->  "one hundred twenty three"
#
set ones {zero one two three four five six seven eight nine

Changes to test/speed1p.explain.

25
26
27
28
29
30
31
32
33
34
35
36
37
38
39

set sqlout [open speed1.txt w]
proc tracesql {sql} {
  puts $::sqlout $sql\;
}
#db trace tracesql

# The number_name procedure below converts its argment (an integer)
# into a string which is the English-language name for that number.
#
# Example:
#
#     puts [number_name 123]   ->  "one hundred twenty three"
#
set ones {zero one two three four five six seven eight nine







|







25
26
27
28
29
30
31
32
33
34
35
36
37
38
39

set sqlout [open speed1.txt w]
proc tracesql {sql} {
  puts $::sqlout $sql\;
}
#db trace tracesql

# The number_name procedure below converts its argument (an integer)
# into a string which is the English-language name for that number.
#
# Example:
#
#     puts [number_name 123]   ->  "one hundred twenty three"
#
set ones {zero one two three four five six seven eight nine

Changes to test/speed1p.test.

25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
speed_trial_init speed1

sqlite3_memdebug_vfs_oom_test 0

# Set a uniform random seed
expr srand(0)

# The number_name procedure below converts its argment (an integer)
# into a string which is the English-language name for that number.
#
# Example:
#
#     puts [number_name 123]   ->  "one hundred twenty three"
#
set ones {zero one two three four five six seven eight nine







|







25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
speed_trial_init speed1

sqlite3_memdebug_vfs_oom_test 0

# Set a uniform random seed
expr srand(0)

# The number_name procedure below converts its argument (an integer)
# into a string which is the English-language name for that number.
#
# Example:
#
#     puts [number_name 123]   ->  "one hundred twenty three"
#
set ones {zero one two three four five six seven eight nine

Changes to test/speed2.test.

23
24
25
26
27
28
29
30
31
32
33
34
35
36
37

set sqlout [open speed2.txt w]
proc tracesql {sql} {
  puts $::sqlout $sql\;
}
#db trace tracesql

# The number_name procedure below converts its argment (an integer)
# into a string which is the English-language name for that number.
#
# Example:
#
#     puts [number_name 123]   ->  "one hundred twenty three"
#
set ones {zero one two three four five six seven eight nine







|







23
24
25
26
27
28
29
30
31
32
33
34
35
36
37

set sqlout [open speed2.txt w]
proc tracesql {sql} {
  puts $::sqlout $sql\;
}
#db trace tracesql

# The number_name procedure below converts its argument (an integer)
# into a string which is the English-language name for that number.
#
# Example:
#
#     puts [number_name 123]   ->  "one hundred twenty three"
#
set ones {zero one two three four five six seven eight nine

Changes to test/speed3.test.

43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
speed_trial_init speed1

# Set a uniform random seed
expr srand(0)

set ::NROW 1000

# The number_name procedure below converts its argment (an integer)
# into a string which is the English-language name for that number.
#
# Example:
#
#     puts [number_name 123]   ->  "one hundred twenty three"
#
set ones {zero one two three four five six seven eight nine







|







43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
speed_trial_init speed1

# Set a uniform random seed
expr srand(0)

set ::NROW 1000

# The number_name procedure below converts its argument (an integer)
# into a string which is the English-language name for that number.
#
# Example:
#
#     puts [number_name 123]   ->  "one hundred twenty three"
#
set ones {zero one two three four five six seven eight nine

Changes to test/speed4.test.

29
30
31
32
33
34
35
36
37
38
39
40
41
42
43

set sqlout [open speed1.txt w]
proc tracesql {sql} {
  puts $::sqlout $sql\;
}
#db trace tracesql

# The number_name procedure below converts its argment (an integer)
# into a string which is the English-language name for that number.
#
# Example:
#
#     puts [number_name 123]   ->  "one hundred twenty three"
#
set ones {zero one two three four five six seven eight nine







|







29
30
31
32
33
34
35
36
37
38
39
40
41
42
43

set sqlout [open speed1.txt w]
proc tracesql {sql} {
  puts $::sqlout $sql\;
}
#db trace tracesql

# The number_name procedure below converts its argument (an integer)
# into a string which is the English-language name for that number.
#
# Example:
#
#     puts [number_name 123]   ->  "one hundred twenty three"
#
set ones {zero one two three four five six seven eight nine

Changes to test/speed4p.explain.

29
30
31
32
33
34
35
36
37
38
39
40
41
42
43

set sqlout [open speed1.txt w]
proc tracesql {sql} {
  puts $::sqlout $sql\;
}
#db trace tracesql

# The number_name procedure below converts its argment (an integer)
# into a string which is the English-language name for that number.
#
# Example:
#
#     puts [number_name 123]   ->  "one hundred twenty three"
#
set ones {zero one two three four five six seven eight nine







|







29
30
31
32
33
34
35
36
37
38
39
40
41
42
43

set sqlout [open speed1.txt w]
proc tracesql {sql} {
  puts $::sqlout $sql\;
}
#db trace tracesql

# The number_name procedure below converts its argument (an integer)
# into a string which is the English-language name for that number.
#
# Example:
#
#     puts [number_name 123]   ->  "one hundred twenty three"
#
set ones {zero one two three four five six seven eight nine

Changes to test/speed4p.test.

29
30
31
32
33
34
35
36
37
38
39
40
41
42
43

set sqlout [open speed1.txt w]
proc tracesql {sql} {
  puts $::sqlout $sql\;
}
#db trace tracesql

# The number_name procedure below converts its argment (an integer)
# into a string which is the English-language name for that number.
#
# Example:
#
#     puts [number_name 123]   ->  "one hundred twenty three"
#
set ones {zero one two three four five six seven eight nine







|







29
30
31
32
33
34
35
36
37
38
39
40
41
42
43

set sqlout [open speed1.txt w]
proc tracesql {sql} {
  puts $::sqlout $sql\;
}
#db trace tracesql

# The number_name procedure below converts its argument (an integer)
# into a string which is the English-language name for that number.
#
# Example:
#
#     puts [number_name 123]   ->  "one hundred twenty three"
#
set ones {zero one two three four five six seven eight nine

Changes to test/speedtest1.c.

7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
...
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
...
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
  "Usage: %s [--options] DATABASE\n"
  "Options:\n"
  "  --autovacuum        Enable AUTOVACUUM mode\n"
  "  --cachesize N       Set the cache size to N\n"
  "  --exclusive         Enable locking_mode=EXCLUSIVE\n"
  "  --explain           Like --sqlonly but with added EXPLAIN keywords\n"
  "  --heap SZ MIN       Memory allocator uses SZ bytes & min allocation MIN\n"
  "  --incrvacuum        Enable incremenatal vacuum mode\n"
  "  --journal M         Set the journal_mode to M\n"
  "  --key KEY           Set the encryption key to KEY\n"
  "  --lookaside N SZ    Configure lookaside for N slots of SZ bytes each\n"
  "  --memdb             Use an in-memory database\n"
  "  --mmap SZ           MMAP the first SZ bytes of the database file\n"
  "  --multithread       Set multithreaded mode\n"
  "  --nomemstat         Disable memory statistics\n"
................................................................................
unsigned int speedtest1_random(void){
  g.x = (g.x>>1) ^ ((1+~(g.x&1)) & 0xd0000001);
  g.y = g.y*1103515245 + 12345;
  return g.x ^ g.y;
}

/* Map the value in within the range of 1...limit into another
** number in a way that is chatic and invertable.
*/
unsigned swizzle(unsigned in, unsigned limit){
  unsigned out = 0;
  while( limit ){
    out = (out<<1) | (in&1);
    in >>= 1;
    limit >>= 1;
................................................................................
*/
unsigned roundup_allones(unsigned limit){
  unsigned m = 1;
  while( m<limit ) m = (m<<1)+1;
  return m;
}

/* The speedtest1_numbername procedure below converts its argment (an integer)
** into a string which is the English-language name for that number.
** The returned string should be freed with sqlite3_free().
**
** Example:
**
**     speedtest1_numbername(123)   ->  "one hundred twenty three"
*/







|







 







|







 







|







7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
...
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
...
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
  "Usage: %s [--options] DATABASE\n"
  "Options:\n"
  "  --autovacuum        Enable AUTOVACUUM mode\n"
  "  --cachesize N       Set the cache size to N\n"
  "  --exclusive         Enable locking_mode=EXCLUSIVE\n"
  "  --explain           Like --sqlonly but with added EXPLAIN keywords\n"
  "  --heap SZ MIN       Memory allocator uses SZ bytes & min allocation MIN\n"
  "  --incrvacuum        Enable incremental vacuum mode\n"
  "  --journal M         Set the journal_mode to M\n"
  "  --key KEY           Set the encryption key to KEY\n"
  "  --lookaside N SZ    Configure lookaside for N slots of SZ bytes each\n"
  "  --memdb             Use an in-memory database\n"
  "  --mmap SZ           MMAP the first SZ bytes of the database file\n"
  "  --multithread       Set multithreaded mode\n"
  "  --nomemstat         Disable memory statistics\n"
................................................................................
unsigned int speedtest1_random(void){
  g.x = (g.x>>1) ^ ((1+~(g.x&1)) & 0xd0000001);
  g.y = g.y*1103515245 + 12345;
  return g.x ^ g.y;
}

/* Map the value in within the range of 1...limit into another
** number in a way that is chaotic and invertable.
*/
unsigned swizzle(unsigned in, unsigned limit){
  unsigned out = 0;
  while( limit ){
    out = (out<<1) | (in&1);
    in >>= 1;
    limit >>= 1;
................................................................................
*/
unsigned roundup_allones(unsigned limit){
  unsigned m = 1;
  while( m<limit ) m = (m<<1)+1;
  return m;
}

/* The speedtest1_numbername procedure below converts its argument (an integer)
** into a string which is the English-language name for that number.
** The returned string should be freed with sqlite3_free().
**
** Example:
**
**     speedtest1_numbername(123)   ->  "one hundred twenty three"
*/

Changes to test/subquery.test.

220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
    INSERT INTO t4 VALUES('10.0');
  }
} {}
do_test subquery-2.5.2 {
  # In the expr "x IN (SELECT a FROM t3)" the RHS of the IN operator
  # has text affinity and the LHS has integer affinity.  The rule is
  # that we try to convert both sides to an integer before doing the
  # comparision.  Hence, the integer value 10 in t3 will compare equal
  # to the string value '10.0' in t4 because the t4 value will be
  # converted into an integer.
  execsql {
    SELECT * FROM t4 WHERE x IN (SELECT a FROM t3);
  }
} {10.0}
do_test subquery-2.5.3.1 {







|







220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
    INSERT INTO t4 VALUES('10.0');
  }
} {}
do_test subquery-2.5.2 {
  # In the expr "x IN (SELECT a FROM t3)" the RHS of the IN operator
  # has text affinity and the LHS has integer affinity.  The rule is
  # that we try to convert both sides to an integer before doing the
  # comparison.  Hence, the integer value 10 in t3 will compare equal
  # to the string value '10.0' in t4 because the t4 value will be
  # converted into an integer.
  execsql {
    SELECT * FROM t4 WHERE x IN (SELECT a FROM t3);
  }
} {10.0}
do_test subquery-2.5.3.1 {

Changes to test/tableapi.test.

226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
ifcapable schema_pragmas {
  do_test tableapi-6.1 {
    sqlite3_get_table_printf $::dbx {PRAGMA user_version} {}
  } {0 1 1 user_version 0}
}

# do_malloc_test closes and deletes the usual db connections and files on
# each iteration.  $::dbx is a seperate connection, and on Windows, will
# cause the file deletion of test.db to fail, so we move the close of $::dbx
# up to here before the do_malloc_test.
do_test tableapi-99.0 {
  sqlite3_close $::dbx
} {SQLITE_OK}

do_malloc_test tableapi-7 -sqlprep {







|







226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
ifcapable schema_pragmas {
  do_test tableapi-6.1 {
    sqlite3_get_table_printf $::dbx {PRAGMA user_version} {}
  } {0 1 1 user_version 0}
}

# do_malloc_test closes and deletes the usual db connections and files on
# each iteration.  $::dbx is a separate connection, and on Windows, will
# cause the file deletion of test.db to fail, so we move the close of $::dbx
# up to here before the do_malloc_test.
do_test tableapi-99.0 {
  sqlite3_close $::dbx
} {SQLITE_OK}

do_malloc_test tableapi-7 -sqlprep {

Changes to test/tester.tcl.

31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
..
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
....
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
....
2328
2329
2330
2331
2332
2333
2334
2335
2336
2337
2338
2339
2340
2341
2342
#
# Test the capability of the SQLite version built into the interpreter to
# determine if a specific test can be run:
#
#      capable                EXPR
#      ifcapable              EXPR
#
# Calulate checksums based on database contents:
#
#      dbcksum                DB DBNAME
#      allcksum               ?DB?
#      cksum                  ?DB?
#
# Commands to execute/explain SQL statements:
#
................................................................................
# Command to test whether or not --verbose=1 was specified on the command
# line (returns 0 for not-verbose, 1 for verbose and 2 for "verbose in the
# output file only").
#
#      verbose
#

# Set the precision of FP arithmatic used by the interpreter. And
# configure SQLite to take database file locks on the page that begins
# 64KB into the database file instead of the one 1GB in. This means
# the code that handles that special case can be tested without creating
# very large database files.
#
set tcl_precision 15
sqlite3_test_control_pending_byte 0x0010000
................................................................................
    set c [catch {uplevel 1 $code} r]
  } else {
    set c [catch {uplevel 1 $elsecode} r]
  }
  return -code $c $r
}

# This proc execs a seperate process that crashes midway through executing
# the SQL script $sql on database test.db.
#
# The crash occurs during a sync() of file $crashfile. When the crash
# occurs a random subset of all unsynced writes made by the process are
# written into the files on disk. Argument $crashdelay indicates the
# number of file syncs to wait before crashing.
#
................................................................................
  # Add some info to the output.
  #
  output2 "Time: $tail $ms ms"
  show_memstats
}

# Open a new connection on database test.db and execute the SQL script
# supplied as an argument. Before returning, close the new conection and
# restore the 4 byte fields starting at header offsets 28, 92 and 96
# to the values they held before the SQL was executed. This simulates
# a write by a pre-3.7.0 client.
#
proc sql36231 {sql} {
  set B [hexio_read test.db 92 8]
  set A [hexio_read test.db 28 4]







|







 







|







 







|







 







|







31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
..
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
....
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
....
2328
2329
2330
2331
2332
2333
2334
2335
2336
2337
2338
2339
2340
2341
2342
#
# Test the capability of the SQLite version built into the interpreter to
# determine if a specific test can be run:
#
#      capable                EXPR
#      ifcapable              EXPR
#
# Calculate checksums based on database contents:
#
#      dbcksum                DB DBNAME
#      allcksum               ?DB?
#      cksum                  ?DB?
#
# Commands to execute/explain SQL statements:
#
................................................................................
# Command to test whether or not --verbose=1 was specified on the command
# line (returns 0 for not-verbose, 1 for verbose and 2 for "verbose in the
# output file only").
#
#      verbose
#

# Set the precision of FP arithmetic used by the interpreter. And
# configure SQLite to take database file locks on the page that begins
# 64KB into the database file instead of the one 1GB in. This means
# the code that handles that special case can be tested without creating
# very large database files.
#
set tcl_precision 15
sqlite3_test_control_pending_byte 0x0010000
................................................................................
    set c [catch {uplevel 1 $code} r]
  } else {
    set c [catch {uplevel 1 $elsecode} r]
  }
  return -code $c $r
}

# This proc execs a separate process that crashes midway through executing
# the SQL script $sql on database test.db.
#
# The crash occurs during a sync() of file $crashfile. When the crash
# occurs a random subset of all unsynced writes made by the process are
# written into the files on disk. Argument $crashdelay indicates the
# number of file syncs to wait before crashing.
#
................................................................................
  # Add some info to the output.
  #
  output2 "Time: $tail $ms ms"
  show_memstats
}

# Open a new connection on database test.db and execute the SQL script
# supplied as an argument. Before returning, close the new connection and
# restore the 4 byte fields starting at header offsets 28, 92 and 96
# to the values they held before the SQL was executed. This simulates
# a write by a pre-3.7.0 client.
#
proc sql36231 {sql} {
  set B [hexio_read test.db 92 8]
  set A [hexio_read test.db 28 4]

Changes to test/thread_common.tcl.

103
104
105
106
107
108
109
110
    return 0
  }
  set ::run_thread_tests_called 1
  return 1;
}

return 0








<
103
104
105
106
107
108
109

    return 0
  }
  set ::run_thread_tests_called 1
  return 1;
}

return 0

Changes to test/triggerupfrom.test.

167
168
169
170
171
172
173
174
  (2,two)->(twelve,two)
  (4,four)->(fourteen,four)
}



finish_test








<
167
168
169
170
171
172
173

  (2,two)->(twelve,two)
  (4,four)->(fourteen,four)
}



finish_test

Changes to test/upfrom1.tcl.

108
109
110
111
112
113
114
115
}
errorsql_test 2.3.2 {
  UPDATE t5 AS apples SET b=1 FROM t5 AS apples;
}


finish_test








<
108
109
110
111
112
113
114

}
errorsql_test 2.3.2 {
  UPDATE t5 AS apples SET b=1 FROM t5 AS apples;
}


finish_test

Changes to test/upfrom2.test.

363
364
365
366
367
368
369
370
371
} {
  do_catchsql_test 5.$tn $update \
    "1 {target object/alias may not appear in FROM clause: $nm}"
}


finish_test









<
<
363
364
365
366
367
368
369


} {
  do_catchsql_test 5.$tn $update \
    "1 {target object/alias may not appear in FROM clause: $nm}"
}


finish_test


Changes to test/upfrom3.test.

255
256
257
258
259
260
261
262
      SELECT * FROM c1
    } {a 1  b 12  c 112}

  }]
}

finish_test








<
255
256
257
258
259
260
261

      SELECT * FROM c1
    } {a 1  b 12  c 112}

  }]
}

finish_test

Changes to test/upfromfault.test.

133
134
135
136
137
138
139
140
      error "unexpected result: $res"
    }
  }
}


finish_test








<
133
134
135
136
137
138
139

      error "unexpected result: $res"
    }
  }
}


finish_test

Changes to test/upsert1.test.

126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
  DELETE FROM t1;
  INSERT OR IGNORE INTO t1(a) VALUES('1'),(1) ON CONFLICT(a) DO NOTHING;
  PRAGMA integrity_check;
} {ok}

# 2018-08-14
# Ticket https://www.sqlite.org/src/info/908f001483982c43
# If there are multiple uniqueness contraints, the UPSERT should fire
# if the one constraint it targets fails, regardless of whether or not
# the other constraints pass or fail.  In other words, the UPSERT constraint
# should be tested first.
#
do_execsql_test upsert1-700 {
  DROP TABLE t1;
  CREATE TABLE t1(a INTEGER PRIMARY KEY, b INT, c INT, d INT, e INT);







|







126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
  DELETE FROM t1;
  INSERT OR IGNORE INTO t1(a) VALUES('1'),(1) ON CONFLICT(a) DO NOTHING;
  PRAGMA integrity_check;
} {ok}

# 2018-08-14
# Ticket https://www.sqlite.org/src/info/908f001483982c43
# If there are multiple uniqueness constraints, the UPSERT should fire
# if the one constraint it targets fails, regardless of whether or not
# the other constraints pass or fail.  In other words, the UPSERT constraint
# should be tested first.
#
do_execsql_test upsert1-700 {
  DROP TABLE t1;
  CREATE TABLE t1(a INTEGER PRIMARY KEY, b INT, c INT, d INT, e INT);

Changes to test/wal2.test.

451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
#               locking_mode=exclusive.
#
#   wal2-6.4.*: Check that xShmLock calls are omitted in exclusive locking
#               mode.
#
#   wal2-6.5.*: 
#
#   wal2-6.6.*: Check that if the xShmLock() to reaquire a WAL read-lock when
#               exiting exclusive mode fails (i.e. SQLITE_IOERR), then the
#               connection silently remains in exclusive mode.
#
do_test wal2-6.1.1 {
  forcedelete test.db test.db-wal test.db-journal
  sqlite3 db test.db
  execsql {







|







451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
#               locking_mode=exclusive.
#
#   wal2-6.4.*: Check that xShmLock calls are omitted in exclusive locking
#               mode.
#
#   wal2-6.5.*: 
#
#   wal2-6.6.*: Check that if the xShmLock() to reacquire a WAL read-lock when
#               exiting exclusive mode fails (i.e. SQLITE_IOERR), then the
#               connection silently remains in exclusive mode.
#
do_test wal2-6.1.1 {
  forcedelete test.db test.db-wal test.db-journal
  sqlite3 db test.db
  execsql {

Changes to test/wal_common.tcl.

85
86
87
88
89
90
91
92
93
  upvar $hdrvar hdr
  set c1 0
  set c2 0
  wal_cksum_intlist c1 c2 [lrange $hdr 0 9]
  lset hdr 10 $c1
  lset hdr 11 $c2
}









<
<
85
86
87
88
89
90
91


  upvar $hdrvar hdr
  set c1 0
  set c2 0
  wal_cksum_intlist c1 c2 [lrange $hdr 0 9]
  lset hdr 10 $c1
  lset hdr 11 $c2
}


Changes to test/walsetlk.test.

191
192
193
194
195
196
197
198
} {wal}

do_test 3.1 {
  list [catch { db2 eval {BEGIN EXCLUSIVE} } msg] $msg
} {1 {database is locked}}

finish_test








<
191
192
193
194
195
196
197

} {wal}

do_test 3.1 {
  list [catch { db2 eval {BEGIN EXCLUSIVE} } msg] $msg
} {1 {database is locked}}

finish_test

Changes to test/wapp.tcl.

338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
        append res "$p:$ln: unsafe \"wapp-$cx\" call: \"[string trim $x]\"\n"
      }
    }
  }
  return $res
}

# Return a string that descripts the current environment.  Applications
# might find this useful for debugging.
#
proc wapp-debug-env {} {
  global wapp
  set out {}
  foreach var [lsort [dict keys $wapp]] {
    if {[string index $var 0]=="."} continue







|







338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
        append res "$p:$ln: unsafe \"wapp-$cx\" call: \"[string trim $x]\"\n"
      }
    }
  }
  return $res
}

# Return a string that describes the current environment.  Applications
# might find this useful for debugging.
#
proc wapp-debug-env {} {
  global wapp
  set out {}
  foreach var [lsort [dict keys $wapp]] {
    if {[string index $var 0]=="."} continue

Changes to test/wapptest.tcl.

890
891
892
893
894
895
896
897
if {$G(noui)==0} {
  wapp-start $lWappArg
} else {
  wapptest_run
  do_some_stuff
  vwait forever
}








<
890
891
892
893
894
895
896

if {$G(noui)==0} {
  wapp-start $lWappArg
} else {
  wapptest_run
  do_some_stuff
  vwait forever
}

Changes to test/win32lock.test.

5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
#
#    May you do good and not evil.
#    May you find forgiveness for yourself and forgive others.
#    May you share freely, never taking more than you give.
#
#***********************************************************************
# This file implements regression tests for SQLite library.  The
# focus of this script is recovery from transient manditory locks
# that sometimes appear on database files due to anti-virus software.
#

if {$tcl_platform(platform)!="windows"} return

set testdir [file dirname $argv0]
source $testdir/tester.tcl







|







5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
#
#    May you do good and not evil.
#    May you find forgiveness for yourself and forgive others.
#    May you share freely, never taking more than you give.
#
#***********************************************************************
# This file implements regression tests for SQLite library.  The
# focus of this script is recovery from transient mandatory locks
# that sometimes appear on database files due to anti-virus software.
#

if {$tcl_platform(platform)!="windows"} return

set testdir [file dirname $argv0]
source $testdir/tester.tcl

Changes to test/window2.tcl.

486
487
488
489
490
491
492
493
494

execsql_test 7.3 {
  SELECT c, sum(c) OVER win1 FROM t1 
  WINDOW win1 AS (ORDER BY 1)
}

finish_test









<
<
486
487
488
489
490
491
492



execsql_test 7.3 {
  SELECT c, sum(c) OVER win1 FROM t1 
  WINDOW win1 AS (ORDER BY 1)
}

finish_test


Changes to test/window3.tcl.

356
357
358
359
360
361
362
363
    FILTER (WHERE a%2=0) OVER win FROM t2
    WINDOW win AS (PARTITION BY (a%10) ORDER BY a $window)
  "

}

finish_test








<
356
357
358
359
360
361
362

    FILTER (WHERE a%2=0) OVER win FROM t2
    WINDOW win AS (PARTITION BY (a%10) ORDER BY a $window)
  "

}

finish_test

Changes to test/window4.tcl.

417
418
419
420
421
422
423
424
  SELECT 
    (SELECT avg(a) UNION SELECT min(a) OVER ()) 
  FROM t2 GROUP BY a
  ORDER BY 1
}

finish_test








<
417
418
419
420
421
422
423

  SELECT 
    (SELECT avg(a) UNION SELECT min(a) OVER ()) 
  FROM t2 GROUP BY a
  ORDER BY 1
}

finish_test

Changes to test/window7.tcl.

84
85
86
87
88
89
90
91
execsql_test 1.8.2 {
  SELECT a, sum(b) OVER (
    ORDER BY a DESC RANGE BETWEEN 0 PRECEDING AND 1 FOLLOWING
  ) FROM t3 ORDER BY 1;
}

finish_test








<
84
85
86
87
88
89
90

execsql_test 1.8.2 {
  SELECT a, sum(b) OVER (
    ORDER BY a DESC RANGE BETWEEN 0 PRECEDING AND 1 FOLLOWING
  ) FROM t3 ORDER BY 1;
}

finish_test

Changes to test/window8.tcl.

417
418
419
420
421
422
423
424
425
  );
"
}



finish_test









<
<
417
418
419
420
421
422
423


  );
"
}



finish_test


Changes to test/windowerr.tcl.

65
66
67
68
69
70
71
72
}

errorsql_test 3.3 {
  SELECT row_number(a) OVER () FROM t1;
}

finish_test








<
65
66
67
68
69
70
71

}

errorsql_test 3.3 {
  SELECT row_number(a) OVER () FROM t1;
}

finish_test

Changes to test/without_rowid1.test.

359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
  CREATE TABLE t1(x INTEGER PRIMARY KEY UNIQUE, b) WITHOUT ROWID;
  CREATE INDEX t1x ON t1(x);
  INSERT INTO t1(x,b) VALUES('funny','buffalo');
  SELECT type, name, '|' FROM sqlite_master;
} {table t1 | index t1x |}

# 2018-04-05: OSSFuzz found that the following was accessing an 
# unintialized memory cell. Which was not actually causing a 
# malfunction, but does cause an assert() to fail.
#
do_execsql_test 9.0 {
  CREATE TABLE t2(b, c, PRIMARY KEY(b,c)) WITHOUT ROWID;
  CREATE UNIQUE INDEX t2b ON t2(b);
  UPDATE t2 SET b=1 WHERE b='';
}







|







359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
  CREATE TABLE t1(x INTEGER PRIMARY KEY UNIQUE, b) WITHOUT ROWID;
  CREATE INDEX t1x ON t1(x);
  INSERT INTO t1(x,b) VALUES('funny','buffalo');
  SELECT type, name, '|' FROM sqlite_master;
} {table t1 | index t1x |}

# 2018-04-05: OSSFuzz found that the following was accessing an 
# uninitialized memory cell. Which was not actually causing a 
# malfunction, but does cause an assert() to fail.
#
do_execsql_test 9.0 {
  CREATE TABLE t2(b, c, PRIMARY KEY(b,c)) WITHOUT ROWID;
  CREATE UNIQUE INDEX t2b ON t2(b);
  UPDATE t2 SET b=1 WHERE b='';
}

Changes to tool/genfkey.README.

59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
...
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
...
125
126
127
128
129
130
131
132
133
134
135
136
137
              row is being updated).

    SET NULL: Instead of throwing an exception, the foreign key fields of
              all corresponding child table rows are set to NULL.

LIMITATIONS

  Apart from those limitiations described above:

    * Implicit mapping to composite primary keys is not supported. If
      a parent table has a composite primary key, then any child table
      that refers to it must explicitly map each column. For example, given
      the following definition of table "parent":

        CREATE TABLE parent(a, b, c, PRIMARY KEY(a, b));
................................................................................
  a foreign key that refers to columns in a parent table that are not
  guaranteed to be unique. If such errors are found and the --ignore-errors
  option was not present, a message for each one is printed to stderr and
  no further processing takes place.

  If errors are found and the --ignore-errors option is passed, then
  no error messages are printed. No "CREATE TRIGGER" statements are generated
  for foriegn-key definitions that contained errors, they are silently
  ignored by subsequent processing.

  All triggers generated by this command have names that match the pattern
  "genfkey*". Unless the --no-drop option is specified, then the program
  also generates a "DROP TRIGGER" statement for each trigger that exists
  in the database with a name that matches this pattern. This allows the
  program to be used to upgrade a database schema for which foreign key
................................................................................

  If the --exec option was passed, then all generated SQL is immediately
  executed on the database. Otherwise, the generated SQL strings are output
  in the same way as the results of SELECT queries are. Normally, this means
  they will be printed to stdout, but this can be configured using other
  dot-commands (i.e. ".output").

  The simplest way to activate the foriegn key definitions in a database
  is simply to open it using the shell tool and enter the command 
  ".genfkey --exec":

    sqlite> .genfkey --exec








|







 







|







 







|




<
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
...
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
...
125
126
127
128
129
130
131
132
133
134
135
136

              row is being updated).

    SET NULL: Instead of throwing an exception, the foreign key fields of
              all corresponding child table rows are set to NULL.

LIMITATIONS

  Apart from those limitations described above:

    * Implicit mapping to composite primary keys is not supported. If
      a parent table has a composite primary key, then any child table
      that refers to it must explicitly map each column. For example, given
      the following definition of table "parent":

        CREATE TABLE parent(a, b, c, PRIMARY KEY(a, b));
................................................................................
  a foreign key that refers to columns in a parent table that are not
  guaranteed to be unique. If such errors are found and the --ignore-errors
  option was not present, a message for each one is printed to stderr and
  no further processing takes place.

  If errors are found and the --ignore-errors option is passed, then
  no error messages are printed. No "CREATE TRIGGER" statements are generated
  for foreign-key definitions that contained errors, they are silently
  ignored by subsequent processing.

  All triggers generated by this command have names that match the pattern
  "genfkey*". Unless the --no-drop option is specified, then the program
  also generates a "DROP TRIGGER" statement for each trigger that exists
  in the database with a name that matches this pattern. This allows the
  program to be used to upgrade a database schema for which foreign key
................................................................................

  If the --exec option was passed, then all generated SQL is immediately
  executed on the database. Otherwise, the generated SQL strings are output
  in the same way as the results of SELECT queries are. Normally, this means
  they will be printed to stdout, but this can be configured using other
  dot-commands (i.e. ".output").

  The simplest way to activate the foreign key definitions in a database
  is simply to open it using the shell tool and enter the command 
  ".genfkey --exec":

    sqlite> .genfkey --exec

Changes to tool/lemon.c.

1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
....
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
....
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
....
3507
3508
3509
3510
3511
3512
3513
3514
3515
3516
3517
3518
3519
3520
3521
....
3562
3563
3564
3565
3566
3567
3568
3569
3570
3571
3572
3573
3574
3575
3576
....
3851
3852
3853
3854
3855
3856
3857
3858
3859
3860
3861
3862
3863
3864
3865
....
4811
4812
4813
4814
4815
4816
4817
4818
4819
4820
4821
4822
4823
4824
4825
  struct config *cfp;  /* For looping thru the config closure of "stp" */
  struct config *bcfp; /* For the inner loop on config closure of "stp" */
  struct config *newcfg;  /* */
  struct symbol *sp;   /* Symbol following the dot in configuration "cfp" */
  struct symbol *bsp;  /* Symbol following the dot in configuration "bcfp" */
  struct state *newstp; /* A pointer to a successor state */

  /* Each configuration becomes complete after it contibutes to a successor
  ** state.  Initially, all configurations are incomplete */
  for(cfp=stp->cfp; cfp; cfp=cfp->next) cfp->status = INCOMPLETE;

  /* Loop through all configurations of the state "stp" */
  for(cfp=stp->cfp; cfp; cfp=cfp->next){
    if( cfp->status==COMPLETE ) continue;    /* Already used by inner loop */
    if( cfp->dot>=cfp->rp->nrhs ) continue;  /* Can't shift this config */
................................................................................
    exit(1);
  }
  lemon_strcpy(*paz, z);
  for(z=*paz; *z && *z!='='; z++){}
  *z = 0;
}

/* Rember the name of the output directory 
*/
static char *outputDir = NULL;
static void handle_d_option(char *z){
  outputDir = (char *) malloc( lemonStrlen(z)+1 );
  if( outputDir==0 ){
    fprintf(stderr,"out of memory\n");
    exit(1);
................................................................................
** Inputs:
**   list:      Pointer to a singly-linked list of structures.
**   next:      Pointer to pointer to the second element of the list.
**   cmp:       A comparison function.
**
** Return Value:
**   A pointer to the head of a sorted list containing the elements
**   orginally in list.
**
** Side effects:
**   The "next" pointers for elements in list are changed.
*/
#define LISTSIZE 30
static char *msort(
  char *list,
................................................................................
    fprintf(fp,"\n");
  }
  fclose(fp);
  return;
}

/* Search for the file "name" which is in the same directory as
** the exacutable */
PRIVATE char *pathsearch(char *argv0, char *name, int modemask)
{
  const char *pathlist;
  char *pathbufptr;
  char *pathbuf;
  char *path,*cp;
  char c;
................................................................................
*/
PRIVATE int compute_action(struct lemon *lemp, struct action *ap)
{
  int act;
  switch( ap->type ){
    case SHIFT:  act = ap->x.stp->statenum;                        break;
    case SHIFTREDUCE: {
      /* Since a SHIFT is inherient after a prior REDUCE, convert any
      ** SHIFTREDUCE action with a nonterminal on the LHS into a simple
      ** REDUCE action: */
      if( ap->sp->index>=lemp->nterminal ){
        act = lemp->minReduce + ap->x.rp->iRule;
      }else{
        act = lemp->minShiftReduce + ap->x.rp->iRule;
      }
................................................................................


  if( rp->nrhs==0 ){
    /* If there are no RHS symbols, then writing directly to the LHS is ok */
    lhsdirect = 1;
  }else if( rp->rhsalias[0]==0 ){
    /* The left-most RHS symbol has no value.  LHS direct is ok.  But
    ** we have to call the distructor on the RHS symbol first. */
    lhsdirect = 1;
    if( has_destructor(rp->rhs[0],lemp) ){
      append_str(0,0,0,0);
      append_str("  yy_destructor(yypParser,%d,&yymsp[%d].minor);\n", 0,
                 rp->rhs[0]->index,1-rp->nrhs);
      rp->codePrefix = Strsafe(append_str(0,0,0,0));
      rp->noCode = 0;
................................................................................
  tplt_print(out,lemp,lemp->overflow,&lineno);
  tplt_xfer(lemp->name,in,out,&lineno);

  /* Generate the tables of rule information.  yyRuleInfoLhs[] and
  ** yyRuleInfoNRhs[].
  **
  ** Note: This code depends on the fact that rules are number
  ** sequentually beginning with 0.
  */
  for(i=0, rp=lemp->rule; rp; rp=rp->next, i++){
    fprintf(out,"  %4d,  /* (%d) ", rp->lhs->index, i);
     rule_print(out, rp);
    fprintf(out," */\n"); lineno++;
  }
  tplt_xfer(lemp->name,in,out,&lineno);







|







 







|







 







|







 







|







 







|







 







|







 







|







1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
....
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
....
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
....
3507
3508
3509
3510
3511
3512
3513
3514
3515
3516
3517
3518
3519
3520
3521
....
3562
3563
3564
3565
3566
3567
3568
3569
3570
3571
3572
3573
3574
3575
3576
....
3851
3852
3853
3854
3855
3856
3857
3858
3859
3860
3861
3862
3863
3864
3865
....
4811
4812
4813
4814
4815
4816
4817
4818
4819
4820
4821
4822
4823
4824
4825
  struct config *cfp;  /* For looping thru the config closure of "stp" */
  struct config *bcfp; /* For the inner loop on config closure of "stp" */
  struct config *newcfg;  /* */
  struct symbol *sp;   /* Symbol following the dot in configuration "cfp" */
  struct symbol *bsp;  /* Symbol following the dot in configuration "bcfp" */
  struct state *newstp; /* A pointer to a successor state */

  /* Each configuration becomes complete after it contributes to a successor
  ** state.  Initially, all configurations are incomplete */
  for(cfp=stp->cfp; cfp; cfp=cfp->next) cfp->status = INCOMPLETE;

  /* Loop through all configurations of the state "stp" */
  for(cfp=stp->cfp; cfp; cfp=cfp->next){
    if( cfp->status==COMPLETE ) continue;    /* Already used by inner loop */
    if( cfp->dot>=cfp->rp->nrhs ) continue;  /* Can't shift this config */
................................................................................
    exit(1);
  }
  lemon_strcpy(*paz, z);
  for(z=*paz; *z && *z!='='; z++){}
  *z = 0;
}

/* Remember the name of the output directory 
*/
static char *outputDir = NULL;
static void handle_d_option(char *z){
  outputDir = (char *) malloc( lemonStrlen(z)+1 );
  if( outputDir==0 ){
    fprintf(stderr,"out of memory\n");
    exit(1);
................................................................................
** Inputs:
**   list:      Pointer to a singly-linked list of structures.
**   next:      Pointer to pointer to the second element of the list.
**   cmp:       A comparison function.
**
** Return Value:
**   A pointer to the head of a sorted list containing the elements
**   originally in list.
**
** Side effects:
**   The "next" pointers for elements in list are changed.
*/
#define LISTSIZE 30
static char *msort(
  char *list,
................................................................................
    fprintf(fp,"\n");
  }
  fclose(fp);
  return;
}

/* Search for the file "name" which is in the same directory as
** the executable */
PRIVATE char *pathsearch(char *argv0, char *name, int modemask)
{
  const char *pathlist;
  char *pathbufptr;
  char *pathbuf;
  char *path,*cp;
  char c;
................................................................................
*/
PRIVATE int compute_action(struct lemon *lemp, struct action *ap)
{
  int act;
  switch( ap->type ){
    case SHIFT:  act = ap->x.stp->statenum;                        break;
    case SHIFTREDUCE: {
      /* Since a SHIFT is inherent after a prior REDUCE, convert any
      ** SHIFTREDUCE action with a nonterminal on the LHS into a simple
      ** REDUCE action: */
      if( ap->sp->index>=lemp->nterminal ){
        act = lemp->minReduce + ap->x.rp->iRule;
      }else{
        act = lemp->minShiftReduce + ap->x.rp->iRule;
      }
................................................................................


  if( rp->nrhs==0 ){
    /* If there are no RHS symbols, then writing directly to the LHS is ok */
    lhsdirect = 1;
  }else if( rp->rhsalias[0]==0 ){
    /* The left-most RHS symbol has no value.  LHS direct is ok.  But
    ** we have to call the destructor on the RHS symbol first. */
    lhsdirect = 1;
    if( has_destructor(rp->rhs[0],lemp) ){
      append_str(0,0,0,0);
      append_str("  yy_destructor(yypParser,%d,&yymsp[%d].minor);\n", 0,
                 rp->rhs[0]->index,1-rp->nrhs);
      rp->codePrefix = Strsafe(append_str(0,0,0,0));
      rp->noCode = 0;
................................................................................
  tplt_print(out,lemp,lemp->overflow,&lineno);
  tplt_xfer(lemp->name,in,out,&lineno);

  /* Generate the tables of rule information.  yyRuleInfoLhs[] and
  ** yyRuleInfoNRhs[].
  **
  ** Note: This code depends on the fact that rules are number
  ** sequentially beginning with 0.
  */
  for(i=0, rp=lemp->rule; rp; rp=rp->next, i++){
    fprintf(out,"  %4d,  /* (%d) ", rp->lhs->index, i);
     rule_print(out, rp);
    fprintf(out," */\n"); lineno++;
  }
  tplt_xfer(lemp->name,in,out,&lineno);

Changes to tool/logest.c.

107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
static void showHelp(const char *zArgv0){
  printf("Usage: %s ARGS...\n", zArgv0);
  printf("Arguments:\n"
    "  NUM    Convert NUM from integer to LogEst and push onto the stack\n"
    " ^NUM    Interpret NUM as a LogEst and push onto stack\n"
    "  x      Multiple the top two elements of the stack\n"
    "  +      Add the top two elements of the stack\n"
    "  dup    Dupliate the top element on the stack\n"
    "  inv    Take the reciprocal of the top of stack.  N = 1/N.\n"
    "  log    Find the LogEst of the number on top of stack\n"
    "  nlogn  Compute NlogN where N is the top of stack\n"
  );
  exit(1);
}








|







107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
static void showHelp(const char *zArgv0){
  printf("Usage: %s ARGS...\n", zArgv0);
  printf("Arguments:\n"
    "  NUM    Convert NUM from integer to LogEst and push onto the stack\n"
    " ^NUM    Interpret NUM as a LogEst and push onto stack\n"
    "  x      Multiple the top two elements of the stack\n"
    "  +      Add the top two elements of the stack\n"
    "  dup    Duplicate the top element on the stack\n"
    "  inv    Take the reciprocal of the top of stack.  N = 1/N.\n"
    "  log    Find the LogEst of the number on top of stack\n"
    "  nlogn  Compute NlogN where N is the top of stack\n"
  );
  exit(1);
}

Changes to tool/mkctimec.tcl.

303
304
305
306
307
308
309
310
311
#endif
}]
}

foreach o [lsort [array names options]] {
  puts [string trim $options($o)]
}









<
<
303
304
305
306
307
308
309


#endif
}]
}

foreach o [lsort [array names options]] {
  puts [string trim $options($o)]
}


Changes to tool/mkshellc.tcl.

16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
** SQLite source tree at tool/mkshellc.tcl.  That script combines source
** code from various constituent source files of SQLite into this single
** "shell.c" file used to implement the SQLite command-line shell.
**
** Most of the code found below comes from the "src/shell.c.in" file in
** the canonical SQLite source tree.  That main file contains "INCLUDE"
** lines that specify other files in the canonical source tree that are
** inserted to getnerate this complete program source file.
**
** The code from multiple files is combined into this single "shell.c"
** source file to help make the command-line program easier to compile.
**
** To modify this program, get a copy of the canonical SQLite source tree,
** edit the src/shell.c.in" and/or some of the other files that are included
** by "src/shell.c.in", then rerun the tool/mkshellc.tcl script.







|







16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
** SQLite source tree at tool/mkshellc.tcl.  That script combines source
** code from various constituent source files of SQLite into this single
** "shell.c" file used to implement the SQLite command-line shell.
**
** Most of the code found below comes from the "src/shell.c.in" file in
** the canonical SQLite source tree.  That main file contains "INCLUDE"
** lines that specify other files in the canonical source tree that are
** inserted to generate this complete program source file.
**
** The code from multiple files is combined into this single "shell.c"
** source file to help make the command-line program easier to compile.
**
** To modify this program, get a copy of the canonical SQLite source tree,
** edit the src/shell.c.in" and/or some of the other files that are included
** by "src/shell.c.in", then rerun the tool/mkshellc.tcl script.

Changes to tool/mkspeedsql.tcl.

12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
#
# $Id: mkspeedsql.tcl,v 1.1 2008/10/09 17:57:34 drh Exp $
#

# Set a uniform random seed
expr srand(0)

# The number_name procedure below converts its argment (an integer)
# into a string which is the English-language name for that number.
#
# Example:
#
#     puts [number_name 123]   ->  "one hundred twenty three"
#
set ones {zero one two three four five six seven eight nine







|







12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
#
# $Id: mkspeedsql.tcl,v 1.1 2008/10/09 17:57:34 drh Exp $
#

# Set a uniform random seed
expr srand(0)

# The number_name procedure below converts its argument (an integer)
# into a string which is the English-language name for that number.
#
# Example:
#
#     puts [number_name 123]   ->  "one hundred twenty three"
#
set ones {zero one two three four five six seven eight nine

Changes to tool/mksqlite3internalh.tcl.

88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
  set nstar [expr {60 - $n}]
  set stars [string range $s78 0 $nstar]
  puts $out "/************** $text $stars/"
}

# Read the source file named $filename and write it into the
# sqlite3.c output file.  If any #include statements are seen,
# process them approprately.
#
proc copy_file {filename} {
  global seen_hdr available_hdr out
  set tail [file tail $filename]
  section_comment "Begin file $tail"
  set in [open $filename r]
  while {![eof $in]} {







|







88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
  set nstar [expr {60 - $n}]
  set stars [string range $s78 0 $nstar]
  puts $out "/************** $text $stars/"
}

# Read the source file named $filename and write it into the
# sqlite3.c output file.  If any #include statements are seen,
# process them appropriately.
#
proc copy_file {filename} {
  global seen_hdr available_hdr out
  set tail [file tail $filename]
  section_comment "Begin file $tail"
  set in [open $filename r]
  while {![eof $in]} {

Changes to tool/omittest.tcl.

9
10
11
12
13
14
15
16
17
18
19
20
21
22
23

The default value for ::MAKEFILE is "../Makefile.linux.gcc".

If -skip_run option is given then only the compile part is attempted.

This script builds the testfixture program and runs the SQLite test suite
once with each SQLITE_OMIT_ option defined and then once with all options
defined together. Each run is performed in a seperate directory created
as a sub-directory of the current directory by the script. The output
of the build is saved in <sub-directory>/build.log. The output of the
test-suite is saved in <sub-directory>/test.log.

Almost any SQLite makefile (except those generated by configure - see below)
should work. The following properties are required:








|







9
10
11
12
13
14
15
16
17
18
19
20
21
22
23

The default value for ::MAKEFILE is "../Makefile.linux.gcc".

If -skip_run option is given then only the compile part is attempted.

This script builds the testfixture program and runs the SQLite test suite
once with each SQLITE_OMIT_ option defined and then once with all options
defined together. Each run is performed in a separate directory created
as a sub-directory of the current directory by the script. The output
of the build is saved in <sub-directory>/build.log. The output of the
test-suite is saved in <sub-directory>/test.log.

Almost any SQLite makefile (except those generated by configure - see below)
should work. The following properties are required:

Changes to tool/run-speed-test.sh.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
#!/bin/bash
#
# This is a template for a script used for day-to-day size and 
# performance monitoring of SQLite.  Typical usage:
#
#     sh run-speed-test.sh trunk  #  Baseline measurement of trunk
#     sh run-speed-test.sh x1     # Measure some experimental change
#     fossil test-diff --tk cout-trunk.txt cout-x1.txt   # View chanages
#
# There are multiple output files, all with a base name given by
# the first argument:
#
#     summary-$BASE.txt           # Copy of standard output
#     cout-$BASE.txt              # cachegrind output
#     explain-$BASE.txt           # EXPLAIN listings (only with --explain)







|







1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
#!/bin/bash
#
# This is a template for a script used for day-to-day size and 
# performance monitoring of SQLite.  Typical usage:
#
#     sh run-speed-test.sh trunk  #  Baseline measurement of trunk
#     sh run-speed-test.sh x1     # Measure some experimental change
#     fossil test-diff --tk cout-trunk.txt cout-x1.txt   # View changes
#
# There are multiple output files, all with a base name given by
# the first argument:
#
#     summary-$BASE.txt           # Copy of standard output
#     cout-$BASE.txt              # cachegrind output
#     explain-$BASE.txt           # EXPLAIN listings (only with --explain)

Changes to tool/speed-check.sh.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
#!/bin/bash
#
# This is a template for a script used for day-to-day size and 
# performance monitoring of SQLite.  Typical usage:
#
#     sh run-speed-test.sh trunk  #  Baseline measurement of trunk
#     sh run-speed-test.sh x1     # Measure some experimental change
#     fossil test-diff --tk cout-trunk.txt cout-x1.txt   # View chanages
#
# There are multiple output files, all with a base name given by
# the first argument:
#
#     summary-$BASE.txt           # Copy of standard output
#     cout-$BASE.txt              # cachegrind output
#     explain-$BASE.txt           # EXPLAIN listings (only with --explain)







|







1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
#!/bin/bash
#
# This is a template for a script used for day-to-day size and 
# performance monitoring of SQLite.  Typical usage:
#
#     sh run-speed-test.sh trunk  #  Baseline measurement of trunk
#     sh run-speed-test.sh x1     # Measure some experimental change
#     fossil test-diff --tk cout-trunk.txt cout-x1.txt   # View changes
#
# There are multiple output files, all with a base name given by
# the first argument:
#
#     summary-$BASE.txt           # Copy of standard output
#     cout-$BASE.txt              # cachegrind output
#     explain-$BASE.txt           # EXPLAIN listings (only with --explain)

Changes to tool/sqldiff.c.

1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
  }
}

/*
** Generate a CHANGESET for all differences from main.zTab to aux.zTab.
*/
static void changeset_one_table(const char *zTab, FILE *out){
  sqlite3_stmt *pStmt;          /* SQL statment */
  char *zId = safeId(zTab);     /* Escaped name of the table */
  char **azCol = 0;             /* List of escaped column names */
  int nCol = 0;                 /* Number of columns */
  int *aiFlg = 0;               /* 0 if column is not part of PK */
  int *aiPk = 0;                /* Column numbers for each PK column */
  int nPk = 0;                  /* Number of PRIMARY KEY columns */
  Str sql;                      /* SQL for the diff query */







|







1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
  }
}

/*
** Generate a CHANGESET for all differences from main.zTab to aux.zTab.
*/
static void changeset_one_table(const char *zTab, FILE *out){
  sqlite3_stmt *pStmt;          /* SQL statement */
  char *zId = safeId(zTab);     /* Escaped name of the table */
  char **azCol = 0;             /* List of escaped column names */
  int nCol = 0;                 /* Number of columns */
  int *aiFlg = 0;               /* 0 if column is not part of PK */
  int *aiPk = 0;                /* Column numbers for each PK column */
  int nPk = 0;                  /* Number of PRIMARY KEY columns */
  Str sql;                      /* SQL for the diff query */