Fossil SCM

Merge trunk into ben-testing

ben 2011-08-04 15:54 ben-testing merge

Commit 3eb07708b1160724e5be464ee1ce5efd65ef3a87

Parent 8320393b2fa0f67…

26 files changed +30 -21 -45 -3 +3 +7 -43 +9 +1 -1 +6 -5 +1 +27 -41 +1 +1 +22 -16 +151 -36 +9 +2 -2 +3 -4 +1212 -471 +21 -4 +278 -29 +278 -29 +74 -13 +7 -7 +1 -1 +1 -1

~ BUILD.txt - GNUmakefile.in - Makefile ~ Makefile.classic + Makefile.classic ~ Makefile.in + Makefile.in ~ auto.def ~ autosetup/autosetup ~ autosetup/find-tclsh ~ autosetup/jimsh0.c ~ src/db.c ~ src/db.c ~ src/diffcmd.c ~ src/export.c ~ src/finfo.c ~ src/makemake.tcl ~ src/manifest.c ~ src/sqlite3.c ~ src/sqlite3.h ~ src/tar.c ~ src/tar.c ~ src/timeline.c ~ src/winhttp.c ~ win/Makefile.dmc ~ win/Makefile.msc

M BUILD.txt

+30 -21

		--- BUILD.txt
		+++ BUILD.txt
		@@ -1,54 +1,60 @@
1		-To do a complete build on most unix systems, just type:
2		-
3		- make
4		-
5		-If you have an unusual unix system for which the standard Makefile
6		-will not work, or if you want to do some non-standard options, you can
7		-also run:
	1	+To do a complete build, just type:
8	2
9	3	./configure; make
10	4
11		-The ./configure script builds GNUmakefile which will be used in place
12		-of Makefile. Run "./configure --help" for a listing of the available
13		-options.
	5	+The ./configure script builds Makefile from Makefile.in based on
	6	+your system and any options you select (run "./configure --help"
	7	+for a listing of the available options.)
	8	+
	9	+If you wish to use the original Makefile with no configuration, you can
	10	+instead use:
	11	+
	12	+ make -f Makefile.classic
14	13
15	14	On a windows box, use one of the Makefiles in the win/ subdirectory,
16	15	according to your compiler and environment. If you have GCC and MSYS
17		-installed on your system, the consider:
	16	+installed on your system, then consider:
18	17
19	18	make -f win/Makefile.mingw
20	19
21	20	If you have VC++ installed on your system, then consider:
22	21
23	22	cd win; nmake /f Makefile.msc
24	23
25	24	If you have trouble, or you want to do something fancy, just look at
26		-top level makefile. There are 6 configuration options that are all well
27		-commented. Instead of editing the Makefile, consider copying the Makefile
28		-to an alternative name such as "GNUMakefile", "BSDMakefile", or "makefile"
29		-and editing the copy.
	25	+Makefile.classic. There are 6 configuration options that are all well
	26	+commented. Instead of editing the Makefile.classic, consider copying
	27	+Makefile.classic to an alternative name such as "GNUMakefile",
	28	+"BSDMakefile", or "makefile" and editing the copy.
30	29
31	30
32	31	BUILDING OUTSIDE THE SOURCE TREE
33	32
34	33	An out of source build is pretty easy:
35	34
36		- 1. Make a new directory to do the builds in.
37		- 2. Copy "Makefile" from the source into the build directory and
38		- modify the SRCDIR macro along the lines of:
	35	+ 1. Make and change to a new directory to do the builds in.
	36	+ 2. Run the "configure" script from this directory.
	37	+ 3. Type: "make"
	38	+
	39	+For example:
39	40
40		- SRCDIR=../src
41		-
42		- 3. type: "make"
	41	+ mkdir build
	42	+ cd build
	43	+ ../configure
	44	+ make
43	45
44	46	This will now keep all generates files seperate from the maintained
45	47	source code.
46	48
47	49	--------------------------------------------------------------------------
48	50
49	51	Here are some notes on what is happening behind the scenes:
	52	+
	53	+* The configure script (if used) examines the options given
	54	+ and runs various tests with the C compiler to create Makefile
	55	+ from the Makefile.in template as well as autoconfig.h
50	56
51	57	* The Makefile just sets up a few macros and then invokes the
52	58	real makefile in src/main.mk. The src/main.mk makefile is
53	59	automatically generated by a TCL script found at src/makemake.tcl.
54	60	Do not edit src/main.mk directly. Update src/makemake.tcl and
		@@ -64,5 +70,8 @@
64	70	A header comment in src/translate.c explains in detail what it does.
65	71
66	72	* The src/mkindex.c program generates some C code that implements
67	73	static lookup tables. See the header comment in the source code
68	74	for details on what it does.
	75	+
	76	+Additional information on the build process is available from
	77	+http://www.fossil-scm.org/fossil/doc/trunk/www/makefile.wiki
69	78
70	79	DELETED GNUmakefile.in
71	80	DELETED Makefile
72	81	ADDED Makefile.classic
73	82	ADDED Makefile.in

	--- BUILD.txt
	+++ BUILD.txt
	@@ -1,54 +1,60 @@
1	To do a complete build on most unix systems, just type:
2
3	make
4
5	If you have an unusual unix system for which the standard Makefile
6	will not work, or if you want to do some non-standard options, you can
7	also run:
8
9	./configure; make
10
11	The ./configure script builds GNUmakefile which will be used in place
12	of Makefile. Run "./configure --help" for a listing of the available
13	options.





14
15	On a windows box, use one of the Makefiles in the win/ subdirectory,
16	according to your compiler and environment. If you have GCC and MSYS
17	installed on your system, the consider:
18
19	make -f win/Makefile.mingw
20
21	If you have VC++ installed on your system, then consider:
22
23	cd win; nmake /f Makefile.msc
24
25	If you have trouble, or you want to do something fancy, just look at
26	top level makefile. There are 6 configuration options that are all well
27	commented. Instead of editing the Makefile, consider copying the Makefile
28	to an alternative name such as "GNUMakefile", "BSDMakefile", or "makefile"
29	and editing the copy.
30
31
32	BUILDING OUTSIDE THE SOURCE TREE
33
34	An out of source build is pretty easy:
35
36	1. Make a new directory to do the builds in.
37	2. Copy "Makefile" from the source into the build directory and
38	modify the SRCDIR macro along the lines of:


39
40	SRCDIR=../src
41
42	3. type: "make"

43
44	This will now keep all generates files seperate from the maintained
45	source code.
46
47	--------------------------------------------------------------------------
48
49	Here are some notes on what is happening behind the scenes:




50
51	* The Makefile just sets up a few macros and then invokes the
52	real makefile in src/main.mk. The src/main.mk makefile is
53	automatically generated by a TCL script found at src/makemake.tcl.
54	Do not edit src/main.mk directly. Update src/makemake.tcl and
	@@ -64,5 +70,8 @@
64	A header comment in src/translate.c explains in detail what it does.
65
66	* The src/mkindex.c program generates some C code that implements
67	static lookup tables. See the header comment in the source code
68	for details on what it does.



69
70	ELETED GNUmakefile.in
71	ELETED Makefile
72	DDED Makefile.classic
73	DDED Makefile.in

	--- BUILD.txt
	+++ BUILD.txt
	@@ -1,54 +1,60 @@
1	To do a complete build, just type:






2
3	./configure; make
4
5	The ./configure script builds Makefile from Makefile.in based on
6	your system and any options you select (run "./configure --help"
7	for a listing of the available options.)
8
9	If you wish to use the original Makefile with no configuration, you can
10	instead use:
11
12	make -f Makefile.classic
13
14	On a windows box, use one of the Makefiles in the win/ subdirectory,
15	according to your compiler and environment. If you have GCC and MSYS
16	installed on your system, then consider:
17
18	make -f win/Makefile.mingw
19
20	If you have VC++ installed on your system, then consider:
21
22	cd win; nmake /f Makefile.msc
23
24	If you have trouble, or you want to do something fancy, just look at
25	Makefile.classic. There are 6 configuration options that are all well
26	commented. Instead of editing the Makefile.classic, consider copying
27	Makefile.classic to an alternative name such as "GNUMakefile",
28	"BSDMakefile", or "makefile" and editing the copy.
29
30
31	BUILDING OUTSIDE THE SOURCE TREE
32
33	An out of source build is pretty easy:
34
35	1. Make and change to a new directory to do the builds in.
36	2. Run the "configure" script from this directory.
37	3. Type: "make"
38
39	For example:
40
41	mkdir build
42	cd build
43	../configure
44	make
45
46	This will now keep all generates files seperate from the maintained
47	source code.
48
49	--------------------------------------------------------------------------
50
51	Here are some notes on what is happening behind the scenes:
52
53	* The configure script (if used) examines the options given
54	and runs various tests with the C compiler to create Makefile
55	from the Makefile.in template as well as autoconfig.h
56
57	* The Makefile just sets up a few macros and then invokes the
58	real makefile in src/main.mk. The src/main.mk makefile is
59	automatically generated by a TCL script found at src/makemake.tcl.
60	Do not edit src/main.mk directly. Update src/makemake.tcl and
	@@ -64,5 +70,8 @@
70	A header comment in src/translate.c explains in detail what it does.
71
72	* The src/mkindex.c program generates some C code that implements
73	static lookup tables. See the header comment in the source code
74	for details on what it does.
75
76	Additional information on the build process is available from
77	http://www.fossil-scm.org/fossil/doc/trunk/www/makefile.wiki
78
79	ELETED GNUmakefile.in
80	ELETED Makefile
81	DDED Makefile.classic
82	DDED Makefile.in

D GNUmakefile.in

-45

		--- a/GNUmakefile.in
		+++ b/GNUmakefile.in
		@@ -1,45 +0,0 @@
1		-#!/usr/bin/make
2		-#
3		-# This is the top-level makefile for Fossil when the build is occurring
4		-# on a unix platform. This works out-of-the-box on most unix platforms.
5		-# But you are free to vary some of the definitions if desired.
6		-#
7		-#### The toplevel directory of the source tree. Fossil can be built
8		-# in a directory that is separate from the source tree. Just change
9		-# the following to point from the build directory to the src/ folder.
10		-#
11		-SRCDIR = @srcdir@/src
12		-
13		-#### The directory into which object code files should be written.
14		-#
15		-#
16		-OBJDIR = ./bld
17		-
18		-#### C Compiler and options for use in building executables that
19		-# will run on the platform that is doing the build. This is used
20		-# to compile code-generator programs as part of the build process.
21		-# See TCC below for the C compiler for building the finished binary.
22		-#
23		-BCC = @CC_FOR_BUILD@
24		-
25		-#### The suffix to add to final executable file. When cross-compiling
26		-# to windows, make this ".exe". Otherwise leave it blank.
27		-#
28		-E = @EXEEXT@
29		-
30		-TCC = @CC@
31		-
32		-#### Tcl shell for use in running the fossil testsuite. If you do not
33		-# care about testing the end result, this can be blank.
34		-#
35		-TCLSH = tclsh
36		-
37		-LIB = @LDFLAGS@ @EXTRA_LDFLAGS@ @LIBS@
38		-TCC += @EXTRA_CFLAGS@ @CPPFLAGS@ @CFLAGS@ -DHAVE_AUTOCONFIG_H
39		-INSTALLDIR = $(DESTDIR)@prefix@/bin
40		-USE_SYSTEM_SQLFORM_CHECK_KIND = disable
41		-
42		-include $(SRCDIR)/main.mk
43		-
44		-distclean: clean
45		- rm -f autoconfi

	--- a/GNUmakefile.in
	+++ b/GNUmakefile.in
	@@ -1,45 +0,0 @@
1	#!/usr/bin/make
2	#
3	# This is the top-level makefile for Fossil when the build is occurring
4	# on a unix platform. This works out-of-the-box on most unix platforms.
5	# But you are free to vary some of the definitions if desired.
6	#
7	#### The toplevel directory of the source tree. Fossil can be built
8	# in a directory that is separate from the source tree. Just change
9	# the following to point from the build directory to the src/ folder.
10	#
11	SRCDIR = @srcdir@/src
12
13	#### The directory into which object code files should be written.
14	#
15	#
16	OBJDIR = ./bld
17
18	#### C Compiler and options for use in building executables that
19	# will run on the platform that is doing the build. This is used
20	# to compile code-generator programs as part of the build process.
21	# See TCC below for the C compiler for building the finished binary.
22	#
23	BCC = @CC_FOR_BUILD@
24
25	#### The suffix to add to final executable file. When cross-compiling
26	# to windows, make this ".exe". Otherwise leave it blank.
27	#
28	E = @EXEEXT@
29
30	TCC = @CC@
31
32	#### Tcl shell for use in running the fossil testsuite. If you do not
33	# care about testing the end result, this can be blank.
34	#
35	TCLSH = tclsh
36
37	LIB = @LDFLAGS@ @EXTRA_LDFLAGS@ @LIBS@
38	TCC += @EXTRA_CFLAGS@ @CPPFLAGS@ @CFLAGS@ -DHAVE_AUTOCONFIG_H
39	INSTALLDIR = $(DESTDIR)@prefix@/bin
40	USE_SYSTEM_SQLFORM_CHECK_KIND = disable
41
42	include $(SRCDIR)/main.mk
43
44	distclean: clean
45	rm -f autoconfi

	--- a/GNUmakefile.in
	+++ b/GNUmakefile.in
	@@ -1,45 +0,0 @@

D Makefile

-3

		--- a/Makefile
		+++ b/Makefile
		@@ -1,3 +0,0 @@
1		-#!/usr/bin/make
2		-#
3		-# This is tg the build. This is used

	--- a/Makefile
	+++ b/Makefile
	@@ -1,3 +0,0 @@
1	#!/usr/bin/make
2	#
3	# This is tg the build. This is used

	--- a/Makefile
	+++ b/Makefile
	@@ -1,3 +0,0 @@

M Makefile.classic

No diff available

A Makefile.classic

		--- a/Makefile.classic
		+++ b/Makefile.classic
		@@ -0,0 +1,3 @@
	1	+#!/usr/bin/make
	2	+#
	3	+# This is tg the build. This is used

	--- a/Makefile.classic
	+++ b/Makefile.classic
	@@ -0,0 +1,3 @@

	--- a/Makefile.classic
	+++ b/Makefile.classic
	@@ -0,0 +1,3 @@
1	#!/usr/bin/make
2	#
3	# This is tg the build. This is used

M Makefile.in

+7 -43

		--- a/Makefile.in
		+++ b/Makefile.in
		@@ -1,45 +1,9 @@
1		-#!/usr/bin/make
2		-#
3		-# This is the top-level makefile for Fossil when the build is occurring
4		-# on a unix platform. This works out-of-the-box on most unix platforms.
5		-# But you are free to vary some of the definitions if desired.
6		-#
7		-#### The toplevel directory of the source tree. Fossil can be built
8		-# in a directory that is separate from the source tree. Just change
9		-# the following to point from the build directory to the src/ folder.
10		-#
11		-SRCDIR = @srcdir@/src
12		-
13		-#### The directory into which object code files should be written.
	1	+ -D_HAVE_SQLITE_ssil-@FOSSIL_CI_PFX@P@1BE,N:fossil:@FOSSIL_CI_PFX-@FOSSIL_CI_PFX@P@1BE,N:fossil:@FOSSIL_CI_PFX -D_HAVE_SQLITE_ssil-@FOSSIL_CI_PFX@P@1BE,N:fossil:@FOSSIL_CI_PFX -fuzz -D_HAVE_SQLITE_ssil-@FOSSIL_CI_PFX@P@1BE,N:fossil:@FOSSIL_CI_PFX-@FOSSIL_CI_PFX@P@1BE,N:fossil:@FOSSIL_CI_PFX -D_HAVE_SQLITE_ssil-@FOSSIL_CI_PFX@P@1BE,N:fossil:@FOSSIL_CI_PFX -fuzzSIL_CI_PFX@P@1BE,N:fossil:@FOSSIL_CI_PFX-@FOSSIL_CI_PFX@P@1BE,N:fossil:@FOSSIL_CI_PFX -D_HAVE_SQLITE_ssitclsh -DFOSSIL_ENABLE_LEGACY_MV_RM=1@CFLAGS@@CFLAGS@Rdi make source
14	2	#
15	3	#
16		-OBJDIR = ./bld
17		-
18		-#### C Compiler and options for use in building executables that
19		-# will run on the platform that is doing the build. This is used
20		-# to compile code-generator programs as part of the build process.
21		-# See TCC below for the C compiler for building the finished binary.
22		-#
23		-BCC = @CC_FOR_BUILD@
24		-
25		-#### The suffix to add to final executable file. When cross-compiling
26		-# to windows, make this ".exe". Otherwise leave it blank.
27		-#
28		-E = @EXEEXT@
29		-
30		-TCC = @CC@
31		-
32		-#### Tcl shell for use in running the fossil testsuite. If you do not
33		-# care about testing the end result, this can be blank.
34		-#
35		-TCLSH = tclsh
36		-
37		-LIB = @LDFLAGS@ @EXTRA_LDFLAGS@ @LIBS@
38		-TCC += @EXTRA_CFLAGS@ @CPPFLAGS@ @CFLAGS@ -DHAVE_AUTOCONFIG_H
39		-INSTALLDIR = $(DESTDIR)@prefix@/bin
40		-USE_SYSTEM_SQLFORM_CHECK_KIND = disable
41		-
42		-include $(SRCDIR)/main.mk
43		-
44		-distclean: clean
45		- rm -f autoconfi
	4	+OBJDIR = ./@P@1BE,N:fossil:@FOSSIL_CI_PFX -fuzz -D_HAVE_SQLITE_ssil-@FOSSILendif
	5	+ifeq ($(findstriSSIL_CI_PFX@P@1BE,N:fossil:@FOSSIL_CI_PFX -fuzz -D_HAVE_SQLITE$(MAKE) reconfig
	6	+endif
	7	+@AUTOREMAKE@ && $(MAKE)
	8	+endif
	9	+FOSSIL_ENABLE_M +

	--- a/Makefile.in
	+++ b/Makefile.in
	@@ -1,45 +1,9 @@
1	#!/usr/bin/make
2	#
3	# This is the top-level makefile for Fossil when the build is occurring
4	# on a unix platform. This works out-of-the-box on most unix platforms.
5	# But you are free to vary some of the definitions if desired.
6	#
7	#### The toplevel directory of the source tree. Fossil can be built
8	# in a directory that is separate from the source tree. Just change
9	# the following to point from the build directory to the src/ folder.
10	#
11	SRCDIR = @srcdir@/src
12
13	#### The directory into which object code files should be written.
14	#
15	#
16	OBJDIR = ./bld
17
18	#### C Compiler and options for use in building executables that
19	# will run on the platform that is doing the build. This is used
20	# to compile code-generator programs as part of the build process.
21	# See TCC below for the C compiler for building the finished binary.
22	#
23	BCC = @CC_FOR_BUILD@
24
25	#### The suffix to add to final executable file. When cross-compiling
26	# to windows, make this ".exe". Otherwise leave it blank.
27	#
28	E = @EXEEXT@
29
30	TCC = @CC@
31
32	#### Tcl shell for use in running the fossil testsuite. If you do not
33	# care about testing the end result, this can be blank.
34	#
35	TCLSH = tclsh
36
37	LIB = @LDFLAGS@ @EXTRA_LDFLAGS@ @LIBS@
38	TCC += @EXTRA_CFLAGS@ @CPPFLAGS@ @CFLAGS@ -DHAVE_AUTOCONFIG_H
39	INSTALLDIR = $(DESTDIR)@prefix@/bin
40	USE_SYSTEM_SQLFORM_CHECK_KIND = disable
41
42	include $(SRCDIR)/main.mk
43
44	distclean: clean
45	rm -f autoconfi

	--- a/Makefile.in
	+++ b/Makefile.in
	@@ -1,45 +1,9 @@
1	-D_HAVE_SQLITE_ssil-@FOSSIL_CI_PFX@P@1BE,N:fossil:@FOSSIL_CI_PFX-@FOSSIL_CI_PFX@P@1BE,N:fossil:@FOSSIL_CI_PFX -D_HAVE_SQLITE_ssil-@FOSSIL_CI_PFX@P@1BE,N:fossil:@FOSSIL_CI_PFX -fuzz -D_HAVE_SQLITE_ssil-@FOSSIL_CI_PFX@P@1BE,N:fossil:@FOSSIL_CI_PFX-@FOSSIL_CI_PFX@P@1BE,N:fossil:@FOSSIL_CI_PFX -D_HAVE_SQLITE_ssil-@FOSSIL_CI_PFX@P@1BE,N:fossil:@FOSSIL_CI_PFX -fuzzSIL_CI_PFX@P@1BE,N:fossil:@FOSSIL_CI_PFX-@FOSSIL_CI_PFX@P@1BE,N:fossil:@FOSSIL_CI_PFX -D_HAVE_SQLITE_ssitclsh -DFOSSIL_ENABLE_LEGACY_MV_RM=1@CFLAGS@@CFLAGS@Rdi make source












2	#
3	#
4	OBJDIR = ./@P@1BE,N:fossil:@FOSSIL_CI_PFX -fuzz -D_HAVE_SQLITE_ssil-@FOSSILendif
5	ifeq ($(findstriSSIL_CI_PFX@P@1BE,N:fossil:@FOSSIL_CI_PFX -fuzz -D_HAVE_SQLITE$(MAKE) reconfig
6	endif
7	@AUTOREMAKE@ && $(MAKE)
8	endif
9	FOSSIL_ENABLE_M +

A Makefile.in

		--- a/Makefile.in
		+++ b/Makefile.in
		@@ -0,0 +1,9 @@
	1	+ -D_HAVE_SQLITE_ssil-@FOSSIL_CI_PFX@P@1BE,N:fossil:@FOSSIL_CI_PFX-@FOSSIL_CI_PFX@P@1BE,N:fossil:@FOSSIL_CI_PFX -D_HAVE_SQLITE_ssil-@FOSSIL_CI_PFX@P@1BE,N:fossil:@FOSSIL_CI_PFX -fuzz -D_HAVE_SQLITE_ssil-@FOSSIL_CI_PFX@P@1BE,N:fossil:@FOSSIL_CI_PFX-@FOSSIL_CI_PFX@P@1BE,N:fossil:@FOSSIL_CI_PFX -D_HAVE_SQLITE_ssil-@FOSSIL_CI_PFX@P@1BE,N:fossil:@FOSSIL_CI_PFX -fuzzSIL_CI_PFX@P@1BE,N:fossil:@FOSSIL_CI_PFX-@FOSSIL_CI_PFX@P@1BE,N:fossil:@FOSSIL_CI_PFX -D_HAVE_SQLITE_ssitclsh -DFOSSIL_ENABLE_LEGACY_MV_RM=1@CFLAGS@@CFLAGS@Rdi make source
	2	+#
	3	+#
	4	+OBJDIR = ./@P@1BE,N:fossil:@FOSSIL_CI_PFX -fuzz -D_HAVE_SQLITE_ssil-@FOSSILendif
	5	+ifeq ($(findstriSSIL_CI_PFX@P@1BE,N:fossil:@FOSSIL_CI_PFX -fuzz -D_HAVE_SQLITE$(MAKE) reconfig
	6	+endif
	7	+@AUTOREMAKE@ && $(MAKE)
	8	+endif
	9	+FOSSIL_ENABLE_M +

	--- a/Makefile.in
	+++ b/Makefile.in
	@@ -0,0 +1,9 @@

	--- a/Makefile.in
	+++ b/Makefile.in
	@@ -0,0 +1,9 @@
1	-D_HAVE_SQLITE_ssil-@FOSSIL_CI_PFX@P@1BE,N:fossil:@FOSSIL_CI_PFX-@FOSSIL_CI_PFX@P@1BE,N:fossil:@FOSSIL_CI_PFX -D_HAVE_SQLITE_ssil-@FOSSIL_CI_PFX@P@1BE,N:fossil:@FOSSIL_CI_PFX -fuzz -D_HAVE_SQLITE_ssil-@FOSSIL_CI_PFX@P@1BE,N:fossil:@FOSSIL_CI_PFX-@FOSSIL_CI_PFX@P@1BE,N:fossil:@FOSSIL_CI_PFX -D_HAVE_SQLITE_ssil-@FOSSIL_CI_PFX@P@1BE,N:fossil:@FOSSIL_CI_PFX -fuzzSIL_CI_PFX@P@1BE,N:fossil:@FOSSIL_CI_PFX-@FOSSIL_CI_PFX@P@1BE,N:fossil:@FOSSIL_CI_PFX -D_HAVE_SQLITE_ssitclsh -DFOSSIL_ENABLE_LEGACY_MV_RM=1@CFLAGS@@CFLAGS@Rdi make source
2	#
3	#
4	OBJDIR = ./@P@1BE,N:fossil:@FOSSIL_CI_PFX -fuzz -D_HAVE_SQLITE_ssil-@FOSSILendif
5	ifeq ($(findstriSSIL_CI_PFX@P@1BE,N:fossil:@FOSSIL_CI_PFX -fuzz -D_HAVE_SQLITE$(MAKE) reconfig
6	endif
7	@AUTOREMAKE@ && $(MAKE)
8	endif
9	FOSSIL_ENABLE_M +

M auto.def

+1 -1

		--- auto.def
		+++ auto.def
		@@ -166,7 +166,7 @@
166	166	if {![cc-check-functions getpassphrase]} {
167	167	# Haiku needs this
168	168	cc-check-function-in-lib getpass bsd
169	169	}
170	170
171		-make-template GNUmakefile.in
	171	+make-template Makefile.in
172	172	make-config-header autoconfig.h -auto {USE_* FOSSIL_*}
173	173

	--- auto.def
	+++ auto.def
	@@ -166,7 +166,7 @@
166	if {![cc-check-functions getpassphrase]} {
167	# Haiku needs this
168	cc-check-function-in-lib getpass bsd
169	}
170
171	make-template GNUmakefile.in
172	make-config-header autoconfig.h -auto {USE_* FOSSIL_*}
173

	--- auto.def
	+++ auto.def
	@@ -166,7 +166,7 @@
166	if {![cc-check-functions getpassphrase]} {
167	# Haiku needs this
168	cc-check-function-in-lib getpass bsd
169	}
170
171	make-template Makefile.in
172	make-config-header autoconfig.h -auto {USE_* FOSSIL_*}
173

M autosetup/autosetup

+6 -5

		--- autosetup/autosetup
		+++ autosetup/autosetup
		@@ -1,7 +1,7 @@
1	1	#!/bin/sh
2		-# Copyright (c) 2006-2010 WorkWare Systems http://www.workware.net.au/
	2	+# Copyright (c) 2006-2011 WorkWare Systems http://www.workware.net.au/
3	3	# All rights reserved
4	4	# vim:se syntax=tcl:
5	5	# \
6	6	dir=`dirname "$0"`; exec `"$dir/find-tclsh" \|\| echo false` "$0" "$@"
7	7
		@@ -84,11 +84,11 @@
84	84	help:=local => "display help and options. Optionally specify a module name, such as --help=system"
85	85	version => "display the version of autosetup"
86	86	ref:=text manual:=text
87	87	reference:=text => "display the autosetup command reference. 'text', 'wiki', 'asciidoc' or 'markdown'"
88	88	debug => "display debugging output as autosetup runs"
89		- install => "install autosetup to the current directory (in the 'autosetup/' subdirectory)"
	89	+ install:=. => "install autosetup to the current or given directory (in the 'autosetup/' subdirectory)"
90	90	force init => "create an initial 'configure' script if none exists"
91	91	# Undocumented options
92	92	option-checking=1
93	93	nopager
94	94	quiet
		@@ -133,13 +133,13 @@
133	133	if {[opt-bool init]} {
134	134	use init
135	135	autosetup_init
136	136	}
137	137
138		- if {[opt-bool install]} {
	138	+ if {[opt-val install] ne ""} {
139	139	use install
140		- autosetup_install
	140	+ autosetup_install [opt-val install]
141	141	}
142	142
143	143	if {![file exists $autosetup(autodef)]} {
144	144	# Check for invalid option first
145	145	options {}
		@@ -1299,12 +1299,13 @@
1299	1299	# Copyright (c) 2006-2010 WorkWare Systems http://www.workware.net.au/
1300	1300	# All rights reserved
1301	1301
1302	1302	# Module which can install autosetup
1303	1303
1304		-proc autosetup_install {} {
	1304	+proc autosetup_install {dir} {
1305	1305	if {[catch {
	1306	+ cd $dir
1306	1307	file mkdir autosetup
1307	1308
1308	1309	set f [open autosetup/autosetup w]
1309	1310
1310	1311	set publicmodules {}
1311	1312

	--- autosetup/autosetup
	+++ autosetup/autosetup
	@@ -1,7 +1,7 @@
1	#!/bin/sh
2	# Copyright (c) 2006-2010 WorkWare Systems http://www.workware.net.au/
3	# All rights reserved
4	# vim:se syntax=tcl:
5	# \
6	dir=`dirname "$0"`; exec `"$dir/find-tclsh" \|\| echo false` "$0" "$@"
7
	@@ -84,11 +84,11 @@
84	help:=local => "display help and options. Optionally specify a module name, such as --help=system"
85	version => "display the version of autosetup"
86	ref:=text manual:=text
87	reference:=text => "display the autosetup command reference. 'text', 'wiki', 'asciidoc' or 'markdown'"
88	debug => "display debugging output as autosetup runs"
89	install => "install autosetup to the current directory (in the 'autosetup/' subdirectory)"
90	force init => "create an initial 'configure' script if none exists"
91	# Undocumented options
92	option-checking=1
93	nopager
94	quiet
	@@ -133,13 +133,13 @@
133	if {[opt-bool init]} {
134	use init
135	autosetup_init
136	}
137
138	if {[opt-bool install]} {
139	use install
140	autosetup_install
141	}
142
143	if {![file exists $autosetup(autodef)]} {
144	# Check for invalid option first
145	options {}
	@@ -1299,12 +1299,13 @@
1299	# Copyright (c) 2006-2010 WorkWare Systems http://www.workware.net.au/
1300	# All rights reserved
1301
1302	# Module which can install autosetup
1303
1304	proc autosetup_install {} {
1305	if {[catch {

1306	file mkdir autosetup
1307
1308	set f [open autosetup/autosetup w]
1309
1310	set publicmodules {}
1311

	--- autosetup/autosetup
	+++ autosetup/autosetup
	@@ -1,7 +1,7 @@
1	#!/bin/sh
2	# Copyright (c) 2006-2011 WorkWare Systems http://www.workware.net.au/
3	# All rights reserved
4	# vim:se syntax=tcl:
5	# \
6	dir=`dirname "$0"`; exec `"$dir/find-tclsh" \|\| echo false` "$0" "$@"
7
	@@ -84,11 +84,11 @@
84	help:=local => "display help and options. Optionally specify a module name, such as --help=system"
85	version => "display the version of autosetup"
86	ref:=text manual:=text
87	reference:=text => "display the autosetup command reference. 'text', 'wiki', 'asciidoc' or 'markdown'"
88	debug => "display debugging output as autosetup runs"
89	install:=. => "install autosetup to the current or given directory (in the 'autosetup/' subdirectory)"
90	force init => "create an initial 'configure' script if none exists"
91	# Undocumented options
92	option-checking=1
93	nopager
94	quiet
	@@ -133,13 +133,13 @@
133	if {[opt-bool init]} {
134	use init
135	autosetup_init
136	}
137
138	if {[opt-val install] ne ""} {
139	use install
140	autosetup_install [opt-val install]
141	}
142
143	if {![file exists $autosetup(autodef)]} {
144	# Check for invalid option first
145	options {}
	@@ -1299,12 +1299,13 @@
1299	# Copyright (c) 2006-2010 WorkWare Systems http://www.workware.net.au/
1300	# All rights reserved
1301
1302	# Module which can install autosetup
1303
1304	proc autosetup_install {dir} {
1305	if {[catch {
1306	cd $dir
1307	file mkdir autosetup
1308
1309	set f [open autosetup/autosetup w]
1310
1311	set publicmodules {}
1312

M autosetup/find-tclsh

		--- autosetup/find-tclsh
		+++ autosetup/find-tclsh
		@@ -10,5 +10,6 @@
10	10	for cc in ${CC_FOR_BUILD:-cc} gcc; do
11	11	{ $cc -o "$d/jimsh0" "$d/jimsh0.c"; } 2>/dev/null \|\| continue
12	12	"$d/jimsh0" "$d/test-tclsh" && exit 0
13	13	done
14	14	echo 1>&2 "No working C compiler found. Tried ${CC_FOR_BUILD:-cc} and gcc."
	15	+echo false
15	16

	--- autosetup/find-tclsh
	+++ autosetup/find-tclsh
	@@ -10,5 +10,6 @@
10	for cc in ${CC_FOR_BUILD:-cc} gcc; do
11	{ $cc -o "$d/jimsh0" "$d/jimsh0.c"; } 2>/dev/null \|\| continue
12	"$d/jimsh0" "$d/test-tclsh" && exit 0
13	done
14	echo 1>&2 "No working C compiler found. Tried ${CC_FOR_BUILD:-cc} and gcc."

15

	--- autosetup/find-tclsh
	+++ autosetup/find-tclsh
	@@ -10,5 +10,6 @@
10	for cc in ${CC_FOR_BUILD:-cc} gcc; do
11	{ $cc -o "$d/jimsh0" "$d/jimsh0.c"; } 2>/dev/null \|\| continue
12	"$d/jimsh0" "$d/test-tclsh" && exit 0
13	done
14	echo 1>&2 "No working C compiler found. Tried ${CC_FOR_BUILD:-cc} and gcc."
15	echo false
16

M autosetup/jimsh0.c

+27 -41

		--- autosetup/jimsh0.c
		+++ autosetup/jimsh0.c
		@@ -2597,11 +2597,11 @@
2597	2597
2598	2598	static int aio_cmd_buffering(Jim_Interp interp, int argc, Jim_Obj const *argv)
2599	2599	{
2600	2600	AioFile *af = Jim_CmdPrivData(interp);
2601	2601
2602		- static const char *options[] = {
	2602	+ static const char * const options[] = {
2603	2603	"none",
2604	2604	"line",
2605	2605	"full",
2606	2606	NULL
2607	2607	};
		@@ -5156,17 +5156,17 @@
5156	5156	return result;
5157	5157	}
5158	5158
5159	5159	void Jim_ReapDetachedPids(struct WaitInfoTable *table)
5160	5160	{
	5161	+ struct WaitInfo *waitPtr;
	5162	+ int count;
	5163	+
5161	5164	if (!table) {
5162	5165	return;
5163	5166	}
5164	5167
5165		- struct WaitInfo *waitPtr;
5166		- int count;
5167		-
5168	5168	for (waitPtr = table->info, count = table->used; count > 0; waitPtr++, count--) {
5169	5169	if (waitPtr->flags & WI_DETACHED) {
5170	5170	int status;
5171	5171	int pid = waitpid(waitPtr->pid, &status, WNOHANG);
5172	5172	if (pid > 0) {
		@@ -6365,30 +6365,30 @@
6365	6365	return JIM_OK;
6366	6366	}
6367	6367	int Jim_InitStaticExtensions(Jim_Interp *interp)
6368	6368	{
6369	6369	extern int Jim_bootstrapInit(Jim_Interp *);
6370		-Jim_bootstrapInit(interp);
6371	6370	extern int Jim_aioInit(Jim_Interp *);
6372		-Jim_aioInit(interp);
6373	6371	extern int Jim_readdirInit(Jim_Interp *);
6374		-Jim_readdirInit(interp);
6375	6372	extern int Jim_globInit(Jim_Interp *);
6376		-Jim_globInit(interp);
6377	6373	extern int Jim_regexpInit(Jim_Interp *);
6378		-Jim_regexpInit(interp);
6379	6374	extern int Jim_fileInit(Jim_Interp *);
	6375	+extern int Jim_execInit(Jim_Interp *);
	6376	+extern int Jim_clockInit(Jim_Interp *);
	6377	+extern int Jim_arrayInit(Jim_Interp *);
	6378	+extern int Jim_stdlibInit(Jim_Interp *);
	6379	+extern int Jim_tclcompatInit(Jim_Interp *);
	6380	+Jim_bootstrapInit(interp);
	6381	+Jim_aioInit(interp);
	6382	+Jim_readdirInit(interp);
	6383	+Jim_globInit(interp);
	6384	+Jim_regexpInit(interp);
6380	6385	Jim_fileInit(interp);
6381		-extern int Jim_execInit(Jim_Interp *);
6382	6386	Jim_execInit(interp);
6383		-extern int Jim_clockInit(Jim_Interp *);
6384	6387	Jim_clockInit(interp);
6385		-extern int Jim_arrayInit(Jim_Interp *);
6386	6388	Jim_arrayInit(interp);
6387		-extern int Jim_stdlibInit(Jim_Interp *);
6388	6389	Jim_stdlibInit(interp);
6389		-extern int Jim_tclcompatInit(Jim_Interp *);
6390	6390	Jim_tclcompatInit(interp);
6391	6391	return JIM_OK;
6392	6392	}
6393	6393
6394	6394	/* Jim - A small embeddable Tcl interpreter
		@@ -16669,15 +16669,16 @@
16669	16669	i = 1;
16670	16670	for (d = 0; d < cmd->u.proc.argListLen; d++) {
16671	16671	Jim_Obj *nameObjPtr = cmd->u.proc.arglist[d].nameObjPtr;
16672	16672	if (d == cmd->u.proc.argsPos) {
16673	16673	/* assign $args */
	16674	+ Jim_Obj *listObjPtr;
16674	16675	int argsLen = 0;
16675	16676	if (cmd->u.proc.reqArity + cmd->u.proc.optArity < argc - 1) {
16676	16677	argsLen = argc - 1 - (cmd->u.proc.reqArity + cmd->u.proc.optArity);
16677	16678	}
16678		- Jim_Obj *listObjPtr = Jim_NewListObj(interp, &argv[i], argsLen);
	16679	+ listObjPtr = Jim_NewListObj(interp, &argv[i], argsLen);
16679	16680
16680	16681	/* It is possible to rename args. */
16681	16682	if (cmd->u.proc.arglist[d].defaultObjPtr) {
16682	16683	nameObjPtr =cmd->u.proc.arglist[d].defaultObjPtr;
16683	16684	}
		@@ -16813,11 +16814,10 @@
16813	16814	{
16814	16815	FILE *fp;
16815	16816	char *buf;
16816	16817	Jim_Obj *scriptObjPtr;
16817	16818	Jim_Obj *prevScriptObj;
16818		- Jim_Stack *prevLocalProcs;
16819	16819	struct stat sb;
16820	16820	int retcode;
16821	16821	int readlen;
16822	16822	struct JimParseResult result;
16823	16823
		@@ -16872,20 +16872,12 @@
16872	16872	}
16873	16873
16874	16874	prevScriptObj = interp->currentScriptObj;
16875	16875	interp->currentScriptObj = scriptObjPtr;
16876	16876
16877		- /* Install a new stack for local procs */
16878		- prevLocalProcs = interp->localProcs;
16879		- interp->localProcs = NULL;
16880		-
16881	16877	retcode = Jim_EvalObj(interp, scriptObjPtr);
16882	16878
16883		- /* Delete any local procs */
16884		- JimDeleteLocalProcs(interp);
16885		- interp->localProcs = prevLocalProcs;
16886		-
16887	16879	/* Handle the JIM_RETURN return code */
16888	16880	if (retcode == JIM_RETURN) {
16889	16881	if (--interp->returnLevel <= 0) {
16890	16882	retcode = interp->returnCode;
16891	16883	interp->returnCode = JIM_OK;
		@@ -18366,11 +18358,11 @@
18366	18358	}
18367	18359
18368	18360	/* [lsort] */
18369	18361	static int Jim_LsortCoreCommand(Jim_Interp interp, int argc, Jim_Obj const argv[])
18370	18362	{
18371		- const char *options[] = {
	18363	+ static const char * const options[] = {
18372	18364	"-ascii", "-nocase", "-increasing", "-decreasing", "-command", "-integer", "-index", NULL
18373	18365	};
18374	18366	enum
18375	18367	{ OPT_ASCII, OPT_NOCASE, OPT_INCREASING, OPT_DECREASING, OPT_COMMAND, OPT_INTEGER, OPT_INDEX };
18376	18368	Jim_Obj *resObj;
		@@ -18488,11 +18480,11 @@
18488	18480
18489	18481	/* [debug] */
18490	18482	static int Jim_DebugCoreCommand(Jim_Interp interp, int argc, Jim_Obj const *argv)
18491	18483	{
18492	18484	#ifdef JIM_DEBUG_COMMAND
18493		- const char *options[] = {
	18485	+ static const char * const options[] = {
18494	18486	"refcount", "objcount", "objects", "invstr", "scriptlen", "exprlen",
18495	18487	"exprbc", "show",
18496	18488	NULL
18497	18489	};
18498	18490	enum
		@@ -18693,32 +18685,23 @@
18693	18685
18694	18686	/* [eval] */
18695	18687	static int Jim_EvalCoreCommand(Jim_Interp interp, int argc, Jim_Obj const *argv)
18696	18688	{
18697	18689	int rc;
18698		- Jim_Stack *prevLocalProcs;
18699	18690
18700	18691	if (argc < 2) {
18701	18692	Jim_WrongNumArgs(interp, 1, argv, "script ?...?");
18702	18693	return JIM_ERR;
18703	18694	}
18704	18695
18705		- /* Install a new stack for local procs */
18706		- prevLocalProcs = interp->localProcs;
18707		- interp->localProcs = NULL;
18708		-
18709	18696	if (argc == 2) {
18710	18697	rc = Jim_EvalObj(interp, argv[1]);
18711	18698	}
18712	18699	else {
18713	18700	rc = Jim_EvalObj(interp, Jim_ConcatObj(interp, argc - 1, argv + 1));
18714	18701	}
18715	18702
18716		- /* Delete any local procs */
18717		- JimDeleteLocalProcs(interp);
18718		- interp->localProcs = prevLocalProcs;
18719		-
18720	18703	if (rc == JIM_ERR) {
18721	18704	/* eval is "interesting", so add a stack frame here */
18722	18705	interp->addStackTrace++;
18723	18706	}
18724	18707	return rc;
		@@ -19504,12 +19487,12 @@
19504	19487	else {
19505	19488	exitCode = Jim_EvalObj(interp, argv[0]);
19506	19489	}
19507	19490	interp->signal_level -= sig;
19508	19491
19509		- /* Catch or pass through? Only the first 64 codes can be passed through */
19510		- if (exitCode >= 0 && exitCode < (int)sizeof(mask) && ((1 << exitCode) & mask) == 0) {
	19492	+ /* Catch or pass through? Only the first 32/64 codes can be passed through */
	19493	+ if (exitCode >= 0 && exitCode < (int)sizeof(mask) * 8 && ((1 << exitCode) & mask) == 0) {
19511	19494	/* Not caught, pass it up */
19512	19495	return exitCode;
19513	19496	}
19514	19497
19515	19498	if (sig && exitCode == JIM_SIGNAL) {
		@@ -19736,11 +19719,11 @@
19736	19719	/* [dict] */
19737	19720	static int Jim_DictCoreCommand(Jim_Interp interp, int argc, Jim_Obj const *argv)
19738	19721	{
19739	19722	Jim_Obj *objPtr;
19740	19723	int option;
19741		- const char *options[] = {
	19724	+ static const char * const options[] = {
19742	19725	"create", "get", "set", "unset", "exists", "keys", "merge", "size", "with", NULL
19743	19726	};
19744	19727	enum
19745	19728	{
19746	19729	OPT_CREATE, OPT_GET, OPT_SET, OPT_UNSET, OPT_EXIST, OPT_KEYS, OPT_MERGE, OPT_SIZE, OPT_WITH,
		@@ -19852,11 +19835,11 @@
19852	19835	}
19853	19836
19854	19837	/* [subst] */
19855	19838	static int Jim_SubstCoreCommand(Jim_Interp interp, int argc, Jim_Obj const *argv)
19856	19839	{
19857		- const char *options[] = {
	19840	+ static const char * const options[] = {
19858	19841	"-nobackslashes", "-nocommands", "-novariables", NULL
19859	19842	};
19860	19843	enum
19861	19844	{ OPT_NOBACKSLASHES, OPT_NOCOMMANDS, OPT_NOVARIABLES };
19862	19845	int i;
		@@ -21399,10 +21382,11 @@
21399	21382	#endif
21400	21383	#include <errno.h>
21401	21384	#include <string.h>
21402	21385
21403	21386	#ifdef USE_LINENOISE
	21387	+#include <unistd.h>
21404	21388	#include "linenoise.h"
21405	21389	#else
21406	21390
21407	21391	#define MAX_LINE_LEN 512
21408	21392
		@@ -21427,11 +21411,11 @@
21427	21411	char *history_file = NULL;
21428	21412	#ifdef USE_LINENOISE
21429	21413	const char *home;
21430	21414
21431	21415	home = getenv("HOME");
21432		- if (home) {
	21416	+ if (home && isatty(STDIN_FILENO)) {
21433	21417	int history_len = strlen(home) + sizeof("/.jim_history");
21434	21418	history_file = Jim_Alloc(history_len);
21435	21419	snprintf(history_file, history_len, "%s/.jim_history", home);
21436	21420	linenoiseHistoryLoad(history_file);
21437	21421	}
		@@ -21504,11 +21488,13 @@
21504	21488	Jim_DecrRefCount(interp, scriptObjPtr);
21505	21489	continue;
21506	21490	}
21507	21491
21508	21492	linenoiseHistoryAdd(Jim_String(scriptObjPtr));
21509		- linenoiseHistorySave(history_file);
	21493	+ if (history_file) {
	21494	+ linenoiseHistorySave(history_file);
	21495	+ }
21510	21496	#endif
21511	21497	retcode = Jim_EvalObj(interp, scriptObjPtr);
21512	21498	Jim_DecrRefCount(interp, scriptObjPtr);
21513	21499
21514	21500
21515	21501

	--- autosetup/jimsh0.c
	+++ autosetup/jimsh0.c
	@@ -2597,11 +2597,11 @@
2597
2598	static int aio_cmd_buffering(Jim_Interp interp, int argc, Jim_Obj const *argv)
2599	{
2600	AioFile *af = Jim_CmdPrivData(interp);
2601
2602	static const char *options[] = {
2603	"none",
2604	"line",
2605	"full",
2606	NULL
2607	};
	@@ -5156,17 +5156,17 @@
5156	return result;
5157	}
5158
5159	void Jim_ReapDetachedPids(struct WaitInfoTable *table)
5160	{



5161	if (!table) {
5162	return;
5163	}
5164
5165	struct WaitInfo *waitPtr;
5166	int count;
5167
5168	for (waitPtr = table->info, count = table->used; count > 0; waitPtr++, count--) {
5169	if (waitPtr->flags & WI_DETACHED) {
5170	int status;
5171	int pid = waitpid(waitPtr->pid, &status, WNOHANG);
5172	if (pid > 0) {
	@@ -6365,30 +6365,30 @@
6365	return JIM_OK;
6366	}
6367	int Jim_InitStaticExtensions(Jim_Interp *interp)
6368	{
6369	extern int Jim_bootstrapInit(Jim_Interp *);
6370	Jim_bootstrapInit(interp);
6371	extern int Jim_aioInit(Jim_Interp *);
6372	Jim_aioInit(interp);
6373	extern int Jim_readdirInit(Jim_Interp *);
6374	Jim_readdirInit(interp);
6375	extern int Jim_globInit(Jim_Interp *);
6376	Jim_globInit(interp);
6377	extern int Jim_regexpInit(Jim_Interp *);
6378	Jim_regexpInit(interp);
6379	extern int Jim_fileInit(Jim_Interp *);










6380	Jim_fileInit(interp);
6381	extern int Jim_execInit(Jim_Interp *);
6382	Jim_execInit(interp);
6383	extern int Jim_clockInit(Jim_Interp *);
6384	Jim_clockInit(interp);
6385	extern int Jim_arrayInit(Jim_Interp *);
6386	Jim_arrayInit(interp);
6387	extern int Jim_stdlibInit(Jim_Interp *);
6388	Jim_stdlibInit(interp);
6389	extern int Jim_tclcompatInit(Jim_Interp *);
6390	Jim_tclcompatInit(interp);
6391	return JIM_OK;
6392	}
6393
6394	/* Jim - A small embeddable Tcl interpreter
	@@ -16669,15 +16669,16 @@
16669	i = 1;
16670	for (d = 0; d < cmd->u.proc.argListLen; d++) {
16671	Jim_Obj *nameObjPtr = cmd->u.proc.arglist[d].nameObjPtr;
16672	if (d == cmd->u.proc.argsPos) {
16673	/* assign $args */

16674	int argsLen = 0;
16675	if (cmd->u.proc.reqArity + cmd->u.proc.optArity < argc - 1) {
16676	argsLen = argc - 1 - (cmd->u.proc.reqArity + cmd->u.proc.optArity);
16677	}
16678	Jim_Obj *listObjPtr = Jim_NewListObj(interp, &argv[i], argsLen);
16679
16680	/* It is possible to rename args. */
16681	if (cmd->u.proc.arglist[d].defaultObjPtr) {
16682	nameObjPtr =cmd->u.proc.arglist[d].defaultObjPtr;
16683	}
	@@ -16813,11 +16814,10 @@
16813	{
16814	FILE *fp;
16815	char *buf;
16816	Jim_Obj *scriptObjPtr;
16817	Jim_Obj *prevScriptObj;
16818	Jim_Stack *prevLocalProcs;
16819	struct stat sb;
16820	int retcode;
16821	int readlen;
16822	struct JimParseResult result;
16823
	@@ -16872,20 +16872,12 @@
16872	}
16873
16874	prevScriptObj = interp->currentScriptObj;
16875	interp->currentScriptObj = scriptObjPtr;
16876
16877	/* Install a new stack for local procs */
16878	prevLocalProcs = interp->localProcs;
16879	interp->localProcs = NULL;
16880
16881	retcode = Jim_EvalObj(interp, scriptObjPtr);
16882
16883	/* Delete any local procs */
16884	JimDeleteLocalProcs(interp);
16885	interp->localProcs = prevLocalProcs;
16886
16887	/* Handle the JIM_RETURN return code */
16888	if (retcode == JIM_RETURN) {
16889	if (--interp->returnLevel <= 0) {
16890	retcode = interp->returnCode;
16891	interp->returnCode = JIM_OK;
	@@ -18366,11 +18358,11 @@
18366	}
18367
18368	/* [lsort] */
18369	static int Jim_LsortCoreCommand(Jim_Interp interp, int argc, Jim_Obj const argv[])
18370	{
18371	const char *options[] = {
18372	"-ascii", "-nocase", "-increasing", "-decreasing", "-command", "-integer", "-index", NULL
18373	};
18374	enum
18375	{ OPT_ASCII, OPT_NOCASE, OPT_INCREASING, OPT_DECREASING, OPT_COMMAND, OPT_INTEGER, OPT_INDEX };
18376	Jim_Obj *resObj;
	@@ -18488,11 +18480,11 @@
18488
18489	/* [debug] */
18490	static int Jim_DebugCoreCommand(Jim_Interp interp, int argc, Jim_Obj const *argv)
18491	{
18492	#ifdef JIM_DEBUG_COMMAND
18493	const char *options[] = {
18494	"refcount", "objcount", "objects", "invstr", "scriptlen", "exprlen",
18495	"exprbc", "show",
18496	NULL
18497	};
18498	enum
	@@ -18693,32 +18685,23 @@
18693
18694	/* [eval] */
18695	static int Jim_EvalCoreCommand(Jim_Interp interp, int argc, Jim_Obj const *argv)
18696	{
18697	int rc;
18698	Jim_Stack *prevLocalProcs;
18699
18700	if (argc < 2) {
18701	Jim_WrongNumArgs(interp, 1, argv, "script ?...?");
18702	return JIM_ERR;
18703	}
18704
18705	/* Install a new stack for local procs */
18706	prevLocalProcs = interp->localProcs;
18707	interp->localProcs = NULL;
18708
18709	if (argc == 2) {
18710	rc = Jim_EvalObj(interp, argv[1]);
18711	}
18712	else {
18713	rc = Jim_EvalObj(interp, Jim_ConcatObj(interp, argc - 1, argv + 1));
18714	}
18715
18716	/* Delete any local procs */
18717	JimDeleteLocalProcs(interp);
18718	interp->localProcs = prevLocalProcs;
18719
18720	if (rc == JIM_ERR) {
18721	/* eval is "interesting", so add a stack frame here */
18722	interp->addStackTrace++;
18723	}
18724	return rc;
	@@ -19504,12 +19487,12 @@
19504	else {
19505	exitCode = Jim_EvalObj(interp, argv[0]);
19506	}
19507	interp->signal_level -= sig;
19508
19509	/* Catch or pass through? Only the first 64 codes can be passed through */
19510	if (exitCode >= 0 && exitCode < (int)sizeof(mask) && ((1 << exitCode) & mask) == 0) {
19511	/* Not caught, pass it up */
19512	return exitCode;
19513	}
19514
19515	if (sig && exitCode == JIM_SIGNAL) {
	@@ -19736,11 +19719,11 @@
19736	/* [dict] */
19737	static int Jim_DictCoreCommand(Jim_Interp interp, int argc, Jim_Obj const *argv)
19738	{
19739	Jim_Obj *objPtr;
19740	int option;
19741	const char *options[] = {
19742	"create", "get", "set", "unset", "exists", "keys", "merge", "size", "with", NULL
19743	};
19744	enum
19745	{
19746	OPT_CREATE, OPT_GET, OPT_SET, OPT_UNSET, OPT_EXIST, OPT_KEYS, OPT_MERGE, OPT_SIZE, OPT_WITH,
	@@ -19852,11 +19835,11 @@
19852	}
19853
19854	/* [subst] */
19855	static int Jim_SubstCoreCommand(Jim_Interp interp, int argc, Jim_Obj const *argv)
19856	{
19857	const char *options[] = {
19858	"-nobackslashes", "-nocommands", "-novariables", NULL
19859	};
19860	enum
19861	{ OPT_NOBACKSLASHES, OPT_NOCOMMANDS, OPT_NOVARIABLES };
19862	int i;
	@@ -21399,10 +21382,11 @@
21399	#endif
21400	#include <errno.h>
21401	#include <string.h>
21402
21403	#ifdef USE_LINENOISE

21404	#include "linenoise.h"
21405	#else
21406
21407	#define MAX_LINE_LEN 512
21408
	@@ -21427,11 +21411,11 @@
21427	char *history_file = NULL;
21428	#ifdef USE_LINENOISE
21429	const char *home;
21430
21431	home = getenv("HOME");
21432	if (home) {
21433	int history_len = strlen(home) + sizeof("/.jim_history");
21434	history_file = Jim_Alloc(history_len);
21435	snprintf(history_file, history_len, "%s/.jim_history", home);
21436	linenoiseHistoryLoad(history_file);
21437	}
	@@ -21504,11 +21488,13 @@
21504	Jim_DecrRefCount(interp, scriptObjPtr);
21505	continue;
21506	}
21507
21508	linenoiseHistoryAdd(Jim_String(scriptObjPtr));
21509	linenoiseHistorySave(history_file);


21510	#endif
21511	retcode = Jim_EvalObj(interp, scriptObjPtr);
21512	Jim_DecrRefCount(interp, scriptObjPtr);
21513
21514
21515

	--- autosetup/jimsh0.c
	+++ autosetup/jimsh0.c
	@@ -2597,11 +2597,11 @@
2597
2598	static int aio_cmd_buffering(Jim_Interp interp, int argc, Jim_Obj const *argv)
2599	{
2600	AioFile *af = Jim_CmdPrivData(interp);
2601
2602	static const char * const options[] = {
2603	"none",
2604	"line",
2605	"full",
2606	NULL
2607	};
	@@ -5156,17 +5156,17 @@
5156	return result;
5157	}
5158
5159	void Jim_ReapDetachedPids(struct WaitInfoTable *table)
5160	{
5161	struct WaitInfo *waitPtr;
5162	int count;
5163
5164	if (!table) {
5165	return;
5166	}
5167



5168	for (waitPtr = table->info, count = table->used; count > 0; waitPtr++, count--) {
5169	if (waitPtr->flags & WI_DETACHED) {
5170	int status;
5171	int pid = waitpid(waitPtr->pid, &status, WNOHANG);
5172	if (pid > 0) {
	@@ -6365,30 +6365,30 @@
6365	return JIM_OK;
6366	}
6367	int Jim_InitStaticExtensions(Jim_Interp *interp)
6368	{
6369	extern int Jim_bootstrapInit(Jim_Interp *);

6370	extern int Jim_aioInit(Jim_Interp *);

6371	extern int Jim_readdirInit(Jim_Interp *);

6372	extern int Jim_globInit(Jim_Interp *);

6373	extern int Jim_regexpInit(Jim_Interp *);

6374	extern int Jim_fileInit(Jim_Interp *);
6375	extern int Jim_execInit(Jim_Interp *);
6376	extern int Jim_clockInit(Jim_Interp *);
6377	extern int Jim_arrayInit(Jim_Interp *);
6378	extern int Jim_stdlibInit(Jim_Interp *);
6379	extern int Jim_tclcompatInit(Jim_Interp *);
6380	Jim_bootstrapInit(interp);
6381	Jim_aioInit(interp);
6382	Jim_readdirInit(interp);
6383	Jim_globInit(interp);
6384	Jim_regexpInit(interp);
6385	Jim_fileInit(interp);

6386	Jim_execInit(interp);

6387	Jim_clockInit(interp);

6388	Jim_arrayInit(interp);

6389	Jim_stdlibInit(interp);

6390	Jim_tclcompatInit(interp);
6391	return JIM_OK;
6392	}
6393
6394	/* Jim - A small embeddable Tcl interpreter
	@@ -16669,15 +16669,16 @@
16669	i = 1;
16670	for (d = 0; d < cmd->u.proc.argListLen; d++) {
16671	Jim_Obj *nameObjPtr = cmd->u.proc.arglist[d].nameObjPtr;
16672	if (d == cmd->u.proc.argsPos) {
16673	/* assign $args */
16674	Jim_Obj *listObjPtr;
16675	int argsLen = 0;
16676	if (cmd->u.proc.reqArity + cmd->u.proc.optArity < argc - 1) {
16677	argsLen = argc - 1 - (cmd->u.proc.reqArity + cmd->u.proc.optArity);
16678	}
16679	listObjPtr = Jim_NewListObj(interp, &argv[i], argsLen);
16680
16681	/* It is possible to rename args. */
16682	if (cmd->u.proc.arglist[d].defaultObjPtr) {
16683	nameObjPtr =cmd->u.proc.arglist[d].defaultObjPtr;
16684	}
	@@ -16813,11 +16814,10 @@
16814	{
16815	FILE *fp;
16816	char *buf;
16817	Jim_Obj *scriptObjPtr;
16818	Jim_Obj *prevScriptObj;

16819	struct stat sb;
16820	int retcode;
16821	int readlen;
16822	struct JimParseResult result;
16823
	@@ -16872,20 +16872,12 @@
16872	}
16873
16874	prevScriptObj = interp->currentScriptObj;
16875	interp->currentScriptObj = scriptObjPtr;
16876




16877	retcode = Jim_EvalObj(interp, scriptObjPtr);
16878




16879	/* Handle the JIM_RETURN return code */
16880	if (retcode == JIM_RETURN) {
16881	if (--interp->returnLevel <= 0) {
16882	retcode = interp->returnCode;
16883	interp->returnCode = JIM_OK;
	@@ -18366,11 +18358,11 @@
18358	}
18359
18360	/* [lsort] */
18361	static int Jim_LsortCoreCommand(Jim_Interp interp, int argc, Jim_Obj const argv[])
18362	{
18363	static const char * const options[] = {
18364	"-ascii", "-nocase", "-increasing", "-decreasing", "-command", "-integer", "-index", NULL
18365	};
18366	enum
18367	{ OPT_ASCII, OPT_NOCASE, OPT_INCREASING, OPT_DECREASING, OPT_COMMAND, OPT_INTEGER, OPT_INDEX };
18368	Jim_Obj *resObj;
	@@ -18488,11 +18480,11 @@
18480
18481	/* [debug] */
18482	static int Jim_DebugCoreCommand(Jim_Interp interp, int argc, Jim_Obj const *argv)
18483	{
18484	#ifdef JIM_DEBUG_COMMAND
18485	static const char * const options[] = {
18486	"refcount", "objcount", "objects", "invstr", "scriptlen", "exprlen",
18487	"exprbc", "show",
18488	NULL
18489	};
18490	enum
	@@ -18693,32 +18685,23 @@
18685
18686	/* [eval] */
18687	static int Jim_EvalCoreCommand(Jim_Interp interp, int argc, Jim_Obj const *argv)
18688	{
18689	int rc;

18690
18691	if (argc < 2) {
18692	Jim_WrongNumArgs(interp, 1, argv, "script ?...?");
18693	return JIM_ERR;
18694	}
18695




18696	if (argc == 2) {
18697	rc = Jim_EvalObj(interp, argv[1]);
18698	}
18699	else {
18700	rc = Jim_EvalObj(interp, Jim_ConcatObj(interp, argc - 1, argv + 1));
18701	}
18702




18703	if (rc == JIM_ERR) {
18704	/* eval is "interesting", so add a stack frame here */
18705	interp->addStackTrace++;
18706	}
18707	return rc;
	@@ -19504,12 +19487,12 @@
19487	else {
19488	exitCode = Jim_EvalObj(interp, argv[0]);
19489	}
19490	interp->signal_level -= sig;
19491
19492	/* Catch or pass through? Only the first 32/64 codes can be passed through */
19493	if (exitCode >= 0 && exitCode < (int)sizeof(mask) * 8 && ((1 << exitCode) & mask) == 0) {
19494	/* Not caught, pass it up */
19495	return exitCode;
19496	}
19497
19498	if (sig && exitCode == JIM_SIGNAL) {
	@@ -19736,11 +19719,11 @@
19719	/* [dict] */
19720	static int Jim_DictCoreCommand(Jim_Interp interp, int argc, Jim_Obj const *argv)
19721	{
19722	Jim_Obj *objPtr;
19723	int option;
19724	static const char * const options[] = {
19725	"create", "get", "set", "unset", "exists", "keys", "merge", "size", "with", NULL
19726	};
19727	enum
19728	{
19729	OPT_CREATE, OPT_GET, OPT_SET, OPT_UNSET, OPT_EXIST, OPT_KEYS, OPT_MERGE, OPT_SIZE, OPT_WITH,
	@@ -19852,11 +19835,11 @@
19835	}
19836
19837	/* [subst] */
19838	static int Jim_SubstCoreCommand(Jim_Interp interp, int argc, Jim_Obj const *argv)
19839	{
19840	static const char * const options[] = {
19841	"-nobackslashes", "-nocommands", "-novariables", NULL
19842	};
19843	enum
19844	{ OPT_NOBACKSLASHES, OPT_NOCOMMANDS, OPT_NOVARIABLES };
19845	int i;
	@@ -21399,10 +21382,11 @@
21382	#endif
21383	#include <errno.h>
21384	#include <string.h>
21385
21386	#ifdef USE_LINENOISE
21387	#include <unistd.h>
21388	#include "linenoise.h"
21389	#else
21390
21391	#define MAX_LINE_LEN 512
21392
	@@ -21427,11 +21411,11 @@
21411	char *history_file = NULL;
21412	#ifdef USE_LINENOISE
21413	const char *home;
21414
21415	home = getenv("HOME");
21416	if (home && isatty(STDIN_FILENO)) {
21417	int history_len = strlen(home) + sizeof("/.jim_history");
21418	history_file = Jim_Alloc(history_len);
21419	snprintf(history_file, history_len, "%s/.jim_history", home);
21420	linenoiseHistoryLoad(history_file);
21421	}
	@@ -21504,11 +21488,13 @@
21488	Jim_DecrRefCount(interp, scriptObjPtr);
21489	continue;
21490	}
21491
21492	linenoiseHistoryAdd(Jim_String(scriptObjPtr));
21493	if (history_file) {
21494	linenoiseHistorySave(history_file);
21495	}
21496	#endif
21497	retcode = Jim_EvalObj(interp, scriptObjPtr);
21498	Jim_DecrRefCount(interp, scriptObjPtr);
21499
21500
21501

M src/db.c

		--- src/db.c
		+++ src/db.c
		@@ -1747,10 +1747,11 @@
1747	1747	{ "self-register", 0, 0, 0, "off" },
1748	1748	{ "ssl-ca-location",0, 40, 0, "" },
1749	1749	{ "ssl-identity", 0, 40, 0, "" },
1750	1750	{ "ssh-command", 0, 32, 0, "" },
1751	1751	{ "web-browser", 0, 32, 0, "" },
	1752	+ { "white-foreground", 0, 0, 0, "off" },
1752	1753	{ 0,0,0,0,0 }
1753	1754	};
1754	1755
1755	1756	/*
1756	1757	** COMMAND: settings
1757	1758

	--- src/db.c
	+++ src/db.c
	@@ -1747,10 +1747,11 @@
1747	{ "self-register", 0, 0, 0, "off" },
1748	{ "ssl-ca-location",0, 40, 0, "" },
1749	{ "ssl-identity", 0, 40, 0, "" },
1750	{ "ssh-command", 0, 32, 0, "" },
1751	{ "web-browser", 0, 32, 0, "" },

1752	{ 0,0,0,0,0 }
1753	};
1754
1755	/*
1756	** COMMAND: settings
1757

	--- src/db.c
	+++ src/db.c
	@@ -1747,10 +1747,11 @@
1747	{ "self-register", 0, 0, 0, "off" },
1748	{ "ssl-ca-location",0, 40, 0, "" },
1749	{ "ssl-identity", 0, 40, 0, "" },
1750	{ "ssh-command", 0, 32, 0, "" },
1751	{ "web-browser", 0, 32, 0, "" },
1752	{ "white-foreground", 0, 0, 0, "off" },
1753	{ 0,0,0,0,0 }
1754	};
1755
1756	/*
1757	** COMMAND: settings
1758

M src/db.c

		--- src/db.c
		+++ src/db.c
		@@ -1747,10 +1747,11 @@
1747	1747	{ "self-register", 0, 0, 0, "off" },
1748	1748	{ "ssl-ca-location",0, 40, 0, "" },
1749	1749	{ "ssl-identity", 0, 40, 0, "" },
1750	1750	{ "ssh-command", 0, 32, 0, "" },
1751	1751	{ "web-browser", 0, 32, 0, "" },
	1752	+ { "white-foreground", 0, 0, 0, "off" },
1752	1753	{ 0,0,0,0,0 }
1753	1754	};
1754	1755
1755	1756	/*
1756	1757	** COMMAND: settings
1757	1758

	--- src/db.c
	+++ src/db.c
	@@ -1747,10 +1747,11 @@
1747	{ "self-register", 0, 0, 0, "off" },
1748	{ "ssl-ca-location",0, 40, 0, "" },
1749	{ "ssl-identity", 0, 40, 0, "" },
1750	{ "ssh-command", 0, 32, 0, "" },
1751	{ "web-browser", 0, 32, 0, "" },

1752	{ 0,0,0,0,0 }
1753	};
1754
1755	/*
1756	** COMMAND: settings
1757

	--- src/db.c
	+++ src/db.c
	@@ -1747,10 +1747,11 @@
1747	{ "self-register", 0, 0, 0, "off" },
1748	{ "ssl-ca-location",0, 40, 0, "" },
1749	{ "ssl-identity", 0, 40, 0, "" },
1750	{ "ssh-command", 0, 32, 0, "" },
1751	{ "web-browser", 0, 32, 0, "" },
1752	{ "white-foreground", 0, 0, 0, "off" },
1753	{ 0,0,0,0,0 }
1754	};
1755
1756	/*
1757	** COMMAND: settings
1758

M src/diffcmd.c

+22 -16

		--- src/diffcmd.c
		+++ src/diffcmd.c
		@@ -175,23 +175,24 @@
175	175	blob_reset(&cmd);
176	176	}
177	177	}
178	178
179	179	/*
180		-** Do a diff against a single file named in g.argv[2] from version zFrom
	180	+** Do a diff against a single file named in zFileTreeName from version zFrom
181	181	** against the same file on disk.
182	182	*/
183	183	static void diff_one_against_disk(
184	184	const char zFrom, / Name of file */
185	185	const char zDiffCmd, / Use this "diff" command */
186		- int ignoreEolWs /* Ignore whitespace changes at end of lines */
	186	+ int ignoreEolWs, /* Ignore whitespace changes at end of lines */
	187	+ const char *zFileTreeName
187	188	){
188	189	Blob fname;
189	190	Blob content;
190		- file_tree_name(g.argv[2], &fname, 1);
	191	+ file_tree_name(zFileTreeName, &fname, 1);
191	192	historical_version_of_file(zFrom, blob_str(&fname), &content, 0, 0);
192		- diff_file(&content, g.argv[2], g.argv[2], zDiffCmd, ignoreEolWs);
	193	+ diff_file(&content, zFileTreeName, zFileTreeName, zDiffCmd, ignoreEolWs);
193	194	blob_reset(&content);
194	195	blob_reset(&fname);
195	196	}
196	197
197	198	/*
		@@ -295,22 +296,22 @@
295	296	}
296	297
297	298	/*
298	299	** Output the differences between two versions of a single file.
299	300	** zFrom and zTo are the check-ins containing the two file versions.
300		-** The filename is contained in g.argv[2].
301	301	*/
302	302	static void diff_one_two_versions(
303	303	const char *zFrom,
304	304	const char *zTo,
305	305	const char *zDiffCmd,
306		- int ignoreEolWs
	306	+ int ignoreEolWs,
	307	+ const char *zFileTreeName
307	308	){
308	309	char *zName;
309	310	Blob fname;
310	311	Blob v1, v2;
311		- file_tree_name(g.argv[2], &fname, 1);
	312	+ file_tree_name(zFileTreeName, &fname, 1);
312	313	zName = blob_str(&fname);
313	314	historical_version_of_file(zFrom, zName, &v1, 0, 0);
314	315	historical_version_of_file(zTo, zName, &v2, 0, 0);
315	316	diff_file_mem(&v1, &v2, zName, zDiffCmd, ignoreEolWs);
316	317	blob_reset(&v1);
		@@ -408,16 +409,16 @@
408	409
409	410	/*
410	411	** COMMAND: diff
411	412	** COMMAND: gdiff
412	413	**
413		-** Usage: %fossil diff\|gdiff ?options? ?FILE?
	414	+** Usage: %fossil diff\|gdiff ?options? ?FILE1? ?FILE2 ...?
414	415	**
415		-** Show the difference between the current version of FILE (as it
416		-** exists on disk) and that same file as it was checked out. Or
417		-** if the FILE argument is omitted, show the unsaved changed currently
418		-** in the working check-out.
	416	+** Show the difference between the current version of each of the FILEs
	417	+** specified (as they exist on disk) and that same file as it was checked
	418	+** out. Or if the FILE arguments are omitted, show the unsaved changed
	419	+** currently in the working check-out.
419	420	**
420	421	** If the "--from VERSION" or "-r VERSION" option is used it specifies
421	422	** the source check-in for the diff operation. If not specified, the
422	423	** source check-in is the base check-in for the current check-out.
423	424	**
		@@ -440,10 +441,11 @@
440	441	int hasNFlag; /* True if -N or --new-file flag is used */
441	442	const char zFrom; / Source version number */
442	443	const char zTo; / Target version number */
443	444	const char zDiffCmd = 0; / External diff command. NULL for internal diff */
444	445	int diffFlags = 0; /* Flags to control the DIFF */
	446	+ int f;
445	447
446	448	isGDiff = g.argv[1][0]=='g';
447	449	isInternDiff = find_option("internal","i",0)!=0;
448	450	zFrom = find_option("from", "r", 1);
449	451	zTo = find_option("to", 0, 1);
		@@ -455,12 +457,14 @@
455	457	db_must_be_within_tree();
456	458	verify_all_options();
457	459	if( !isInternDiff ){
458	460	zDiffCmd = db_get(isGDiff ? "gdiff-command" : "diff-command", 0);
459	461	}
460		- if( g.argc==3 ){
461		- diff_one_against_disk(zFrom, zDiffCmd, 0);
	462	+ if( g.argc>=3 ){
	463	+ for(f=2; f<g.argc; ++f){
	464	+ diff_one_against_disk(zFrom, zDiffCmd, 0, g.argv[f]);
	465	+ }
462	466	}else{
463	467	diff_all_against_disk(zFrom, zDiffCmd, diffFlags);
464	468	}
465	469	}else if( zFrom==0 ){
466	470	fossil_fatal("must use --from if --to is present");
		@@ -468,12 +472,14 @@
468	472	db_find_and_open_repository(0, 0);
469	473	verify_all_options();
470	474	if( !isInternDiff ){
471	475	zDiffCmd = db_get(isGDiff ? "gdiff-command" : "diff-command", 0);
472	476	}
473		- if( g.argc==3 ){
474		- diff_one_two_versions(zFrom, zTo, zDiffCmd, 0);
	477	+ if( g.argc>=3 ){
	478	+ for(f=2; f<g.argc; ++f){
	479	+ diff_one_two_versions(zFrom, zTo, zDiffCmd, 0, g.argv[f]);
	480	+ }
475	481	}else{
476	482	diff_all_two_versions(zFrom, zTo, zDiffCmd, diffFlags);
477	483	}
478	484	}
479	485	}
480	486

	--- src/diffcmd.c
	+++ src/diffcmd.c
	@@ -175,23 +175,24 @@
175	blob_reset(&cmd);
176	}
177	}
178
179	/*
180	** Do a diff against a single file named in g.argv[2] from version zFrom
181	** against the same file on disk.
182	*/
183	static void diff_one_against_disk(
184	const char zFrom, / Name of file */
185	const char zDiffCmd, / Use this "diff" command */
186	int ignoreEolWs /* Ignore whitespace changes at end of lines */

187	){
188	Blob fname;
189	Blob content;
190	file_tree_name(g.argv[2], &fname, 1);
191	historical_version_of_file(zFrom, blob_str(&fname), &content, 0, 0);
192	diff_file(&content, g.argv[2], g.argv[2], zDiffCmd, ignoreEolWs);
193	blob_reset(&content);
194	blob_reset(&fname);
195	}
196
197	/*
	@@ -295,22 +296,22 @@
295	}
296
297	/*
298	** Output the differences between two versions of a single file.
299	** zFrom and zTo are the check-ins containing the two file versions.
300	** The filename is contained in g.argv[2].
301	*/
302	static void diff_one_two_versions(
303	const char *zFrom,
304	const char *zTo,
305	const char *zDiffCmd,
306	int ignoreEolWs

307	){
308	char *zName;
309	Blob fname;
310	Blob v1, v2;
311	file_tree_name(g.argv[2], &fname, 1);
312	zName = blob_str(&fname);
313	historical_version_of_file(zFrom, zName, &v1, 0, 0);
314	historical_version_of_file(zTo, zName, &v2, 0, 0);
315	diff_file_mem(&v1, &v2, zName, zDiffCmd, ignoreEolWs);
316	blob_reset(&v1);
	@@ -408,16 +409,16 @@
408
409	/*
410	** COMMAND: diff
411	** COMMAND: gdiff
412	**
413	** Usage: %fossil diff\|gdiff ?options? ?FILE?
414	**
415	** Show the difference between the current version of FILE (as it
416	** exists on disk) and that same file as it was checked out. Or
417	** if the FILE argument is omitted, show the unsaved changed currently
418	** in the working check-out.
419	**
420	** If the "--from VERSION" or "-r VERSION" option is used it specifies
421	** the source check-in for the diff operation. If not specified, the
422	** source check-in is the base check-in for the current check-out.
423	**
	@@ -440,10 +441,11 @@
440	int hasNFlag; /* True if -N or --new-file flag is used */
441	const char zFrom; / Source version number */
442	const char zTo; / Target version number */
443	const char zDiffCmd = 0; / External diff command. NULL for internal diff */
444	int diffFlags = 0; /* Flags to control the DIFF */

445
446	isGDiff = g.argv[1][0]=='g';
447	isInternDiff = find_option("internal","i",0)!=0;
448	zFrom = find_option("from", "r", 1);
449	zTo = find_option("to", 0, 1);
	@@ -455,12 +457,14 @@
455	db_must_be_within_tree();
456	verify_all_options();
457	if( !isInternDiff ){
458	zDiffCmd = db_get(isGDiff ? "gdiff-command" : "diff-command", 0);
459	}
460	if( g.argc==3 ){
461	diff_one_against_disk(zFrom, zDiffCmd, 0);


462	}else{
463	diff_all_against_disk(zFrom, zDiffCmd, diffFlags);
464	}
465	}else if( zFrom==0 ){
466	fossil_fatal("must use --from if --to is present");
	@@ -468,12 +472,14 @@
468	db_find_and_open_repository(0, 0);
469	verify_all_options();
470	if( !isInternDiff ){
471	zDiffCmd = db_get(isGDiff ? "gdiff-command" : "diff-command", 0);
472	}
473	if( g.argc==3 ){
474	diff_one_two_versions(zFrom, zTo, zDiffCmd, 0);


475	}else{
476	diff_all_two_versions(zFrom, zTo, zDiffCmd, diffFlags);
477	}
478	}
479	}
480

	--- src/diffcmd.c
	+++ src/diffcmd.c
	@@ -175,23 +175,24 @@
175	blob_reset(&cmd);
176	}
177	}
178
179	/*
180	** Do a diff against a single file named in zFileTreeName from version zFrom
181	** against the same file on disk.
182	*/
183	static void diff_one_against_disk(
184	const char zFrom, / Name of file */
185	const char zDiffCmd, / Use this "diff" command */
186	int ignoreEolWs, /* Ignore whitespace changes at end of lines */
187	const char *zFileTreeName
188	){
189	Blob fname;
190	Blob content;
191	file_tree_name(zFileTreeName, &fname, 1);
192	historical_version_of_file(zFrom, blob_str(&fname), &content, 0, 0);
193	diff_file(&content, zFileTreeName, zFileTreeName, zDiffCmd, ignoreEolWs);
194	blob_reset(&content);
195	blob_reset(&fname);
196	}
197
198	/*
	@@ -295,22 +296,22 @@
296	}
297
298	/*
299	** Output the differences between two versions of a single file.
300	** zFrom and zTo are the check-ins containing the two file versions.

301	*/
302	static void diff_one_two_versions(
303	const char *zFrom,
304	const char *zTo,
305	const char *zDiffCmd,
306	int ignoreEolWs,
307	const char *zFileTreeName
308	){
309	char *zName;
310	Blob fname;
311	Blob v1, v2;
312	file_tree_name(zFileTreeName, &fname, 1);
313	zName = blob_str(&fname);
314	historical_version_of_file(zFrom, zName, &v1, 0, 0);
315	historical_version_of_file(zTo, zName, &v2, 0, 0);
316	diff_file_mem(&v1, &v2, zName, zDiffCmd, ignoreEolWs);
317	blob_reset(&v1);
	@@ -408,16 +409,16 @@
409
410	/*
411	** COMMAND: diff
412	** COMMAND: gdiff
413	**
414	** Usage: %fossil diff\|gdiff ?options? ?FILE1? ?FILE2 ...?
415	**
416	** Show the difference between the current version of each of the FILEs
417	** specified (as they exist on disk) and that same file as it was checked
418	** out. Or if the FILE arguments are omitted, show the unsaved changed
419	** currently in the working check-out.
420	**
421	** If the "--from VERSION" or "-r VERSION" option is used it specifies
422	** the source check-in for the diff operation. If not specified, the
423	** source check-in is the base check-in for the current check-out.
424	**
	@@ -440,10 +441,11 @@
441	int hasNFlag; /* True if -N or --new-file flag is used */
442	const char zFrom; / Source version number */
443	const char zTo; / Target version number */
444	const char zDiffCmd = 0; / External diff command. NULL for internal diff */
445	int diffFlags = 0; /* Flags to control the DIFF */
446	int f;
447
448	isGDiff = g.argv[1][0]=='g';
449	isInternDiff = find_option("internal","i",0)!=0;
450	zFrom = find_option("from", "r", 1);
451	zTo = find_option("to", 0, 1);
	@@ -455,12 +457,14 @@
457	db_must_be_within_tree();
458	verify_all_options();
459	if( !isInternDiff ){
460	zDiffCmd = db_get(isGDiff ? "gdiff-command" : "diff-command", 0);
461	}
462	if( g.argc>=3 ){
463	for(f=2; f<g.argc; ++f){
464	diff_one_against_disk(zFrom, zDiffCmd, 0, g.argv[f]);
465	}
466	}else{
467	diff_all_against_disk(zFrom, zDiffCmd, diffFlags);
468	}
469	}else if( zFrom==0 ){
470	fossil_fatal("must use --from if --to is present");
	@@ -468,12 +472,14 @@
472	db_find_and_open_repository(0, 0);
473	verify_all_options();
474	if( !isInternDiff ){
475	zDiffCmd = db_get(isGDiff ? "gdiff-command" : "diff-command", 0);
476	}
477	if( g.argc>=3 ){
478	for(f=2; f<g.argc; ++f){
479	diff_one_two_versions(zFrom, zTo, zDiffCmd, 0, g.argv[f]);
480	}
481	}else{
482	diff_all_two_versions(zFrom, zTo, zDiffCmd, diffFlags);
483	}
484	}
485	}
486

M src/export.c

+151 -36

		--- src/export.c
		+++ src/export.c
		@@ -78,15 +78,17 @@
78	78	free(zName);
79	79	free(zEmail);
80	80	db_reset(&q);
81	81	}
82	82
	83	+#define BLOBMARK(rid) ((rid) * 2)
	84	+#define COMMITMARK(rid) ((rid) * 2 + 1)
83	85
84	86	/*
85	87	** COMMAND: export
86	88	**
87		-** Usage: %fossil export --git ?REPOSITORY?
	89	+** Usage: %fossil export --git ?options? ?REPOSITORY?
88	90	**
89	91	** Write an export of all check-ins to standard output. The export is
90	92	** written in the git-fast-export file format assuming the --git option is
91	93	** provided. The git-fast-export format is currently the only VCS
92	94	** interchange format supported, though other formats may be added in
		@@ -95,98 +97,190 @@
95	97	** Run this command within a checkout. Or use the -R or --repository
96	98	** option to specify a Fossil repository to be exported.
97	99	**
98	100	** Only check-ins are exported using --git. Git does not support tickets
99	101	** or wiki or events or attachments, so none of those are exported.
	102	+**
	103	+** If the "--import-marks FILE" option is used, it contains a list of
	104	+** rids to skip.
	105	+**
	106	+** If the "--export-marks FILE" option is used, the rid of all commits and
	107	+** blobs written on exit for use with "--import-marks" on the next run.
100	108	*/
101	109	void export_cmd(void){
102		- Stmt q;
	110	+ Stmt q, q2, q3;
103	111	int i;
104		- int firstCkin; /* Integer offset to check-in marks */
105	112	Bag blobs, vers;
	113	+ const char *markfile_in;
	114	+ const char *markfile_out;
	115	+
106	116	bag_init(&blobs);
107	117	bag_init(&vers);
108	118
109	119	find_option("git", 0, 0); /* Ignore the --git option for now */
	120	+ markfile_in = find_option("import-marks", 0, 1);
	121	+ markfile_out = find_option("export-marks", 0, 1);
	122	+
110	123	db_find_and_open_repository(0, 2);
111	124	verify_all_options();
112	125	if( g.argc!=2 && g.argc!=3 ){ usage("--git ?REPOSITORY?"); }
	126	+
	127	+ db_multi_exec("CREATE TEMPORARY TABLE oldblob(rid INTEGER PRIMARY KEY)");
	128	+ db_multi_exec("CREATE TEMPORARY TABLE oldcommit(rid INTEGER PRIMARY KEY)");
	129	+ if( markfile_in!=0 ){
	130	+ Stmt qb,qc;
	131	+ char line[100];
	132	+ FILE *f;
	133	+
	134	+ f = fopen(markfile_in, "r");
	135	+ if( f==0 ){
	136	+ fossil_panic("cannot open %s for reading", markfile_in);
	137	+ }
	138	+ db_prepare(&qb, "INSERT OR IGNORE INTO oldblob VALUES (:rid)");
	139	+ db_prepare(&qc, "INSERT OR IGNORE INTO oldcommit VALUES (:rid)");
	140	+ while( fgets(line, sizeof(line), f)!=0 ){
	141	+ if( *line == 'b' ){
	142	+ db_bind_text(&qb, ":rid", line + 1);
	143	+ db_step(&qb);
	144	+ db_reset(&qb);
	145	+ bag_insert(&blobs, atoi(line + 1));
	146	+ }else if( *line == 'c' ){
	147	+ db_bind_text(&qc, ":rid", line + 1);
	148	+ db_step(&qc);
	149	+ db_reset(&qc);
	150	+ bag_insert(&vers, atoi(line + 1));
	151	+ }else{
	152	+ fossil_panic("bad input from %s: %s", markfile_in, line);
	153	+ }
	154	+ }
	155	+ db_finalize(&qb);
	156	+ db_finalize(&qc);
	157	+ fclose(f);
	158	+ }
113	159
114	160	/* Step 1: Generate "blob" records for every artifact that is part
115	161	** of a check-in
116	162	*/
117	163	fossil_binary_mode(stdout);
118		- db_prepare(&q, "SELECT DISTINCT fid FROM mlink WHERE fid>0");
	164	+ db_multi_exec("CREATE TEMPORARY TABLE newblob(rid INTEGER KEY, srcid INTEGER)");
	165	+ db_multi_exec("CREATE INDEX newblob_src ON newblob(srcid)");
	166	+ db_multi_exec(
	167	+ "INSERT INTO newblob"
	168	+ " SELECT DISTINCT fid,"
	169	+ " CASE WHEN EXISTS(SELECT 1 FROM delta WHERE rid=fid AND NOT EXISTS(SELECT 1 FROM oldblob WHERE srcid=fid))"
	170	+ " THEN (SELECT srcid FROM delta WHERE rid=fid)"
	171	+ " ELSE 0"
	172	+ " END"
	173	+ " FROM mlink"
	174	+ " WHERE fid>0 AND NOT EXISTS(SELECT 1 FROM oldblob WHERE rid=fid)");
	175	+ db_prepare(&q,
	176	+ "SELECT DISTINCT fid FROM mlink"
	177	+ " WHERE fid>0 AND NOT EXISTS(SELECT 1 FROM oldblob WHERE rid=fid)");
	178	+ db_prepare(&q2, "INSERT INTO oldblob VALUES (:rid)");
	179	+ db_prepare(&q3, "SELECT rid FROM newblob WHERE srcid= (:srcid)");
119	180	while( db_step(&q)==SQLITE_ROW ){
120	181	int rid = db_column_int(&q, 0);
121	182	Blob content;
122		- content_get(rid, &content);
123		- printf("blob\nmark :%d\ndata %d\n", rid, blob_size(&content));
124		- bag_insert(&blobs, rid);
125		- fwrite(blob_buffer(&content), 1, blob_size(&content), stdout);
126		- printf("\n");
127		- blob_reset(&content);
	183	+
	184	+ while( !bag_find(&blobs, rid) ){
	185	+ content_get(rid, &content);
	186	+ db_bind_int(&q2, ":rid", rid);
	187	+ db_step(&q2);
	188	+ db_reset(&q2);
	189	+ printf("blob\nmark :%d\ndata %d\n", BLOBMARK(rid), blob_size(&content));
	190	+ bag_insert(&blobs, rid);
	191	+ fwrite(blob_buffer(&content), 1, blob_size(&content), stdout);
	192	+ printf("\n");
	193	+ blob_reset(&content);
	194	+
	195	+ db_bind_int(&q3, ":srcid", rid);
	196	+ if( db_step(&q3) != SQLITE_ROW ){
	197	+ db_reset(&q3);
	198	+ break;
	199	+ }
	200	+ rid = db_column_int(&q3, 0);
	201	+ db_reset(&q3);
	202	+ }
128	203	}
129	204	db_finalize(&q);
	205	+ db_finalize(&q2);
	206	+ db_finalize(&q3);
130	207
131	208	/* Output the commit records.
132	209	*/
133		- firstCkin = db_int(0, "SELECT max(rid) FROM blob")+1;
134	210	db_prepare(&q,
135	211	"SELECT strftime('%%s',mtime), objid, coalesce(comment,ecomment),"
136	212	" coalesce(user,euser),"
137	213	" (SELECT value FROM tagxref WHERE rid=objid AND tagid=%d)"
138	214	" FROM event"
139		- " WHERE type='ci'"
	215	+ " WHERE type='ci' AND NOT EXISTS (SELECT 1 FROM oldcommit WHERE objid=rid)"
140	216	" ORDER BY mtime ASC",
141	217	TAG_BRANCH
142	218	);
	219	+ db_prepare(&q2, "INSERT INTO oldcommit VALUES (:rid)");
143	220	while( db_step(&q)==SQLITE_ROW ){
	221	+ Stmt q4;
144	222	const char *zSecondsSince1970 = db_column_text(&q, 0);
145	223	int ckinId = db_column_int(&q, 1);
146	224	const char *zComment = db_column_text(&q, 2);
147	225	const char *zUser = db_column_text(&q, 3);
148	226	const char *zBranch = db_column_text(&q, 4);
149	227	char *zBr;
150		- Manifest *p;
151		- ManifestFile *pFile;
152		- const char *zFromType;
153	228
154	229	bag_insert(&vers, ckinId);
	230	+ db_bind_int(&q2, ":rid", ckinId);
	231	+ db_step(&q2);
	232	+ db_reset(&q2);
155	233	if( zBranch==0 ) zBranch = "trunk";
156	234	zBr = mprintf("%s", zBranch);
157	235	for(i=0; zBr[i]; i++){
158	236	if( !fossil_isalnum(zBr[i]) ) zBr[i] = '_';
159	237	}
160		- printf("commit refs/heads/%s\nmark :%d\n", zBr, ckinId+firstCkin);
	238	+ printf("commit refs/heads/%s\nmark :%d\n", zBr, COMMITMARK(ckinId));
161	239	free(zBr);
162	240	printf("committer");
163	241	print_person(zUser);
164	242	printf(" %s +0000\n", zSecondsSince1970);
165	243	if( zComment==0 ) zComment = "null comment";
166	244	printf("data %d\n%s\n", (int)strlen(zComment), zComment);
167		- p = manifest_get(ckinId, CFTYPE_ANY);
168		- zFromType = "from";
169		- for(i=0; i<p->nParent; i++){
170		- int pid = fast_uuid_to_rid(p->azParent[i]);
171		- if( pid==0 \|\| !bag_find(&vers, pid) ) continue;
172		- printf("%s :%d\n", zFromType, fast_uuid_to_rid(p->azParent[i])+firstCkin);
173		- zFromType = "merge";
174		- }
175		- printf("deleteall\n");
176		- manifest_file_rewind(p);
177		- while( (pFile=manifest_file_next(p, 0))!=0 ){
178		- int fid = fast_uuid_to_rid(pFile->zUuid);
179		- const char *zPerm = "100644";
180		- if( fid==0 ) continue;
181		- if( pFile->zPerm && strstr(pFile->zPerm,"x") ) zPerm = "100755";
182		- if( !bag_find(&blobs, fid) ) continue;
183		- printf("M %s :%d %s\n", zPerm, fid, pFile->zName);
184		- }
185		- manifest_cache_insert(p);
	245	+ db_prepare(&q3, "SELECT pid FROM plink WHERE cid=%d AND isprim", ckinId);
	246	+ if( db_step(&q3) == SQLITE_ROW ){
	247	+ printf("from :%d\n", COMMITMARK(db_column_int(&q3, 0)));
	248	+ db_prepare(&q4,
	249	+ "SELECT pid FROM plink"
	250	+ " WHERE cid=%d AND NOT isprim"
	251	+ " AND NOT EXISTS(SELECT 1 FROM phantom WHERE rid=pid)"
	252	+ " ORDER BY pid",
	253	+ ckinId);
	254	+ while( db_step(&q4)==SQLITE_ROW ){
	255	+ printf("merge :%d\n", COMMITMARK(db_column_int(&q4,0)));
	256	+ }
	257	+ db_finalize(&q4);
	258	+ }else{
	259	+ printf("deleteall\n");
	260	+ }
	261	+
	262	+ db_prepare(&q4,
	263	+ "SELECT filename.name, mlink.fid, mlink.mperm FROM mlink"
	264	+ " JOIN filename ON filename.fnid=mlink.fnid"
	265	+ " WHERE mlink.mid=%d",
	266	+ ckinId
	267	+ );
	268	+ while( db_step(&q4)==SQLITE_ROW ){
	269	+ const char *zName = db_column_text(&q4,0);
	270	+ int zNew = db_column_int(&q4,1);
	271	+ int mPerm = db_column_int(&q4,2);
	272	+ if( zNew==0)
	273	+ printf("D %s\n", zName);
	274	+ else if( bag_find(&blobs, zNew) )
	275	+ printf("M %s :%d %s\n", mPerm ? "100755" : "100644", BLOBMARK(zNew), zName);
	276	+ }
	277	+ db_finalize(&q4);
	278	+ db_finalize(&q3);
186	279	printf("\n");
187	280	}
	281	+ db_finalize(&q2);
188	282	db_finalize(&q);
189	283	bag_clear(&blobs);
190	284	manifest_cache_clear();
191	285
192	286
		@@ -207,13 +301,34 @@
207	301	zEncoded = mprintf("%s", zTagname);
208	302	for(i=0; zEncoded[i]; i++){
209	303	if( !fossil_isalnum(zEncoded[i]) ) zEncoded[i] = '_';
210	304	}
211	305	printf("tag %s\n", zEncoded);
212		- printf("from :%d\n", rid+firstCkin);
	306	+ printf("from :%d\n", COMMITMARK(rid));
213	307	printf("tagger <tagger> %s +0000\n", zSecSince1970);
214	308	printf("data 0\n");
215	309	fossil_free(zEncoded);
216	310	}
217	311	db_finalize(&q);
218	312	bag_clear(&vers);
	313	+
	314	+ if( markfile_out!=0 ){
	315	+ FILE *f;
	316	+ f = fopen(markfile_out, "w");
	317	+ if( f == 0 ){
	318	+ fossil_panic("cannot open %s for writing", markfile_out);
	319	+ }
	320	+ db_prepare(&q, "SELECT rid FROM oldblob");
	321	+ while( db_step(&q)==SQLITE_ROW ){
	322	+ fprintf(f, "b%d\n", db_column_int(&q, 0));
	323	+ }
	324	+ db_finalize(&q);
	325	+ db_prepare(&q, "SELECT rid FROM oldcommit");
	326	+ while( db_step(&q)==SQLITE_ROW ){
	327	+ fprintf(f, "c%d\n", db_column_int(&q, 0));
	328	+ }
	329	+ db_finalize(&q);
	330	+ if( ferror(f)!=0 \|\| fclose(f)!=0 ) {
	331	+ fossil_panic("error while writing %s", markfile_out);
	332	+ }
	333	+ }
219	334	}
220	335

	--- src/export.c
	+++ src/export.c
	@@ -78,15 +78,17 @@
78	free(zName);
79	free(zEmail);
80	db_reset(&q);
81	}
82


83
84	/*
85	** COMMAND: export
86	**
87	** Usage: %fossil export --git ?REPOSITORY?
88	**
89	** Write an export of all check-ins to standard output. The export is
90	** written in the git-fast-export file format assuming the --git option is
91	** provided. The git-fast-export format is currently the only VCS
92	** interchange format supported, though other formats may be added in
	@@ -95,98 +97,190 @@
95	** Run this command within a checkout. Or use the -R or --repository
96	** option to specify a Fossil repository to be exported.
97	**
98	** Only check-ins are exported using --git. Git does not support tickets
99	** or wiki or events or attachments, so none of those are exported.






100	*/
101	void export_cmd(void){
102	Stmt q;
103	int i;
104	int firstCkin; /* Integer offset to check-in marks */
105	Bag blobs, vers;



106	bag_init(&blobs);
107	bag_init(&vers);
108
109	find_option("git", 0, 0); /* Ignore the --git option for now */



110	db_find_and_open_repository(0, 2);
111	verify_all_options();
112	if( g.argc!=2 && g.argc!=3 ){ usage("--git ?REPOSITORY?"); }

































113
114	/* Step 1: Generate "blob" records for every artifact that is part
115	** of a check-in
116	*/
117	fossil_binary_mode(stdout);
118	db_prepare(&q, "SELECT DISTINCT fid FROM mlink WHERE fid>0");















119	while( db_step(&q)==SQLITE_ROW ){
120	int rid = db_column_int(&q, 0);
121	Blob content;
122	content_get(rid, &content);
123	printf("blob\nmark :%d\ndata %d\n", rid, blob_size(&content));
124	bag_insert(&blobs, rid);
125	fwrite(blob_buffer(&content), 1, blob_size(&content), stdout);
126	printf("\n");
127	blob_reset(&content);














128	}
129	db_finalize(&q);


130
131	/* Output the commit records.
132	*/
133	firstCkin = db_int(0, "SELECT max(rid) FROM blob")+1;
134	db_prepare(&q,
135	"SELECT strftime('%%s',mtime), objid, coalesce(comment,ecomment),"
136	" coalesce(user,euser),"
137	" (SELECT value FROM tagxref WHERE rid=objid AND tagid=%d)"
138	" FROM event"
139	" WHERE type='ci'"
140	" ORDER BY mtime ASC",
141	TAG_BRANCH
142	);

143	while( db_step(&q)==SQLITE_ROW ){

144	const char *zSecondsSince1970 = db_column_text(&q, 0);
145	int ckinId = db_column_int(&q, 1);
146	const char *zComment = db_column_text(&q, 2);
147	const char *zUser = db_column_text(&q, 3);
148	const char *zBranch = db_column_text(&q, 4);
149	char *zBr;
150	Manifest *p;
151	ManifestFile *pFile;
152	const char *zFromType;
153
154	bag_insert(&vers, ckinId);



155	if( zBranch==0 ) zBranch = "trunk";
156	zBr = mprintf("%s", zBranch);
157	for(i=0; zBr[i]; i++){
158	if( !fossil_isalnum(zBr[i]) ) zBr[i] = '_';
159	}
160	printf("commit refs/heads/%s\nmark :%d\n", zBr, ckinId+firstCkin);
161	free(zBr);
162	printf("committer");
163	print_person(zUser);
164	printf(" %s +0000\n", zSecondsSince1970);
165	if( zComment==0 ) zComment = "null comment";
166	printf("data %d\n%s\n", (int)strlen(zComment), zComment);
167	p = manifest_get(ckinId, CFTYPE_ANY);
168	zFromType = "from";
169	for(i=0; i<p->nParent; i++){
170	int pid = fast_uuid_to_rid(p->azParent[i]);
171	if( pid==0 \|\| !bag_find(&vers, pid) ) continue;
172	printf("%s :%d\n", zFromType, fast_uuid_to_rid(p->azParent[i])+firstCkin);
173	zFromType = "merge";
174	}
175	printf("deleteall\n");
176	manifest_file_rewind(p);
177	while( (pFile=manifest_file_next(p, 0))!=0 ){
178	int fid = fast_uuid_to_rid(pFile->zUuid);
179	const char *zPerm = "100644";
180	if( fid==0 ) continue;
181	if( pFile->zPerm && strstr(pFile->zPerm,"x") ) zPerm = "100755";
182	if( !bag_find(&blobs, fid) ) continue;
183	printf("M %s :%d %s\n", zPerm, fid, pFile->zName);
184	}
185	manifest_cache_insert(p);















186	printf("\n");
187	}

188	db_finalize(&q);
189	bag_clear(&blobs);
190	manifest_cache_clear();
191
192
	@@ -207,13 +301,34 @@
207	zEncoded = mprintf("%s", zTagname);
208	for(i=0; zEncoded[i]; i++){
209	if( !fossil_isalnum(zEncoded[i]) ) zEncoded[i] = '_';
210	}
211	printf("tag %s\n", zEncoded);
212	printf("from :%d\n", rid+firstCkin);
213	printf("tagger <tagger> %s +0000\n", zSecSince1970);
214	printf("data 0\n");
215	fossil_free(zEncoded);
216	}
217	db_finalize(&q);
218	bag_clear(&vers);





















219	}
220

	--- src/export.c
	+++ src/export.c
	@@ -78,15 +78,17 @@
78	free(zName);
79	free(zEmail);
80	db_reset(&q);
81	}
82
83	#define BLOBMARK(rid) ((rid) * 2)
84	#define COMMITMARK(rid) ((rid) * 2 + 1)
85
86	/*
87	** COMMAND: export
88	**
89	** Usage: %fossil export --git ?options? ?REPOSITORY?
90	**
91	** Write an export of all check-ins to standard output. The export is
92	** written in the git-fast-export file format assuming the --git option is
93	** provided. The git-fast-export format is currently the only VCS
94	** interchange format supported, though other formats may be added in
	@@ -95,98 +97,190 @@
97	** Run this command within a checkout. Or use the -R or --repository
98	** option to specify a Fossil repository to be exported.
99	**
100	** Only check-ins are exported using --git. Git does not support tickets
101	** or wiki or events or attachments, so none of those are exported.
102	**
103	** If the "--import-marks FILE" option is used, it contains a list of
104	** rids to skip.
105	**
106	** If the "--export-marks FILE" option is used, the rid of all commits and
107	** blobs written on exit for use with "--import-marks" on the next run.
108	*/
109	void export_cmd(void){
110	Stmt q, q2, q3;
111	int i;

112	Bag blobs, vers;
113	const char *markfile_in;
114	const char *markfile_out;
115
116	bag_init(&blobs);
117	bag_init(&vers);
118
119	find_option("git", 0, 0); /* Ignore the --git option for now */
120	markfile_in = find_option("import-marks", 0, 1);
121	markfile_out = find_option("export-marks", 0, 1);
122
123	db_find_and_open_repository(0, 2);
124	verify_all_options();
125	if( g.argc!=2 && g.argc!=3 ){ usage("--git ?REPOSITORY?"); }
126
127	db_multi_exec("CREATE TEMPORARY TABLE oldblob(rid INTEGER PRIMARY KEY)");
128	db_multi_exec("CREATE TEMPORARY TABLE oldcommit(rid INTEGER PRIMARY KEY)");
129	if( markfile_in!=0 ){
130	Stmt qb,qc;
131	char line[100];
132	FILE *f;
133
134	f = fopen(markfile_in, "r");
135	if( f==0 ){
136	fossil_panic("cannot open %s for reading", markfile_in);
137	}
138	db_prepare(&qb, "INSERT OR IGNORE INTO oldblob VALUES (:rid)");
139	db_prepare(&qc, "INSERT OR IGNORE INTO oldcommit VALUES (:rid)");
140	while( fgets(line, sizeof(line), f)!=0 ){
141	if( *line == 'b' ){
142	db_bind_text(&qb, ":rid", line + 1);
143	db_step(&qb);
144	db_reset(&qb);
145	bag_insert(&blobs, atoi(line + 1));
146	}else if( *line == 'c' ){
147	db_bind_text(&qc, ":rid", line + 1);
148	db_step(&qc);
149	db_reset(&qc);
150	bag_insert(&vers, atoi(line + 1));
151	}else{
152	fossil_panic("bad input from %s: %s", markfile_in, line);
153	}
154	}
155	db_finalize(&qb);
156	db_finalize(&qc);
157	fclose(f);
158	}
159
160	/* Step 1: Generate "blob" records for every artifact that is part
161	** of a check-in
162	*/
163	fossil_binary_mode(stdout);
164	db_multi_exec("CREATE TEMPORARY TABLE newblob(rid INTEGER KEY, srcid INTEGER)");
165	db_multi_exec("CREATE INDEX newblob_src ON newblob(srcid)");
166	db_multi_exec(
167	"INSERT INTO newblob"
168	" SELECT DISTINCT fid,"
169	" CASE WHEN EXISTS(SELECT 1 FROM delta WHERE rid=fid AND NOT EXISTS(SELECT 1 FROM oldblob WHERE srcid=fid))"
170	" THEN (SELECT srcid FROM delta WHERE rid=fid)"
171	" ELSE 0"
172	" END"
173	" FROM mlink"
174	" WHERE fid>0 AND NOT EXISTS(SELECT 1 FROM oldblob WHERE rid=fid)");
175	db_prepare(&q,
176	"SELECT DISTINCT fid FROM mlink"
177	" WHERE fid>0 AND NOT EXISTS(SELECT 1 FROM oldblob WHERE rid=fid)");
178	db_prepare(&q2, "INSERT INTO oldblob VALUES (:rid)");
179	db_prepare(&q3, "SELECT rid FROM newblob WHERE srcid= (:srcid)");
180	while( db_step(&q)==SQLITE_ROW ){
181	int rid = db_column_int(&q, 0);
182	Blob content;
183
184	while( !bag_find(&blobs, rid) ){
185	content_get(rid, &content);
186	db_bind_int(&q2, ":rid", rid);
187	db_step(&q2);
188	db_reset(&q2);
189	printf("blob\nmark :%d\ndata %d\n", BLOBMARK(rid), blob_size(&content));
190	bag_insert(&blobs, rid);
191	fwrite(blob_buffer(&content), 1, blob_size(&content), stdout);
192	printf("\n");
193	blob_reset(&content);
194
195	db_bind_int(&q3, ":srcid", rid);
196	if( db_step(&q3) != SQLITE_ROW ){
197	db_reset(&q3);
198	break;
199	}
200	rid = db_column_int(&q3, 0);
201	db_reset(&q3);
202	}
203	}
204	db_finalize(&q);
205	db_finalize(&q2);
206	db_finalize(&q3);
207
208	/* Output the commit records.
209	*/

210	db_prepare(&q,
211	"SELECT strftime('%%s',mtime), objid, coalesce(comment,ecomment),"
212	" coalesce(user,euser),"
213	" (SELECT value FROM tagxref WHERE rid=objid AND tagid=%d)"
214	" FROM event"
215	" WHERE type='ci' AND NOT EXISTS (SELECT 1 FROM oldcommit WHERE objid=rid)"
216	" ORDER BY mtime ASC",
217	TAG_BRANCH
218	);
219	db_prepare(&q2, "INSERT INTO oldcommit VALUES (:rid)");
220	while( db_step(&q)==SQLITE_ROW ){
221	Stmt q4;
222	const char *zSecondsSince1970 = db_column_text(&q, 0);
223	int ckinId = db_column_int(&q, 1);
224	const char *zComment = db_column_text(&q, 2);
225	const char *zUser = db_column_text(&q, 3);
226	const char *zBranch = db_column_text(&q, 4);
227	char *zBr;



228
229	bag_insert(&vers, ckinId);
230	db_bind_int(&q2, ":rid", ckinId);
231	db_step(&q2);
232	db_reset(&q2);
233	if( zBranch==0 ) zBranch = "trunk";
234	zBr = mprintf("%s", zBranch);
235	for(i=0; zBr[i]; i++){
236	if( !fossil_isalnum(zBr[i]) ) zBr[i] = '_';
237	}
238	printf("commit refs/heads/%s\nmark :%d\n", zBr, COMMITMARK(ckinId));
239	free(zBr);
240	printf("committer");
241	print_person(zUser);
242	printf(" %s +0000\n", zSecondsSince1970);
243	if( zComment==0 ) zComment = "null comment";
244	printf("data %d\n%s\n", (int)strlen(zComment), zComment);
245	db_prepare(&q3, "SELECT pid FROM plink WHERE cid=%d AND isprim", ckinId);
246	if( db_step(&q3) == SQLITE_ROW ){
247	printf("from :%d\n", COMMITMARK(db_column_int(&q3, 0)));
248	db_prepare(&q4,
249	"SELECT pid FROM plink"
250	" WHERE cid=%d AND NOT isprim"
251	" AND NOT EXISTS(SELECT 1 FROM phantom WHERE rid=pid)"
252	" ORDER BY pid",
253	ckinId);
254	while( db_step(&q4)==SQLITE_ROW ){
255	printf("merge :%d\n", COMMITMARK(db_column_int(&q4,0)));
256	}
257	db_finalize(&q4);
258	}else{
259	printf("deleteall\n");
260	}
261
262	db_prepare(&q4,
263	"SELECT filename.name, mlink.fid, mlink.mperm FROM mlink"
264	" JOIN filename ON filename.fnid=mlink.fnid"
265	" WHERE mlink.mid=%d",
266	ckinId
267	);
268	while( db_step(&q4)==SQLITE_ROW ){
269	const char *zName = db_column_text(&q4,0);
270	int zNew = db_column_int(&q4,1);
271	int mPerm = db_column_int(&q4,2);
272	if( zNew==0)
273	printf("D %s\n", zName);
274	else if( bag_find(&blobs, zNew) )
275	printf("M %s :%d %s\n", mPerm ? "100755" : "100644", BLOBMARK(zNew), zName);
276	}
277	db_finalize(&q4);
278	db_finalize(&q3);
279	printf("\n");
280	}
281	db_finalize(&q2);
282	db_finalize(&q);
283	bag_clear(&blobs);
284	manifest_cache_clear();
285
286
	@@ -207,13 +301,34 @@
301	zEncoded = mprintf("%s", zTagname);
302	for(i=0; zEncoded[i]; i++){
303	if( !fossil_isalnum(zEncoded[i]) ) zEncoded[i] = '_';
304	}
305	printf("tag %s\n", zEncoded);
306	printf("from :%d\n", COMMITMARK(rid));
307	printf("tagger <tagger> %s +0000\n", zSecSince1970);
308	printf("data 0\n");
309	fossil_free(zEncoded);
310	}
311	db_finalize(&q);
312	bag_clear(&vers);
313
314	if( markfile_out!=0 ){
315	FILE *f;
316	f = fopen(markfile_out, "w");
317	if( f == 0 ){
318	fossil_panic("cannot open %s for writing", markfile_out);
319	}
320	db_prepare(&q, "SELECT rid FROM oldblob");
321	while( db_step(&q)==SQLITE_ROW ){
322	fprintf(f, "b%d\n", db_column_int(&q, 0));
323	}
324	db_finalize(&q);
325	db_prepare(&q, "SELECT rid FROM oldcommit");
326	while( db_step(&q)==SQLITE_ROW ){
327	fprintf(f, "c%d\n", db_column_int(&q, 0));
328	}
329	db_finalize(&q);
330	if( ferror(f)!=0 \|\| fclose(f)!=0 ) {
331	fossil_panic("error while writing %s", markfile_out);
332	}
333	}
334	}
335

M src/finfo.c

		--- src/finfo.c
		+++ src/finfo.c
		@@ -193,10 +193,12 @@
193	193	** Additional query parameters:
194	194	**
195	195	** a=DATE Only show changes after DATE
196	196	** b=DATE Only show changes before DATE
197	197	** n=NUM Show the first NUM changes only
	198	+** brbg Background color by branch name
	199	+** ubg Background color by user name
198	200	*/
199	201	void finfo_page(void){
200	202	Stmt q;
201	203	const char *zFilename;
202	204	char zPrevDate[20];
		@@ -204,10 +206,12 @@
204	206	const char *zB;
205	207	int n;
206	208	Blob title;
207	209	Blob sql;
208	210	GraphContext *pGraph;
	211	+ int brBg = P("brbg")!=0;
	212	+ int uBg = P("ubg")!=0;
209	213
210	214	login_check_credentials();
211	215	if( !g.okRead ){ login_needed(); return; }
212	216	style_header("File History");
213	217	login_anonymous_available();
		@@ -268,10 +272,15 @@
268	272	int gidx;
269	273	char zTime[10];
270	274	char zShort[20];
271	275	char zShortCkin[20];
272	276	if( zBr==0 ) zBr = "trunk";
	277	+ if( uBg ){
	278	+ zBgClr = hash_color(zUser);
	279	+ }else if( brBg \|\| zBgClr==0 \|\| zBgClr[0]==0 ){
	280	+ zBgClr = strcmp(zBr,"trunk")==0 ? "white" : hash_color(zBr);
	281	+ }
273	282	gidx = graph_add_row(pGraph, frid, fpid>0 ? 1 : 0, &fpid, zBr, zBgClr, 0);
274	283	if( memcmp(zDate, zPrevDate, 10) ){
275	284	sqlite3_snprintf(sizeof(zPrevDate), zPrevDate, "%.10s", zDate);
276	285	@ <tr><td>
277	286	@ <div class="divider">%s(zPrevDate)</div>
278	287

	--- src/finfo.c
	+++ src/finfo.c
	@@ -193,10 +193,12 @@
193	** Additional query parameters:
194	**
195	** a=DATE Only show changes after DATE
196	** b=DATE Only show changes before DATE
197	** n=NUM Show the first NUM changes only


198	*/
199	void finfo_page(void){
200	Stmt q;
201	const char *zFilename;
202	char zPrevDate[20];
	@@ -204,10 +206,12 @@
204	const char *zB;
205	int n;
206	Blob title;
207	Blob sql;
208	GraphContext *pGraph;


209
210	login_check_credentials();
211	if( !g.okRead ){ login_needed(); return; }
212	style_header("File History");
213	login_anonymous_available();
	@@ -268,10 +272,15 @@
268	int gidx;
269	char zTime[10];
270	char zShort[20];
271	char zShortCkin[20];
272	if( zBr==0 ) zBr = "trunk";





273	gidx = graph_add_row(pGraph, frid, fpid>0 ? 1 : 0, &fpid, zBr, zBgClr, 0);
274	if( memcmp(zDate, zPrevDate, 10) ){
275	sqlite3_snprintf(sizeof(zPrevDate), zPrevDate, "%.10s", zDate);
276	@ <tr><td>
277	@ <div class="divider">%s(zPrevDate)</div>
278

	--- src/finfo.c
	+++ src/finfo.c
	@@ -193,10 +193,12 @@
193	** Additional query parameters:
194	**
195	** a=DATE Only show changes after DATE
196	** b=DATE Only show changes before DATE
197	** n=NUM Show the first NUM changes only
198	** brbg Background color by branch name
199	** ubg Background color by user name
200	*/
201	void finfo_page(void){
202	Stmt q;
203	const char *zFilename;
204	char zPrevDate[20];
	@@ -204,10 +206,12 @@
206	const char *zB;
207	int n;
208	Blob title;
209	Blob sql;
210	GraphContext *pGraph;
211	int brBg = P("brbg")!=0;
212	int uBg = P("ubg")!=0;
213
214	login_check_credentials();
215	if( !g.okRead ){ login_needed(); return; }
216	style_header("File History");
217	login_anonymous_available();
	@@ -268,10 +272,15 @@
272	int gidx;
273	char zTime[10];
274	char zShort[20];
275	char zShortCkin[20];
276	if( zBr==0 ) zBr = "trunk";
277	if( uBg ){
278	zBgClr = hash_color(zUser);
279	}else if( brBg \|\| zBgClr==0 \|\| zBgClr[0]==0 ){
280	zBgClr = strcmp(zBr,"trunk")==0 ? "white" : hash_color(zBr);
281	}
282	gidx = graph_add_row(pGraph, frid, fpid>0 ? 1 : 0, &fpid, zBr, zBgClr, 0);
283	if( memcmp(zDate, zPrevDate, 10) ){
284	sqlite3_snprintf(sizeof(zPrevDate), zPrevDate, "%.10s", zDate);
285	@ <tr><td>
286	@ <div class="divider">%s(zPrevDate)</div>
287

M src/makemake.tcl

+2 -2

		--- src/makemake.tcl
		+++ src/makemake.tcl
		@@ -589,11 +589,11 @@
589	589	clean:
590	590	-del $(OBJDIR)\*.obj
591	591	-del .obj _.c .h .map
592	592
593	593	realclean:
594		- -del $(APPNAME) translate$E mkindex$E makeheaders$E version$E
	594	+ -del $(APPNAME) translate$E mkindex$E makeheaders$E mkversion$E
595	595
596	596	}
597	597	foreach s [lsort $src] {
598	598	writeln "\$(OBJDIR)\\$s\$O : ${s}_.c ${s}.h"
599	599	writeln "\t\$(TCC) -o\$@ -c ${s}_.c\n"
		@@ -727,11 +727,11 @@
727	727	-del $(OX)\*.obj
728	728	-del .obj _.c .h .map
729	729	-del headers linkopts
730	730
731	731	realclean:
732		- -del $(APPNAME) translate$E mkindex$E makeheaders$E version$E
	732	+ -del $(APPNAME) translate$E mkindex$E makeheaders$E mkversion$E
733	733
734	734	}
735	735	foreach s [lsort $src] {
736	736	writeln "\$(OX)\\$s\$O : ${s}_.c ${s}.h"
737	737	writeln "\t\$(TCC) /Fo\$@ -c ${s}_.c\n"
738	738

	--- src/makemake.tcl
	+++ src/makemake.tcl
	@@ -589,11 +589,11 @@
589	clean:
590	-del $(OBJDIR)\*.obj
591	-del .obj _.c .h .map
592
593	realclean:
594	-del $(APPNAME) translate$E mkindex$E makeheaders$E version$E
595
596	}
597	foreach s [lsort $src] {
598	writeln "\$(OBJDIR)\\$s\$O : ${s}_.c ${s}.h"
599	writeln "\t\$(TCC) -o\$@ -c ${s}_.c\n"
	@@ -727,11 +727,11 @@
727	-del $(OX)\*.obj
728	-del .obj _.c .h .map
729	-del headers linkopts
730
731	realclean:
732	-del $(APPNAME) translate$E mkindex$E makeheaders$E version$E
733
734	}
735	foreach s [lsort $src] {
736	writeln "\$(OX)\\$s\$O : ${s}_.c ${s}.h"
737	writeln "\t\$(TCC) /Fo\$@ -c ${s}_.c\n"
738

	--- src/makemake.tcl
	+++ src/makemake.tcl
	@@ -589,11 +589,11 @@
589	clean:
590	-del $(OBJDIR)\*.obj
591	-del .obj _.c .h .map
592
593	realclean:
594	-del $(APPNAME) translate$E mkindex$E makeheaders$E mkversion$E
595
596	}
597	foreach s [lsort $src] {
598	writeln "\$(OBJDIR)\\$s\$O : ${s}_.c ${s}.h"
599	writeln "\t\$(TCC) -o\$@ -c ${s}_.c\n"
	@@ -727,11 +727,11 @@
727	-del $(OX)\*.obj
728	-del .obj _.c .h .map
729	-del headers linkopts
730
731	realclean:
732	-del $(APPNAME) translate$E mkindex$E makeheaders$E mkversion$E
733
734	}
735	foreach s [lsort $src] {
736	writeln "\$(OX)\\$s\$O : ${s}_.c ${s}.h"
737	writeln "\t\$(TCC) /Fo\$@ -c ${s}_.c\n"
738

M src/manifest.c

+3 -4

		--- src/manifest.c
		+++ src/manifest.c
		@@ -1323,14 +1323,13 @@
1323	1323	if( pChildFile ){
1324	1324	add_one_mlink(cid, 0, pChildFile->zUuid, pChildFile->zName, 0,
1325	1325	isPublic, manifest_file_mperm(pChildFile));
1326	1326	}
1327	1327	}
1328		- }else if( pChild->zBaseline==0 && pParent->zBaseline!=0 ){
1329		- /* Parent is a delta but pChild is a baseline. Look for files that are
1330		- ** present in pParent but which are missing from pChild and mark them
1331		- ** has having been deleted. */
	1328	+ }else if( pChild->zBaseline==0 ){
	1329	+ /* pChild is a baseline. Look for files that are present in pParent
	1330	+ ** but are missing from pChild and mark them as having been deleted. */
1332	1331	manifest_file_rewind(pParent);
1333	1332	while( (pParentFile = manifest_file_next(pParent,0))!=0 ){
1334	1333	pChildFile = manifest_file_seek(pChild, pParentFile->zName);
1335	1334	if( pChildFile==0 && pParentFile->zUuid!=0 ){
1336	1335	add_one_mlink(cid, pParentFile->zUuid, 0, pParentFile->zName, 0,
1337	1336

	--- src/manifest.c
	+++ src/manifest.c
	@@ -1323,14 +1323,13 @@
1323	if( pChildFile ){
1324	add_one_mlink(cid, 0, pChildFile->zUuid, pChildFile->zName, 0,
1325	isPublic, manifest_file_mperm(pChildFile));
1326	}
1327	}
1328	}else if( pChild->zBaseline==0 && pParent->zBaseline!=0 ){
1329	/* Parent is a delta but pChild is a baseline. Look for files that are
1330	** present in pParent but which are missing from pChild and mark them
1331	** has having been deleted. */
1332	manifest_file_rewind(pParent);
1333	while( (pParentFile = manifest_file_next(pParent,0))!=0 ){
1334	pChildFile = manifest_file_seek(pChild, pParentFile->zName);
1335	if( pChildFile==0 && pParentFile->zUuid!=0 ){
1336	add_one_mlink(cid, pParentFile->zUuid, 0, pParentFile->zName, 0,
1337

	--- src/manifest.c
	+++ src/manifest.c
	@@ -1323,14 +1323,13 @@
1323	if( pChildFile ){
1324	add_one_mlink(cid, 0, pChildFile->zUuid, pChildFile->zName, 0,
1325	isPublic, manifest_file_mperm(pChildFile));
1326	}
1327	}
1328	}else if( pChild->zBaseline==0 ){
1329	/* pChild is a baseline. Look for files that are present in pParent
1330	** but are missing from pChild and mark them as having been deleted. */

1331	manifest_file_rewind(pParent);
1332	while( (pParentFile = manifest_file_next(pParent,0))!=0 ){
1333	pChildFile = manifest_file_seek(pChild, pParentFile->zName);
1334	if( pChildFile==0 && pParentFile->zUuid!=0 ){
1335	add_one_mlink(cid, pParentFile->zUuid, 0, pParentFile->zName, 0,
1336

M src/sqlite3.c

+1212 -471

		--- src/sqlite3.c
		+++ src/sqlite3.c
		@@ -1,8 +1,8 @@
1	1	/******************************************************************************
2	2	** This file is an amalgamation of many separate C source files from SQLite
3		-** version 3.7.7. By combining all the individual C code files into this
	3	+** version 3.7.8. By combining all the individual C code files into this
4	4	** single large file, the entire code can be compiled as a single translation
5	5	** unit. This allows many compilers to do optimizations that would not be
6	6	** possible if the files were compiled separately. Performance improvements
7	7	** of 5% or more are commonly seen when SQLite is compiled as a single
8	8	** translation unit.
		@@ -648,13 +648,13 @@
648	648	**
649	649	** See also: [sqlite3_libversion()],
650	650	** [sqlite3_libversion_number()], [sqlite3_sourceid()],
651	651	** [sqlite_version()] and [sqlite_source_id()].
652	652	*/
653		-#define SQLITE_VERSION "3.7.7"
654		-#define SQLITE_VERSION_NUMBER 3007007
655		-#define SQLITE_SOURCE_ID "2011-06-24 11:29:51 9b191bb4c7c1e1b12b188c0b3eee1f8f587887c8"
	653	+#define SQLITE_VERSION "3.7.8"
	654	+#define SQLITE_VERSION_NUMBER 3007008
	655	+#define SQLITE_SOURCE_ID "2011-07-19 18:29:00 ed5f0aad6b21066bacd01521e82c22e96991f400"
656	656
657	657	/*
658	658	** CAPI3REF: Run-Time Library Version Numbers
659	659	** KEYWORDS: sqlite3_version, sqlite3_sourceid
660	660	**
		@@ -1282,20 +1282,37 @@
1282	1282	** when [PRAGMA synchronous \| PRAGMA synchronous=OFF] is set, but most
1283	1283	** VFSes do not need this signal and should silently ignore this opcode.
1284	1284	** Applications should not call [sqlite3_file_control()] with this
1285	1285	** opcode as doing so may disrupt the operation of the specialized VFSes
1286	1286	** that do require it.
	1287	+**
	1288	+** ^The [SQLITE_FCNTL_WIN32_AV_RETRY] opcode is used to configure automatic
	1289	+** retry counts and intervals for certain disk I/O operations for the
	1290	+** windows [VFS] in order to work to provide robustness against
	1291	+** anti-virus programs. By default, the windows VFS will retry file read,
	1292	+** file write, and file delete opertions up to 10 times, with a delay
	1293	+** of 25 milliseconds before the first retry and with the delay increasing
	1294	+** by an additional 25 milliseconds with each subsequent retry. This
	1295	+** opcode allows those to values (10 retries and 25 milliseconds of delay)
	1296	+** to be adjusted. The values are changed for all database connections
	1297	+** within the same process. The argument is a pointer to an array of two
	1298	+** integers where the first integer i the new retry count and the second
	1299	+** integer is the delay. If either integer is negative, then the setting
	1300	+** is not changed but instead the prior value of that setting is written
	1301	+** into the array entry, allowing the current retry settings to be
	1302	+** interrogated. The zDbName parameter is ignored.
	1303	+**
1287	1304	*/
1288	1305	#define SQLITE_FCNTL_LOCKSTATE 1
1289	1306	#define SQLITE_GET_LOCKPROXYFILE 2
1290	1307	#define SQLITE_SET_LOCKPROXYFILE 3
1291	1308	#define SQLITE_LAST_ERRNO 4
1292	1309	#define SQLITE_FCNTL_SIZE_HINT 5
1293	1310	#define SQLITE_FCNTL_CHUNK_SIZE 6
1294	1311	#define SQLITE_FCNTL_FILE_POINTER 7
1295	1312	#define SQLITE_FCNTL_SYNC_OMITTED 8
1296		-
	1313	+#define SQLITE_FCNTL_WIN32_AV_RETRY 9
1297	1314
1298	1315	/*
1299	1316	** CAPI3REF: Mutex Handle
1300	1317	**
1301	1318	** The mutex module within SQLite defines [sqlite3_mutex] to be an
		@@ -9558,10 +9575,11 @@
9558	9575	#define SQLITE_IndexSearch 0x08 /* Disable indexes for searching */
9559	9576	#define SQLITE_IndexCover 0x10 /* Disable index covering table */
9560	9577	#define SQLITE_GroupByOrder 0x20 /* Disable GROUPBY cover of ORDERBY */
9561	9578	#define SQLITE_FactorOutConst 0x40 /* Disable factoring out constants */
9562	9579	#define SQLITE_IdxRealAsInt 0x80 /* Store REAL as INT in indices */
	9580	+#define SQLITE_DistinctOpt 0x80 /* DISTINCT using indexes */
9563	9581	#define SQLITE_OptMask 0xff /* Mask of all disablable opts */
9564	9582
9565	9583	/*
9566	9584	** Possible values for the sqlite.magic field.
9567	9585	** The numbers are obtained at random and have no special meaning, other
		@@ -10449,10 +10467,11 @@
10449	10467	Table pTab; / An SQL table corresponding to zName */
10450	10468	Select pSelect; / A SELECT statement used in place of a table name */
10451	10469	u8 isPopulated; /* Temporary table associated with SELECT is populated */
10452	10470	u8 jointype; /* Type of join between this able and the previous */
10453	10471	u8 notIndexed; /* True if there is a NOT INDEXED clause */
	10472	+ u8 isCorrelated; /* True if sub-query is correlated */
10454	10473	#ifndef SQLITE_OMIT_EXPLAIN
10455	10474	u8 iSelectId; /* If pSelect!=0, the id of the sub-select in EQP */
10456	10475	#endif
10457	10476	int iCursor; /* The VDBE cursor number used to access this table */
10458	10477	Expr pOn; / The ON clause of a join */
		@@ -10568,10 +10587,11 @@
10568	10587	struct WhereInfo {
10569	10588	Parse pParse; / Parsing and code generating context */
10570	10589	u16 wctrlFlags; /* Flags originally passed to sqlite3WhereBegin() */
10571	10590	u8 okOnePass; /* Ok to use one-pass algorithm for UPDATE or DELETE */
10572	10591	u8 untestedTerms; /* Not all WHERE terms resolved by outer loop */
	10592	+ u8 eDistinct;
10573	10593	SrcList pTabList; / List of tables in the join */
10574	10594	int iTop; /* The very beginning of the WHERE loop */
10575	10595	int iContinue; /* Jump here to continue with next record */
10576	10596	int iBreak; /* Jump here to break out of the loop */
10577	10597	int nLevel; /* Number of nested loop */
		@@ -10579,10 +10599,13 @@
10579	10599	double savedNQueryLoop; /* pParse->nQueryLoop outside the WHERE loop */
10580	10600	double nRowOut; /* Estimated number of output rows */
10581	10601	WhereLevel a[1]; /* Information about each nest loop in WHERE */
10582	10602	};
10583	10603
	10604	+#define WHERE_DISTINCT_UNIQUE 1
	10605	+#define WHERE_DISTINCT_ORDERED 2
	10606	+
10584	10607	/*
10585	10608	** A NameContext defines a context in which to resolve table and column
10586	10609	** names. The context consists of a list of tables (the pSrcList) field and
10587	10610	** a list of named expression (pEList). The named expression list may
10588	10611	** be NULL. The pSrc corresponds to the FROM clause of a SELECT or
		@@ -11340,11 +11363,11 @@
11340	11363	#if defined(SQLITE_ENABLE_UPDATE_DELETE_LIMIT) && !defined(SQLITE_OMIT_SUBQUERY)
11341	11364	SQLITE_PRIVATE Expr sqlite3LimitWhere(Parse , SrcList , Expr , ExprList , Expr , Expr , char );
11342	11365	#endif
11343	11366	SQLITE_PRIVATE void sqlite3DeleteFrom(Parse, SrcList, Expr*);
11344	11367	SQLITE_PRIVATE void sqlite3Update(Parse, SrcList, ExprList, Expr, int);
11345		-SQLITE_PRIVATE WhereInfo sqlite3WhereBegin(Parse, SrcList, Expr, ExprList**, u16);
	11368	+SQLITE_PRIVATE WhereInfo sqlite3WhereBegin(Parse, SrcList, Expr, ExprList*,ExprList,u16);
11346	11369	SQLITE_PRIVATE void sqlite3WhereEnd(WhereInfo*);
11347	11370	SQLITE_PRIVATE int sqlite3ExprCodeGetColumn(Parse, Table, int, int, int);
11348	11371	SQLITE_PRIVATE void sqlite3ExprCodeGetColumnOfTable(Vdbe, Table, int, int, int);
11349	11372	SQLITE_PRIVATE void sqlite3ExprCodeMove(Parse*, int, int, int);
11350	11373	SQLITE_PRIVATE void sqlite3ExprCodeCopy(Parse*, int, int, int);
		@@ -15841,11 +15864,11 @@
15841	15864	** Free an outstanding memory allocation.
15842	15865	**
15843	15866	** This function assumes that the necessary mutexes, if any, are
15844	15867	** already held by the caller. Hence "Unsafe".
15845	15868	*/
15846		-void memsys3FreeUnsafe(void *pOld){
	15869	+static void memsys3FreeUnsafe(void *pOld){
15847	15870	Mem3Block p = (Mem3Block)pOld;
15848	15871	int i;
15849	15872	u32 size, x;
15850	15873	assert( sqlite3_mutex_held(mem3.mutex) );
15851	15874	assert( p>mem3.aPool && p<&mem3.aPool[mem3.nPool] );
		@@ -15916,21 +15939,21 @@
15916	15939	}
15917	15940
15918	15941	/*
15919	15942	** Free memory.
15920	15943	*/
15921		-void memsys3Free(void *pPrior){
	15944	+static void memsys3Free(void *pPrior){
15922	15945	assert( pPrior );
15923	15946	memsys3Enter();
15924	15947	memsys3FreeUnsafe(pPrior);
15925	15948	memsys3Leave();
15926	15949	}
15927	15950
15928	15951	/*
15929	15952	** Change the size of an existing memory allocation
15930	15953	*/
15931		-void memsys3Realloc(void pPrior, int nBytes){
	15954	+static void memsys3Realloc(void pPrior, int nBytes){
15932	15955	int nOld;
15933	15956	void *p;
15934	15957	if( pPrior==0 ){
15935	15958	return sqlite3_malloc(nBytes);
15936	15959	}
		@@ -25143,11 +25166,13 @@
25143	25166	case EINVAL:
25144	25167	case ENOTCONN:
25145	25168	case ENODEV:
25146	25169	case ENXIO:
25147	25170	case ENOENT:
	25171	+#ifdef ESTALE /* ESTALE is not defined on Interix systems */
25148	25172	case ESTALE:
	25173	+#endif
25149	25174	case ENOSYS:
25150	25175	/* these should force the client to close the file and reconnect */
25151	25176
25152	25177	default:
25153	25178	return sqliteIOErr;
		@@ -31820,10 +31845,58 @@
31820	31845	iLine, iErrno, zFunc, zPath, zMsg
31821	31846	);
31822	31847
31823	31848	return errcode;
31824	31849	}
	31850	+
	31851	+/*
	31852	+** The number of times that a ReadFile(), WriteFile(), and DeleteFile()
	31853	+** will be retried following a locking error - probably caused by
	31854	+** antivirus software. Also the initial delay before the first retry.
	31855	+** The delay increases linearly with each retry.
	31856	+*/
	31857	+#ifndef SQLITE_WIN32_IOERR_RETRY
	31858	+# define SQLITE_WIN32_IOERR_RETRY 10
	31859	+#endif
	31860	+#ifndef SQLITE_WIN32_IOERR_RETRY_DELAY
	31861	+# define SQLITE_WIN32_IOERR_RETRY_DELAY 25
	31862	+#endif
	31863	+static int win32IoerrRetry = SQLITE_WIN32_IOERR_RETRY;
	31864	+static int win32IoerrRetryDelay = SQLITE_WIN32_IOERR_RETRY_DELAY;
	31865	+
	31866	+/*
	31867	+** If a ReadFile() or WriteFile() error occurs, invoke this routine
	31868	+** to see if it should be retried. Return TRUE to retry. Return FALSE
	31869	+** to give up with an error.
	31870	+*/
	31871	+static int retryIoerr(int *pnRetry){
	31872	+ DWORD e;
	31873	+ if( *pnRetry>=win32IoerrRetry ){
	31874	+ return 0;
	31875	+ }
	31876	+ e = GetLastError();
	31877	+ if( e==ERROR_ACCESS_DENIED \|\|
	31878	+ e==ERROR_LOCK_VIOLATION \|\|
	31879	+ e==ERROR_SHARING_VIOLATION ){
	31880	+ Sleep(win32IoerrRetryDelay(1+pnRetry));
	31881	+ ++*pnRetry;
	31882	+ return 1;
	31883	+ }
	31884	+ return 0;
	31885	+}
	31886	+
	31887	+/*
	31888	+** Log a I/O error retry episode.
	31889	+*/
	31890	+static void logIoerr(int nRetry){
	31891	+ if( nRetry ){
	31892	+ sqlite3_log(SQLITE_IOERR,
	31893	+ "delayed %dms for lock/sharing conflict",
	31894	+ win32IoerrRetryDelaynRetry(nRetry+1)/2
	31895	+ );
	31896	+ }
	31897	+}
31825	31898
31826	31899	#if SQLITE_OS_WINCE
31827	31900	/*************************************************************************
31828	31901	** This section contains code for WinCE only.
31829	31902	*/
		@@ -32238,22 +32311,25 @@
32238	32311	int amt, /* Number of bytes to read */
32239	32312	sqlite3_int64 offset /* Begin reading at this offset */
32240	32313	){
32241	32314	winFile pFile = (winFile)id; /* file handle */
32242	32315	DWORD nRead; /* Number of bytes actually read from file */
	32316	+ int nRetry = 0; /* Number of retrys */
32243	32317
32244	32318	assert( id!=0 );
32245	32319	SimulateIOError(return SQLITE_IOERR_READ);
32246	32320	OSTRACE(("READ %d lock=%d\n", pFile->h, pFile->locktype));
32247	32321
32248	32322	if( seekWinFile(pFile, offset) ){
32249	32323	return SQLITE_FULL;
32250	32324	}
32251		- if( !ReadFile(pFile->h, pBuf, amt, &nRead, 0) ){
	32325	+ while( !ReadFile(pFile->h, pBuf, amt, &nRead, 0) ){
	32326	+ if( retryIoerr(&nRetry) ) continue;
32252	32327	pFile->lastErrno = GetLastError();
32253	32328	return winLogError(SQLITE_IOERR_READ, "winRead", pFile->zPath);
32254	32329	}
	32330	+ logIoerr(nRetry);
32255	32331	if( nRead<(DWORD)amt ){
32256	32332	/* Unread parts of the buffer must be zero-filled */
32257	32333	memset(&((char*)pBuf)[nRead], 0, amt-nRead);
32258	32334	return SQLITE_IOERR_SHORT_READ;
32259	32335	}
		@@ -32271,10 +32347,11 @@
32271	32347	int amt, /* Number of bytes to write */
32272	32348	sqlite3_int64 offset /* Offset into the file to begin writing at */
32273	32349	){
32274	32350	int rc; /* True if error has occured, else false */
32275	32351	winFile pFile = (winFile)id; /* File handle */
	32352	+ int nRetry = 0; /* Number of retries */
32276	32353
32277	32354	assert( amt>0 );
32278	32355	assert( pFile );
32279	32356	SimulateIOError(return SQLITE_IOERR_WRITE);
32280	32357	SimulateDiskfullError(return SQLITE_FULL);
		@@ -32285,11 +32362,16 @@
32285	32362	if( rc==0 ){
32286	32363	u8 aRem = (u8 )pBuf; /* Data yet to be written */
32287	32364	int nRem = amt; /* Number of bytes yet to be written */
32288	32365	DWORD nWrite; /* Bytes written by each WriteFile() call */
32289	32366
32290		- while( nRem>0 && WriteFile(pFile->h, aRem, nRem, &nWrite, 0) && nWrite>0 ){
	32367	+ while( nRem>0 ){
	32368	+ if( !WriteFile(pFile->h, aRem, nRem, &nWrite, 0) ){
	32369	+ if( retryIoerr(&nRetry) ) continue;
	32370	+ break;
	32371	+ }
	32372	+ if( nWrite<=0 ) break;
32291	32373	aRem += nWrite;
32292	32374	nRem -= nWrite;
32293	32375	}
32294	32376	if( nRem>0 ){
32295	32377	pFile->lastErrno = GetLastError();
		@@ -32301,10 +32383,12 @@
32301	32383	if( ( pFile->lastErrno==ERROR_HANDLE_DISK_FULL )
32302	32384	\|\| ( pFile->lastErrno==ERROR_DISK_FULL )){
32303	32385	return SQLITE_FULL;
32304	32386	}
32305	32387	return winLogError(SQLITE_IOERR_WRITE, "winWrite", pFile->zPath);
	32388	+ }else{
	32389	+ logIoerr(nRetry);
32306	32390	}
32307	32391	return SQLITE_OK;
32308	32392	}
32309	32393
32310	32394	/*
		@@ -32716,10 +32800,24 @@
32716	32800	SimulateIOErrorBenign(0);
32717	32801	return SQLITE_OK;
32718	32802	}
32719	32803	case SQLITE_FCNTL_SYNC_OMITTED: {
32720	32804	return SQLITE_OK;
	32805	+ }
	32806	+ case SQLITE_FCNTL_WIN32_AV_RETRY: {
	32807	+ int a = (int)pArg;
	32808	+ if( a[0]>0 ){
	32809	+ win32IoerrRetry = a[0];
	32810	+ }else{
	32811	+ a[0] = win32IoerrRetry;
	32812	+ }
	32813	+ if( a[1]>0 ){
	32814	+ win32IoerrRetryDelay = a[1];
	32815	+ }else{
	32816	+ a[1] = win32IoerrRetryDelay;
	32817	+ }
	32818	+ return SQLITE_OK;
32721	32819	}
32722	32820	}
32723	32821	return SQLITE_NOTFOUND;
32724	32822	}
32725	32823
		@@ -33734,19 +33832,17 @@
33734	33832	** file open, we will be unable to delete it. To work around this
33735	33833	** problem, we delay 100 milliseconds and try to delete again. Up
33736	33834	** to MX_DELETION_ATTEMPTs deletion attempts are run before giving
33737	33835	** up and returning an error.
33738	33836	*/
33739		-#define MX_DELETION_ATTEMPTS 5
33740	33837	static int winDelete(
33741	33838	sqlite3_vfs pVfs, / Not used on win32 */
33742	33839	const char zFilename, / Name of file to delete */
33743	33840	int syncDir /* Not used on win32 */
33744	33841	){
33745	33842	int cnt = 0;
33746		- DWORD rc;
33747		- DWORD error = 0;
	33843	+ int rc;
33748	33844	void *zConverted;
33749	33845	UNUSED_PARAMETER(pVfs);
33750	33846	UNUSED_PARAMETER(syncDir);
33751	33847
33752	33848	SimulateIOError(return SQLITE_IOERR_DELETE);
		@@ -33753,38 +33849,34 @@
33753	33849	zConverted = convertUtf8Filename(zFilename);
33754	33850	if( zConverted==0 ){
33755	33851	return SQLITE_NOMEM;
33756	33852	}
33757	33853	if( isNT() ){
33758		- do{
33759		- DeleteFileW(zConverted);
33760		- }while( ( ((rc = GetFileAttributesW(zConverted)) != INVALID_FILE_ATTRIBUTES)
33761		- \|\| ((error = GetLastError()) == ERROR_ACCESS_DENIED))
33762		- && (++cnt < MX_DELETION_ATTEMPTS)
33763		- && (Sleep(100), 1) );
	33854	+ rc = 1;
	33855	+ while( GetFileAttributesW(zConverted)!=INVALID_FILE_ATTRIBUTES &&
	33856	+ (rc = DeleteFileW(zConverted))==0 && retryIoerr(&cnt) ){}
	33857	+ rc = rc ? SQLITE_OK : SQLITE_ERROR;
33764	33858	/* isNT() is 1 if SQLITE_OS_WINCE==1, so this else is never executed.
33765	33859	** Since the ASCII version of these Windows API do not exist for WINCE,
33766	33860	** it's important to not reference them for WINCE builds.
33767	33861	*/
33768	33862	#if SQLITE_OS_WINCE==0
33769	33863	}else{
33770		- do{
33771		- DeleteFileA(zConverted);
33772		- }while( ( ((rc = GetFileAttributesA(zConverted)) != INVALID_FILE_ATTRIBUTES)
33773		- \|\| ((error = GetLastError()) == ERROR_ACCESS_DENIED))
33774		- && (++cnt < MX_DELETION_ATTEMPTS)
33775		- && (Sleep(100), 1) );
	33864	+ rc = 1;
	33865	+ while( GetFileAttributesA(zConverted)!=INVALID_FILE_ATTRIBUTES &&
	33866	+ (rc = DeleteFileA(zConverted))==0 && retryIoerr(&cnt) ){}
	33867	+ rc = rc ? SQLITE_OK : SQLITE_ERROR;
33776	33868	#endif
33777	33869	}
	33870	+ if( rc ){
	33871	+ rc = winLogError(SQLITE_IOERR_DELETE, "winDelete", zFilename);
	33872	+ }else{
	33873	+ logIoerr(cnt);
	33874	+ }
33778	33875	free(zConverted);
33779		- OSTRACE(("DELETE \"%s\" %s\n", zFilename,
33780		- ( (rc==INVALID_FILE_ATTRIBUTES) && (error==ERROR_FILE_NOT_FOUND)) ?
33781		- "ok" : "failed" ));
33782		-
33783		- return ( (rc == INVALID_FILE_ATTRIBUTES)
33784		- && (error == ERROR_FILE_NOT_FOUND)) ? SQLITE_OK :
33785		- winLogError(SQLITE_IOERR_DELETE, "winDelete", zFilename);
	33876	+ OSTRACE(("DELETE \"%s\" %s\n", zFilename, (rc ? "failed" : "ok" )));
	33877	+ return rc;
33786	33878	}
33787	33879
33788	33880	/*
33789	33881	** Check the existance and status of a file.
33790	33882	*/
		@@ -59049,10 +59141,11 @@
59049	59141	nMem = 10;
59050	59142	}
59051	59143	memset(zCsr, 0, zEnd-zCsr);
59052	59144	zCsr += (zCsr - (u8*)0)&7;
59053	59145	assert( EIGHT_BYTE_ALIGNMENT(zCsr) );
	59146	+ p->expired = 0;
59054	59147
59055	59148	/* Memory for registers, parameters, cursor, etc, is allocated in two
59056	59149	** passes. On the first pass, we try to reuse unused space at the
59057	59150	** end of the opcode array. If we are unable to satisfy all memory
59058	59151	** requirements by reusing the opcode array tail, then the second
		@@ -61277,11 +61370,11 @@
61277	61370	sqlite3_mutex_enter(db->mutex);
61278	61371	while( (rc = sqlite3Step(v))==SQLITE_SCHEMA
61279	61372	&& cnt++ < SQLITE_MAX_SCHEMA_RETRY
61280	61373	&& (rc2 = rc = sqlite3Reprepare(v))==SQLITE_OK ){
61281	61374	sqlite3_reset(pStmt);
61282		- v->expired = 0;
	61375	+ assert( v->expired==0 );
61283	61376	}
61284	61377	if( rc2!=SQLITE_OK && ALWAYS(v->isPrepareV2) && ALWAYS(db->pErr) ){
61285	61378	/* This case occurs after failing to recompile an sql statement.
61286	61379	** The error message from the SQL compiler has already been loaded
61287	61380	** into the database handle. This block copies the error message
		@@ -65968,11 +66061,11 @@
65968	66061	** automatically created table with root-page 1 (an BLOB_INTKEY table).
65969	66062	*/
65970	66063	if( pOp->p4.pKeyInfo ){
65971	66064	int pgno;
65972	66065	assert( pOp->p4type==P4_KEYINFO );
65973		- rc = sqlite3BtreeCreateTable(u.ax.pCx->pBt, &pgno, BTREE_BLOBKEY);
	66066	+ rc = sqlite3BtreeCreateTable(u.ax.pCx->pBt, &pgno, BTREE_BLOBKEY \| pOp->p5);
65974	66067	if( rc==SQLITE_OK ){
65975	66068	assert( pgno==MASTER_ROOT+1 );
65976	66069	rc = sqlite3BtreeCursor(u.ax.pCx->pBt, pgno, 1,
65977	66070	(KeyInfo*)pOp->p4.z, u.ax.pCx->pCursor);
65978	66071	u.ax.pCx->pKeyInfo = pOp->p4.pKeyInfo;
		@@ -71007,15 +71100,29 @@
71007	71100	/* Recursively resolve names in all subqueries
71008	71101	*/
71009	71102	for(i=0; i<p->pSrc->nSrc; i++){
71010	71103	struct SrcList_item *pItem = &p->pSrc->a[i];
71011	71104	if( pItem->pSelect ){
	71105	+ NameContext pNC; / Used to iterate name contexts */
	71106	+ int nRef = 0; /* Refcount for pOuterNC and outer contexts */
71012	71107	const char *zSavedContext = pParse->zAuthContext;
	71108	+
	71109	+ /* Count the total number of references to pOuterNC and all of its
	71110	+ ** parent contexts. After resolving references to expressions in
	71111	+ ** pItem->pSelect, check if this value has changed. If so, then
	71112	+ ** SELECT statement pItem->pSelect must be correlated. Set the
	71113	+ ** pItem->isCorrelated flag if this is the case. */
	71114	+ for(pNC=pOuterNC; pNC; pNC=pNC->pNext) nRef += pNC->nRef;
	71115	+
71013	71116	if( pItem->zName ) pParse->zAuthContext = pItem->zName;
71014	71117	sqlite3ResolveSelectNames(pParse, pItem->pSelect, pOuterNC);
71015	71118	pParse->zAuthContext = zSavedContext;
71016	71119	if( pParse->nErr \|\| db->mallocFailed ) return WRC_Abort;
	71120	+
	71121	+ for(pNC=pOuterNC; pNC; pNC=pNC->pNext) nRef -= pNC->nRef;
	71122	+ assert( pItem->isCorrelated==0 && nRef<=0 );
	71123	+ pItem->isCorrelated = (nRef!=0);
71017	71124	}
71018	71125	}
71019	71126
71020	71127	/* If there are no aggregate functions in the result-set, and no GROUP BY
71021	71128	** expression, do not allow aggregates in any of the other expressions.
		@@ -72119,10 +72226,11 @@
72119	72226	pNewItem->zName = sqlite3DbStrDup(db, pOldItem->zName);
72120	72227	pNewItem->zAlias = sqlite3DbStrDup(db, pOldItem->zAlias);
72121	72228	pNewItem->jointype = pOldItem->jointype;
72122	72229	pNewItem->iCursor = pOldItem->iCursor;
72123	72230	pNewItem->isPopulated = pOldItem->isPopulated;
	72231	+ pNewItem->isCorrelated = pOldItem->isCorrelated;
72124	72232	pNewItem->zIndex = sqlite3DbStrDup(db, pOldItem->zIndex);
72125	72233	pNewItem->notIndexed = pOldItem->notIndexed;
72126	72234	pNewItem->pIndex = pOldItem->pIndex;
72127	72235	pTab = pNewItem->pTab = pOldItem->pTab;
72128	72236	if( pTab ){
		@@ -80126,11 +80234,11 @@
80126	80234	if( pStart ){
80127	80235	assert( pEnd!=0 );
80128	80236	/* A named index with an explicit CREATE INDEX statement */
80129	80237	zStmt = sqlite3MPrintf(db, "CREATE%s INDEX %.*s",
80130	80238	onError==OE_None ? "" : " UNIQUE",
80131		- pEnd->z - pName->z + 1,
	80239	+ (int)(pEnd->z - pName->z) + 1,
80132	80240	pName->z);
80133	80241	}else{
80134	80242	/* An automatic index created by a PRIMARY KEY or UNIQUE constraint */
80135	80243	/* zStmt = sqlite3MPrintf(""); */
80136	80244	zStmt = 0;
		@@ -81921,11 +82029,13 @@
81921	82029	int regRowid; /* Actual register containing rowids */
81922	82030
81923	82031	/* Collect rowids of every row to be deleted.
81924	82032	*/
81925	82033	sqlite3VdbeAddOp2(v, OP_Null, 0, iRowSet);
81926		- pWInfo = sqlite3WhereBegin(pParse, pTabList, pWhere,0,WHERE_DUPLICATES_OK);
	82034	+ pWInfo = sqlite3WhereBegin(
	82035	+ pParse, pTabList, pWhere, 0, 0, WHERE_DUPLICATES_OK
	82036	+ );
81927	82037	if( pWInfo==0 ) goto delete_from_cleanup;
81928	82038	regRowid = sqlite3ExprCodeGetColumn(pParse, pTab, -1, iCur, iRowid);
81929	82039	sqlite3VdbeAddOp2(v, OP_RowSetAdd, iRowSet, regRowid);
81930	82040	if( db->flags & SQLITE_CountRows ){
81931	82041	sqlite3VdbeAddOp2(v, OP_AddImm, memCnt, 1);
		@@ -84368,11 +84478,11 @@
84368	84478
84369	84479	/* Create VDBE to loop through the entries in pSrc that match the WHERE
84370	84480	** clause. If the constraint is not deferred, throw an exception for
84371	84481	** each row found. Otherwise, for deferred constraints, increment the
84372	84482	** deferred constraint counter by nIncr for each row selected. */
84373		- pWInfo = sqlite3WhereBegin(pParse, pSrc, pWhere, 0, 0);
	84483	+ pWInfo = sqlite3WhereBegin(pParse, pSrc, pWhere, 0, 0, 0);
84374	84484	if( nIncr>0 && pFKey->isDeferred==0 ){
84375	84485	sqlite3ParseToplevel(pParse)->mayAbort = 1;
84376	84486	}
84377	84487	sqlite3VdbeAddOp2(v, OP_FkCounter, pFKey->isDeferred, nIncr);
84378	84488	if( pWInfo ){
		@@ -87235,10 +87345,13 @@
87235	87345	int (strnicmp)(const char,const char*,int);
87236	87346	int (unlock_notify)(sqlite3,void()(void,int),void);
87237	87347	int (wal_autocheckpoint)(sqlite3,int);
87238	87348	int (wal_checkpoint)(sqlite3,const char*);
87239	87349	void (wal_hook)(sqlite3,int()(void,sqlite3,const char,int),void);
	87350	+ int (blob_reopen)(sqlite3_blob,sqlite3_int64);
	87351	+ int (vtab_config)(sqlite3,int op,...);
	87352	+ int (vtab_on_conflict)(sqlite3);
87240	87353	};
87241	87354
87242	87355	/*
87243	87356	** The following macros redefine the API routines so that they are
87244	87357	** redirected throught the global sqlite3_api structure.
		@@ -87435,10 +87548,13 @@
87435	87548	#define sqlite3_strnicmp sqlite3_api->strnicmp
87436	87549	#define sqlite3_unlock_notify sqlite3_api->unlock_notify
87437	87550	#define sqlite3_wal_autocheckpoint sqlite3_api->wal_autocheckpoint
87438	87551	#define sqlite3_wal_checkpoint sqlite3_api->wal_checkpoint
87439	87552	#define sqlite3_wal_hook sqlite3_api->wal_hook
	87553	+#define sqlite3_blob_reopen sqlite3_api->blob_reopen
	87554	+#define sqlite3_vtab_config sqlite3_api->vtab_config
	87555	+#define sqlite3_vtab_on_conflict sqlite3_api->vtab_on_conflict
87440	87556	#endif /* SQLITE_CORE */
87441	87557
87442	87558	#define SQLITE_EXTENSION_INIT1 const sqlite3_api_routines *sqlite3_api = 0;
87443	87559	#define SQLITE_EXTENSION_INIT2(v) sqlite3_api = v;
87444	87560
		@@ -87509,10 +87625,12 @@
87509	87625
87510	87626	#ifdef SQLITE_OMIT_VIRTUALTABLE
87511	87627	# define sqlite3_create_module 0
87512	87628	# define sqlite3_create_module_v2 0
87513	87629	# define sqlite3_declare_vtab 0
	87630	+# define sqlite3_vtab_config 0
	87631	+# define sqlite3_vtab_on_conflict 0
87514	87632	#endif
87515	87633
87516	87634	#ifdef SQLITE_OMIT_SHARED_CACHE
87517	87635	# define sqlite3_enable_shared_cache 0
87518	87636	#endif
		@@ -87532,10 +87650,11 @@
87532	87650	#define sqlite3_blob_bytes 0
87533	87651	#define sqlite3_blob_close 0
87534	87652	#define sqlite3_blob_open 0
87535	87653	#define sqlite3_blob_read 0
87536	87654	#define sqlite3_blob_write 0
	87655	+#define sqlite3_blob_reopen 0
87537	87656	#endif
87538	87657
87539	87658	/*
87540	87659	** The following structure contains pointers to all SQLite API routines.
87541	87660	** A pointer to this structure is passed into extensions when they are
		@@ -87797,10 +87916,13 @@
87797	87916	#else
87798	87917	0,
87799	87918	0,
87800	87919	0,
87801	87920	#endif
	87921	+ sqlite3_blob_reopen,
	87922	+ sqlite3_vtab_config,
	87923	+ sqlite3_vtab_on_conflict,
87802	87924	};
87803	87925
87804	87926	/*
87805	87927	** Attempt to load an SQLite extension library contained in the file
87806	87928	** zFile. The entry point is zProc. zProc may be 0 in which case a
		@@ -94186,10 +94308,11 @@
94186	94308	Expr pHaving; / The HAVING clause. May be NULL */
94187	94309	int isDistinct; /* True if the DISTINCT keyword is present */
94188	94310	int distinct; /* Table to use for the distinct set */
94189	94311	int rc = 1; /* Value to return from this function */
94190	94312	int addrSortIndex; /* Address of an OP_OpenEphemeral instruction */
	94313	+ int addrDistinctIndex; /* Address of an OP_OpenEphemeral instruction */
94191	94314	AggInfo sAggInfo; /* Information used by aggregate queries */
94192	94315	int iEnd; /* Address of the end of the query */
94193	94316	sqlite3 db; / The database connection */
94194	94317
94195	94318	#ifndef SQLITE_OMIT_EXPLAIN
		@@ -94312,20 +94435,10 @@
94312	94435	explainSetInteger(pParse->iSelectId, iRestoreSelectId);
94313	94436	return rc;
94314	94437	}
94315	94438	#endif
94316	94439
94317		- /* If possible, rewrite the query to use GROUP BY instead of DISTINCT.
94318		- ** GROUP BY might use an index, DISTINCT never does.
94319		- */
94320		- assert( p->pGroupBy==0 \|\| (p->selFlags & SF_Aggregate)!=0 );
94321		- if( (p->selFlags & (SF_Distinct\|SF_Aggregate))==SF_Distinct ){
94322		- p->pGroupBy = sqlite3ExprListDup(db, p->pEList, 0);
94323		- pGroupBy = p->pGroupBy;
94324		- p->selFlags &= ~SF_Distinct;
94325		- }
94326		-
94327	94440	/* If there is both a GROUP BY and an ORDER BY clause and they are
94328	94441	** identical, then disable the ORDER BY clause since the GROUP BY
94329	94442	** will cause elements to come out in the correct order. This is
94330	94443	** an optimization - the correct answer should result regardless.
94331	94444	** Use the SQLITE_GroupByOrder flag with SQLITE_TESTCTRL_OPTIMIZER
		@@ -94333,10 +94446,34 @@
94333	94446	*/
94334	94447	if( sqlite3ExprListCompare(p->pGroupBy, pOrderBy)==0
94335	94448	&& (db->flags & SQLITE_GroupByOrder)==0 ){
94336	94449	pOrderBy = 0;
94337	94450	}
	94451	+
	94452	+ /* If the query is DISTINCT with an ORDER BY but is not an aggregate, and
	94453	+ ** if the select-list is the same as the ORDER BY list, then this query
	94454	+ ** can be rewritten as a GROUP BY. In other words, this:
	94455	+ **
	94456	+ ** SELECT DISTINCT xyz FROM ... ORDER BY xyz
	94457	+ **
	94458	+ ** is transformed to:
	94459	+ **
	94460	+ ** SELECT xyz FROM ... GROUP BY xyz
	94461	+ **
	94462	+ ** The second form is preferred as a single index (or temp-table) may be
	94463	+ ** used for both the ORDER BY and DISTINCT processing. As originally
	94464	+ ** written the query must use a temp-table for at least one of the ORDER
	94465	+ ** BY and DISTINCT, and an index or separate temp-table for the other.
	94466	+ */
	94467	+ if( (p->selFlags & (SF_Distinct\|SF_Aggregate))==SF_Distinct
	94468	+ && sqlite3ExprListCompare(pOrderBy, p->pEList)==0
	94469	+ ){
	94470	+ p->selFlags &= ~SF_Distinct;
	94471	+ p->pGroupBy = sqlite3ExprListDup(db, p->pEList, 0);
	94472	+ pGroupBy = p->pGroupBy;
	94473	+ pOrderBy = 0;
	94474	+ }
94338	94475
94339	94476	/* If there is an ORDER BY clause, then this sorting
94340	94477	** index might end up being unused if the data can be
94341	94478	** extracted in pre-sorted order. If that is the case, then the
94342	94479	** OP_OpenEphemeral instruction will be changed to an OP_Noop once
		@@ -94369,26 +94506,25 @@
94369	94506
94370	94507	/* Open a virtual index to use for the distinct set.
94371	94508	*/
94372	94509	if( p->selFlags & SF_Distinct ){
94373	94510	KeyInfo *pKeyInfo;
94374		- assert( isAgg \|\| pGroupBy );
94375	94511	distinct = pParse->nTab++;
94376	94512	pKeyInfo = keyInfoFromExprList(pParse, p->pEList);
94377		- sqlite3VdbeAddOp4(v, OP_OpenEphemeral, distinct, 0, 0,
94378		- (char*)pKeyInfo, P4_KEYINFO_HANDOFF);
	94513	+ addrDistinctIndex = sqlite3VdbeAddOp4(v, OP_OpenEphemeral, distinct, 0, 0,
	94514	+ (char*)pKeyInfo, P4_KEYINFO_HANDOFF);
94379	94515	sqlite3VdbeChangeP5(v, BTREE_UNORDERED);
94380	94516	}else{
94381		- distinct = -1;
	94517	+ distinct = addrDistinctIndex = -1;
94382	94518	}
94383	94519
94384	94520	/* Aggregate and non-aggregate queries are handled differently */
94385	94521	if( !isAgg && pGroupBy==0 ){
94386		- /* This case is for non-aggregate queries
94387		- ** Begin the database scan
94388		- */
94389		- pWInfo = sqlite3WhereBegin(pParse, pTabList, pWhere, &pOrderBy, 0);
	94522	+ ExprList *pDist = (isDistinct ? p->pEList : 0);
	94523	+
	94524	+ /* Begin the database scan. */
	94525	+ pWInfo = sqlite3WhereBegin(pParse, pTabList, pWhere, &pOrderBy, pDist, 0);
94390	94526	if( pWInfo==0 ) goto select_end;
94391	94527	if( pWInfo->nRowOut < p->nSelectRow ) p->nSelectRow = pWInfo->nRowOut;
94392	94528
94393	94529	/* If sorting index that was created by a prior OP_OpenEphemeral
94394	94530	** instruction ended up not being needed, then change the OP_OpenEphemeral
		@@ -94397,14 +94533,56 @@
94397	94533	if( addrSortIndex>=0 && pOrderBy==0 ){
94398	94534	sqlite3VdbeChangeToNoop(v, addrSortIndex, 1);
94399	94535	p->addrOpenEphm[2] = -1;
94400	94536	}
94401	94537
94402		- /* Use the standard inner loop
94403		- */
94404		- assert(!isDistinct);
94405		- selectInnerLoop(pParse, p, pEList, 0, 0, pOrderBy, -1, pDest,
	94538	+ if( pWInfo->eDistinct ){
	94539	+ VdbeOp pOp; / No longer required OpenEphemeral instr. */
	94540	+
	94541	+ assert( addrDistinctIndex>0 );
	94542	+ pOp = sqlite3VdbeGetOp(v, addrDistinctIndex);
	94543	+
	94544	+ assert( isDistinct );
	94545	+ assert( pWInfo->eDistinct==WHERE_DISTINCT_ORDERED
	94546	+ \|\| pWInfo->eDistinct==WHERE_DISTINCT_UNIQUE
	94547	+ );
	94548	+ distinct = -1;
	94549	+ if( pWInfo->eDistinct==WHERE_DISTINCT_ORDERED ){
	94550	+ int iJump;
	94551	+ int iExpr;
	94552	+ int iFlag = ++pParse->nMem;
	94553	+ int iBase = pParse->nMem+1;
	94554	+ int iBase2 = iBase + pEList->nExpr;
	94555	+ pParse->nMem += (pEList->nExpr*2);
	94556	+
	94557	+ /* Change the OP_OpenEphemeral coded earlier to an OP_Integer. The
	94558	+ ** OP_Integer initializes the "first row" flag. */
	94559	+ pOp->opcode = OP_Integer;
	94560	+ pOp->p1 = 1;
	94561	+ pOp->p2 = iFlag;
	94562	+
	94563	+ sqlite3ExprCodeExprList(pParse, pEList, iBase, 1);
	94564	+ iJump = sqlite3VdbeCurrentAddr(v) + 1 + pEList->nExpr + 1 + 1;
	94565	+ sqlite3VdbeAddOp2(v, OP_If, iFlag, iJump-1);
	94566	+ for(iExpr=0; iExpr<pEList->nExpr; iExpr++){
	94567	+ CollSeq *pColl = sqlite3ExprCollSeq(pParse, pEList->a[iExpr].pExpr);
	94568	+ sqlite3VdbeAddOp3(v, OP_Ne, iBase+iExpr, iJump, iBase2+iExpr);
	94569	+ sqlite3VdbeChangeP4(v, -1, (const char *)pColl, P4_COLLSEQ);
	94570	+ sqlite3VdbeChangeP5(v, SQLITE_NULLEQ);
	94571	+ }
	94572	+ sqlite3VdbeAddOp2(v, OP_Goto, 0, pWInfo->iContinue);
	94573	+
	94574	+ sqlite3VdbeAddOp2(v, OP_Integer, 0, iFlag);
	94575	+ assert( sqlite3VdbeCurrentAddr(v)==iJump );
	94576	+ sqlite3VdbeAddOp3(v, OP_Move, iBase, iBase2, pEList->nExpr);
	94577	+ }else{
	94578	+ pOp->opcode = OP_Noop;
	94579	+ }
	94580	+ }
	94581	+
	94582	+ /* Use the standard inner loop. */
	94583	+ selectInnerLoop(pParse, p, pEList, 0, 0, pOrderBy, distinct, pDest,
94406	94584	pWInfo->iContinue, pWInfo->iBreak);
94407	94585
94408	94586	/* End the database scan loop.
94409	94587	*/
94410	94588	sqlite3WhereEnd(pWInfo);
		@@ -94510,11 +94688,11 @@
94510	94688	** This might involve two separate loops with an OP_Sort in between, or
94511	94689	** it might be a single loop that uses an index to extract information
94512	94690	** in the right order to begin with.
94513	94691	*/
94514	94692	sqlite3VdbeAddOp2(v, OP_Gosub, regReset, addrReset);
94515		- pWInfo = sqlite3WhereBegin(pParse, pTabList, pWhere, &pGroupBy, 0);
	94693	+ pWInfo = sqlite3WhereBegin(pParse, pTabList, pWhere, &pGroupBy, 0, 0);
94516	94694	if( pWInfo==0 ) goto select_end;
94517	94695	if( pGroupBy==0 ){
94518	94696	/* The optimizer is able to deliver rows in group by order so
94519	94697	** we do not have to sort. The OP_OpenEphemeral table will be
94520	94698	** cancelled later because we still need to use the pKeyInfo
		@@ -94772,11 +94950,11 @@
94772	94950	/* This case runs if the aggregate has no GROUP BY clause. The
94773	94951	** processing is much simpler since there is only a single row
94774	94952	** of output.
94775	94953	*/
94776	94954	resetAccumulator(pParse, &sAggInfo);
94777		- pWInfo = sqlite3WhereBegin(pParse, pTabList, pWhere, &pMinMax, flag);
	94955	+ pWInfo = sqlite3WhereBegin(pParse, pTabList, pWhere, &pMinMax, 0, flag);
94778	94956	if( pWInfo==0 ){
94779	94957	sqlite3ExprListDelete(db, pDel);
94780	94958	goto select_end;
94781	94959	}
94782	94960	updateAccumulator(pParse, &sAggInfo);
		@@ -95248,19 +95426,32 @@
95248	95426	iDb = sqlite3TwoPartName(pParse, pName1, pName2, &pName);
95249	95427	if( iDb<0 ){
95250	95428	goto trigger_cleanup;
95251	95429	}
95252	95430	}
	95431	+ if( !pTableName \|\| db->mallocFailed ){
	95432	+ goto trigger_cleanup;
	95433	+ }
	95434	+
	95435	+ /* A long-standing parser bug is that this syntax was allowed:
	95436	+ **
	95437	+ ** CREATE TRIGGER attached.demo AFTER INSERT ON attached.tab ....
	95438	+ ** ^^^^^^^^
	95439	+ **
	95440	+ ** To maintain backwards compatibility, ignore the database
	95441	+ ** name on pTableName if we are reparsing our of SQLITE_MASTER.
	95442	+ */
	95443	+ if( db->init.busy && iDb!=1 ){
	95444	+ sqlite3DbFree(db, pTableName->a[0].zDatabase);
	95445	+ pTableName->a[0].zDatabase = 0;
	95446	+ }
95253	95447
95254	95448	/* If the trigger name was unqualified, and the table is a temp table,
95255	95449	** then set iDb to 1 to create the trigger in the temporary database.
95256	95450	** If sqlite3SrcListLookup() returns 0, indicating the table does not
95257	95451	** exist, the error is caught by the block below.
95258	95452	*/
95259		- if( !pTableName \|\| db->mallocFailed ){
95260		- goto trigger_cleanup;
95261		- }
95262	95453	pTab = sqlite3SrcListLookup(pParse, pTableName);
95263	95454	if( db->init.busy==0 && pName2->n==0 && pTab
95264	95455	&& pTab->pSchema==db->aDb[1].pSchema ){
95265	95456	iDb = 1;
95266	95457	}
		@@ -96554,11 +96745,13 @@
96554	96745	}
96555	96746
96556	96747	/* Begin the database scan
96557	96748	*/
96558	96749	sqlite3VdbeAddOp2(v, OP_Null, 0, regOldRowid);
96559		- pWInfo = sqlite3WhereBegin(pParse, pTabList, pWhere,0, WHERE_ONEPASS_DESIRED);
	96750	+ pWInfo = sqlite3WhereBegin(
	96751	+ pParse, pTabList, pWhere, 0, 0, WHERE_ONEPASS_DESIRED
	96752	+ );
96560	96753	if( pWInfo==0 ) goto update_cleanup;
96561	96754	okOnePass = pWInfo->okOnePass;
96562	96755
96563	96756	/* Remember the rowid of every item to be updated.
96564	96757	*/
		@@ -98580,10 +98773,11 @@
98580	98773	#define WHERE_REVERSE 0x02000000 /* Scan in reverse order */
98581	98774	#define WHERE_UNIQUE 0x04000000 /* Selects no more than one row */
98582	98775	#define WHERE_VIRTUALTABLE 0x08000000 /* Use virtual-table processing */
98583	98776	#define WHERE_MULTI_OR 0x10000000 /* OR using multiple indices */
98584	98777	#define WHERE_TEMP_INDEX 0x20000000 /* Uses an ephemeral index */
	98778	+#define WHERE_DISTINCT 0x40000000 /* Correct order for DISTINCT */
98585	98779
98586	98780	/*
98587	98781	** Initialize a preallocated WhereClause structure.
98588	98782	*/
98589	98783	static void whereClauseInit(
		@@ -99724,10 +99918,166 @@
99724	99918	}
99725	99919	}
99726	99920	return 0;
99727	99921	}
99728	99922
	99923	+/*
	99924	+** This function searches the expression list passed as the second argument
	99925	+** for an expression of type TK_COLUMN that refers to the same column and
	99926	+** uses the same collation sequence as the iCol'th column of index pIdx.
	99927	+** Argument iBase is the cursor number used for the table that pIdx refers
	99928	+** to.
	99929	+**
	99930	+** If such an expression is found, its index in pList->a[] is returned. If
	99931	+** no expression is found, -1 is returned.
	99932	+*/
	99933	+static int findIndexCol(
	99934	+ Parse pParse, / Parse context */
	99935	+ ExprList pList, / Expression list to search */
	99936	+ int iBase, /* Cursor for table associated with pIdx */
	99937	+ Index pIdx, / Index to match column of */
	99938	+ int iCol /* Column of index to match */
	99939	+){
	99940	+ int i;
	99941	+ const char *zColl = pIdx->azColl[iCol];
	99942	+
	99943	+ for(i=0; i<pList->nExpr; i++){
	99944	+ Expr *p = pList->a[i].pExpr;
	99945	+ if( p->op==TK_COLUMN
	99946	+ && p->iColumn==pIdx->aiColumn[iCol]
	99947	+ && p->iTable==iBase
	99948	+ ){
	99949	+ CollSeq *pColl = sqlite3ExprCollSeq(pParse, p);
	99950	+ if( ALWAYS(pColl) && 0==sqlite3StrICmp(pColl->zName, zColl) ){
	99951	+ return i;
	99952	+ }
	99953	+ }
	99954	+ }
	99955	+
	99956	+ return -1;
	99957	+}
	99958	+
	99959	+/*
	99960	+** This routine determines if pIdx can be used to assist in processing a
	99961	+** DISTINCT qualifier. In other words, it tests whether or not using this
	99962	+** index for the outer loop guarantees that rows with equal values for
	99963	+** all expressions in the pDistinct list are delivered grouped together.
	99964	+**
	99965	+** For example, the query
	99966	+**
	99967	+** SELECT DISTINCT a, b, c FROM tbl WHERE a = ?
	99968	+**
	99969	+** can benefit from any index on columns "b" and "c".
	99970	+*/
	99971	+static int isDistinctIndex(
	99972	+ Parse pParse, / Parsing context */
	99973	+ WhereClause pWC, / The WHERE clause */
	99974	+ Index pIdx, / The index being considered */
	99975	+ int base, /* Cursor number for the table pIdx is on */
	99976	+ ExprList pDistinct, / The DISTINCT expressions */
	99977	+ int nEqCol /* Number of index columns with == */
	99978	+){
	99979	+ Bitmask mask = 0; /* Mask of unaccounted for pDistinct exprs */
	99980	+ int i; /* Iterator variable */
	99981	+
	99982	+ if( pIdx->zName==0 \|\| pDistinct==0 \|\| pDistinct->nExpr>=BMS ) return 0;
	99983	+ testcase( pDistinct->nExpr==BMS-1 );
	99984	+
	99985	+ /* Loop through all the expressions in the distinct list. If any of them
	99986	+ ** are not simple column references, return early. Otherwise, test if the
	99987	+ ** WHERE clause contains a "col=X" clause. If it does, the expression
	99988	+ ** can be ignored. If it does not, and the column does not belong to the
	99989	+ ** same table as index pIdx, return early. Finally, if there is no
	99990	+ ** matching "col=X" expression and the column is on the same table as pIdx,
	99991	+ ** set the corresponding bit in variable mask.
	99992	+ */
	99993	+ for(i=0; i<pDistinct->nExpr; i++){
	99994	+ WhereTerm *pTerm;
	99995	+ Expr *p = pDistinct->a[i].pExpr;
	99996	+ if( p->op!=TK_COLUMN ) return 0;
	99997	+ pTerm = findTerm(pWC, p->iTable, p->iColumn, ~(Bitmask)0, WO_EQ, 0);
	99998	+ if( pTerm ){
	99999	+ Expr *pX = pTerm->pExpr;
	100000	+ CollSeq *p1 = sqlite3BinaryCompareCollSeq(pParse, pX->pLeft, pX->pRight);
	100001	+ CollSeq *p2 = sqlite3ExprCollSeq(pParse, p);
	100002	+ if( p1==p2 ) continue;
	100003	+ }
	100004	+ if( p->iTable!=base ) return 0;
	100005	+ mask \|= (((Bitmask)1) << i);
	100006	+ }
	100007	+
	100008	+ for(i=nEqCol; mask && i<pIdx->nColumn; i++){
	100009	+ int iExpr = findIndexCol(pParse, pDistinct, base, pIdx, i);
	100010	+ if( iExpr<0 ) break;
	100011	+ mask &= ~(((Bitmask)1) << iExpr);
	100012	+ }
	100013	+
	100014	+ return (mask==0);
	100015	+}
	100016	+
	100017	+
	100018	+/*
	100019	+** Return true if the DISTINCT expression-list passed as the third argument
	100020	+** is redundant. A DISTINCT list is redundant if the database contains a
	100021	+** UNIQUE index that guarantees that the result of the query will be distinct
	100022	+** anyway.
	100023	+*/
	100024	+static int isDistinctRedundant(
	100025	+ Parse *pParse,
	100026	+ SrcList *pTabList,
	100027	+ WhereClause *pWC,
	100028	+ ExprList *pDistinct
	100029	+){
	100030	+ Table *pTab;
	100031	+ Index *pIdx;
	100032	+ int i;
	100033	+ int iBase;
	100034	+
	100035	+ /* If there is more than one table or sub-select in the FROM clause of
	100036	+ ** this query, then it will not be possible to show that the DISTINCT
	100037	+ ** clause is redundant. */
	100038	+ if( pTabList->nSrc!=1 ) return 0;
	100039	+ iBase = pTabList->a[0].iCursor;
	100040	+ pTab = pTabList->a[0].pTab;
	100041	+
	100042	+ /* If any of the expressions is an IPK column on table iBase, then return
	100043	+ ** true. Note: The (p->iTable==iBase) part of this test may be false if the
	100044	+ ** current SELECT is a correlated sub-query.
	100045	+ */
	100046	+ for(i=0; i<pDistinct->nExpr; i++){
	100047	+ Expr *p = pDistinct->a[i].pExpr;
	100048	+ if( p->op==TK_COLUMN && p->iTable==iBase && p->iColumn<0 ) return 1;
	100049	+ }
	100050	+
	100051	+ /* Loop through all indices on the table, checking each to see if it makes
	100052	+ ** the DISTINCT qualifier redundant. It does so if:
	100053	+ **
	100054	+ ** 1. The index is itself UNIQUE, and
	100055	+ **
	100056	+ ** 2. All of the columns in the index are either part of the pDistinct
	100057	+ ** list, or else the WHERE clause contains a term of the form "col=X",
	100058	+ ** where X is a constant value. The collation sequences of the
	100059	+ ** comparison and select-list expressions must match those of the index.
	100060	+ */
	100061	+ for(pIdx=pTab->pIndex; pIdx; pIdx=pIdx->pNext){
	100062	+ if( pIdx->onError==OE_None ) continue;
	100063	+ for(i=0; i<pIdx->nColumn; i++){
	100064	+ int iCol = pIdx->aiColumn[i];
	100065	+ if( 0==findTerm(pWC, iBase, iCol, ~(Bitmask)0, WO_EQ, pIdx)
	100066	+ && 0>findIndexCol(pParse, pDistinct, iBase, pIdx, i)
	100067	+ ){
	100068	+ break;
	100069	+ }
	100070	+ }
	100071	+ if( i==pIdx->nColumn ){
	100072	+ /* This index implies that the DISTINCT qualifier is redundant. */
	100073	+ return 1;
	100074	+ }
	100075	+ }
	100076	+
	100077	+ return 0;
	100078	+}
99729	100079
99730	100080	/*
99731	100081	** This routine decides if pIdx can be used to satisfy the ORDER BY
99732	100082	** clause. If it can, it returns 1. If pIdx cannot satisfy the
99733	100083	** ORDER BY clause, this routine returns 0.
		@@ -99760,11 +100110,14 @@
99760	100110	int sortOrder = 0; /* XOR of index and ORDER BY sort direction */
99761	100111	int nTerm; /* Number of ORDER BY terms */
99762	100112	struct ExprList_item pTerm; / A term of the ORDER BY clause */
99763	100113	sqlite3 *db = pParse->db;
99764	100114
99765		- assert( pOrderBy!=0 );
	100115	+ if( !pOrderBy ) return 0;
	100116	+ if( wsFlags & WHERE_COLUMN_IN ) return 0;
	100117	+ if( pIdx->bUnordered ) return 0;
	100118	+
99766	100119	nTerm = pOrderBy->nExpr;
99767	100120	assert( nTerm>0 );
99768	100121
99769	100122	/* Argument pIdx must either point to a 'real' named index structure,
99770	100123	** or an index structure allocated on the stack by bestBtreeIndex() to
		@@ -100073,10 +100426,14 @@
100073	100426	double costTempIdx; /* per-query cost of the transient index */
100074	100427	WhereTerm pTerm; / A single term of the WHERE clause */
100075	100428	WhereTerm pWCEnd; / End of pWC->a[] */
100076	100429	Table pTable; / Table tht might be indexed */
100077	100430
	100431	+ if( pParse->nQueryLoop<=(double)1 ){
	100432	+ /* There is no point in building an automatic index for a single scan */
	100433	+ return;
	100434	+ }
100078	100435	if( (pParse->db->flags & SQLITE_AutoIndex)==0 ){
100079	100436	/* Automatic indices are disabled at run-time */
100080	100437	return;
100081	100438	}
100082	100439	if( (pCost->plan.wsFlags & WHERE_NOT_FULLSCAN)!=0 ){
		@@ -100084,10 +100441,14 @@
100084	100441	return;
100085	100442	}
100086	100443	if( pSrc->notIndexed ){
100087	100444	/* The NOT INDEXED clause appears in the SQL. */
100088	100445	return;
	100446	+ }
	100447	+ if( pSrc->isCorrelated ){
	100448	+ /* The source is a correlated sub-query. No point in indexing it. */
	100449	+ return;
100089	100450	}
100090	100451
100091	100452	assert( pParse->nQueryLoop >= (double)1 );
100092	100453	pTable = pSrc->pTab;
100093	100454	nTableRow = pTable->nRowEst;
		@@ -101016,10 +101377,11 @@
101016	101377	WhereClause pWC, / The WHERE clause */
101017	101378	struct SrcList_item pSrc, / The FROM clause term to search */
101018	101379	Bitmask notReady, /* Mask of cursors not available for indexing */
101019	101380	Bitmask notValid, /* Cursors not available for any purpose */
101020	101381	ExprList pOrderBy, / The ORDER BY clause */
	101382	+ ExprList pDistinct, / The select-list if query is DISTINCT */
101021	101383	WhereCost pCost / Lowest cost query plan */
101022	101384	){
101023	101385	int iCur = pSrc->iCursor; /* The cursor of the table to be accessed */
101024	101386	Index pProbe; / An index we are evaluating */
101025	101387	Index pIdx; / Copy of pProbe, or zero for IPK index */
		@@ -101156,11 +101518,12 @@
101156	101518	int nEq; /* Number of == or IN terms matching index */
101157	101519	int bInEst = 0; /* True if "x IN (SELECT...)" seen */
101158	101520	int nInMul = 1; /* Number of distinct equalities to lookup */
101159	101521	int estBound = 100; /* Estimated reduction in search space */
101160	101522	int nBound = 0; /* Number of range constraints seen */
101161		- int bSort = 0; /* True if external sort required */
	101523	+ int bSort = !!pOrderBy; /* True if external sort required */
	101524	+ int bDist = !!pDistinct; /* True if index cannot help with DISTINCT */
101162	101525	int bLookup = 0; /* True if not a covering index */
101163	101526	WhereTerm pTerm; / A single term of the WHERE clause */
101164	101527	#ifdef SQLITE_ENABLE_STAT2
101165	101528	WhereTerm pFirstTerm = 0; / First term matching the index */
101166	101529	#endif
		@@ -101220,21 +101583,24 @@
101220	101583
101221	101584	/* If there is an ORDER BY clause and the index being considered will
101222	101585	** naturally scan rows in the required order, set the appropriate flags
101223	101586	** in wsFlags. Otherwise, if there is an ORDER BY clause but the index
101224	101587	** will scan rows in a different order, set the bSort variable. */
101225		- if( pOrderBy ){
101226		- if( (wsFlags & WHERE_COLUMN_IN)==0
101227		- && pProbe->bUnordered==0
101228		- && isSortingIndex(pParse, pWC->pMaskSet, pProbe, iCur, pOrderBy,
101229		- nEq, wsFlags, &rev)
101230		- ){
101231		- wsFlags \|= WHERE_ROWID_RANGE\|WHERE_COLUMN_RANGE\|WHERE_ORDERBY;
101232		- wsFlags \|= (rev ? WHERE_REVERSE : 0);
101233		- }else{
101234		- bSort = 1;
101235		- }
	101588	+ if( isSortingIndex(
	101589	+ pParse, pWC->pMaskSet, pProbe, iCur, pOrderBy, nEq, wsFlags, &rev)
	101590	+ ){
	101591	+ bSort = 0;
	101592	+ wsFlags \|= WHERE_ROWID_RANGE\|WHERE_COLUMN_RANGE\|WHERE_ORDERBY;
	101593	+ wsFlags \|= (rev ? WHERE_REVERSE : 0);
	101594	+ }
	101595	+
	101596	+ /* If there is a DISTINCT qualifier and this index will scan rows in
	101597	+ ** order of the DISTINCT expressions, clear bDist and set the appropriate
	101598	+ ** flags in wsFlags. */
	101599	+ if( isDistinctIndex(pParse, pWC, pProbe, iCur, pDistinct, nEq) ){
	101600	+ bDist = 0;
	101601	+ wsFlags \|= WHERE_ROWID_RANGE\|WHERE_COLUMN_RANGE\|WHERE_DISTINCT;
101236	101602	}
101237	101603
101238	101604	/* If currently calculating the cost of using an index (not the IPK
101239	101605	** index), determine if all required column data may be obtained without
101240	101606	** using the main table (i.e. if the index is a covering
		@@ -101265,16 +101631,17 @@
101265	101631	nRow = aiRowEst[0]/2;
101266	101632	nInMul = (int)(nRow / aiRowEst[nEq]);
101267	101633	}
101268	101634
101269	101635	#ifdef SQLITE_ENABLE_STAT2
101270		- /* If the constraint is of the form x=VALUE and histogram
	101636	+ /* If the constraint is of the form x=VALUE or x IN (E1,E2,...)
	101637	+ ** and we do not think that values of x are unique and if histogram
101271	101638	** data is available for column x, then it might be possible
101272	101639	** to get a better estimate on the number of rows based on
101273	101640	** VALUE and how common that value is according to the histogram.
101274	101641	*/
101275		- if( nRow>(double)1 && nEq==1 && pFirstTerm!=0 ){
	101642	+ if( nRow>(double)1 && nEq==1 && pFirstTerm!=0 && aiRowEst[1]>1 ){
101276	101643	if( pFirstTerm->eOperator & (WO_EQ\|WO_ISNULL) ){
101277	101644	testcase( pFirstTerm->eOperator==WO_EQ );
101278	101645	testcase( pFirstTerm->eOperator==WO_ISNULL );
101279	101646	whereEqualScanEst(pParse, pProbe, pFirstTerm->pExpr->pRight, &nRow);
101280	101647	}else if( pFirstTerm->eOperator==WO_IN && bInEst==0 ){
		@@ -101347,10 +101714,13 @@
101347	101714	** difference and select C of 3.0.
101348	101715	*/
101349	101716	if( bSort ){
101350	101717	cost += nRowestLog(nRow)3;
101351	101718	}
	101719	+ if( bDist ){
	101720	+ cost += nRowestLog(nRow)3;
	101721	+ }
101352	101722
101353	101723	/** Cost of using this index has now been computed **/
101354	101724
101355	101725	/* If there are additional constraints on this table that cannot
101356	101726	** be used with the current index, but which might lower the number
		@@ -101492,11 +101862,11 @@
101492	101862	}
101493	101863	sqlite3DbFree(pParse->db, p);
101494	101864	}else
101495	101865	#endif
101496	101866	{
101497		- bestBtreeIndex(pParse, pWC, pSrc, notReady, notValid, pOrderBy, pCost);
	101867	+ bestBtreeIndex(pParse, pWC, pSrc, notReady, notValid, pOrderBy, 0, pCost);
101498	101868	}
101499	101869	}
101500	101870
101501	101871	/*
101502	101872	** Disable a term in the WHERE clause. Except, do not disable the term
		@@ -102454,11 +102824,11 @@
102454	102824	for(ii=0; ii<pOrWc->nTerm; ii++){
102455	102825	WhereTerm *pOrTerm = &pOrWc->a[ii];
102456	102826	if( pOrTerm->leftCursor==iCur \|\| pOrTerm->eOperator==WO_AND ){
102457	102827	WhereInfo pSubWInfo; / Info for single OR-term scan */
102458	102828	/* Loop through table entries that match term pOrTerm. */
102459		- pSubWInfo = sqlite3WhereBegin(pParse, pOrTab, pOrTerm->pExpr, 0,
	102829	+ pSubWInfo = sqlite3WhereBegin(pParse, pOrTab, pOrTerm->pExpr, 0, 0,
102460	102830	WHERE_OMIT_OPEN \| WHERE_OMIT_CLOSE \|
102461	102831	WHERE_FORCE_TABLE \| WHERE_ONETABLE_ONLY);
102462	102832	if( pSubWInfo ){
102463	102833	explainOneScan(
102464	102834	pParse, pOrTab, &pSubWInfo->a[0], iLevel, pLevel->iFrom, 0
		@@ -102695,10 +103065,11 @@
102695	103065	SQLITE_PRIVATE WhereInfo *sqlite3WhereBegin(
102696	103066	Parse pParse, / The parser context */
102697	103067	SrcList pTabList, / A list of all tables to be scanned */
102698	103068	Expr pWhere, / The WHERE clause */
102699	103069	ExprList *ppOrderBy, / An ORDER BY clause, or NULL */
	103070	+ ExprList pDistinct, / The select-list for DISTINCT queries - or NULL */
102700	103071	u16 wctrlFlags /* One of the WHERE_* flags defined in sqliteInt.h */
102701	103072	){
102702	103073	int i; /* Loop counter */
102703	103074	int nByteWInfo; /* Num. bytes allocated for WhereInfo struct */
102704	103075	int nTabList; /* Number of elements in pTabList */
		@@ -102754,10 +103125,14 @@
102754	103125	pWInfo->iBreak = sqlite3VdbeMakeLabel(v);
102755	103126	pWInfo->pWC = pWC = (WhereClause )&((u8 )pWInfo)[nByteWInfo];
102756	103127	pWInfo->wctrlFlags = wctrlFlags;
102757	103128	pWInfo->savedNQueryLoop = pParse->nQueryLoop;
102758	103129	pMaskSet = (WhereMaskSet*)&pWC[1];
	103130	+
	103131	+ /* Disable the DISTINCT optimization if SQLITE_DistinctOpt is set via
	103132	+ ** sqlite3_test_ctrl(SQLITE_TESTCTRL_OPTIMIZATIONS,...) */
	103133	+ if( db->flags & SQLITE_DistinctOpt ) pDistinct = 0;
102759	103134
102760	103135	/* Split the WHERE clause into separate subexpressions where each
102761	103136	** subexpression is separated by an AND operator.
102762	103137	*/
102763	103138	initMaskSet(pMaskSet);
		@@ -102821,10 +103196,19 @@
102821	103196	*/
102822	103197	exprAnalyzeAll(pTabList, pWC);
102823	103198	if( db->mallocFailed ){
102824	103199	goto whereBeginError;
102825	103200	}
	103201	+
	103202	+ /* Check if the DISTINCT qualifier, if there is one, is redundant.
	103203	+ ** If it is, then set pDistinct to NULL and WhereInfo.eDistinct to
	103204	+ ** WHERE_DISTINCT_UNIQUE to tell the caller to ignore the DISTINCT.
	103205	+ */
	103206	+ if( pDistinct && isDistinctRedundant(pParse, pTabList, pWC, pDistinct) ){
	103207	+ pDistinct = 0;
	103208	+ pWInfo->eDistinct = WHERE_DISTINCT_UNIQUE;
	103209	+ }
102826	103210
102827	103211	/* Chose the best index to use for each table in the FROM clause.
102828	103212	**
102829	103213	** This loop fills in the following fields:
102830	103214	**
		@@ -102905,10 +103289,11 @@
102905	103289	Bitmask mask; /* Mask of tables not yet ready */
102906	103290	for(j=iFrom, pTabItem=&pTabList->a[j]; j<nTabList; j++, pTabItem++){
102907	103291	int doNotReorder; /* True if this table should not be reordered */
102908	103292	WhereCost sCost; /* Cost information from best[Virtual]Index() */
102909	103293	ExprList pOrderBy; / ORDER BY clause for index to optimize */
	103294	+ ExprList pDist; / DISTINCT clause for index to optimize */
102910	103295
102911	103296	doNotReorder = (pTabItem->jointype & (JT_LEFT\|JT_CROSS))!=0;
102912	103297	if( j!=iFrom && doNotReorder ) break;
102913	103298	m = getMask(pMaskSet, pTabItem->iCursor);
102914	103299	if( (m & notReady)==0 ){
		@@ -102915,10 +103300,11 @@
102915	103300	if( j==iFrom ) iFrom++;
102916	103301	continue;
102917	103302	}
102918	103303	mask = (isOptimal ? m : notReady);
102919	103304	pOrderBy = ((i==0 && ppOrderBy )?*ppOrderBy:0);
	103305	+ pDist = (i==0 ? pDistinct : 0);
102920	103306	if( pTabItem->pIndex==0 ) nUnconstrained++;
102921	103307
102922	103308	WHERETRACE(("=== trying table %d with isOptimal=%d ===\n",
102923	103309	j, isOptimal));
102924	103310	assert( pTabItem->pTab );
		@@ -102929,11 +103315,11 @@
102929	103315	&sCost, pp);
102930	103316	}else
102931	103317	#endif
102932	103318	{
102933	103319	bestBtreeIndex(pParse, pWC, pTabItem, mask, notReady, pOrderBy,
102934		- &sCost);
	103320	+ pDist, &sCost);
102935	103321	}
102936	103322	assert( isOptimal \|\| (sCost.used&notReady)==0 );
102937	103323
102938	103324	/* If an INDEXED BY clause is present, then the plan must use that
102939	103325	** index if it uses any index at all */
		@@ -102989,10 +103375,14 @@
102989	103375	WHERETRACE(("*** Optimizer selects table %d for loop %d"
102990	103376	" with cost=%g and nRow=%g\n",
102991	103377	bestJ, pLevel-pWInfo->a, bestPlan.rCost, bestPlan.plan.nRow));
102992	103378	if( (bestPlan.plan.wsFlags & WHERE_ORDERBY)!=0 ){
102993	103379	*ppOrderBy = 0;
	103380	+ }
	103381	+ if( (bestPlan.plan.wsFlags & WHERE_DISTINCT)!=0 ){
	103382	+ assert( pWInfo->eDistinct==0 );
	103383	+ pWInfo->eDistinct = WHERE_DISTINCT_ORDERED;
102994	103384	}
102995	103385	andFlags &= bestPlan.plan.wsFlags;
102996	103386	pLevel->plan = bestPlan.plan;
102997	103387	testcase( bestPlan.plan.wsFlags & WHERE_INDEXED );
102998	103388	testcase( bestPlan.plan.wsFlags & WHERE_TEMP_INDEX );
		@@ -111519,11 +111909,17 @@
111519	111909	*/
111520	111910	#if defined(SQLITE_ENABLE_FTS4) && !defined(SQLITE_ENABLE_FTS3)
111521	111911	# define SQLITE_ENABLE_FTS3
111522	111912	#endif
111523	111913
111524		-#ifdef SQLITE_ENABLE_FTS3
	111914	+#if !defined(SQLITE_CORE) \|\| defined(SQLITE_ENABLE_FTS3)
	111915	+
	111916	+/* If not building as part of the core, include sqlite3ext.h. */
	111917	+#ifndef SQLITE_CORE
	111918	+SQLITE_API extern const sqlite3_api_routines *sqlite3_api;
	111919	+#endif
	111920	+
111525	111921	/************ Include fts3_tokenizer.h in the middle of fts3Int.h ********/
111526	111922	/************ Begin file fts3_tokenizer.h ********************************/
111527	111923	/*
111528	111924	** 2006 July 10
111529	111925	**
		@@ -112052,11 +112448,11 @@
112052	112448
112053	112449	sqlite3_int64 iDocid; /* Current docid (if pList!=0) */
112054	112450	int bFreeList; /* True if pList should be sqlite3_free()d */
112055	112451	char pList; / Pointer to position list following iDocid */
112056	112452	int nList; /* Length of position list */
112057		-} doclist;
	112453	+};
112058	112454
112059	112455	/*
112060	112456	** A "phrase" is a sequence of one or more tokens that must match in
112061	112457	** sequence. A single token is the base case and the most common case.
112062	112458	** For a sequence of tokens contained in double-quotes (i.e. "one two three")
		@@ -112252,23 +112648,12 @@
112252	112648	#endif
112253	112649
112254	112650	/* fts3_aux.c */
112255	112651	SQLITE_PRIVATE int sqlite3Fts3InitAux(sqlite3 *db);
112256	112652
112257		-SQLITE_PRIVATE int sqlite3Fts3TermSegReaderCursor(
112258		- Fts3Cursor pCsr, / Virtual table cursor handle */
112259		- const char zTerm, / Term to query for */
112260		- int nTerm, /* Size of zTerm in bytes */
112261		- int isPrefix, /* True for a prefix search */
112262		- Fts3MultiSegReader *ppSegcsr / OUT: Allocated seg-reader cursor */
112263		-);
112264		-
112265	112653	SQLITE_PRIVATE void sqlite3Fts3EvalPhraseCleanup(Fts3Phrase *);
112266	112654
112267		-SQLITE_PRIVATE int sqlite3Fts3EvalStart(Fts3Cursor , Fts3Expr , int);
112268		-SQLITE_PRIVATE int sqlite3Fts3EvalNext(Fts3Cursor *pCsr);
112269		-
112270	112655	SQLITE_PRIVATE int sqlite3Fts3MsrIncrStart(
112271	112656	Fts3Table, Fts3MultiSegReader, int, const char*, int);
112272	112657	SQLITE_PRIVATE int sqlite3Fts3MsrIncrNext(
112273	112658	Fts3Table , Fts3MultiSegReader , sqlite3_int64 , char , int );
112274	112659	SQLITE_PRIVATE char sqlite3Fts3EvalPhrasePoslist(Fts3Cursor , Fts3Expr *, int iCol);
		@@ -112275,11 +112660,11 @@
112275	112660	SQLITE_PRIVATE int sqlite3Fts3MsrOvfl(Fts3Cursor , Fts3MultiSegReader , int *);
112276	112661	SQLITE_PRIVATE int sqlite3Fts3MsrIncrRestart(Fts3MultiSegReader *pCsr);
112277	112662
112278	112663	SQLITE_PRIVATE int sqlite3Fts3DeferredTokenList(Fts3DeferredToken , char , int );
112279	112664
112280		-#endif /* SQLITE_ENABLE_FTS3 */
	112665	+#endif /* !SQLITE_CORE \|\| SQLITE_ENABLE_FTS3 */
112281	112666	#endif /* _FTSINT_H */
112282	112667
112283	112668	/************ End of fts3Int.h *******************************************/
112284	112669	/************ Continuing where we left off in fts3.c *********************/
112285	112670	#if !defined(SQLITE_CORE) \|\| defined(SQLITE_ENABLE_FTS3)
		@@ -112291,10 +112676,15 @@
112291	112676
112292	112677	#ifndef SQLITE_CORE
112293	112678	SQLITE_EXTENSION_INIT1
112294	112679	#endif
112295	112680
	112681	+static int fts3EvalNext(Fts3Cursor *pCsr);
	112682	+static int fts3EvalStart(Fts3Cursor *pCsr);
	112683	+static int fts3TermSegReaderCursor(
	112684	+ Fts3Cursor , const char , int, int, Fts3MultiSegReader **);
	112685	+
112296	112686	/*
112297	112687	** Write a 64-bit variable-length integer to memory starting at p[0].
112298	112688	** The length of data written will be between 1 and FTS3_VARINT_MAX bytes.
112299	112689	** The number of bytes written is returned.
112300	112690	*/
		@@ -112799,31 +113189,60 @@
112799	113189	}
112800	113190	sqlite3_free(zFree);
112801	113191	return zRet;
112802	113192	}
112803	113193
	113194	+/*
	113195	+** This function interprets the string at (*pp) as a non-negative integer
	113196	+** value. It reads the integer and sets *pnOut to the value read, then
	113197	+** sets *pp to point to the byte immediately following the last byte of
	113198	+** the integer value.
	113199	+**
	113200	+** Only decimal digits ('0'..'9') may be part of an integer value.
	113201	+**
	113202	+** If *pp does not being with a decimal digit SQLITE_ERROR is returned and
	113203	+** the output value undefined. Otherwise SQLITE_OK is returned.
	113204	+**
	113205	+** This function is used when parsing the "prefix=" FTS4 parameter.
	113206	+*/
112804	113207	static int fts3GobbleInt(const char *pp, int pnOut){
112805		- const char p = pp;
112806		- int nInt = 0;
	113208	+ const char p = pp; /* Iterator pointer */
	113209	+ int nInt = 0; /* Output value */
	113210	+
112807	113211	for(p=*pp; p[0]>='0' && p[0]<='9'; p++){
112808	113212	nInt = nInt * 10 + (p[0] - '0');
112809	113213	}
112810	113214	if( p==*pp ) return SQLITE_ERROR;
112811	113215	*pnOut = nInt;
112812	113216	*pp = p;
112813	113217	return SQLITE_OK;
112814	113218	}
112815	113219
112816		-
	113220	+/*
	113221	+** This function is called to allocate an array of Fts3Index structures
	113222	+** representing the indexes maintained by the current FTS table. FTS tables
	113223	+** always maintain the main "terms" index, but may also maintain one or
	113224	+** more "prefix" indexes, depending on the value of the "prefix=" parameter
	113225	+** (if any) specified as part of the CREATE VIRTUAL TABLE statement.
	113226	+**
	113227	+** Argument zParam is passed the value of the "prefix=" option if one was
	113228	+** specified, or NULL otherwise.
	113229	+**
	113230	+** If no error occurs, SQLITE_OK is returned and *apIndex set to point to
	113231	+** the allocated array. *pnIndex is set to the number of elements in the
	113232	+** array. If an error does occur, an SQLite error code is returned.
	113233	+**
	113234	+** Regardless of whether or not an error is returned, it is the responsibility
	113235	+** of the caller to call sqlite3_free() on the output array to free it.
	113236	+*/
112817	113237	static int fts3PrefixParameter(
112818	113238	const char zParam, / ABC in prefix=ABC parameter to parse */
112819	113239	int pnIndex, / OUT: size of apIndex[] array /
112820		- struct Fts3Index *apIndex, / OUT: Array of indexes for this table */
112821		- struct Fts3Index *apFree / OUT: Free this with sqlite3_free() */
	113240	+ struct Fts3Index *apIndex / OUT: Array of indexes for this table */
112822	113241	){
112823		- struct Fts3Index *aIndex;
112824		- int nIndex = 1;
	113242	+ struct Fts3Index aIndex; / Allocated array */
	113243	+ int nIndex = 1; /* Number of entries in array */
112825	113244
112826	113245	if( zParam && zParam[0] ){
112827	113246	const char *p;
112828	113247	nIndex++;
112829	113248	for(p=zParam; *p; p++){
		@@ -112830,11 +113249,11 @@
112830	113249	if( *p==',' ) nIndex++;
112831	113250	}
112832	113251	}
112833	113252
112834	113253	aIndex = sqlite3_malloc(sizeof(struct Fts3Index) * nIndex);
112835		- apIndex = apFree = aIndex;
	113254	+ *apIndex = aIndex;
112836	113255	*pnIndex = nIndex;
112837	113256	if( !aIndex ){
112838	113257	return SQLITE_NOMEM;
112839	113258	}
112840	113259
		@@ -112887,12 +113306,11 @@
112887	113306	int isFts4 = (argv[0][3]=='4'); /* True for FTS4, false for FTS3 */
112888	113307	const char *aCol; / Array of column names */
112889	113308	sqlite3_tokenizer pTokenizer = 0; / Tokenizer for this table */
112890	113309
112891	113310	int nIndex; /* Size of aIndex[] array */
112892		- struct Fts3Index aIndex; / Array of indexes for this table */
112893		- struct Fts3Index aFree = 0; / Free this before returning */
	113311	+ struct Fts3Index aIndex = 0; / Array of indexes for this table */
112894	113312
112895	113313	/* The results of parsing supported FTS4 key=value options: */
112896	113314	int bNoDocsize = 0; /* True to omit %_docsize table */
112897	113315	int bDescIdx = 0; /* True to store descending indexes */
112898	113316	char zPrefix = 0; / Prefix parameter value (or NULL) */
		@@ -113025,11 +113443,11 @@
113025	113443	rc = sqlite3Fts3InitTokenizer(pHash, "simple", &pTokenizer, pzErr);
113026	113444	if( rc!=SQLITE_OK ) goto fts3_init_out;
113027	113445	}
113028	113446	assert( pTokenizer );
113029	113447
113030		- rc = fts3PrefixParameter(zPrefix, &nIndex, &aIndex, &aFree);
	113448	+ rc = fts3PrefixParameter(zPrefix, &nIndex, &aIndex);
113031	113449	if( rc==SQLITE_ERROR ){
113032	113450	assert( zPrefix );
113033	113451	*pzErr = sqlite3_mprintf("error parsing prefix parameter: %s", zPrefix);
113034	113452	}
113035	113453	if( rc!=SQLITE_OK ) goto fts3_init_out;
		@@ -113112,11 +113530,11 @@
113112	113530	/* Declare the table schema to SQLite. */
113113	113531	fts3DeclareVtab(&rc, p);
113114	113532
113115	113533	fts3_init_out:
113116	113534	sqlite3_free(zPrefix);
113117		- sqlite3_free(aFree);
	113535	+ sqlite3_free(aIndex);
113118	113536	sqlite3_free(zCompress);
113119	113537	sqlite3_free(zUncompress);
113120	113538	sqlite3_free((void *)aCol);
113121	113539	if( rc!=SQLITE_OK ){
113122	113540	if( p ){
		@@ -113703,12 +114121,10 @@
113703	114121	*pp1 = p1 + 1;
113704	114122	*pp2 = p2 + 1;
113705	114123	}
113706	114124
113707	114125	/*
113708		-** nToken==1 searches for adjacent positions.
113709		-**
113710	114126	** This function is used to merge two position lists into one. When it is
113711	114127	** called, pp1 and pp2 must both point to position lists. A position-list is
113712	114128	** the part of a doclist that follows each document id. For example, if a row
113713	114129	** contains:
113714	114130	**
		@@ -113724,10 +114140,12 @@
113724	114140	** If isSaveLeft is 0, an entry is added to the output position list for
113725	114141	** each position in *pp2 for which there exists one or more positions in
113726	114142	** pp1 so that (pos(pp2)>pos(pp1) && pos(pp2)-pos(*pp1)<=nToken). i.e.
113727	114143	** when the pp1 token appears before the pp2 token, but not more than nToken
113728	114144	** slots before it.
	114145	+**
	114146	+** e.g. nToken==1 searches for adjacent positions.
113729	114147	*/
113730	114148	static int fts3PoslistPhraseMerge(
113731	114149	char *pp, / IN/OUT: Preallocated output buffer */
113732	114150	int nToken, /* Maximum difference in token positions */
113733	114151	int isSaveLeft, /* Save the left position */
		@@ -113890,26 +114308,38 @@
113890	114308
113891	114309	return res;
113892	114310	}
113893	114311
113894	114312	/*
113895		-** A pointer to an instance of this structure is used as the context
113896		-** argument to sqlite3Fts3SegReaderIterate()
	114313	+** An instance of this function is used to merge together the (potentially
	114314	+** large number of) doclists for each term that matches a prefix query.
	114315	+** See function fts3TermSelectMerge() for details.
113897	114316	*/
113898	114317	typedef struct TermSelect TermSelect;
113899	114318	struct TermSelect {
113900		- int isReqPos;
113901		- char aaOutput[16]; / Malloc'd output buffer */
113902		- int anOutput[16]; /* Size of output in bytes */
	114319	+ char aaOutput[16]; / Malloc'd output buffers */
	114320	+ int anOutput[16]; /* Size each output buffer in bytes */
113903	114321	};
113904	114322
113905		-
	114323	+/*
	114324	+** This function is used to read a single varint from a buffer. Parameter
	114325	+** pEnd points 1 byte past the end of the buffer. When this function is
	114326	+** called, if *pp points to pEnd or greater, then the end of the buffer
	114327	+** has been reached. In this case *pp is set to 0 and the function returns.
	114328	+**
	114329	+** If *pp does not point to or past pEnd, then a single varint is read
	114330	+** from pp. pp is then set to point 1 byte past the end of the read varint.
	114331	+**
	114332	+** If bDescIdx is false, the value read is added to *pVal before returning.
	114333	+** If it is true, the value read is subtracted from *pVal before this
	114334	+** function returns.
	114335	+*/
113906	114336	static void fts3GetDeltaVarint3(
113907		- char **pp,
113908		- char *pEnd,
113909		- int bDescIdx,
113910		- sqlite3_int64 *pVal
	114337	+ char *pp, / IN/OUT: Point to read varint from */
	114338	+ char pEnd, / End of buffer */
	114339	+ int bDescIdx, /* True if docids are descending */
	114340	+ sqlite3_int64 pVal / IN/OUT: Integer value */
113911	114341	){
113912	114342	if( *pp>=pEnd ){
113913	114343	*pp = 0;
113914	114344	}else{
113915	114345	sqlite3_int64 iVal;
		@@ -113920,10 +114350,25 @@
113920	114350	*pVal += iVal;
113921	114351	}
113922	114352	}
113923	114353	}
113924	114354
	114355	+/*
	114356	+** This function is used to write a single varint to a buffer. The varint
	114357	+** is written to pp. Before returning, pp is set to point 1 byte past the
	114358	+** end of the value written.
	114359	+**
	114360	+** If *pbFirst is zero when this function is called, the value written to
	114361	+** the buffer is that of parameter iVal.
	114362	+**
	114363	+** If *pbFirst is non-zero when this function is called, then the value
	114364	+** written is either (iVal-piPrev) (if bDescIdx is zero) or (piPrev-iVal)
	114365	+** (if bDescIdx is non-zero).
	114366	+**
	114367	+** Before returning, this function always sets pbFirst to 1 and piPrev
	114368	+** to the value of parameter iVal.
	114369	+*/
113925	114370	static void fts3PutDeltaVarint3(
113926	114371	char *pp, / IN/OUT: Output pointer */
113927	114372	int bDescIdx, /* True for descending docids */
113928	114373	sqlite3_int64 piPrev, / IN/OUT: Previous value written to list */
113929	114374	int pbFirst, / IN/OUT: True after first int written */
		@@ -113940,14 +114385,38 @@
113940	114385	pp += sqlite3Fts3PutVarint(pp, iWrite);
113941	114386	*piPrev = iVal;
113942	114387	*pbFirst = 1;
113943	114388	}
113944	114389
113945		-#define COMPARE_DOCID(i1, i2) ((bDescIdx?-1:1) * (i1-i2))
113946	114390
	114391	+/*
	114392	+** This macro is used by various functions that merge doclists. The two
	114393	+** arguments are 64-bit docid values. If the value of the stack variable
	114394	+** bDescDoclist is 0 when this macro is invoked, then it returns (i1-i2).
	114395	+** Otherwise, (i2-i1).
	114396	+**
	114397	+** Using this makes it easier to write code that can merge doclists that are
	114398	+** sorted in either ascending or descending order.
	114399	+*/
	114400	+#define DOCID_CMP(i1, i2) ((bDescDoclist?-1:1) * (i1-i2))
	114401	+
	114402	+/*
	114403	+** This function does an "OR" merge of two doclists (output contains all
	114404	+** positions contained in either argument doclist). If the docids in the
	114405	+** input doclists are sorted in ascending order, parameter bDescDoclist
	114406	+** should be false. If they are sorted in ascending order, it should be
	114407	+** passed a non-zero value.
	114408	+**
	114409	+** If no error occurs, *paOut is set to point at an sqlite3_malloc'd buffer
	114410	+** containing the output doclist and SQLITE_OK is returned. In this case
	114411	+** *pnOut is set to the number of bytes in the output doclist.
	114412	+**
	114413	+** If an error occurs, an SQLite error code is returned. The output values
	114414	+** are undefined in this case.
	114415	+*/
113947	114416	static int fts3DoclistOrMerge(
113948		- int bDescIdx, /* True if arguments are desc */
	114417	+ int bDescDoclist, /* True if arguments are desc */
113949	114418	char a1, int n1, / First doclist */
113950	114419	char a2, int n2, / Second doclist */
113951	114420	char *paOut, int pnOut /* OUT: Malloc'd doclist */
113952	114421	){
113953	114422	sqlite3_int64 i1 = 0;
		@@ -113968,35 +114437,47 @@
113968	114437
113969	114438	p = aOut;
113970	114439	fts3GetDeltaVarint3(&p1, pEnd1, 0, &i1);
113971	114440	fts3GetDeltaVarint3(&p2, pEnd2, 0, &i2);
113972	114441	while( p1 \|\| p2 ){
113973		- sqlite3_int64 iDiff = COMPARE_DOCID(i1, i2);
	114442	+ sqlite3_int64 iDiff = DOCID_CMP(i1, i2);
113974	114443
113975	114444	if( p2 && p1 && iDiff==0 ){
113976		- fts3PutDeltaVarint3(&p, bDescIdx, &iPrev, &bFirstOut, i1);
	114445	+ fts3PutDeltaVarint3(&p, bDescDoclist, &iPrev, &bFirstOut, i1);
113977	114446	fts3PoslistMerge(&p, &p1, &p2);
113978		- fts3GetDeltaVarint3(&p1, pEnd1, bDescIdx, &i1);
113979		- fts3GetDeltaVarint3(&p2, pEnd2, bDescIdx, &i2);
	114447	+ fts3GetDeltaVarint3(&p1, pEnd1, bDescDoclist, &i1);
	114448	+ fts3GetDeltaVarint3(&p2, pEnd2, bDescDoclist, &i2);
113980	114449	}else if( !p2 \|\| (p1 && iDiff<0) ){
113981		- fts3PutDeltaVarint3(&p, bDescIdx, &iPrev, &bFirstOut, i1);
	114450	+ fts3PutDeltaVarint3(&p, bDescDoclist, &iPrev, &bFirstOut, i1);
113982	114451	fts3PoslistCopy(&p, &p1);
113983		- fts3GetDeltaVarint3(&p1, pEnd1, bDescIdx, &i1);
	114452	+ fts3GetDeltaVarint3(&p1, pEnd1, bDescDoclist, &i1);
113984	114453	}else{
113985		- fts3PutDeltaVarint3(&p, bDescIdx, &iPrev, &bFirstOut, i2);
	114454	+ fts3PutDeltaVarint3(&p, bDescDoclist, &iPrev, &bFirstOut, i2);
113986	114455	fts3PoslistCopy(&p, &p2);
113987		- fts3GetDeltaVarint3(&p2, pEnd2, bDescIdx, &i2);
	114456	+ fts3GetDeltaVarint3(&p2, pEnd2, bDescDoclist, &i2);
113988	114457	}
113989	114458	}
113990	114459
113991	114460	*paOut = aOut;
113992	114461	*pnOut = (p-aOut);
113993	114462	return SQLITE_OK;
113994	114463	}
113995	114464
	114465	+/*
	114466	+** This function does a "phrase" merge of two doclists. In a phrase merge,
	114467	+** the output contains a copy of each position from the right-hand input
	114468	+** doclist for which there is a position in the left-hand input doclist
	114469	+** exactly nDist tokens before it.
	114470	+**
	114471	+** If the docids in the input doclists are sorted in ascending order,
	114472	+** parameter bDescDoclist should be false. If they are sorted in ascending
	114473	+** order, it should be passed a non-zero value.
	114474	+**
	114475	+** The right-hand input doclist is overwritten by this function.
	114476	+*/
113996	114477	static void fts3DoclistPhraseMerge(
113997		- int bDescIdx, /* True if arguments are desc */
	114478	+ int bDescDoclist, /* True if arguments are desc */
113998	114479	int nDist, /* Distance from left to right (1=adjacent) */
113999	114480	char aLeft, int nLeft, / Left doclist */
114000	114481	char aRight, int pnRight /* IN/OUT: Right/output doclist */
114001	114482	){
114002	114483	sqlite3_int64 i1 = 0;
		@@ -114015,30 +114496,30 @@
114015	114496	p = aOut;
114016	114497	fts3GetDeltaVarint3(&p1, pEnd1, 0, &i1);
114017	114498	fts3GetDeltaVarint3(&p2, pEnd2, 0, &i2);
114018	114499
114019	114500	while( p1 && p2 ){
114020		- sqlite3_int64 iDiff = COMPARE_DOCID(i1, i2);
	114501	+ sqlite3_int64 iDiff = DOCID_CMP(i1, i2);
114021	114502	if( iDiff==0 ){
114022	114503	char *pSave = p;
114023	114504	sqlite3_int64 iPrevSave = iPrev;
114024	114505	int bFirstOutSave = bFirstOut;
114025	114506
114026		- fts3PutDeltaVarint3(&p, bDescIdx, &iPrev, &bFirstOut, i1);
	114507	+ fts3PutDeltaVarint3(&p, bDescDoclist, &iPrev, &bFirstOut, i1);
114027	114508	if( 0==fts3PoslistPhraseMerge(&p, nDist, 0, 1, &p1, &p2) ){
114028	114509	p = pSave;
114029	114510	iPrev = iPrevSave;
114030	114511	bFirstOut = bFirstOutSave;
114031	114512	}
114032		- fts3GetDeltaVarint3(&p1, pEnd1, bDescIdx, &i1);
114033		- fts3GetDeltaVarint3(&p2, pEnd2, bDescIdx, &i2);
	114513	+ fts3GetDeltaVarint3(&p1, pEnd1, bDescDoclist, &i1);
	114514	+ fts3GetDeltaVarint3(&p2, pEnd2, bDescDoclist, &i2);
114034	114515	}else if( iDiff<0 ){
114035	114516	fts3PoslistCopy(0, &p1);
114036		- fts3GetDeltaVarint3(&p1, pEnd1, bDescIdx, &i1);
	114517	+ fts3GetDeltaVarint3(&p1, pEnd1, bDescDoclist, &i1);
114037	114518	}else{
114038	114519	fts3PoslistCopy(0, &p2);
114039		- fts3GetDeltaVarint3(&p2, pEnd2, bDescIdx, &i2);
	114520	+ fts3GetDeltaVarint3(&p2, pEnd2, bDescDoclist, &i2);
114040	114521	}
114041	114522	}
114042	114523
114043	114524	*pnRight = p - aOut;
114044	114525	}
		@@ -114051,11 +114532,11 @@
114051	114532	**
114052	114533	** If an OOM error occurs, return SQLITE_NOMEM. In this case it is
114053	114534	** the responsibility of the caller to free any doclists left in the
114054	114535	** TermSelect.aaOutput[] array.
114055	114536	*/
114056		-static int fts3TermSelectMerge(Fts3Table p, TermSelect pTS){
	114537	+static int fts3TermSelectFinishMerge(Fts3Table p, TermSelect pTS){
114057	114538	char *aOut = 0;
114058	114539	int nOut = 0;
114059	114540	int i;
114060	114541
114061	114542	/* Loop through the doclists in the aaOutput[] array. Merge them all
		@@ -114092,28 +114573,29 @@
114092	114573	pTS->anOutput[0] = nOut;
114093	114574	return SQLITE_OK;
114094	114575	}
114095	114576
114096	114577	/*
114097		-** This function is used as the sqlite3Fts3SegReaderIterate() callback when
114098		-** querying the full-text index for a doclist associated with a term or
114099		-** term-prefix.
	114578	+** Merge the doclist aDoclist/nDoclist into the TermSelect object passed
	114579	+** as the first argument. The merge is an "OR" merge (see function
	114580	+** fts3DoclistOrMerge() for details).
	114581	+**
	114582	+** This function is called with the doclist for each term that matches
	114583	+** a queried prefix. It merges all these doclists into one, the doclist
	114584	+** for the specified prefix. Since there can be a very large number of
	114585	+** doclists to merge, the merging is done pair-wise using the TermSelect
	114586	+** object.
	114587	+**
	114588	+** This function returns SQLITE_OK if the merge is successful, or an
	114589	+** SQLite error code (SQLITE_NOMEM) if an error occurs.
114100	114590	*/
114101		-static int fts3TermSelectCb(
114102		- Fts3Table p, / Virtual table object */
114103		- void pContext, / Pointer to TermSelect structure */
114104		- char *zTerm,
114105		- int nTerm,
114106		- char *aDoclist,
114107		- int nDoclist
	114591	+static int fts3TermSelectMerge(
	114592	+ Fts3Table p, / FTS table handle */
	114593	+ TermSelect pTS, / TermSelect object to merge into */
	114594	+ char aDoclist, / Pointer to doclist */
	114595	+ int nDoclist /* Size of aDoclist in bytes */
114108	114596	){
114109		- TermSelect pTS = (TermSelect )pContext;
114110		-
114111		- UNUSED_PARAMETER(p);
114112		- UNUSED_PARAMETER(zTerm);
114113		- UNUSED_PARAMETER(nTerm);
114114		-
114115	114597	if( pTS->aaOutput[0]==0 ){
114116	114598	/* If this is the first term selected, copy the doclist to the output
114117	114599	** buffer using memcpy(). */
114118	114600	pTS->aaOutput[0] = sqlite3_malloc(nDoclist);
114119	114601	pTS->anOutput[0] = nDoclist;
		@@ -114180,23 +114662,30 @@
114180	114662	}
114181	114663	pCsr->apSegment[pCsr->nSegment++] = pNew;
114182	114664	return SQLITE_OK;
114183	114665	}
114184	114666
	114667	+/*
	114668	+** Add seg-reader objects to the Fts3MultiSegReader object passed as the
	114669	+** 8th argument.
	114670	+**
	114671	+** This function returns SQLITE_OK if successful, or an SQLite error code
	114672	+** otherwise.
	114673	+*/
114185	114674	static int fts3SegReaderCursor(
114186	114675	Fts3Table p, / FTS3 table handle */
114187	114676	int iIndex, /* Index to search (from 0 to p->nIndex-1) */
114188	114677	int iLevel, /* Level of segments to scan */
114189	114678	const char zTerm, / Term to query for */
114190	114679	int nTerm, /* Size of zTerm in bytes */
114191	114680	int isPrefix, /* True for a prefix search */
114192	114681	int isScan, /* True to scan from zTerm to EOF */
114193		- Fts3MultiSegReader pCsr / Cursor object to populate */
	114682	+ Fts3MultiSegReader pCsr / Cursor object to populate */
114194	114683	){
114195		- int rc = SQLITE_OK;
114196		- int rc2;
114197		- sqlite3_stmt *pStmt = 0;
	114684	+ int rc = SQLITE_OK; /* Error code */
	114685	+ sqlite3_stmt pStmt = 0; / Statement to iterate through segments */
	114686	+ int rc2; /* Result of sqlite3_reset() */
114198	114687
114199	114688	/* If iLevel is less than 0 and this is not a scan, include a seg-reader
114200	114689	** for the pending-terms. If this is a scan, then this call must be being
114201	114690	** made by an fts4aux module, not an FTS table. In this case calling
114202	114691	** Fts3SegReaderPending might segfault, as the data structures used by
		@@ -114281,28 +114770,46 @@
114281	114770	return fts3SegReaderCursor(
114282	114771	p, iIndex, iLevel, zTerm, nTerm, isPrefix, isScan, pCsr
114283	114772	);
114284	114773	}
114285	114774
	114775	+/*
	114776	+** In addition to its current configuration, have the Fts3MultiSegReader
	114777	+** passed as the 4th argument also scan the doclist for term zTerm/nTerm.
	114778	+**
	114779	+** SQLITE_OK is returned if no error occurs, otherwise an SQLite error code.
	114780	+*/
114286	114781	static int fts3SegReaderCursorAddZero(
114287		- Fts3Table *p,
114288		- const char *zTerm,
114289		- int nTerm,
114290		- Fts3MultiSegReader *pCsr
	114782	+ Fts3Table p, / FTS virtual table handle */
	114783	+ const char zTerm, / Term to scan doclist of */
	114784	+ int nTerm, /* Number of bytes in zTerm */
	114785	+ Fts3MultiSegReader pCsr / Fts3MultiSegReader to modify */
114291	114786	){
114292	114787	return fts3SegReaderCursor(p, 0, FTS3_SEGCURSOR_ALL, zTerm, nTerm, 0, 0,pCsr);
114293	114788	}
114294	114789
114295		-
114296		-SQLITE_PRIVATE int sqlite3Fts3TermSegReaderCursor(
	114790	+/*
	114791	+** Open an Fts3MultiSegReader to scan the doclist for term zTerm/nTerm. Or,
	114792	+** if isPrefix is true, to scan the doclist for all terms for which
	114793	+** zTerm/nTerm is a prefix. If successful, return SQLITE_OK and write
	114794	+** a pointer to the new Fts3MultiSegReader to *ppSegcsr. Otherwise, return
	114795	+** an SQLite error code.
	114796	+**
	114797	+** It is the responsibility of the caller to free this object by eventually
	114798	+** passing it to fts3SegReaderCursorFree()
	114799	+**
	114800	+** SQLITE_OK is returned if no error occurs, otherwise an SQLite error code.
	114801	+** Output parameter *ppSegcsr is set to 0 if an error occurs.
	114802	+*/
	114803	+static int fts3TermSegReaderCursor(
114297	114804	Fts3Cursor pCsr, / Virtual table cursor handle */
114298	114805	const char zTerm, / Term to query for */
114299	114806	int nTerm, /* Size of zTerm in bytes */
114300	114807	int isPrefix, /* True for a prefix search */
114301	114808	Fts3MultiSegReader *ppSegcsr / OUT: Allocated seg-reader cursor */
114302	114809	){
114303		- Fts3MultiSegReader pSegcsr; / Object to allocate and return */
	114810	+ Fts3MultiSegReader pSegcsr; / Object to allocate and return */
114304	114811	int rc = SQLITE_NOMEM; /* Return code */
114305	114812
114306	114813	pSegcsr = sqlite3_malloc(sizeof(Fts3MultiSegReader));
114307	114814	if( pSegcsr ){
114308	114815	int i;
		@@ -114342,62 +114849,53 @@
114342	114849
114343	114850	*ppSegcsr = pSegcsr;
114344	114851	return rc;
114345	114852	}
114346	114853
	114854	+/*
	114855	+** Free an Fts3MultiSegReader allocated by fts3TermSegReaderCursor().
	114856	+*/
114347	114857	static void fts3SegReaderCursorFree(Fts3MultiSegReader *pSegcsr){
114348	114858	sqlite3Fts3SegReaderFinish(pSegcsr);
114349	114859	sqlite3_free(pSegcsr);
114350	114860	}
114351	114861
114352	114862	/*
114353	114863	** This function retreives the doclist for the specified term (or term
114354		-** prefix) from the database.
114355		-**
114356		-** The returned doclist may be in one of two formats, depending on the
114357		-** value of parameter isReqPos. If isReqPos is zero, then the doclist is
114358		-** a sorted list of delta-compressed docids (a bare doclist). If isReqPos
114359		-** is non-zero, then the returned list is in the same format as is stored
114360		-** in the database without the found length specifier at the start of on-disk
114361		-** doclists.
	114864	+** prefix) from the database.
114362	114865	*/
114363	114866	static int fts3TermSelect(
114364	114867	Fts3Table p, / Virtual table handle */
114365	114868	Fts3PhraseToken pTok, / Token to query for */
114366	114869	int iColumn, /* Column to query (or -ve for all columns) */
114367		- int isReqPos, /* True to include position lists in output */
114368	114870	int pnOut, / OUT: Size of buffer at ppOut /
114369	114871	char *ppOut / OUT: Malloced result buffer */
114370	114872	){
114371	114873	int rc; /* Return code */
114372		- Fts3MultiSegReader pSegcsr; / Seg-reader cursor for this term */
114373		- TermSelect tsc; /* Context object for fts3TermSelectCb() */
	114874	+ Fts3MultiSegReader pSegcsr; / Seg-reader cursor for this term */
	114875	+ TermSelect tsc; /* Object for pair-wise doclist merging */
114374	114876	Fts3SegFilter filter; /* Segment term filter configuration */
114375	114877
114376	114878	pSegcsr = pTok->pSegcsr;
114377	114879	memset(&tsc, 0, sizeof(TermSelect));
114378		- tsc.isReqPos = isReqPos;
114379	114880
114380		- filter.flags = FTS3_SEGMENT_IGNORE_EMPTY
	114881	+ filter.flags = FTS3_SEGMENT_IGNORE_EMPTY \| FTS3_SEGMENT_REQUIRE_POS
114381	114882	\| (pTok->isPrefix ? FTS3_SEGMENT_PREFIX : 0)
114382		- \| (isReqPos ? FTS3_SEGMENT_REQUIRE_POS : 0)
114383	114883	\| (iColumn<p->nColumn ? FTS3_SEGMENT_COLUMN_FILTER : 0);
114384	114884	filter.iCol = iColumn;
114385	114885	filter.zTerm = pTok->z;
114386	114886	filter.nTerm = pTok->n;
114387	114887
114388	114888	rc = sqlite3Fts3SegReaderStart(p, pSegcsr, &filter);
114389	114889	while( SQLITE_OK==rc
114390	114890	&& SQLITE_ROW==(rc = sqlite3Fts3SegReaderStep(p, pSegcsr))
114391	114891	){
114392		- rc = fts3TermSelectCb(p, (void *)&tsc,
114393		- pSegcsr->zTerm, pSegcsr->nTerm, pSegcsr->aDoclist, pSegcsr->nDoclist
114394		- );
	114892	+ rc = fts3TermSelectMerge(p, &tsc, pSegcsr->aDoclist, pSegcsr->nDoclist);
114395	114893	}
114396	114894
114397	114895	if( rc==SQLITE_OK ){
114398		- rc = fts3TermSelectMerge(p, &tsc);
	114896	+ rc = fts3TermSelectFinishMerge(p, &tsc);
114399	114897	}
114400	114898	if( rc==SQLITE_OK ){
114401	114899	*ppOut = tsc.aaOutput[0];
114402	114900	*pnOut = tsc.anOutput[0];
114403	114901	}else{
		@@ -114419,28 +114917,19 @@
114419	114917	** If the isPoslist argument is true, then it is assumed that the doclist
114420	114918	** contains a position-list following each docid. Otherwise, it is assumed
114421	114919	** that the doclist is simply a list of docids stored as delta encoded
114422	114920	** varints.
114423	114921	*/
114424		-static int fts3DoclistCountDocids(int isPoslist, char *aList, int nList){
	114922	+static int fts3DoclistCountDocids(char *aList, int nList){
114425	114923	int nDoc = 0; /* Return value */
114426	114924	if( aList ){
114427	114925	char aEnd = &aList[nList]; / Pointer to one byte after EOF */
114428	114926	char p = aList; / Cursor */
114429		- if( !isPoslist ){
114430		- /* The number of docids in the list is the same as the number of
114431		- ** varints. In FTS3 a varint consists of a single byte with the 0x80
114432		- ** bit cleared and zero or more bytes with the 0x80 bit set. So to
114433		- ** count the varints in the buffer, just count the number of bytes
114434		- ** with the 0x80 bit clear. */
114435		- while( p<aEnd ) nDoc += (((*p++)&0x80)==0);
114436		- }else{
114437		- while( p<aEnd ){
114438		- nDoc++;
114439		- while( (p++)&0x80 ); / Skip docid varint */
114440		- fts3PoslistCopy(0, &p); /* Skip over position list */
114441		- }
	114927	+ while( p<aEnd ){
	114928	+ nDoc++;
	114929	+ while( (p++)&0x80 ); / Skip docid varint */
	114930	+ fts3PoslistCopy(0, &p); /* Skip over position list */
114442	114931	}
114443	114932	}
114444	114933
114445	114934	return nDoc;
114446	114935	}
		@@ -114466,11 +114955,11 @@
114466	114955	}else{
114467	114956	pCsr->iPrevId = sqlite3_column_int64(pCsr->pStmt, 0);
114468	114957	rc = SQLITE_OK;
114469	114958	}
114470	114959	}else{
114471		- rc = sqlite3Fts3EvalNext((Fts3Cursor *)pCursor);
	114960	+ rc = fts3EvalNext((Fts3Cursor *)pCursor);
114472	114961	}
114473	114962	assert( ((Fts3Table *)pCsr->base.pVtab)->pSegments==0 );
114474	114963	return rc;
114475	114964	}
114476	114965
		@@ -114543,11 +115032,11 @@
114543	115032	}
114544	115033
114545	115034	rc = sqlite3Fts3ReadLock(p);
114546	115035	if( rc!=SQLITE_OK ) return rc;
114547	115036
114548		- rc = sqlite3Fts3EvalStart(pCsr, pCsr->pExpr, 1);
	115037	+ rc = fts3EvalStart(pCsr);
114549	115038
114550	115039	sqlite3Fts3SegmentsClose(p);
114551	115040	if( rc!=SQLITE_OK ) return rc;
114552	115041	pCsr->pNextId = pCsr->aDoclist;
114553	115042	pCsr->iPrevId = 0;
		@@ -114950,26 +115439,43 @@
114950	115439	p->zDb, p->zName, zName
114951	115440	);
114952	115441	return rc;
114953	115442	}
114954	115443
	115444	+/*
	115445	+** The xSavepoint() method.
	115446	+**
	115447	+** Flush the contents of the pending-terms table to disk.
	115448	+*/
114955	115449	static int fts3SavepointMethod(sqlite3_vtab *pVtab, int iSavepoint){
114956	115450	UNUSED_PARAMETER(iSavepoint);
114957	115451	assert( ((Fts3Table *)pVtab)->inTransaction );
114958	115452	assert( ((Fts3Table *)pVtab)->mxSavepoint < iSavepoint );
114959	115453	TESTONLY( ((Fts3Table *)pVtab)->mxSavepoint = iSavepoint );
114960	115454	return fts3SyncMethod(pVtab);
114961	115455	}
	115456	+
	115457	+/*
	115458	+** The xRelease() method.
	115459	+**
	115460	+** This is a no-op.
	115461	+*/
114962	115462	static int fts3ReleaseMethod(sqlite3_vtab *pVtab, int iSavepoint){
114963	115463	TESTONLY( Fts3Table p = (Fts3Table)pVtab );
114964	115464	UNUSED_PARAMETER(iSavepoint);
114965	115465	UNUSED_PARAMETER(pVtab);
114966	115466	assert( p->inTransaction );
114967	115467	assert( p->mxSavepoint >= iSavepoint );
114968	115468	TESTONLY( p->mxSavepoint = iSavepoint-1 );
114969	115469	return SQLITE_OK;
114970	115470	}
	115471	+
	115472	+/*
	115473	+** The xRollbackTo() method.
	115474	+**
	115475	+** Discard the contents of the pending terms table.
	115476	+*/
114971	115477	static int fts3RollbackToMethod(sqlite3_vtab *pVtab, int iSavepoint){
114972	115478	Fts3Table p = (Fts3Table)pVtab;
114973	115479	UNUSED_PARAMETER(iSavepoint);
114974	115480	assert( p->inTransaction );
114975	115481	assert( p->mxSavepoint >= iSavepoint );
		@@ -115114,22 +115620,10 @@
115114	115620	sqlite3Fts3HashClear(pHash);
115115	115621	sqlite3_free(pHash);
115116	115622	}
115117	115623	return rc;
115118	115624	}
115119		-
115120		-#if !SQLITE_CORE
115121		-SQLITE_API int sqlite3_extension_init(
115122		- sqlite3 *db,
115123		- char **pzErrMsg,
115124		- const sqlite3_api_routines *pApi
115125		-){
115126		- SQLITE_EXTENSION_INIT2(pApi)
115127		- return sqlite3Fts3Init(db);
115128		-}
115129		-#endif
115130		-
115131	115625
115132	115626	/*
115133	115627	** Allocate an Fts3MultiSegReader for each token in the expression headed
115134	115628	** by pExpr.
115135	115629	**
		@@ -115143,24 +115637,24 @@
115143	115637	** there exists prefix b-tree of the right length) then it may be traversed
115144	115638	** and merged incrementally. Otherwise, it has to be merged into an in-memory
115145	115639	** doclist and then traversed.
115146	115640	*/
115147	115641	static void fts3EvalAllocateReaders(
115148		- Fts3Cursor *pCsr,
115149		- Fts3Expr *pExpr,
	115642	+ Fts3Cursor pCsr, / FTS cursor handle */
	115643	+ Fts3Expr pExpr, / Allocate readers for this expression */
115150	115644	int pnToken, / OUT: Total number of tokens in phrase. */
115151	115645	int pnOr, / OUT: Total number of OR nodes in expr. */
115152		- int *pRc
	115646	+ int pRc / IN/OUT: Error code */
115153	115647	){
115154	115648	if( pExpr && SQLITE_OK==*pRc ){
115155	115649	if( pExpr->eType==FTSQUERY_PHRASE ){
115156	115650	int i;
115157	115651	int nToken = pExpr->pPhrase->nToken;
115158	115652	*pnToken += nToken;
115159	115653	for(i=0; i<nToken; i++){
115160	115654	Fts3PhraseToken *pToken = &pExpr->pPhrase->aToken[i];
115161		- int rc = sqlite3Fts3TermSegReaderCursor(pCsr,
	115655	+ int rc = fts3TermSegReaderCursor(pCsr,
115162	115656	pToken->z, pToken->n, pToken->isPrefix, &pToken->pSegcsr
115163	115657	);
115164	115658	if( rc!=SQLITE_OK ){
115165	115659	*pRc = rc;
115166	115660	return;
		@@ -115174,16 +115668,24 @@
115174	115668	fts3EvalAllocateReaders(pCsr, pExpr->pRight, pnToken, pnOr, pRc);
115175	115669	}
115176	115670	}
115177	115671	}
115178	115672
	115673	+/*
	115674	+** Arguments pList/nList contain the doclist for token iToken of phrase p.
	115675	+** It is merged into the main doclist stored in p->doclist.aAll/nAll.
	115676	+**
	115677	+** This function assumes that pList points to a buffer allocated using
	115678	+** sqlite3_malloc(). This function takes responsibility for eventually
	115679	+** freeing the buffer.
	115680	+*/
115179	115681	static void fts3EvalPhraseMergeToken(
115180		- Fts3Table *pTab,
115181		- Fts3Phrase *p,
115182		- int iToken,
115183		- char *pList,
115184		- int nList
	115682	+ Fts3Table pTab, / FTS Table pointer */
	115683	+ Fts3Phrase p, / Phrase to merge pList/nList into */
	115684	+ int iToken, /* Token pList/nList corresponds to */
	115685	+ char pList, / Pointer to doclist */
	115686	+ int nList /* Number of bytes in pList */
115185	115687	){
115186	115688	assert( iToken!=p->iDoclistToken );
115187	115689
115188	115690	if( pList==0 ){
115189	115691	sqlite3_free(p->doclist.aAll);
		@@ -115228,13 +115730,19 @@
115228	115730	}
115229	115731
115230	115732	if( iToken>p->iDoclistToken ) p->iDoclistToken = iToken;
115231	115733	}
115232	115734
	115735	+/*
	115736	+** Load the doclist for phrase p into p->doclist.aAll/nAll. The loaded doclist
	115737	+** does not take deferred tokens into account.
	115738	+**
	115739	+** SQLITE_OK is returned if no error occurs, otherwise an SQLite error code.
	115740	+*/
115233	115741	static int fts3EvalPhraseLoad(
115234		- Fts3Cursor *pCsr,
115235		- Fts3Phrase *p
	115742	+ Fts3Cursor pCsr, / FTS Cursor handle */
	115743	+ Fts3Phrase p / Phrase object */
115236	115744	){
115237	115745	Fts3Table pTab = (Fts3Table )pCsr->base.pVtab;
115238	115746	int iToken;
115239	115747	int rc = SQLITE_OK;
115240	115748
		@@ -115243,11 +115751,11 @@
115243	115751	assert( pToken->pDeferred==0 \|\| pToken->pSegcsr==0 );
115244	115752
115245	115753	if( pToken->pSegcsr ){
115246	115754	int nThis = 0;
115247	115755	char *pThis = 0;
115248		- rc = fts3TermSelect(pTab, pToken, p->iColumn, 1, &nThis, &pThis);
	115756	+ rc = fts3TermSelect(pTab, pToken, p->iColumn, &nThis, &pThis);
115249	115757	if( rc==SQLITE_OK ){
115250	115758	fts3EvalPhraseMergeToken(pTab, p, iToken, pThis, nThis);
115251	115759	}
115252	115760	}
115253	115761	assert( pToken->pSegcsr==0 );
		@@ -115254,18 +115762,26 @@
115254	115762	}
115255	115763
115256	115764	return rc;
115257	115765	}
115258	115766
	115767	+/*
	115768	+** This function is called on each phrase after the position lists for
	115769	+** any deferred tokens have been loaded into memory. It updates the phrases
	115770	+** current position list to include only those positions that are really
	115771	+** instances of the phrase (after considering deferred tokens). If this
	115772	+** means that the phrase does not appear in the current row, doclist.pList
	115773	+** and doclist.nList are both zeroed.
	115774	+**
	115775	+** SQLITE_OK is returned if no error occurs, otherwise an SQLite error code.
	115776	+*/
115259	115777	static int fts3EvalDeferredPhrase(Fts3Cursor pCsr, Fts3Phrase pPhrase){
115260		- int iToken;
115261		- int rc = SQLITE_OK;
115262		-
115263		- int nMaxUndeferred = pPhrase->iDoclistToken;
115264		- char *aPoslist = 0;
115265		- int nPoslist = 0;
115266		- int iPrev = -1;
	115778	+ int iToken; /* Used to iterate through phrase tokens */
	115779	+ int rc = SQLITE_OK; /* Return code */
	115780	+ char aPoslist = 0; / Position list for deferred tokens */
	115781	+ int nPoslist = 0; /* Number of bytes in aPoslist */
	115782	+ int iPrev = -1; /* Token number of previous deferred token */
115267	115783
115268	115784	assert( pPhrase->doclist.bFreeList==0 );
115269	115785
115270	115786	for(iToken=0; rc==SQLITE_OK && iToken<pPhrase->nToken; iToken++){
115271	115787	Fts3PhraseToken *pToken = &pPhrase->aToken[iToken];
		@@ -115307,10 +115823,11 @@
115307	115823	iPrev = iToken;
115308	115824	}
115309	115825	}
115310	115826
115311	115827	if( iPrev>=0 ){
	115828	+ int nMaxUndeferred = pPhrase->iDoclistToken;
115312	115829	if( nMaxUndeferred<0 ){
115313	115830	pPhrase->doclist.pList = aPoslist;
115314	115831	pPhrase->doclist.nList = nPoslist;
115315	115832	pPhrase->doclist.iDocid = pCsr->iPrevId;
115316	115833	pPhrase->doclist.bFreeList = 1;
		@@ -115355,13 +115872,19 @@
115355	115872	/*
115356	115873	** This function is called for each Fts3Phrase in a full-text query
115357	115874	** expression to initialize the mechanism for returning rows. Once this
115358	115875	** function has been called successfully on an Fts3Phrase, it may be
115359	115876	** used with fts3EvalPhraseNext() to iterate through the matching docids.
	115877	+**
	115878	+** If parameter bOptOk is true, then the phrase may (or may not) use the
	115879	+** incremental loading strategy. Otherwise, the entire doclist is loaded into
	115880	+** memory within this call.
	115881	+**
	115882	+** SQLITE_OK is returned if no error occurs, otherwise an SQLite error code.
115360	115883	*/
115361	115884	static int fts3EvalPhraseStart(Fts3Cursor pCsr, int bOptOk, Fts3Phrase p){
115362		- int rc;
	115885	+ int rc; /* Error code */
115363	115886	Fts3PhraseToken *pFirst = &p->aToken[0];
115364	115887	Fts3Table pTab = (Fts3Table )pCsr->base.pVtab;
115365	115888
115366	115889	if( pCsr->bDesc==pTab->bDescIdx
115367	115890	&& bOptOk==1
		@@ -115385,11 +115908,17 @@
115385	115908	return rc;
115386	115909	}
115387	115910
115388	115911	/*
115389	115912	** This function is used to iterate backwards (from the end to start)
115390		-** through doclists.
	115913	+** through doclists. It is used by this module to iterate through phrase
	115914	+** doclists in reverse and by the fts3_write.c module to iterate through
	115915	+** pending-terms lists when writing to databases with "order=desc".
	115916	+**
	115917	+** The doclist may be sorted in ascending (parameter bDescIdx==0) or
	115918	+** descending (parameter bDescIdx==1) order of docid. Regardless, this
	115919	+** function iterates from the end of the doclist to the beginning.
115391	115920	*/
115392	115921	SQLITE_PRIVATE void sqlite3Fts3DoclistPrev(
115393	115922	int bDescIdx, /* True if the doclist is desc */
115394	115923	char aDoclist, / Pointer to entire doclist */
115395	115924	int nDoclist, /* Length of aDoclist in bytes */
		@@ -115450,13 +115979,13 @@
115450	115979	** If there is no "next" entry and no error occurs, then *pbEof is set to
115451	115980	** 1 before returning. Otherwise, if no error occurs and the iterator is
115452	115981	** successfully advanced, *pbEof is set to 0.
115453	115982	*/
115454	115983	static int fts3EvalPhraseNext(
115455		- Fts3Cursor *pCsr,
115456		- Fts3Phrase *p,
115457		- u8 *pbEof
	115984	+ Fts3Cursor pCsr, / FTS Cursor handle */
	115985	+ Fts3Phrase p, / Phrase object to advance to next docid */
	115986	+ u8 pbEof / OUT: Set to 1 if EOF */
115458	115987	){
115459	115988	int rc = SQLITE_OK;
115460	115989	Fts3Doclist *pDL = &p->doclist;
115461	115990	Fts3Table pTab = (Fts3Table )pCsr->base.pVtab;
115462	115991
		@@ -115498,14 +116027,14 @@
115498	116027	fts3PoslistCopy(0, &pIter);
115499	116028	pDL->nList = (pIter - pDL->pList);
115500	116029
115501	116030	/* pIter now points just past the 0x00 that terminates the position-
115502	116031	** list for document pDL->iDocid. However, if this position-list was
115503		- ** edited in place by fts3EvalNearTrim2(), then pIter may not actually
	116032	+ ** edited in place by fts3EvalNearTrim(), then pIter may not actually
115504	116033	** point to the start of the next docid value. The following line deals
115505	116034	** with this case by advancing pIter past the zero-padding added by
115506		- ** fts3EvalNearTrim2(). */
	116035	+ ** fts3EvalNearTrim(). */
115507	116036	while( pIter<pEnd && *pIter==0 ) pIter++;
115508	116037
115509	116038	pDL->pNextDocid = pIter;
115510	116039	assert( pIter>=&pDL->aAll[pDL->nAll] \|\| *pIter );
115511	116040	*pbEof = 0;
		@@ -115513,15 +116042,31 @@
115513	116042	}
115514	116043
115515	116044	return rc;
115516	116045	}
115517	116046
	116047	+/*
	116048	+**
	116049	+** If *pRc is not SQLITE_OK when this function is called, it is a no-op.
	116050	+** Otherwise, fts3EvalPhraseStart() is called on all phrases within the
	116051	+** expression. Also the Fts3Expr.bDeferred variable is set to true for any
	116052	+** expressions for which all descendent tokens are deferred.
	116053	+**
	116054	+** If parameter bOptOk is zero, then it is guaranteed that the
	116055	+** Fts3Phrase.doclist.aAll/nAll variables contain the entire doclist for
	116056	+** each phrase in the expression (subject to deferred token processing).
	116057	+** Or, if bOptOk is non-zero, then one or more tokens within the expression
	116058	+** may be loaded incrementally, meaning doclist.aAll/nAll is not available.
	116059	+**
	116060	+** If an error occurs within this function, *pRc is set to an SQLite error
	116061	+** code before returning.
	116062	+*/
115518	116063	static void fts3EvalStartReaders(
115519		- Fts3Cursor *pCsr,
115520		- Fts3Expr *pExpr,
115521		- int bOptOk,
115522		- int *pRc
	116064	+ Fts3Cursor pCsr, / FTS Cursor handle */
	116065	+ Fts3Expr pExpr, / Expression to initialize phrases in */
	116066	+ int bOptOk, /* True to enable incremental loading */
	116067	+ int pRc / IN/OUT: Error code */
115523	116068	){
115524	116069	if( pExpr && SQLITE_OK==*pRc ){
115525	116070	if( pExpr->eType==FTSQUERY_PHRASE ){
115526	116071	int i;
115527	116072	int nToken = pExpr->pPhrase->nToken;
		@@ -115536,27 +116081,46 @@
115536	116081	pExpr->bDeferred = (pExpr->pLeft->bDeferred && pExpr->pRight->bDeferred);
115537	116082	}
115538	116083	}
115539	116084	}
115540	116085
	116086	+/*
	116087	+** An array of the following structures is assembled as part of the process
	116088	+** of selecting tokens to defer before the query starts executing (as part
	116089	+** of the xFilter() method). There is one element in the array for each
	116090	+** token in the FTS expression.
	116091	+**
	116092	+** Tokens are divided into AND/NEAR clusters. All tokens in a cluster belong
	116093	+** to phrases that are connected only by AND and NEAR operators (not OR or
	116094	+** NOT). When determining tokens to defer, each AND/NEAR cluster is considered
	116095	+** separately. The root of a tokens AND/NEAR cluster is stored in
	116096	+** Fts3TokenAndCost.pRoot.
	116097	+*/
115541	116098	typedef struct Fts3TokenAndCost Fts3TokenAndCost;
115542	116099	struct Fts3TokenAndCost {
115543	116100	Fts3Phrase pPhrase; / The phrase the token belongs to */
115544	116101	int iToken; /* Position of token in phrase */
115545	116102	Fts3PhraseToken pToken; / The token itself */
115546		- Fts3Expr *pRoot;
115547		- int nOvfl;
	116103	+ Fts3Expr pRoot; / Root of NEAR/AND cluster */
	116104	+ int nOvfl; /* Number of overflow pages to load doclist */
115548	116105	int iCol; /* The column the token must match */
115549	116106	};
115550	116107
	116108	+/*
	116109	+** This function is used to populate an allocated Fts3TokenAndCost array.
	116110	+**
	116111	+** If *pRc is not SQLITE_OK when this function is called, it is a no-op.
	116112	+** Otherwise, if an error occurs during execution, *pRc is set to an
	116113	+** SQLite error code.
	116114	+*/
115551	116115	static void fts3EvalTokenCosts(
115552		- Fts3Cursor *pCsr,
115553		- Fts3Expr *pRoot,
115554		- Fts3Expr *pExpr,
115555		- Fts3TokenAndCost **ppTC,
115556		- Fts3Expr ***ppOr,
115557		- int *pRc
	116116	+ Fts3Cursor pCsr, / FTS Cursor handle */
	116117	+ Fts3Expr pRoot, / Root of current AND/NEAR cluster */
	116118	+ Fts3Expr pExpr, / Expression to consider */
	116119	+ Fts3TokenAndCost *ppTC, / Write new entries to (ppTC)++ */
	116120	+ Fts3Expr **ppOr, / Write new OR root to (ppOr)++ */
	116121	+ int pRc / IN/OUT: Error code */
115558	116122	){
115559	116123	if( *pRc==SQLITE_OK && pExpr ){
115560	116124	if( pExpr->eType==FTSQUERY_PHRASE ){
115561	116125	Fts3Phrase *pPhrase = pExpr->pPhrase;
115562	116126	int i;
		@@ -115584,23 +116148,34 @@
115584	116148	fts3EvalTokenCosts(pCsr, pRoot, pExpr->pRight, ppTC, ppOr, pRc);
115585	116149	}
115586	116150	}
115587	116151	}
115588	116152
	116153	+/*
	116154	+** Determine the average document (row) size in pages. If successful,
	116155	+** write this value to *pnPage and return SQLITE_OK. Otherwise, return
	116156	+** an SQLite error code.
	116157	+**
	116158	+** The average document size in pages is calculated by first calculating
	116159	+** determining the average size in bytes, B. If B is less than the amount
	116160	+** of data that will fit on a single leaf page of an intkey table in
	116161	+** this database, then the average docsize is 1. Otherwise, it is 1 plus
	116162	+** the number of overflow pages consumed by a record B bytes in size.
	116163	+*/
115589	116164	static int fts3EvalAverageDocsize(Fts3Cursor pCsr, int pnPage){
115590	116165	if( pCsr->nRowAvg==0 ){
115591	116166	/* The average document size, which is required to calculate the cost
115592		- ** of each doclist, has not yet been determined. Read the required
115593		- ** data from the %_stat table to calculate it.
115594		- **
115595		- ** Entry 0 of the %_stat table is a blob containing (nCol+1) FTS3
115596		- ** varints, where nCol is the number of columns in the FTS3 table.
115597		- ** The first varint is the number of documents currently stored in
115598		- ** the table. The following nCol varints contain the total amount of
115599		- ** data stored in all rows of each column of the table, from left
115600		- ** to right.
115601		- */
	116167	+ ** of each doclist, has not yet been determined. Read the required
	116168	+ ** data from the %_stat table to calculate it.
	116169	+ **
	116170	+ ** Entry 0 of the %_stat table is a blob containing (nCol+1) FTS3
	116171	+ ** varints, where nCol is the number of columns in the FTS3 table.
	116172	+ ** The first varint is the number of documents currently stored in
	116173	+ ** the table. The following nCol varints contain the total amount of
	116174	+ ** data stored in all rows of each column of the table, from left
	116175	+ ** to right.
	116176	+ */
115602	116177	int rc;
115603	116178	Fts3Table p = (Fts3Table)pCsr->base.pVtab;
115604	116179	sqlite3_stmt *pStmt;
115605	116180	sqlite3_int64 nDoc = 0;
115606	116181	sqlite3_int64 nByte = 0;
		@@ -115631,109 +116206,151 @@
115631	116206
115632	116207	*pnPage = pCsr->nRowAvg;
115633	116208	return SQLITE_OK;
115634	116209	}
115635	116210
	116211	+/*
	116212	+** This function is called to select the tokens (if any) that will be
	116213	+** deferred. The array aTC[] has already been populated when this is
	116214	+** called.
	116215	+**
	116216	+** This function is called once for each AND/NEAR cluster in the
	116217	+** expression. Each invocation determines which tokens to defer within
	116218	+** the cluster with root node pRoot. See comments above the definition
	116219	+** of struct Fts3TokenAndCost for more details.
	116220	+**
	116221	+** If no error occurs, SQLITE_OK is returned and sqlite3Fts3DeferToken()
	116222	+** called on each token to defer. Otherwise, an SQLite error code is
	116223	+** returned.
	116224	+*/
115636	116225	static int fts3EvalSelectDeferred(
115637		- Fts3Cursor *pCsr,
115638		- Fts3Expr *pRoot,
115639		- Fts3TokenAndCost *aTC,
115640		- int nTC
	116226	+ Fts3Cursor pCsr, / FTS Cursor handle */
	116227	+ Fts3Expr pRoot, / Consider tokens with this root node */
	116228	+ Fts3TokenAndCost aTC, / Array of expression tokens and costs */
	116229	+ int nTC /* Number of entries in aTC[] */
115641	116230	){
115642		- int nDocSize = 0;
115643		- int nDocEst = 0;
115644		- int rc = SQLITE_OK;
115645	116231	Fts3Table pTab = (Fts3Table )pCsr->base.pVtab;
115646		- int ii;
	116232	+ int nDocSize = 0; /* Number of pages per doc loaded */
	116233	+ int rc = SQLITE_OK; /* Return code */
	116234	+ int ii; /* Iterator variable for various purposes */
	116235	+ int nOvfl = 0; /* Total overflow pages used by doclists */
	116236	+ int nToken = 0; /* Total number of tokens in cluster */
115647	116237
115648		- int nOvfl = 0;
115649		- int nTerm = 0;
	116238	+ int nMinEst = 0; /* The minimum count for any phrase so far. */
	116239	+ int nLoad4 = 1; /* (Phrases that will be loaded)^4. */
115650	116240
	116241	+ /* Count the tokens in this AND/NEAR cluster. If none of the doclists
	116242	+ ** associated with the tokens spill onto overflow pages, or if there is
	116243	+ ** only 1 token, exit early. No tokens to defer in this case. */
115651	116244	for(ii=0; ii<nTC; ii++){
115652	116245	if( aTC[ii].pRoot==pRoot ){
115653	116246	nOvfl += aTC[ii].nOvfl;
115654		- nTerm++;
	116247	+ nToken++;
115655	116248	}
115656	116249	}
115657		- if( nOvfl==0 \|\| nTerm<2 ) return SQLITE_OK;
	116250	+ if( nOvfl==0 \|\| nToken<2 ) return SQLITE_OK;
115658	116251
	116252	+ /* Obtain the average docsize (in pages). */
115659	116253	rc = fts3EvalAverageDocsize(pCsr, &nDocSize);
115660		-
115661		- for(ii=0; ii<nTerm && rc==SQLITE_OK; ii++){
115662		- int jj;
115663		- Fts3TokenAndCost *pTC = 0;
115664		-
115665		- for(jj=0; jj<nTC; jj++){
115666		- if( aTC[jj].pToken && aTC[jj].pRoot==pRoot
115667		- && (!pTC \|\| aTC[jj].nOvfl<pTC->nOvfl)
	116254	+ assert( rc!=SQLITE_OK \|\| nDocSize>0 );
	116255	+
	116256	+
	116257	+ /* Iterate through all tokens in this AND/NEAR cluster, in ascending order
	116258	+ ** of the number of overflow pages that will be loaded by the pager layer
	116259	+ ** to retrieve the entire doclist for the token from the full-text index.
	116260	+ ** Load the doclists for tokens that are either:
	116261	+ **
	116262	+ ** a. The cheapest token in the entire query (i.e. the one visited by the
	116263	+ ** first iteration of this loop), or
	116264	+ **
	116265	+ ** b. Part of a multi-token phrase.
	116266	+ **
	116267	+ ** After each token doclist is loaded, merge it with the others from the
	116268	+ ** same phrase and count the number of documents that the merged doclist
	116269	+ ** contains. Set variable "nMinEst" to the smallest number of documents in
	116270	+ ** any phrase doclist for which 1 or more token doclists have been loaded.
	116271	+ ** Let nOther be the number of other phrases for which it is certain that
	116272	+ ** one or more tokens will not be deferred.
	116273	+ **
	116274	+ ** Then, for each token, defer it if loading the doclist would result in
	116275	+ ** loading N or more overflow pages into memory, where N is computed as:
	116276	+ **
	116277	+ ** (nMinEst + 4^nOther - 1) / (4^nOther)
	116278	+ */
	116279	+ for(ii=0; ii<nToken && rc==SQLITE_OK; ii++){
	116280	+ int iTC; /* Used to iterate through aTC[] array. */
	116281	+ Fts3TokenAndCost pTC = 0; / Set to cheapest remaining token. */
	116282	+
	116283	+ /* Set pTC to point to the cheapest remaining token. */
	116284	+ for(iTC=0; iTC<nTC; iTC++){
	116285	+ if( aTC[iTC].pToken && aTC[iTC].pRoot==pRoot
	116286	+ && (!pTC \|\| aTC[iTC].nOvfl<pTC->nOvfl)
115668	116287	){
115669		- pTC = &aTC[jj];
	116288	+ pTC = &aTC[iTC];
115670	116289	}
115671	116290	}
115672	116291	assert( pTC );
115673	116292
115674		- /* At this point pTC points to the cheapest remaining token. */
115675		- if( ii==0 ){
115676		- if( pTC->nOvfl ){
115677		- nDocEst = (pTC->nOvfl * pTab->nPgsz + pTab->nPgsz) / 10;
115678		- }else{
	116293	+ if( ii && pTC->nOvfl>=((nMinEst+(nLoad4/4)-1)/(nLoad4/4))*nDocSize ){
	116294	+ /* The number of overflow pages to load for this (and therefore all
	116295	+ ** subsequent) tokens is greater than the estimated number of pages
	116296	+ ** that will be loaded if all subsequent tokens are deferred.
	116297	+ */
	116298	+ Fts3PhraseToken *pToken = pTC->pToken;
	116299	+ rc = sqlite3Fts3DeferToken(pCsr, pToken, pTC->iCol);
	116300	+ fts3SegReaderCursorFree(pToken->pSegcsr);
	116301	+ pToken->pSegcsr = 0;
	116302	+ }else{
	116303	+ nLoad4 = nLoad4*4;
	116304	+ if( ii==0 \|\| pTC->pPhrase->nToken>1 ){
	116305	+ /* Either this is the cheapest token in the entire query, or it is
	116306	+ ** part of a multi-token phrase. Either way, the entire doclist will
	116307	+ ** (eventually) be loaded into memory. It may as well be now. */
115679	116308	Fts3PhraseToken *pToken = pTC->pToken;
115680	116309	int nList = 0;
115681	116310	char *pList = 0;
115682		- rc = fts3TermSelect(pTab, pToken, pTC->iCol, 1, &nList, &pList);
	116311	+ rc = fts3TermSelect(pTab, pToken, pTC->iCol, &nList, &pList);
115683	116312	assert( rc==SQLITE_OK \|\| pList==0 );
115684		-
115685	116313	if( rc==SQLITE_OK ){
115686		- nDocEst = fts3DoclistCountDocids(1, pList, nList);
	116314	+ int nCount;
115687	116315	fts3EvalPhraseMergeToken(pTab, pTC->pPhrase, pTC->iToken,pList,nList);
	116316	+ nCount = fts3DoclistCountDocids(
	116317	+ pTC->pPhrase->doclist.aAll, pTC->pPhrase->doclist.nAll
	116318	+ );
	116319	+ if( ii==0 \|\| nCount<nMinEst ) nMinEst = nCount;
115688	116320	}
115689	116321	}
115690		- }else{
115691		- if( pTC->nOvfl>=(nDocEst*nDocSize) ){
115692		- Fts3PhraseToken *pToken = pTC->pToken;
115693		- rc = sqlite3Fts3DeferToken(pCsr, pToken, pTC->iCol);
115694		- fts3SegReaderCursorFree(pToken->pSegcsr);
115695		- pToken->pSegcsr = 0;
115696		- }
115697		- nDocEst = 1 + (nDocEst/4);
115698	116322	}
115699	116323	pTC->pToken = 0;
115700	116324	}
115701	116325
115702	116326	return rc;
115703	116327	}
115704	116328
115705		-SQLITE_PRIVATE int sqlite3Fts3EvalStart(Fts3Cursor pCsr, Fts3Expr pExpr, int bOptOk){
	116329	+/*
	116330	+** This function is called from within the xFilter method. It initializes
	116331	+** the full-text query currently stored in pCsr->pExpr. To iterate through
	116332	+** the results of a query, the caller does:
	116333	+**
	116334	+** fts3EvalStart(pCsr);
	116335	+** while( 1 ){
	116336	+** fts3EvalNext(pCsr);
	116337	+** if( pCsr->bEof ) break;
	116338	+** ... return row pCsr->iPrevId to the caller ...
	116339	+** }
	116340	+*/
	116341	+static int fts3EvalStart(Fts3Cursor *pCsr){
115706	116342	Fts3Table pTab = (Fts3Table )pCsr->base.pVtab;
115707	116343	int rc = SQLITE_OK;
115708	116344	int nToken = 0;
115709	116345	int nOr = 0;
115710	116346
115711	116347	/* Allocate a MultiSegReader for each token in the expression. */
115712		- fts3EvalAllocateReaders(pCsr, pExpr, &nToken, &nOr, &rc);
115713		-
115714		- /* Call fts3EvalPhraseStart() on all phrases in the expression. TODO:
115715		- ** This call will eventually also be responsible for determining which
115716		- ** tokens are 'deferred' until the document text is loaded into memory.
115717		- **
115718		- ** Each token in each phrase is dealt with using one of the following
115719		- ** three strategies:
115720		- **
115721		- ** 1. Entire doclist loaded into memory as part of the
115722		- ** fts3EvalStartReaders() call.
115723		- **
115724		- ** 2. Doclist loaded into memory incrementally, as part of each
115725		- ** sqlite3Fts3EvalNext() call.
115726		- **
115727		- ** 3. Token doclist is never loaded. Instead, documents are loaded into
115728		- ** memory and scanned for the token as part of the sqlite3Fts3EvalNext()
115729		- ** call. This is known as a "deferred" token.
115730		- */
115731		-
115732		- /* If bOptOk is true, check if there are any tokens that should be deferred.
115733		- */
115734		- if( rc==SQLITE_OK && bOptOk && nToken>1 && pTab->bHasStat ){
	116348	+ fts3EvalAllocateReaders(pCsr, pCsr->pExpr, &nToken, &nOr, &rc);
	116349	+
	116350	+ /* Determine which, if any, tokens in the expression should be deferred. */
	116351	+ if( rc==SQLITE_OK && nToken>1 && pTab->bHasStat ){
115735	116352	Fts3TokenAndCost *aTC;
115736	116353	Fts3Expr **apOr;
115737	116354	aTC = (Fts3TokenAndCost *)sqlite3_malloc(
115738	116355	sizeof(Fts3TokenAndCost) * nToken
115739	116356	+ sizeof(Fts3Expr ) nOr * 2
		@@ -115745,11 +116362,11 @@
115745	116362	}else{
115746	116363	int ii;
115747	116364	Fts3TokenAndCost *pTC = aTC;
115748	116365	Fts3Expr **ppOr = apOr;
115749	116366
115750		- fts3EvalTokenCosts(pCsr, 0, pExpr, &pTC, &ppOr, &rc);
	116367	+ fts3EvalTokenCosts(pCsr, 0, pCsr->pExpr, &pTC, &ppOr, &rc);
115751	116368	nToken = pTC-aTC;
115752	116369	nOr = ppOr-apOr;
115753	116370
115754	116371	if( rc==SQLITE_OK ){
115755	116372	rc = fts3EvalSelectDeferred(pCsr, 0, aTC, nToken);
		@@ -115760,25 +116377,50 @@
115760	116377
115761	116378	sqlite3_free(aTC);
115762	116379	}
115763	116380	}
115764	116381
115765		- fts3EvalStartReaders(pCsr, pExpr, bOptOk, &rc);
	116382	+ fts3EvalStartReaders(pCsr, pCsr->pExpr, 1, &rc);
115766	116383	return rc;
115767	116384	}
115768	116385
115769		-static void fts3EvalZeroPoslist(Fts3Phrase *pPhrase){
	116386	+/*
	116387	+** Invalidate the current position list for phrase pPhrase.
	116388	+*/
	116389	+static void fts3EvalInvalidatePoslist(Fts3Phrase *pPhrase){
115770	116390	if( pPhrase->doclist.bFreeList ){
115771	116391	sqlite3_free(pPhrase->doclist.pList);
115772	116392	}
115773	116393	pPhrase->doclist.pList = 0;
115774	116394	pPhrase->doclist.nList = 0;
115775	116395	pPhrase->doclist.bFreeList = 0;
115776	116396	}
115777	116397
115778		-static int fts3EvalNearTrim2(
115779		- int nNear,
	116398	+/*
	116399	+** This function is called to edit the position list associated with
	116400	+** the phrase object passed as the fifth argument according to a NEAR
	116401	+** condition. For example:
	116402	+**
	116403	+** abc NEAR/5 "def ghi"
	116404	+**
	116405	+** Parameter nNear is passed the NEAR distance of the expression (5 in
	116406	+** the example above). When this function is called, *paPoslist points to
	116407	+** the position list, and *pnToken is the number of phrase tokens in, the
	116408	+** phrase on the other side of the NEAR operator to pPhrase. For example,
	116409	+** if pPhrase refers to the "def ghi" phrase, then *paPoslist points to
	116410	+** the position list associated with phrase "abc".
	116411	+**
	116412	+** All positions in the pPhrase position list that are not sufficiently
	116413	+** close to a position in the *paPoslist position list are removed. If this
	116414	+** leaves 0 positions, zero is returned. Otherwise, non-zero.
	116415	+**
	116416	+** Before returning, *paPoslist is set to point to the position lsit
	116417	+** associated with pPhrase. And *pnToken is set to the number of tokens in
	116418	+** pPhrase.
	116419	+*/
	116420	+static int fts3EvalNearTrim(
	116421	+ int nNear, /* NEAR distance. As in "NEAR/nNear". */
115780	116422	char aTmp, / Temporary space to use */
115781	116423	char *paPoslist, / IN/OUT: Position list */
115782	116424	int pnToken, / IN/OUT: Tokens in phrase of paPoslist /
115783	116425	Fts3Phrase pPhrase / The phrase object to trim the doclist of */
115784	116426	){
		@@ -115806,10 +116448,176 @@
115806	116448	}
115807	116449
115808	116450	return res;
115809	116451	}
115810	116452
	116453	+/*
	116454	+** This function is a no-op if *pRc is other than SQLITE_OK when it is called.
	116455	+** Otherwise, it advances the expression passed as the second argument to
	116456	+** point to the next matching row in the database. Expressions iterate through
	116457	+** matching rows in docid order. Ascending order if Fts3Cursor.bDesc is zero,
	116458	+** or descending if it is non-zero.
	116459	+**
	116460	+** If an error occurs, *pRc is set to an SQLite error code. Otherwise, if
	116461	+** successful, the following variables in pExpr are set:
	116462	+**
	116463	+** Fts3Expr.bEof (non-zero if EOF - there is no next row)
	116464	+** Fts3Expr.iDocid (valid if bEof==0. The docid of the next row)
	116465	+**
	116466	+** If the expression is of type FTSQUERY_PHRASE, and the expression is not
	116467	+** at EOF, then the following variables are populated with the position list
	116468	+** for the phrase for the visited row:
	116469	+**
	116470	+** FTs3Expr.pPhrase->doclist.nList (length of pList in bytes)
	116471	+** FTs3Expr.pPhrase->doclist.pList (pointer to position list)
	116472	+**
	116473	+** It says above that this function advances the expression to the next
	116474	+** matching row. This is usually true, but there are the following exceptions:
	116475	+**
	116476	+** 1. Deferred tokens are not taken into account. If a phrase consists
	116477	+** entirely of deferred tokens, it is assumed to match every row in
	116478	+** the db. In this case the position-list is not populated at all.
	116479	+**
	116480	+** Or, if a phrase contains one or more deferred tokens and one or
	116481	+** more non-deferred tokens, then the expression is advanced to the
	116482	+** next possible match, considering only non-deferred tokens. In other
	116483	+** words, if the phrase is "A B C", and "B" is deferred, the expression
	116484	+** is advanced to the next row that contains an instance of "A * C",
	116485	+** where "*" may match any single token. The position list in this case
	116486	+** is populated as for "A * C" before returning.
	116487	+**
	116488	+** 2. NEAR is treated as AND. If the expression is "x NEAR y", it is
	116489	+** advanced to point to the next row that matches "x AND y".
	116490	+**
	116491	+** See fts3EvalTestDeferredAndNear() for details on testing if a row is
	116492	+** really a match, taking into account deferred tokens and NEAR operators.
	116493	+*/
	116494	+static void fts3EvalNextRow(
	116495	+ Fts3Cursor pCsr, / FTS Cursor handle */
	116496	+ Fts3Expr pExpr, / Expr. to advance to next matching row */
	116497	+ int pRc / IN/OUT: Error code */
	116498	+){
	116499	+ if( *pRc==SQLITE_OK ){
	116500	+ int bDescDoclist = pCsr->bDesc; /* Used by DOCID_CMP() macro */
	116501	+ assert( pExpr->bEof==0 );
	116502	+ pExpr->bStart = 1;
	116503	+
	116504	+ switch( pExpr->eType ){
	116505	+ case FTSQUERY_NEAR:
	116506	+ case FTSQUERY_AND: {
	116507	+ Fts3Expr *pLeft = pExpr->pLeft;
	116508	+ Fts3Expr *pRight = pExpr->pRight;
	116509	+ assert( !pLeft->bDeferred \|\| !pRight->bDeferred );
	116510	+
	116511	+ if( pLeft->bDeferred ){
	116512	+ /* LHS is entirely deferred. So we assume it matches every row.
	116513	+ ** Advance the RHS iterator to find the next row visited. */
	116514	+ fts3EvalNextRow(pCsr, pRight, pRc);
	116515	+ pExpr->iDocid = pRight->iDocid;
	116516	+ pExpr->bEof = pRight->bEof;
	116517	+ }else if( pRight->bDeferred ){
	116518	+ /* RHS is entirely deferred. So we assume it matches every row.
	116519	+ ** Advance the LHS iterator to find the next row visited. */
	116520	+ fts3EvalNextRow(pCsr, pLeft, pRc);
	116521	+ pExpr->iDocid = pLeft->iDocid;
	116522	+ pExpr->bEof = pLeft->bEof;
	116523	+ }else{
	116524	+ /* Neither the RHS or LHS are deferred. */
	116525	+ fts3EvalNextRow(pCsr, pLeft, pRc);
	116526	+ fts3EvalNextRow(pCsr, pRight, pRc);
	116527	+ while( !pLeft->bEof && !pRight->bEof && *pRc==SQLITE_OK ){
	116528	+ sqlite3_int64 iDiff = DOCID_CMP(pLeft->iDocid, pRight->iDocid);
	116529	+ if( iDiff==0 ) break;
	116530	+ if( iDiff<0 ){
	116531	+ fts3EvalNextRow(pCsr, pLeft, pRc);
	116532	+ }else{
	116533	+ fts3EvalNextRow(pCsr, pRight, pRc);
	116534	+ }
	116535	+ }
	116536	+ pExpr->iDocid = pLeft->iDocid;
	116537	+ pExpr->bEof = (pLeft->bEof \|\| pRight->bEof);
	116538	+ }
	116539	+ break;
	116540	+ }
	116541	+
	116542	+ case FTSQUERY_OR: {
	116543	+ Fts3Expr *pLeft = pExpr->pLeft;
	116544	+ Fts3Expr *pRight = pExpr->pRight;
	116545	+ sqlite3_int64 iCmp = DOCID_CMP(pLeft->iDocid, pRight->iDocid);
	116546	+
	116547	+ assert( pLeft->bStart \|\| pLeft->iDocid==pRight->iDocid );
	116548	+ assert( pRight->bStart \|\| pLeft->iDocid==pRight->iDocid );
	116549	+
	116550	+ if( pRight->bEof \|\| (pLeft->bEof==0 && iCmp<0) ){
	116551	+ fts3EvalNextRow(pCsr, pLeft, pRc);
	116552	+ }else if( pLeft->bEof \|\| (pRight->bEof==0 && iCmp>0) ){
	116553	+ fts3EvalNextRow(pCsr, pRight, pRc);
	116554	+ }else{
	116555	+ fts3EvalNextRow(pCsr, pLeft, pRc);
	116556	+ fts3EvalNextRow(pCsr, pRight, pRc);
	116557	+ }
	116558	+
	116559	+ pExpr->bEof = (pLeft->bEof && pRight->bEof);
	116560	+ iCmp = DOCID_CMP(pLeft->iDocid, pRight->iDocid);
	116561	+ if( pRight->bEof \|\| (pLeft->bEof==0 && iCmp<0) ){
	116562	+ pExpr->iDocid = pLeft->iDocid;
	116563	+ }else{
	116564	+ pExpr->iDocid = pRight->iDocid;
	116565	+ }
	116566	+
	116567	+ break;
	116568	+ }
	116569	+
	116570	+ case FTSQUERY_NOT: {
	116571	+ Fts3Expr *pLeft = pExpr->pLeft;
	116572	+ Fts3Expr *pRight = pExpr->pRight;
	116573	+
	116574	+ if( pRight->bStart==0 ){
	116575	+ fts3EvalNextRow(pCsr, pRight, pRc);
	116576	+ assert( *pRc!=SQLITE_OK \|\| pRight->bStart );
	116577	+ }
	116578	+
	116579	+ fts3EvalNextRow(pCsr, pLeft, pRc);
	116580	+ if( pLeft->bEof==0 ){
	116581	+ while( !*pRc
	116582	+ && !pRight->bEof
	116583	+ && DOCID_CMP(pLeft->iDocid, pRight->iDocid)>0
	116584	+ ){
	116585	+ fts3EvalNextRow(pCsr, pRight, pRc);
	116586	+ }
	116587	+ }
	116588	+ pExpr->iDocid = pLeft->iDocid;
	116589	+ pExpr->bEof = pLeft->bEof;
	116590	+ break;
	116591	+ }
	116592	+
	116593	+ default: {
	116594	+ Fts3Phrase *pPhrase = pExpr->pPhrase;
	116595	+ fts3EvalInvalidatePoslist(pPhrase);
	116596	+ *pRc = fts3EvalPhraseNext(pCsr, pPhrase, &pExpr->bEof);
	116597	+ pExpr->iDocid = pPhrase->doclist.iDocid;
	116598	+ break;
	116599	+ }
	116600	+ }
	116601	+ }
	116602	+}
	116603	+
	116604	+/*
	116605	+** If *pRc is not SQLITE_OK, or if pExpr is not the root node of a NEAR
	116606	+** cluster, then this function returns 1 immediately.
	116607	+**
	116608	+** Otherwise, it checks if the current row really does match the NEAR
	116609	+** expression, using the data currently stored in the position lists
	116610	+** (Fts3Expr->pPhrase.doclist.pList/nList) for each phrase in the expression.
	116611	+**
	116612	+** If the current row is a match, the position list associated with each
	116613	+** phrase in the NEAR expression is edited in place to contain only those
	116614	+** phrase instances sufficiently close to their peers to satisfy all NEAR
	116615	+** constraints. In this case it returns 1. If the NEAR expression does not
	116616	+** match the current row, 0 is returned. The position lists may or may not
	116617	+** be edited if 0 is returned.
	116618	+*/
115811	116619	static int fts3EvalNearTest(Fts3Expr pExpr, int pRc){
115812	116620	int res = 1;
115813	116621
115814	116622	/* The following block runs if pExpr is the root of a NEAR query.
115815	116623	** For example, the query:
		@@ -115827,11 +116635,11 @@
115827	116635	** \| \|
115828	116636	** "w" "x"
115829	116637	**
115830	116638	** The right-hand child of a NEAR node is always a phrase. The
115831	116639	** left-hand child may be either a phrase or a NEAR node. There are
115832		- ** no exceptions to this.
	116640	+ ** no exceptions to this - it's the way the parser in fts3_expr.c works.
115833	116641	*/
115834	116642	if( *pRc==SQLITE_OK
115835	116643	&& pExpr->eType==FTSQUERY_NEAR
115836	116644	&& pExpr->bEof==0
115837	116645	&& (pExpr->pParent==0 \|\| pExpr->pParent->eType!=FTSQUERY_NEAR)
		@@ -115854,21 +116662,21 @@
115854	116662	int nToken = p->pPhrase->nToken;
115855	116663
115856	116664	for(p=p->pParent;res && p && p->eType==FTSQUERY_NEAR; p=p->pParent){
115857	116665	Fts3Phrase *pPhrase = p->pRight->pPhrase;
115858	116666	int nNear = p->nNear;
115859		- res = fts3EvalNearTrim2(nNear, aTmp, &aPoslist, &nToken, pPhrase);
	116667	+ res = fts3EvalNearTrim(nNear, aTmp, &aPoslist, &nToken, pPhrase);
115860	116668	}
115861	116669
115862	116670	aPoslist = pExpr->pRight->pPhrase->doclist.pList;
115863	116671	nToken = pExpr->pRight->pPhrase->nToken;
115864	116672	for(p=pExpr->pLeft; p && res; p=p->pLeft){
115865	116673	int nNear = p->pParent->nNear;
115866	116674	Fts3Phrase *pPhrase = (
115867	116675	p->eType==FTSQUERY_NEAR ? p->pRight->pPhrase : p->pPhrase
115868	116676	);
115869		- res = fts3EvalNearTrim2(nNear, aTmp, &aPoslist, &nToken, pPhrase);
	116677	+ res = fts3EvalNearTrim(nNear, aTmp, &aPoslist, &nToken, pPhrase);
115870	116678	}
115871	116679	}
115872	116680
115873	116681	sqlite3_free(aTmp);
115874	116682	}
		@@ -115875,132 +116683,33 @@
115875	116683
115876	116684	return res;
115877	116685	}
115878	116686
115879	116687	/*
115880		-** This macro is used by the fts3EvalNext() function. The two arguments are
115881		-** 64-bit docid values. If the current query is "ORDER BY docid ASC", then
115882		-** the macro returns (i1 - i2). Or if it is "ORDER BY docid DESC", then
115883		-** it returns (i2 - i1). This allows the same code to be used for merging
115884		-** doclists in ascending or descending order.
	116688	+** This function is a helper function for fts3EvalTestDeferredAndNear().
	116689	+** Assuming no error occurs or has occurred, It returns non-zero if the
	116690	+** expression passed as the second argument matches the row that pCsr
	116691	+** currently points to, or zero if it does not.
	116692	+**
	116693	+** If *pRc is not SQLITE_OK when this function is called, it is a no-op.
	116694	+** If an error occurs during execution of this function, *pRc is set to
	116695	+** the appropriate SQLite error code. In this case the returned value is
	116696	+** undefined.
115885	116697	*/
115886		-#define DOCID_CMP(i1, i2) ((pCsr->bDesc?-1:1) * (i1-i2))
115887		-
115888		-static void fts3EvalNext(
115889		- Fts3Cursor *pCsr,
115890		- Fts3Expr *pExpr,
115891		- int *pRc
115892		-){
115893		- if( *pRc==SQLITE_OK ){
115894		- assert( pExpr->bEof==0 );
115895		- pExpr->bStart = 1;
115896		-
115897		- switch( pExpr->eType ){
115898		- case FTSQUERY_NEAR:
115899		- case FTSQUERY_AND: {
115900		- Fts3Expr *pLeft = pExpr->pLeft;
115901		- Fts3Expr *pRight = pExpr->pRight;
115902		- assert( !pLeft->bDeferred \|\| !pRight->bDeferred );
115903		- if( pLeft->bDeferred ){
115904		- fts3EvalNext(pCsr, pRight, pRc);
115905		- pExpr->iDocid = pRight->iDocid;
115906		- pExpr->bEof = pRight->bEof;
115907		- }else if( pRight->bDeferred ){
115908		- fts3EvalNext(pCsr, pLeft, pRc);
115909		- pExpr->iDocid = pLeft->iDocid;
115910		- pExpr->bEof = pLeft->bEof;
115911		- }else{
115912		- fts3EvalNext(pCsr, pLeft, pRc);
115913		- fts3EvalNext(pCsr, pRight, pRc);
115914		-
115915		- while( !pLeft->bEof && !pRight->bEof && *pRc==SQLITE_OK ){
115916		- sqlite3_int64 iDiff = DOCID_CMP(pLeft->iDocid, pRight->iDocid);
115917		- if( iDiff==0 ) break;
115918		- if( iDiff<0 ){
115919		- fts3EvalNext(pCsr, pLeft, pRc);
115920		- }else{
115921		- fts3EvalNext(pCsr, pRight, pRc);
115922		- }
115923		- }
115924		-
115925		- pExpr->iDocid = pLeft->iDocid;
115926		- pExpr->bEof = (pLeft->bEof \|\| pRight->bEof);
115927		- }
115928		- break;
115929		- }
115930		-
115931		- case FTSQUERY_OR: {
115932		- Fts3Expr *pLeft = pExpr->pLeft;
115933		- Fts3Expr *pRight = pExpr->pRight;
115934		- sqlite3_int64 iCmp = DOCID_CMP(pLeft->iDocid, pRight->iDocid);
115935		-
115936		- assert( pLeft->bStart \|\| pLeft->iDocid==pRight->iDocid );
115937		- assert( pRight->bStart \|\| pLeft->iDocid==pRight->iDocid );
115938		-
115939		- if( pRight->bEof \|\| (pLeft->bEof==0 && iCmp<0) ){
115940		- fts3EvalNext(pCsr, pLeft, pRc);
115941		- }else if( pLeft->bEof \|\| (pRight->bEof==0 && iCmp>0) ){
115942		- fts3EvalNext(pCsr, pRight, pRc);
115943		- }else{
115944		- fts3EvalNext(pCsr, pLeft, pRc);
115945		- fts3EvalNext(pCsr, pRight, pRc);
115946		- }
115947		-
115948		- pExpr->bEof = (pLeft->bEof && pRight->bEof);
115949		- iCmp = DOCID_CMP(pLeft->iDocid, pRight->iDocid);
115950		- if( pRight->bEof \|\| (pLeft->bEof==0 && iCmp<0) ){
115951		- pExpr->iDocid = pLeft->iDocid;
115952		- }else{
115953		- pExpr->iDocid = pRight->iDocid;
115954		- }
115955		-
115956		- break;
115957		- }
115958		-
115959		- case FTSQUERY_NOT: {
115960		- Fts3Expr *pLeft = pExpr->pLeft;
115961		- Fts3Expr *pRight = pExpr->pRight;
115962		-
115963		- if( pRight->bStart==0 ){
115964		- fts3EvalNext(pCsr, pRight, pRc);
115965		- assert( *pRc!=SQLITE_OK \|\| pRight->bStart );
115966		- }
115967		-
115968		- fts3EvalNext(pCsr, pLeft, pRc);
115969		- if( pLeft->bEof==0 ){
115970		- while( !*pRc
115971		- && !pRight->bEof
115972		- && DOCID_CMP(pLeft->iDocid, pRight->iDocid)>0
115973		- ){
115974		- fts3EvalNext(pCsr, pRight, pRc);
115975		- }
115976		- }
115977		- pExpr->iDocid = pLeft->iDocid;
115978		- pExpr->bEof = pLeft->bEof;
115979		- break;
115980		- }
115981		-
115982		- default: {
115983		- Fts3Phrase *pPhrase = pExpr->pPhrase;
115984		- fts3EvalZeroPoslist(pPhrase);
115985		- *pRc = fts3EvalPhraseNext(pCsr, pPhrase, &pExpr->bEof);
115986		- pExpr->iDocid = pPhrase->doclist.iDocid;
115987		- break;
115988		- }
115989		- }
115990		- }
115991		-}
115992		-
115993		-static int fts3EvalDeferredTest(Fts3Cursor pCsr, Fts3Expr pExpr, int *pRc){
115994		- int bHit = 1;
	116698	+static int fts3EvalTestExpr(
	116699	+ Fts3Cursor pCsr, / FTS cursor handle */
	116700	+ Fts3Expr pExpr, / Expr to test. May or may not be root. */
	116701	+ int pRc / IN/OUT: Error code */
	116702	+){
	116703	+ int bHit = 1; /* Return value */
115995	116704	if( *pRc==SQLITE_OK ){
115996	116705	switch( pExpr->eType ){
115997	116706	case FTSQUERY_NEAR:
115998	116707	case FTSQUERY_AND:
115999	116708	bHit = (
116000		- fts3EvalDeferredTest(pCsr, pExpr->pLeft, pRc)
116001		- && fts3EvalDeferredTest(pCsr, pExpr->pRight, pRc)
	116709	+ fts3EvalTestExpr(pCsr, pExpr->pLeft, pRc)
	116710	+ && fts3EvalTestExpr(pCsr, pExpr->pRight, pRc)
116002	116711	&& fts3EvalNearTest(pExpr, pRc)
116003	116712	);
116004	116713
116005	116714	/* If the NEAR expression does not match any rows, zero the doclist for
116006	116715	** all phrases involved in the NEAR. This is because the snippet(),
		@@ -116022,31 +116731,31 @@
116022	116731	&& (pExpr->pParent==0 \|\| pExpr->pParent->eType!=FTSQUERY_NEAR)
116023	116732	){
116024	116733	Fts3Expr *p;
116025	116734	for(p=pExpr; p->pPhrase==0; p=p->pLeft){
116026	116735	if( p->pRight->iDocid==pCsr->iPrevId ){
116027		- fts3EvalZeroPoslist(p->pRight->pPhrase);
	116736	+ fts3EvalInvalidatePoslist(p->pRight->pPhrase);
116028	116737	}
116029	116738	}
116030	116739	if( p->iDocid==pCsr->iPrevId ){
116031		- fts3EvalZeroPoslist(p->pPhrase);
	116740	+ fts3EvalInvalidatePoslist(p->pPhrase);
116032	116741	}
116033	116742	}
116034	116743
116035	116744	break;
116036	116745
116037	116746	case FTSQUERY_OR: {
116038		- int bHit1 = fts3EvalDeferredTest(pCsr, pExpr->pLeft, pRc);
116039		- int bHit2 = fts3EvalDeferredTest(pCsr, pExpr->pRight, pRc);
	116747	+ int bHit1 = fts3EvalTestExpr(pCsr, pExpr->pLeft, pRc);
	116748	+ int bHit2 = fts3EvalTestExpr(pCsr, pExpr->pRight, pRc);
116040	116749	bHit = bHit1 \|\| bHit2;
116041	116750	break;
116042	116751	}
116043	116752
116044	116753	case FTSQUERY_NOT:
116045	116754	bHit = (
116046		- fts3EvalDeferredTest(pCsr, pExpr->pLeft, pRc)
116047		- && !fts3EvalDeferredTest(pCsr, pExpr->pRight, pRc)
	116755	+ fts3EvalTestExpr(pCsr, pExpr->pLeft, pRc)
	116756	+ && !fts3EvalTestExpr(pCsr, pExpr->pRight, pRc)
116048	116757	);
116049	116758	break;
116050	116759
116051	116760	default: {
116052	116761	if( pCsr->pDeferred
		@@ -116053,11 +116762,11 @@
116053	116762	&& (pExpr->iDocid==pCsr->iPrevId \|\| pExpr->bDeferred)
116054	116763	){
116055	116764	Fts3Phrase *pPhrase = pExpr->pPhrase;
116056	116765	assert( pExpr->bDeferred \|\| pPhrase->doclist.bFreeList==0 );
116057	116766	if( pExpr->bDeferred ){
116058		- fts3EvalZeroPoslist(pPhrase);
	116767	+ fts3EvalInvalidatePoslist(pPhrase);
116059	116768	}
116060	116769	*pRc = fts3EvalDeferredPhrase(pCsr, pPhrase);
116061	116770	bHit = (pPhrase->doclist.pList!=0);
116062	116771	pExpr->iDocid = pCsr->iPrevId;
116063	116772	}else{
		@@ -116069,31 +116778,53 @@
116069	116778	}
116070	116779	return bHit;
116071	116780	}
116072	116781
116073	116782	/*
116074		-** Return 1 if both of the following are true:
	116783	+** This function is called as the second part of each xNext operation when
	116784	+** iterating through the results of a full-text query. At this point the
	116785	+** cursor points to a row that matches the query expression, with the
	116786	+** following caveats:
	116787	+**
	116788	+** * Up until this point, "NEAR" operators in the expression have been
	116789	+** treated as "AND".
	116790	+**
	116791	+** * Deferred tokens have not yet been considered.
	116792	+**
	116793	+** If *pRc is not SQLITE_OK when this function is called, it immediately
	116794	+** returns 0. Otherwise, it tests whether or not after considering NEAR
	116795	+** operators and deferred tokens the current row is still a match for the
	116796	+** expression. It returns 1 if both of the following are true:
116075	116797	**
116076	116798	** 1. *pRc is SQLITE_OK when this function returns, and
116077	116799	**
116078	116800	** 2. After scanning the current FTS table row for the deferred tokens,
116079		-** it is determined that the row does not match the query.
	116801	+** it is determined that the row does not match the query.
116080	116802	**
116081	116803	** Or, if no error occurs and it seems the current row does match the FTS
116082	116804	** query, return 0.
116083	116805	*/
116084		-static int fts3EvalLoadDeferred(Fts3Cursor pCsr, int pRc){
	116806	+static int fts3EvalTestDeferredAndNear(Fts3Cursor pCsr, int pRc){
116085	116807	int rc = *pRc;
116086	116808	int bMiss = 0;
116087	116809	if( rc==SQLITE_OK ){
	116810	+
	116811	+ /* If there are one or more deferred tokens, load the current row into
	116812	+ ** memory and scan it to determine the position list for each deferred
	116813	+ ** token. Then, see if this row is really a match, considering deferred
	116814	+ ** tokens and NEAR operators (neither of which were taken into account
	116815	+ ** earlier, by fts3EvalNextRow()).
	116816	+ */
116088	116817	if( pCsr->pDeferred ){
116089	116818	rc = fts3CursorSeek(0, pCsr);
116090	116819	if( rc==SQLITE_OK ){
116091	116820	rc = sqlite3Fts3CacheDeferredDoclists(pCsr);
116092	116821	}
116093	116822	}
116094		- bMiss = (0==fts3EvalDeferredTest(pCsr, pCsr->pExpr, &rc));
	116823	+ bMiss = (0==fts3EvalTestExpr(pCsr, pCsr->pExpr, &rc));
	116824	+
	116825	+ /* Free the position-lists accumulated for each deferred token above. */
116095	116826	sqlite3Fts3FreeDeferredDoclists(pCsr);
116096	116827	*pRc = rc;
116097	116828	}
116098	116829	return (rc==SQLITE_OK && bMiss);
116099	116830	}
		@@ -116100,11 +116831,11 @@
116100	116831
116101	116832	/*
116102	116833	** Advance to the next document that matches the FTS expression in
116103	116834	** Fts3Cursor.pExpr.
116104	116835	*/
116105		-SQLITE_PRIVATE int sqlite3Fts3EvalNext(Fts3Cursor *pCsr){
	116836	+static int fts3EvalNext(Fts3Cursor *pCsr){
116106	116837	int rc = SQLITE_OK; /* Return Code */
116107	116838	Fts3Expr *pExpr = pCsr->pExpr;
116108	116839	assert( pCsr->isEof==0 );
116109	116840	if( pExpr==0 ){
116110	116841	pCsr->isEof = 1;
		@@ -116112,23 +116843,23 @@
116112	116843	do {
116113	116844	if( pCsr->isRequireSeek==0 ){
116114	116845	sqlite3_reset(pCsr->pStmt);
116115	116846	}
116116	116847	assert( sqlite3_data_count(pCsr->pStmt)==0 );
116117		- fts3EvalNext(pCsr, pExpr, &rc);
	116848	+ fts3EvalNextRow(pCsr, pExpr, &rc);
116118	116849	pCsr->isEof = pExpr->bEof;
116119	116850	pCsr->isRequireSeek = 1;
116120	116851	pCsr->isMatchinfoNeeded = 1;
116121	116852	pCsr->iPrevId = pExpr->iDocid;
116122		- }while( pCsr->isEof==0 && fts3EvalLoadDeferred(pCsr, &rc) );
	116853	+ }while( pCsr->isEof==0 && fts3EvalTestDeferredAndNear(pCsr, &rc) );
116123	116854	}
116124	116855	return rc;
116125	116856	}
116126	116857
116127	116858	/*
116128	116859	** Restart interation for expression pExpr so that the next call to
116129		-** sqlite3Fts3EvalNext() visits the first row. Do not allow incremental
	116860	+** fts3EvalNext() visits the first row. Do not allow incremental
116130	116861	** loading or merging of phrase doclists for this iteration.
116131	116862	**
116132	116863	** If *pRc is other than SQLITE_OK when this function is called, it is
116133	116864	** a no-op. If an error occurs within this function, *pRc is set to an
116134	116865	** SQLite error code before returning.
		@@ -116140,11 +116871,11 @@
116140	116871	){
116141	116872	if( pExpr && *pRc==SQLITE_OK ){
116142	116873	Fts3Phrase *pPhrase = pExpr->pPhrase;
116143	116874
116144	116875	if( pPhrase ){
116145		- fts3EvalZeroPoslist(pPhrase);
	116876	+ fts3EvalInvalidatePoslist(pPhrase);
116146	116877	if( pPhrase->bIncr ){
116147	116878	assert( pPhrase->nToken==1 );
116148	116879	assert( pPhrase->aToken[0].pSegcsr );
116149	116880	sqlite3Fts3MsrIncrRestart(pPhrase->aToken[0].pSegcsr);
116150	116881	*pRc = fts3EvalPhraseStart(pCsr, 0, pPhrase);
		@@ -116256,18 +116987,18 @@
116256	116987	/* Ensure the %_content statement is reset. */
116257	116988	if( pCsr->isRequireSeek==0 ) sqlite3_reset(pCsr->pStmt);
116258	116989	assert( sqlite3_data_count(pCsr->pStmt)==0 );
116259	116990
116260	116991	/* Advance to the next document */
116261		- fts3EvalNext(pCsr, pRoot, &rc);
	116992	+ fts3EvalNextRow(pCsr, pRoot, &rc);
116262	116993	pCsr->isEof = pRoot->bEof;
116263	116994	pCsr->isRequireSeek = 1;
116264	116995	pCsr->isMatchinfoNeeded = 1;
116265	116996	pCsr->iPrevId = pRoot->iDocid;
116266	116997	}while( pCsr->isEof==0
116267	116998	&& pRoot->eType==FTSQUERY_NEAR
116268		- && fts3EvalLoadDeferred(pCsr, &rc)
	116999	+ && fts3EvalTestDeferredAndNear(pCsr, &rc)
116269	117000	);
116270	117001
116271	117002	if( rc==SQLITE_OK && pCsr->isEof==0 ){
116272	117003	fts3EvalUpdateCounts(pRoot);
116273	117004	}
		@@ -116285,14 +117016,14 @@
116285	117016	**
116286	117017	** do {...} while( pRoot->iDocid<iDocid && rc==SQLITE_OK );
116287	117018	*/
116288	117019	fts3EvalRestart(pCsr, pRoot, &rc);
116289	117020	do {
116290		- fts3EvalNext(pCsr, pRoot, &rc);
	117021	+ fts3EvalNextRow(pCsr, pRoot, &rc);
116291	117022	assert( pRoot->bEof==0 );
116292	117023	}while( pRoot->iDocid!=iDocid && rc==SQLITE_OK );
116293		- fts3EvalLoadDeferred(pCsr, &rc);
	117024	+ fts3EvalTestDeferredAndNear(pCsr, &rc);
116294	117025	}
116295	117026	}
116296	117027	return rc;
116297	117028	}
116298	117029
		@@ -116419,18 +117150,32 @@
116419	117150	*/
116420	117151	SQLITE_PRIVATE void sqlite3Fts3EvalPhraseCleanup(Fts3Phrase *pPhrase){
116421	117152	if( pPhrase ){
116422	117153	int i;
116423	117154	sqlite3_free(pPhrase->doclist.aAll);
116424		- fts3EvalZeroPoslist(pPhrase);
	117155	+ fts3EvalInvalidatePoslist(pPhrase);
116425	117156	memset(&pPhrase->doclist, 0, sizeof(Fts3Doclist));
116426	117157	for(i=0; i<pPhrase->nToken; i++){
116427	117158	fts3SegReaderCursorFree(pPhrase->aToken[i].pSegcsr);
116428	117159	pPhrase->aToken[i].pSegcsr = 0;
116429	117160	}
116430	117161	}
116431	117162	}
	117163	+
	117164	+#if !SQLITE_CORE
	117165	+/*
	117166	+** Initialize API pointer table, if required.
	117167	+*/
	117168	+SQLITE_API int sqlite3_extension_init(
	117169	+ sqlite3 *db,
	117170	+ char **pzErrMsg,
	117171	+ const sqlite3_api_routines *pApi
	117172	+){
	117173	+ SQLITE_EXTENSION_INIT2(pApi)
	117174	+ return sqlite3Fts3Init(db);
	117175	+}
	117176	+#endif
116432	117177
116433	117178	#endif
116434	117179
116435	117180	/************ End of fts3.c **********************************************/
116436	117181	/************ Begin file fts3_aux.c **************************************/
		@@ -118917,14 +119662,10 @@
118917	119662	** (in which case SQLITE_CORE is not defined), or
118918	119663	**
118919	119664	** * The FTS3 module is being built into the core of
118920	119665	** SQLite (in which case SQLITE_ENABLE_FTS3 is defined).
118921	119666	*/
118922		-#ifndef SQLITE_CORE
118923		- SQLITE_EXTENSION_INIT1
118924		-#endif
118925		-
118926	119667	#if !defined(SQLITE_CORE) \|\| defined(SQLITE_ENABLE_FTS3)
118927	119668
118928	119669
118929	119670	/*
118930	119671	** Implementation of the SQL scalar function for accessing the underlying
118931	119672

	--- src/sqlite3.c
	+++ src/sqlite3.c
	@@ -1,8 +1,8 @@
1	/******************************************************************************
2	** This file is an amalgamation of many separate C source files from SQLite
3	** version 3.7.7. By combining all the individual C code files into this
4	** single large file, the entire code can be compiled as a single translation
5	** unit. This allows many compilers to do optimizations that would not be
6	** possible if the files were compiled separately. Performance improvements
7	** of 5% or more are commonly seen when SQLite is compiled as a single
8	** translation unit.
	@@ -648,13 +648,13 @@
648	**
649	** See also: [sqlite3_libversion()],
650	** [sqlite3_libversion_number()], [sqlite3_sourceid()],
651	** [sqlite_version()] and [sqlite_source_id()].
652	*/
653	#define SQLITE_VERSION "3.7.7"
654	#define SQLITE_VERSION_NUMBER 3007007
655	#define SQLITE_SOURCE_ID "2011-06-24 11:29:51 9b191bb4c7c1e1b12b188c0b3eee1f8f587887c8"
656
657	/*
658	** CAPI3REF: Run-Time Library Version Numbers
659	** KEYWORDS: sqlite3_version, sqlite3_sourceid
660	**
	@@ -1282,20 +1282,37 @@
1282	** when [PRAGMA synchronous \| PRAGMA synchronous=OFF] is set, but most
1283	** VFSes do not need this signal and should silently ignore this opcode.
1284	** Applications should not call [sqlite3_file_control()] with this
1285	** opcode as doing so may disrupt the operation of the specialized VFSes
1286	** that do require it.

















1287	*/
1288	#define SQLITE_FCNTL_LOCKSTATE 1
1289	#define SQLITE_GET_LOCKPROXYFILE 2
1290	#define SQLITE_SET_LOCKPROXYFILE 3
1291	#define SQLITE_LAST_ERRNO 4
1292	#define SQLITE_FCNTL_SIZE_HINT 5
1293	#define SQLITE_FCNTL_CHUNK_SIZE 6
1294	#define SQLITE_FCNTL_FILE_POINTER 7
1295	#define SQLITE_FCNTL_SYNC_OMITTED 8
1296
1297
1298	/*
1299	** CAPI3REF: Mutex Handle
1300	**
1301	** The mutex module within SQLite defines [sqlite3_mutex] to be an
	@@ -9558,10 +9575,11 @@
9558	#define SQLITE_IndexSearch 0x08 /* Disable indexes for searching */
9559	#define SQLITE_IndexCover 0x10 /* Disable index covering table */
9560	#define SQLITE_GroupByOrder 0x20 /* Disable GROUPBY cover of ORDERBY */
9561	#define SQLITE_FactorOutConst 0x40 /* Disable factoring out constants */
9562	#define SQLITE_IdxRealAsInt 0x80 /* Store REAL as INT in indices */

9563	#define SQLITE_OptMask 0xff /* Mask of all disablable opts */
9564
9565	/*
9566	** Possible values for the sqlite.magic field.
9567	** The numbers are obtained at random and have no special meaning, other
	@@ -10449,10 +10467,11 @@
10449	Table pTab; / An SQL table corresponding to zName */
10450	Select pSelect; / A SELECT statement used in place of a table name */
10451	u8 isPopulated; /* Temporary table associated with SELECT is populated */
10452	u8 jointype; /* Type of join between this able and the previous */
10453	u8 notIndexed; /* True if there is a NOT INDEXED clause */

10454	#ifndef SQLITE_OMIT_EXPLAIN
10455	u8 iSelectId; /* If pSelect!=0, the id of the sub-select in EQP */
10456	#endif
10457	int iCursor; /* The VDBE cursor number used to access this table */
10458	Expr pOn; / The ON clause of a join */
	@@ -10568,10 +10587,11 @@
10568	struct WhereInfo {
10569	Parse pParse; / Parsing and code generating context */
10570	u16 wctrlFlags; /* Flags originally passed to sqlite3WhereBegin() */
10571	u8 okOnePass; /* Ok to use one-pass algorithm for UPDATE or DELETE */
10572	u8 untestedTerms; /* Not all WHERE terms resolved by outer loop */

10573	SrcList pTabList; / List of tables in the join */
10574	int iTop; /* The very beginning of the WHERE loop */
10575	int iContinue; /* Jump here to continue with next record */
10576	int iBreak; /* Jump here to break out of the loop */
10577	int nLevel; /* Number of nested loop */
	@@ -10579,10 +10599,13 @@
10579	double savedNQueryLoop; /* pParse->nQueryLoop outside the WHERE loop */
10580	double nRowOut; /* Estimated number of output rows */
10581	WhereLevel a[1]; /* Information about each nest loop in WHERE */
10582	};
10583



10584	/*
10585	** A NameContext defines a context in which to resolve table and column
10586	** names. The context consists of a list of tables (the pSrcList) field and
10587	** a list of named expression (pEList). The named expression list may
10588	** be NULL. The pSrc corresponds to the FROM clause of a SELECT or
	@@ -11340,11 +11363,11 @@
11340	#if defined(SQLITE_ENABLE_UPDATE_DELETE_LIMIT) && !defined(SQLITE_OMIT_SUBQUERY)
11341	SQLITE_PRIVATE Expr sqlite3LimitWhere(Parse , SrcList , Expr , ExprList , Expr , Expr , char );
11342	#endif
11343	SQLITE_PRIVATE void sqlite3DeleteFrom(Parse, SrcList, Expr*);
11344	SQLITE_PRIVATE void sqlite3Update(Parse, SrcList, ExprList, Expr, int);
11345	SQLITE_PRIVATE WhereInfo sqlite3WhereBegin(Parse, SrcList, Expr, ExprList**, u16);
11346	SQLITE_PRIVATE void sqlite3WhereEnd(WhereInfo*);
11347	SQLITE_PRIVATE int sqlite3ExprCodeGetColumn(Parse, Table, int, int, int);
11348	SQLITE_PRIVATE void sqlite3ExprCodeGetColumnOfTable(Vdbe, Table, int, int, int);
11349	SQLITE_PRIVATE void sqlite3ExprCodeMove(Parse*, int, int, int);
11350	SQLITE_PRIVATE void sqlite3ExprCodeCopy(Parse*, int, int, int);
	@@ -15841,11 +15864,11 @@
15841	** Free an outstanding memory allocation.
15842	**
15843	** This function assumes that the necessary mutexes, if any, are
15844	** already held by the caller. Hence "Unsafe".
15845	*/
15846	void memsys3FreeUnsafe(void *pOld){
15847	Mem3Block p = (Mem3Block)pOld;
15848	int i;
15849	u32 size, x;
15850	assert( sqlite3_mutex_held(mem3.mutex) );
15851	assert( p>mem3.aPool && p<&mem3.aPool[mem3.nPool] );
	@@ -15916,21 +15939,21 @@
15916	}
15917
15918	/*
15919	** Free memory.
15920	*/
15921	void memsys3Free(void *pPrior){
15922	assert( pPrior );
15923	memsys3Enter();
15924	memsys3FreeUnsafe(pPrior);
15925	memsys3Leave();
15926	}
15927
15928	/*
15929	** Change the size of an existing memory allocation
15930	*/
15931	void memsys3Realloc(void pPrior, int nBytes){
15932	int nOld;
15933	void *p;
15934	if( pPrior==0 ){
15935	return sqlite3_malloc(nBytes);
15936	}
	@@ -25143,11 +25166,13 @@
25143	case EINVAL:
25144	case ENOTCONN:
25145	case ENODEV:
25146	case ENXIO:
25147	case ENOENT:

25148	case ESTALE:

25149	case ENOSYS:
25150	/* these should force the client to close the file and reconnect */
25151
25152	default:
25153	return sqliteIOErr;
	@@ -31820,10 +31845,58 @@
31820	iLine, iErrno, zFunc, zPath, zMsg
31821	);
31822
31823	return errcode;
31824	}
















































31825
31826	#if SQLITE_OS_WINCE
31827	/*************************************************************************
31828	** This section contains code for WinCE only.
31829	*/
	@@ -32238,22 +32311,25 @@
32238	int amt, /* Number of bytes to read */
32239	sqlite3_int64 offset /* Begin reading at this offset */
32240	){
32241	winFile pFile = (winFile)id; /* file handle */
32242	DWORD nRead; /* Number of bytes actually read from file */

32243
32244	assert( id!=0 );
32245	SimulateIOError(return SQLITE_IOERR_READ);
32246	OSTRACE(("READ %d lock=%d\n", pFile->h, pFile->locktype));
32247
32248	if( seekWinFile(pFile, offset) ){
32249	return SQLITE_FULL;
32250	}
32251	if( !ReadFile(pFile->h, pBuf, amt, &nRead, 0) ){

32252	pFile->lastErrno = GetLastError();
32253	return winLogError(SQLITE_IOERR_READ, "winRead", pFile->zPath);
32254	}

32255	if( nRead<(DWORD)amt ){
32256	/* Unread parts of the buffer must be zero-filled */
32257	memset(&((char*)pBuf)[nRead], 0, amt-nRead);
32258	return SQLITE_IOERR_SHORT_READ;
32259	}
	@@ -32271,10 +32347,11 @@
32271	int amt, /* Number of bytes to write */
32272	sqlite3_int64 offset /* Offset into the file to begin writing at */
32273	){
32274	int rc; /* True if error has occured, else false */
32275	winFile pFile = (winFile)id; /* File handle */

32276
32277	assert( amt>0 );
32278	assert( pFile );
32279	SimulateIOError(return SQLITE_IOERR_WRITE);
32280	SimulateDiskfullError(return SQLITE_FULL);
	@@ -32285,11 +32362,16 @@
32285	if( rc==0 ){
32286	u8 aRem = (u8 )pBuf; /* Data yet to be written */
32287	int nRem = amt; /* Number of bytes yet to be written */
32288	DWORD nWrite; /* Bytes written by each WriteFile() call */
32289
32290	while( nRem>0 && WriteFile(pFile->h, aRem, nRem, &nWrite, 0) && nWrite>0 ){





32291	aRem += nWrite;
32292	nRem -= nWrite;
32293	}
32294	if( nRem>0 ){
32295	pFile->lastErrno = GetLastError();
	@@ -32301,10 +32383,12 @@
32301	if( ( pFile->lastErrno==ERROR_HANDLE_DISK_FULL )
32302	\|\| ( pFile->lastErrno==ERROR_DISK_FULL )){
32303	return SQLITE_FULL;
32304	}
32305	return winLogError(SQLITE_IOERR_WRITE, "winWrite", pFile->zPath);


32306	}
32307	return SQLITE_OK;
32308	}
32309
32310	/*
	@@ -32716,10 +32800,24 @@
32716	SimulateIOErrorBenign(0);
32717	return SQLITE_OK;
32718	}
32719	case SQLITE_FCNTL_SYNC_OMITTED: {
32720	return SQLITE_OK;














32721	}
32722	}
32723	return SQLITE_NOTFOUND;
32724	}
32725
	@@ -33734,19 +33832,17 @@
33734	** file open, we will be unable to delete it. To work around this
33735	** problem, we delay 100 milliseconds and try to delete again. Up
33736	** to MX_DELETION_ATTEMPTs deletion attempts are run before giving
33737	** up and returning an error.
33738	*/
33739	#define MX_DELETION_ATTEMPTS 5
33740	static int winDelete(
33741	sqlite3_vfs pVfs, / Not used on win32 */
33742	const char zFilename, / Name of file to delete */
33743	int syncDir /* Not used on win32 */
33744	){
33745	int cnt = 0;
33746	DWORD rc;
33747	DWORD error = 0;
33748	void *zConverted;
33749	UNUSED_PARAMETER(pVfs);
33750	UNUSED_PARAMETER(syncDir);
33751
33752	SimulateIOError(return SQLITE_IOERR_DELETE);
	@@ -33753,38 +33849,34 @@
33753	zConverted = convertUtf8Filename(zFilename);
33754	if( zConverted==0 ){
33755	return SQLITE_NOMEM;
33756	}
33757	if( isNT() ){
33758	do{
33759	DeleteFileW(zConverted);
33760	}while( ( ((rc = GetFileAttributesW(zConverted)) != INVALID_FILE_ATTRIBUTES)
33761	\|\| ((error = GetLastError()) == ERROR_ACCESS_DENIED))
33762	&& (++cnt < MX_DELETION_ATTEMPTS)
33763	&& (Sleep(100), 1) );
33764	/* isNT() is 1 if SQLITE_OS_WINCE==1, so this else is never executed.
33765	** Since the ASCII version of these Windows API do not exist for WINCE,
33766	** it's important to not reference them for WINCE builds.
33767	*/
33768	#if SQLITE_OS_WINCE==0
33769	}else{
33770	do{
33771	DeleteFileA(zConverted);
33772	}while( ( ((rc = GetFileAttributesA(zConverted)) != INVALID_FILE_ATTRIBUTES)
33773	\|\| ((error = GetLastError()) == ERROR_ACCESS_DENIED))
33774	&& (++cnt < MX_DELETION_ATTEMPTS)
33775	&& (Sleep(100), 1) );
33776	#endif
33777	}





33778	free(zConverted);
33779	OSTRACE(("DELETE \"%s\" %s\n", zFilename,
33780	( (rc==INVALID_FILE_ATTRIBUTES) && (error==ERROR_FILE_NOT_FOUND)) ?
33781	"ok" : "failed" ));
33782
33783	return ( (rc == INVALID_FILE_ATTRIBUTES)
33784	&& (error == ERROR_FILE_NOT_FOUND)) ? SQLITE_OK :
33785	winLogError(SQLITE_IOERR_DELETE, "winDelete", zFilename);
33786	}
33787
33788	/*
33789	** Check the existance and status of a file.
33790	*/
	@@ -59049,10 +59141,11 @@
59049	nMem = 10;
59050	}
59051	memset(zCsr, 0, zEnd-zCsr);
59052	zCsr += (zCsr - (u8*)0)&7;
59053	assert( EIGHT_BYTE_ALIGNMENT(zCsr) );

59054
59055	/* Memory for registers, parameters, cursor, etc, is allocated in two
59056	** passes. On the first pass, we try to reuse unused space at the
59057	** end of the opcode array. If we are unable to satisfy all memory
59058	** requirements by reusing the opcode array tail, then the second
	@@ -61277,11 +61370,11 @@
61277	sqlite3_mutex_enter(db->mutex);
61278	while( (rc = sqlite3Step(v))==SQLITE_SCHEMA
61279	&& cnt++ < SQLITE_MAX_SCHEMA_RETRY
61280	&& (rc2 = rc = sqlite3Reprepare(v))==SQLITE_OK ){
61281	sqlite3_reset(pStmt);
61282	v->expired = 0;
61283	}
61284	if( rc2!=SQLITE_OK && ALWAYS(v->isPrepareV2) && ALWAYS(db->pErr) ){
61285	/* This case occurs after failing to recompile an sql statement.
61286	** The error message from the SQL compiler has already been loaded
61287	** into the database handle. This block copies the error message
	@@ -65968,11 +66061,11 @@
65968	** automatically created table with root-page 1 (an BLOB_INTKEY table).
65969	*/
65970	if( pOp->p4.pKeyInfo ){
65971	int pgno;
65972	assert( pOp->p4type==P4_KEYINFO );
65973	rc = sqlite3BtreeCreateTable(u.ax.pCx->pBt, &pgno, BTREE_BLOBKEY);
65974	if( rc==SQLITE_OK ){
65975	assert( pgno==MASTER_ROOT+1 );
65976	rc = sqlite3BtreeCursor(u.ax.pCx->pBt, pgno, 1,
65977	(KeyInfo*)pOp->p4.z, u.ax.pCx->pCursor);
65978	u.ax.pCx->pKeyInfo = pOp->p4.pKeyInfo;
	@@ -71007,15 +71100,29 @@
71007	/* Recursively resolve names in all subqueries
71008	*/
71009	for(i=0; i<p->pSrc->nSrc; i++){
71010	struct SrcList_item *pItem = &p->pSrc->a[i];
71011	if( pItem->pSelect ){


71012	const char *zSavedContext = pParse->zAuthContext;








71013	if( pItem->zName ) pParse->zAuthContext = pItem->zName;
71014	sqlite3ResolveSelectNames(pParse, pItem->pSelect, pOuterNC);
71015	pParse->zAuthContext = zSavedContext;
71016	if( pParse->nErr \|\| db->mallocFailed ) return WRC_Abort;




71017	}
71018	}
71019
71020	/* If there are no aggregate functions in the result-set, and no GROUP BY
71021	** expression, do not allow aggregates in any of the other expressions.
	@@ -72119,10 +72226,11 @@
72119	pNewItem->zName = sqlite3DbStrDup(db, pOldItem->zName);
72120	pNewItem->zAlias = sqlite3DbStrDup(db, pOldItem->zAlias);
72121	pNewItem->jointype = pOldItem->jointype;
72122	pNewItem->iCursor = pOldItem->iCursor;
72123	pNewItem->isPopulated = pOldItem->isPopulated;

72124	pNewItem->zIndex = sqlite3DbStrDup(db, pOldItem->zIndex);
72125	pNewItem->notIndexed = pOldItem->notIndexed;
72126	pNewItem->pIndex = pOldItem->pIndex;
72127	pTab = pNewItem->pTab = pOldItem->pTab;
72128	if( pTab ){
	@@ -80126,11 +80234,11 @@
80126	if( pStart ){
80127	assert( pEnd!=0 );
80128	/* A named index with an explicit CREATE INDEX statement */
80129	zStmt = sqlite3MPrintf(db, "CREATE%s INDEX %.*s",
80130	onError==OE_None ? "" : " UNIQUE",
80131	pEnd->z - pName->z + 1,
80132	pName->z);
80133	}else{
80134	/* An automatic index created by a PRIMARY KEY or UNIQUE constraint */
80135	/* zStmt = sqlite3MPrintf(""); */
80136	zStmt = 0;
	@@ -81921,11 +82029,13 @@
81921	int regRowid; /* Actual register containing rowids */
81922
81923	/* Collect rowids of every row to be deleted.
81924	*/
81925	sqlite3VdbeAddOp2(v, OP_Null, 0, iRowSet);
81926	pWInfo = sqlite3WhereBegin(pParse, pTabList, pWhere,0,WHERE_DUPLICATES_OK);


81927	if( pWInfo==0 ) goto delete_from_cleanup;
81928	regRowid = sqlite3ExprCodeGetColumn(pParse, pTab, -1, iCur, iRowid);
81929	sqlite3VdbeAddOp2(v, OP_RowSetAdd, iRowSet, regRowid);
81930	if( db->flags & SQLITE_CountRows ){
81931	sqlite3VdbeAddOp2(v, OP_AddImm, memCnt, 1);
	@@ -84368,11 +84478,11 @@
84368
84369	/* Create VDBE to loop through the entries in pSrc that match the WHERE
84370	** clause. If the constraint is not deferred, throw an exception for
84371	** each row found. Otherwise, for deferred constraints, increment the
84372	** deferred constraint counter by nIncr for each row selected. */
84373	pWInfo = sqlite3WhereBegin(pParse, pSrc, pWhere, 0, 0);
84374	if( nIncr>0 && pFKey->isDeferred==0 ){
84375	sqlite3ParseToplevel(pParse)->mayAbort = 1;
84376	}
84377	sqlite3VdbeAddOp2(v, OP_FkCounter, pFKey->isDeferred, nIncr);
84378	if( pWInfo ){
	@@ -87235,10 +87345,13 @@
87235	int (strnicmp)(const char,const char*,int);
87236	int (unlock_notify)(sqlite3,void()(void,int),void);
87237	int (wal_autocheckpoint)(sqlite3,int);
87238	int (wal_checkpoint)(sqlite3,const char*);
87239	void (wal_hook)(sqlite3,int()(void,sqlite3,const char,int),void);



87240	};
87241
87242	/*
87243	** The following macros redefine the API routines so that they are
87244	** redirected throught the global sqlite3_api structure.
	@@ -87435,10 +87548,13 @@
87435	#define sqlite3_strnicmp sqlite3_api->strnicmp
87436	#define sqlite3_unlock_notify sqlite3_api->unlock_notify
87437	#define sqlite3_wal_autocheckpoint sqlite3_api->wal_autocheckpoint
87438	#define sqlite3_wal_checkpoint sqlite3_api->wal_checkpoint
87439	#define sqlite3_wal_hook sqlite3_api->wal_hook



87440	#endif /* SQLITE_CORE */
87441
87442	#define SQLITE_EXTENSION_INIT1 const sqlite3_api_routines *sqlite3_api = 0;
87443	#define SQLITE_EXTENSION_INIT2(v) sqlite3_api = v;
87444
	@@ -87509,10 +87625,12 @@
87509
87510	#ifdef SQLITE_OMIT_VIRTUALTABLE
87511	# define sqlite3_create_module 0
87512	# define sqlite3_create_module_v2 0
87513	# define sqlite3_declare_vtab 0


87514	#endif
87515
87516	#ifdef SQLITE_OMIT_SHARED_CACHE
87517	# define sqlite3_enable_shared_cache 0
87518	#endif
	@@ -87532,10 +87650,11 @@
87532	#define sqlite3_blob_bytes 0
87533	#define sqlite3_blob_close 0
87534	#define sqlite3_blob_open 0
87535	#define sqlite3_blob_read 0
87536	#define sqlite3_blob_write 0

87537	#endif
87538
87539	/*
87540	** The following structure contains pointers to all SQLite API routines.
87541	** A pointer to this structure is passed into extensions when they are
	@@ -87797,10 +87916,13 @@
87797	#else
87798	0,
87799	0,
87800	0,
87801	#endif



87802	};
87803
87804	/*
87805	** Attempt to load an SQLite extension library contained in the file
87806	** zFile. The entry point is zProc. zProc may be 0 in which case a
	@@ -94186,10 +94308,11 @@
94186	Expr pHaving; / The HAVING clause. May be NULL */
94187	int isDistinct; /* True if the DISTINCT keyword is present */
94188	int distinct; /* Table to use for the distinct set */
94189	int rc = 1; /* Value to return from this function */
94190	int addrSortIndex; /* Address of an OP_OpenEphemeral instruction */

94191	AggInfo sAggInfo; /* Information used by aggregate queries */
94192	int iEnd; /* Address of the end of the query */
94193	sqlite3 db; / The database connection */
94194
94195	#ifndef SQLITE_OMIT_EXPLAIN
	@@ -94312,20 +94435,10 @@
94312	explainSetInteger(pParse->iSelectId, iRestoreSelectId);
94313	return rc;
94314	}
94315	#endif
94316
94317	/* If possible, rewrite the query to use GROUP BY instead of DISTINCT.
94318	** GROUP BY might use an index, DISTINCT never does.
94319	*/
94320	assert( p->pGroupBy==0 \|\| (p->selFlags & SF_Aggregate)!=0 );
94321	if( (p->selFlags & (SF_Distinct\|SF_Aggregate))==SF_Distinct ){
94322	p->pGroupBy = sqlite3ExprListDup(db, p->pEList, 0);
94323	pGroupBy = p->pGroupBy;
94324	p->selFlags &= ~SF_Distinct;
94325	}
94326
94327	/* If there is both a GROUP BY and an ORDER BY clause and they are
94328	** identical, then disable the ORDER BY clause since the GROUP BY
94329	** will cause elements to come out in the correct order. This is
94330	** an optimization - the correct answer should result regardless.
94331	** Use the SQLITE_GroupByOrder flag with SQLITE_TESTCTRL_OPTIMIZER
	@@ -94333,10 +94446,34 @@
94333	*/
94334	if( sqlite3ExprListCompare(p->pGroupBy, pOrderBy)==0
94335	&& (db->flags & SQLITE_GroupByOrder)==0 ){
94336	pOrderBy = 0;
94337	}
























94338
94339	/* If there is an ORDER BY clause, then this sorting
94340	** index might end up being unused if the data can be
94341	** extracted in pre-sorted order. If that is the case, then the
94342	** OP_OpenEphemeral instruction will be changed to an OP_Noop once
	@@ -94369,26 +94506,25 @@
94369
94370	/* Open a virtual index to use for the distinct set.
94371	*/
94372	if( p->selFlags & SF_Distinct ){
94373	KeyInfo *pKeyInfo;
94374	assert( isAgg \|\| pGroupBy );
94375	distinct = pParse->nTab++;
94376	pKeyInfo = keyInfoFromExprList(pParse, p->pEList);
94377	sqlite3VdbeAddOp4(v, OP_OpenEphemeral, distinct, 0, 0,
94378	(char*)pKeyInfo, P4_KEYINFO_HANDOFF);
94379	sqlite3VdbeChangeP5(v, BTREE_UNORDERED);
94380	}else{
94381	distinct = -1;
94382	}
94383
94384	/* Aggregate and non-aggregate queries are handled differently */
94385	if( !isAgg && pGroupBy==0 ){
94386	/* This case is for non-aggregate queries
94387	** Begin the database scan
94388	*/
94389	pWInfo = sqlite3WhereBegin(pParse, pTabList, pWhere, &pOrderBy, 0);
94390	if( pWInfo==0 ) goto select_end;
94391	if( pWInfo->nRowOut < p->nSelectRow ) p->nSelectRow = pWInfo->nRowOut;
94392
94393	/* If sorting index that was created by a prior OP_OpenEphemeral
94394	** instruction ended up not being needed, then change the OP_OpenEphemeral
	@@ -94397,14 +94533,56 @@
94397	if( addrSortIndex>=0 && pOrderBy==0 ){
94398	sqlite3VdbeChangeToNoop(v, addrSortIndex, 1);
94399	p->addrOpenEphm[2] = -1;
94400	}
94401
94402	/* Use the standard inner loop
94403	*/
94404	assert(!isDistinct);
94405	selectInnerLoop(pParse, p, pEList, 0, 0, pOrderBy, -1, pDest,










































94406	pWInfo->iContinue, pWInfo->iBreak);
94407
94408	/* End the database scan loop.
94409	*/
94410	sqlite3WhereEnd(pWInfo);
	@@ -94510,11 +94688,11 @@
94510	** This might involve two separate loops with an OP_Sort in between, or
94511	** it might be a single loop that uses an index to extract information
94512	** in the right order to begin with.
94513	*/
94514	sqlite3VdbeAddOp2(v, OP_Gosub, regReset, addrReset);
94515	pWInfo = sqlite3WhereBegin(pParse, pTabList, pWhere, &pGroupBy, 0);
94516	if( pWInfo==0 ) goto select_end;
94517	if( pGroupBy==0 ){
94518	/* The optimizer is able to deliver rows in group by order so
94519	** we do not have to sort. The OP_OpenEphemeral table will be
94520	** cancelled later because we still need to use the pKeyInfo
	@@ -94772,11 +94950,11 @@
94772	/* This case runs if the aggregate has no GROUP BY clause. The
94773	** processing is much simpler since there is only a single row
94774	** of output.
94775	*/
94776	resetAccumulator(pParse, &sAggInfo);
94777	pWInfo = sqlite3WhereBegin(pParse, pTabList, pWhere, &pMinMax, flag);
94778	if( pWInfo==0 ){
94779	sqlite3ExprListDelete(db, pDel);
94780	goto select_end;
94781	}
94782	updateAccumulator(pParse, &sAggInfo);
	@@ -95248,19 +95426,32 @@
95248	iDb = sqlite3TwoPartName(pParse, pName1, pName2, &pName);
95249	if( iDb<0 ){
95250	goto trigger_cleanup;
95251	}
95252	}
















95253
95254	/* If the trigger name was unqualified, and the table is a temp table,
95255	** then set iDb to 1 to create the trigger in the temporary database.
95256	** If sqlite3SrcListLookup() returns 0, indicating the table does not
95257	** exist, the error is caught by the block below.
95258	*/
95259	if( !pTableName \|\| db->mallocFailed ){
95260	goto trigger_cleanup;
95261	}
95262	pTab = sqlite3SrcListLookup(pParse, pTableName);
95263	if( db->init.busy==0 && pName2->n==0 && pTab
95264	&& pTab->pSchema==db->aDb[1].pSchema ){
95265	iDb = 1;
95266	}
	@@ -96554,11 +96745,13 @@
96554	}
96555
96556	/* Begin the database scan
96557	*/
96558	sqlite3VdbeAddOp2(v, OP_Null, 0, regOldRowid);
96559	pWInfo = sqlite3WhereBegin(pParse, pTabList, pWhere,0, WHERE_ONEPASS_DESIRED);


96560	if( pWInfo==0 ) goto update_cleanup;
96561	okOnePass = pWInfo->okOnePass;
96562
96563	/* Remember the rowid of every item to be updated.
96564	*/
	@@ -98580,10 +98773,11 @@
98580	#define WHERE_REVERSE 0x02000000 /* Scan in reverse order */
98581	#define WHERE_UNIQUE 0x04000000 /* Selects no more than one row */
98582	#define WHERE_VIRTUALTABLE 0x08000000 /* Use virtual-table processing */
98583	#define WHERE_MULTI_OR 0x10000000 /* OR using multiple indices */
98584	#define WHERE_TEMP_INDEX 0x20000000 /* Uses an ephemeral index */

98585
98586	/*
98587	** Initialize a preallocated WhereClause structure.
98588	*/
98589	static void whereClauseInit(
	@@ -99724,10 +99918,166 @@
99724	}
99725	}
99726	return 0;
99727	}
99728




























































































































































99729
99730	/*
99731	** This routine decides if pIdx can be used to satisfy the ORDER BY
99732	** clause. If it can, it returns 1. If pIdx cannot satisfy the
99733	** ORDER BY clause, this routine returns 0.
	@@ -99760,11 +100110,14 @@
99760	int sortOrder = 0; /* XOR of index and ORDER BY sort direction */
99761	int nTerm; /* Number of ORDER BY terms */
99762	struct ExprList_item pTerm; / A term of the ORDER BY clause */
99763	sqlite3 *db = pParse->db;
99764
99765	assert( pOrderBy!=0 );



99766	nTerm = pOrderBy->nExpr;
99767	assert( nTerm>0 );
99768
99769	/* Argument pIdx must either point to a 'real' named index structure,
99770	** or an index structure allocated on the stack by bestBtreeIndex() to
	@@ -100073,10 +100426,14 @@
100073	double costTempIdx; /* per-query cost of the transient index */
100074	WhereTerm pTerm; / A single term of the WHERE clause */
100075	WhereTerm pWCEnd; / End of pWC->a[] */
100076	Table pTable; / Table tht might be indexed */
100077




100078	if( (pParse->db->flags & SQLITE_AutoIndex)==0 ){
100079	/* Automatic indices are disabled at run-time */
100080	return;
100081	}
100082	if( (pCost->plan.wsFlags & WHERE_NOT_FULLSCAN)!=0 ){
	@@ -100084,10 +100441,14 @@
100084	return;
100085	}
100086	if( pSrc->notIndexed ){
100087	/* The NOT INDEXED clause appears in the SQL. */
100088	return;




100089	}
100090
100091	assert( pParse->nQueryLoop >= (double)1 );
100092	pTable = pSrc->pTab;
100093	nTableRow = pTable->nRowEst;
	@@ -101016,10 +101377,11 @@
101016	WhereClause pWC, / The WHERE clause */
101017	struct SrcList_item pSrc, / The FROM clause term to search */
101018	Bitmask notReady, /* Mask of cursors not available for indexing */
101019	Bitmask notValid, /* Cursors not available for any purpose */
101020	ExprList pOrderBy, / The ORDER BY clause */

101021	WhereCost pCost / Lowest cost query plan */
101022	){
101023	int iCur = pSrc->iCursor; /* The cursor of the table to be accessed */
101024	Index pProbe; / An index we are evaluating */
101025	Index pIdx; / Copy of pProbe, or zero for IPK index */
	@@ -101156,11 +101518,12 @@
101156	int nEq; /* Number of == or IN terms matching index */
101157	int bInEst = 0; /* True if "x IN (SELECT...)" seen */
101158	int nInMul = 1; /* Number of distinct equalities to lookup */
101159	int estBound = 100; /* Estimated reduction in search space */
101160	int nBound = 0; /* Number of range constraints seen */
101161	int bSort = 0; /* True if external sort required */

101162	int bLookup = 0; /* True if not a covering index */
101163	WhereTerm pTerm; / A single term of the WHERE clause */
101164	#ifdef SQLITE_ENABLE_STAT2
101165	WhereTerm pFirstTerm = 0; / First term matching the index */
101166	#endif
	@@ -101220,21 +101583,24 @@
101220
101221	/* If there is an ORDER BY clause and the index being considered will
101222	** naturally scan rows in the required order, set the appropriate flags
101223	** in wsFlags. Otherwise, if there is an ORDER BY clause but the index
101224	** will scan rows in a different order, set the bSort variable. */
101225	if( pOrderBy ){
101226	if( (wsFlags & WHERE_COLUMN_IN)==0
101227	&& pProbe->bUnordered==0
101228	&& isSortingIndex(pParse, pWC->pMaskSet, pProbe, iCur, pOrderBy,
101229	nEq, wsFlags, &rev)
101230	){
101231	wsFlags \|= WHERE_ROWID_RANGE\|WHERE_COLUMN_RANGE\|WHERE_ORDERBY;
101232	wsFlags \|= (rev ? WHERE_REVERSE : 0);
101233	}else{
101234	bSort = 1;
101235	}



101236	}
101237
101238	/* If currently calculating the cost of using an index (not the IPK
101239	** index), determine if all required column data may be obtained without
101240	** using the main table (i.e. if the index is a covering
	@@ -101265,16 +101631,17 @@
101265	nRow = aiRowEst[0]/2;
101266	nInMul = (int)(nRow / aiRowEst[nEq]);
101267	}
101268
101269	#ifdef SQLITE_ENABLE_STAT2
101270	/* If the constraint is of the form x=VALUE and histogram

101271	** data is available for column x, then it might be possible
101272	** to get a better estimate on the number of rows based on
101273	** VALUE and how common that value is according to the histogram.
101274	*/
101275	if( nRow>(double)1 && nEq==1 && pFirstTerm!=0 ){
101276	if( pFirstTerm->eOperator & (WO_EQ\|WO_ISNULL) ){
101277	testcase( pFirstTerm->eOperator==WO_EQ );
101278	testcase( pFirstTerm->eOperator==WO_ISNULL );
101279	whereEqualScanEst(pParse, pProbe, pFirstTerm->pExpr->pRight, &nRow);
101280	}else if( pFirstTerm->eOperator==WO_IN && bInEst==0 ){
	@@ -101347,10 +101714,13 @@
101347	** difference and select C of 3.0.
101348	*/
101349	if( bSort ){
101350	cost += nRowestLog(nRow)3;
101351	}



101352
101353	/** Cost of using this index has now been computed **/
101354
101355	/* If there are additional constraints on this table that cannot
101356	** be used with the current index, but which might lower the number
	@@ -101492,11 +101862,11 @@
101492	}
101493	sqlite3DbFree(pParse->db, p);
101494	}else
101495	#endif
101496	{
101497	bestBtreeIndex(pParse, pWC, pSrc, notReady, notValid, pOrderBy, pCost);
101498	}
101499	}
101500
101501	/*
101502	** Disable a term in the WHERE clause. Except, do not disable the term
	@@ -102454,11 +102824,11 @@
102454	for(ii=0; ii<pOrWc->nTerm; ii++){
102455	WhereTerm *pOrTerm = &pOrWc->a[ii];
102456	if( pOrTerm->leftCursor==iCur \|\| pOrTerm->eOperator==WO_AND ){
102457	WhereInfo pSubWInfo; / Info for single OR-term scan */
102458	/* Loop through table entries that match term pOrTerm. */
102459	pSubWInfo = sqlite3WhereBegin(pParse, pOrTab, pOrTerm->pExpr, 0,
102460	WHERE_OMIT_OPEN \| WHERE_OMIT_CLOSE \|
102461	WHERE_FORCE_TABLE \| WHERE_ONETABLE_ONLY);
102462	if( pSubWInfo ){
102463	explainOneScan(
102464	pParse, pOrTab, &pSubWInfo->a[0], iLevel, pLevel->iFrom, 0
	@@ -102695,10 +103065,11 @@
102695	SQLITE_PRIVATE WhereInfo *sqlite3WhereBegin(
102696	Parse pParse, / The parser context */
102697	SrcList pTabList, / A list of all tables to be scanned */
102698	Expr pWhere, / The WHERE clause */
102699	ExprList *ppOrderBy, / An ORDER BY clause, or NULL */

102700	u16 wctrlFlags /* One of the WHERE_* flags defined in sqliteInt.h */
102701	){
102702	int i; /* Loop counter */
102703	int nByteWInfo; /* Num. bytes allocated for WhereInfo struct */
102704	int nTabList; /* Number of elements in pTabList */
	@@ -102754,10 +103125,14 @@
102754	pWInfo->iBreak = sqlite3VdbeMakeLabel(v);
102755	pWInfo->pWC = pWC = (WhereClause )&((u8 )pWInfo)[nByteWInfo];
102756	pWInfo->wctrlFlags = wctrlFlags;
102757	pWInfo->savedNQueryLoop = pParse->nQueryLoop;
102758	pMaskSet = (WhereMaskSet*)&pWC[1];




102759
102760	/* Split the WHERE clause into separate subexpressions where each
102761	** subexpression is separated by an AND operator.
102762	*/
102763	initMaskSet(pMaskSet);
	@@ -102821,10 +103196,19 @@
102821	*/
102822	exprAnalyzeAll(pTabList, pWC);
102823	if( db->mallocFailed ){
102824	goto whereBeginError;
102825	}









102826
102827	/* Chose the best index to use for each table in the FROM clause.
102828	**
102829	** This loop fills in the following fields:
102830	**
	@@ -102905,10 +103289,11 @@
102905	Bitmask mask; /* Mask of tables not yet ready */
102906	for(j=iFrom, pTabItem=&pTabList->a[j]; j<nTabList; j++, pTabItem++){
102907	int doNotReorder; /* True if this table should not be reordered */
102908	WhereCost sCost; /* Cost information from best[Virtual]Index() */
102909	ExprList pOrderBy; / ORDER BY clause for index to optimize */

102910
102911	doNotReorder = (pTabItem->jointype & (JT_LEFT\|JT_CROSS))!=0;
102912	if( j!=iFrom && doNotReorder ) break;
102913	m = getMask(pMaskSet, pTabItem->iCursor);
102914	if( (m & notReady)==0 ){
	@@ -102915,10 +103300,11 @@
102915	if( j==iFrom ) iFrom++;
102916	continue;
102917	}
102918	mask = (isOptimal ? m : notReady);
102919	pOrderBy = ((i==0 && ppOrderBy )?*ppOrderBy:0);

102920	if( pTabItem->pIndex==0 ) nUnconstrained++;
102921
102922	WHERETRACE(("=== trying table %d with isOptimal=%d ===\n",
102923	j, isOptimal));
102924	assert( pTabItem->pTab );
	@@ -102929,11 +103315,11 @@
102929	&sCost, pp);
102930	}else
102931	#endif
102932	{
102933	bestBtreeIndex(pParse, pWC, pTabItem, mask, notReady, pOrderBy,
102934	&sCost);
102935	}
102936	assert( isOptimal \|\| (sCost.used&notReady)==0 );
102937
102938	/* If an INDEXED BY clause is present, then the plan must use that
102939	** index if it uses any index at all */
	@@ -102989,10 +103375,14 @@
102989	WHERETRACE(("*** Optimizer selects table %d for loop %d"
102990	" with cost=%g and nRow=%g\n",
102991	bestJ, pLevel-pWInfo->a, bestPlan.rCost, bestPlan.plan.nRow));
102992	if( (bestPlan.plan.wsFlags & WHERE_ORDERBY)!=0 ){
102993	*ppOrderBy = 0;




102994	}
102995	andFlags &= bestPlan.plan.wsFlags;
102996	pLevel->plan = bestPlan.plan;
102997	testcase( bestPlan.plan.wsFlags & WHERE_INDEXED );
102998	testcase( bestPlan.plan.wsFlags & WHERE_TEMP_INDEX );
	@@ -111519,11 +111909,17 @@
111519	*/
111520	#if defined(SQLITE_ENABLE_FTS4) && !defined(SQLITE_ENABLE_FTS3)
111521	# define SQLITE_ENABLE_FTS3
111522	#endif
111523
111524	#ifdef SQLITE_ENABLE_FTS3






111525	/************ Include fts3_tokenizer.h in the middle of fts3Int.h ********/
111526	/************ Begin file fts3_tokenizer.h ********************************/
111527	/*
111528	** 2006 July 10
111529	**
	@@ -112052,11 +112448,11 @@
112052
112053	sqlite3_int64 iDocid; /* Current docid (if pList!=0) */
112054	int bFreeList; /* True if pList should be sqlite3_free()d */
112055	char pList; / Pointer to position list following iDocid */
112056	int nList; /* Length of position list */
112057	} doclist;
112058
112059	/*
112060	** A "phrase" is a sequence of one or more tokens that must match in
112061	** sequence. A single token is the base case and the most common case.
112062	** For a sequence of tokens contained in double-quotes (i.e. "one two three")
	@@ -112252,23 +112648,12 @@
112252	#endif
112253
112254	/* fts3_aux.c */
112255	SQLITE_PRIVATE int sqlite3Fts3InitAux(sqlite3 *db);
112256
112257	SQLITE_PRIVATE int sqlite3Fts3TermSegReaderCursor(
112258	Fts3Cursor pCsr, / Virtual table cursor handle */
112259	const char zTerm, / Term to query for */
112260	int nTerm, /* Size of zTerm in bytes */
112261	int isPrefix, /* True for a prefix search */
112262	Fts3MultiSegReader *ppSegcsr / OUT: Allocated seg-reader cursor */
112263	);
112264
112265	SQLITE_PRIVATE void sqlite3Fts3EvalPhraseCleanup(Fts3Phrase *);
112266
112267	SQLITE_PRIVATE int sqlite3Fts3EvalStart(Fts3Cursor , Fts3Expr , int);
112268	SQLITE_PRIVATE int sqlite3Fts3EvalNext(Fts3Cursor *pCsr);
112269
112270	SQLITE_PRIVATE int sqlite3Fts3MsrIncrStart(
112271	Fts3Table, Fts3MultiSegReader, int, const char*, int);
112272	SQLITE_PRIVATE int sqlite3Fts3MsrIncrNext(
112273	Fts3Table , Fts3MultiSegReader , sqlite3_int64 , char , int );
112274	SQLITE_PRIVATE char sqlite3Fts3EvalPhrasePoslist(Fts3Cursor , Fts3Expr *, int iCol);
	@@ -112275,11 +112660,11 @@
112275	SQLITE_PRIVATE int sqlite3Fts3MsrOvfl(Fts3Cursor , Fts3MultiSegReader , int *);
112276	SQLITE_PRIVATE int sqlite3Fts3MsrIncrRestart(Fts3MultiSegReader *pCsr);
112277
112278	SQLITE_PRIVATE int sqlite3Fts3DeferredTokenList(Fts3DeferredToken , char , int );
112279
112280	#endif /* SQLITE_ENABLE_FTS3 */
112281	#endif /* _FTSINT_H */
112282
112283	/************ End of fts3Int.h *******************************************/
112284	/************ Continuing where we left off in fts3.c *********************/
112285	#if !defined(SQLITE_CORE) \|\| defined(SQLITE_ENABLE_FTS3)
	@@ -112291,10 +112676,15 @@
112291
112292	#ifndef SQLITE_CORE
112293	SQLITE_EXTENSION_INIT1
112294	#endif
112295





112296	/*
112297	** Write a 64-bit variable-length integer to memory starting at p[0].
112298	** The length of data written will be between 1 and FTS3_VARINT_MAX bytes.
112299	** The number of bytes written is returned.
112300	*/
	@@ -112799,31 +113189,60 @@
112799	}
112800	sqlite3_free(zFree);
112801	return zRet;
112802	}
112803













112804	static int fts3GobbleInt(const char *pp, int pnOut){
112805	const char p = pp;
112806	int nInt = 0;

112807	for(p=*pp; p[0]>='0' && p[0]<='9'; p++){
112808	nInt = nInt * 10 + (p[0] - '0');
112809	}
112810	if( p==*pp ) return SQLITE_ERROR;
112811	*pnOut = nInt;
112812	*pp = p;
112813	return SQLITE_OK;
112814	}
112815
112816
















112817	static int fts3PrefixParameter(
112818	const char zParam, / ABC in prefix=ABC parameter to parse */
112819	int pnIndex, / OUT: size of apIndex[] array /
112820	struct Fts3Index *apIndex, / OUT: Array of indexes for this table */
112821	struct Fts3Index *apFree / OUT: Free this with sqlite3_free() */
112822	){
112823	struct Fts3Index *aIndex;
112824	int nIndex = 1;
112825
112826	if( zParam && zParam[0] ){
112827	const char *p;
112828	nIndex++;
112829	for(p=zParam; *p; p++){
	@@ -112830,11 +113249,11 @@
112830	if( *p==',' ) nIndex++;
112831	}
112832	}
112833
112834	aIndex = sqlite3_malloc(sizeof(struct Fts3Index) * nIndex);
112835	apIndex = apFree = aIndex;
112836	*pnIndex = nIndex;
112837	if( !aIndex ){
112838	return SQLITE_NOMEM;
112839	}
112840
	@@ -112887,12 +113306,11 @@
112887	int isFts4 = (argv[0][3]=='4'); /* True for FTS4, false for FTS3 */
112888	const char *aCol; / Array of column names */
112889	sqlite3_tokenizer pTokenizer = 0; / Tokenizer for this table */
112890
112891	int nIndex; /* Size of aIndex[] array */
112892	struct Fts3Index aIndex; / Array of indexes for this table */
112893	struct Fts3Index aFree = 0; / Free this before returning */
112894
112895	/* The results of parsing supported FTS4 key=value options: */
112896	int bNoDocsize = 0; /* True to omit %_docsize table */
112897	int bDescIdx = 0; /* True to store descending indexes */
112898	char zPrefix = 0; / Prefix parameter value (or NULL) */
	@@ -113025,11 +113443,11 @@
113025	rc = sqlite3Fts3InitTokenizer(pHash, "simple", &pTokenizer, pzErr);
113026	if( rc!=SQLITE_OK ) goto fts3_init_out;
113027	}
113028	assert( pTokenizer );
113029
113030	rc = fts3PrefixParameter(zPrefix, &nIndex, &aIndex, &aFree);
113031	if( rc==SQLITE_ERROR ){
113032	assert( zPrefix );
113033	*pzErr = sqlite3_mprintf("error parsing prefix parameter: %s", zPrefix);
113034	}
113035	if( rc!=SQLITE_OK ) goto fts3_init_out;
	@@ -113112,11 +113530,11 @@
113112	/* Declare the table schema to SQLite. */
113113	fts3DeclareVtab(&rc, p);
113114
113115	fts3_init_out:
113116	sqlite3_free(zPrefix);
113117	sqlite3_free(aFree);
113118	sqlite3_free(zCompress);
113119	sqlite3_free(zUncompress);
113120	sqlite3_free((void *)aCol);
113121	if( rc!=SQLITE_OK ){
113122	if( p ){
	@@ -113703,12 +114121,10 @@
113703	*pp1 = p1 + 1;
113704	*pp2 = p2 + 1;
113705	}
113706
113707	/*
113708	** nToken==1 searches for adjacent positions.
113709	**
113710	** This function is used to merge two position lists into one. When it is
113711	** called, pp1 and pp2 must both point to position lists. A position-list is
113712	** the part of a doclist that follows each document id. For example, if a row
113713	** contains:
113714	**
	@@ -113724,10 +114140,12 @@
113724	** If isSaveLeft is 0, an entry is added to the output position list for
113725	** each position in *pp2 for which there exists one or more positions in
113726	** pp1 so that (pos(pp2)>pos(pp1) && pos(pp2)-pos(*pp1)<=nToken). i.e.
113727	** when the pp1 token appears before the pp2 token, but not more than nToken
113728	** slots before it.


113729	*/
113730	static int fts3PoslistPhraseMerge(
113731	char *pp, / IN/OUT: Preallocated output buffer */
113732	int nToken, /* Maximum difference in token positions */
113733	int isSaveLeft, /* Save the left position */
	@@ -113890,26 +114308,38 @@
113890
113891	return res;
113892	}
113893
113894	/*
113895	** A pointer to an instance of this structure is used as the context
113896	** argument to sqlite3Fts3SegReaderIterate()

113897	*/
113898	typedef struct TermSelect TermSelect;
113899	struct TermSelect {
113900	int isReqPos;
113901	char aaOutput[16]; / Malloc'd output buffer */
113902	int anOutput[16]; /* Size of output in bytes */
113903	};
113904
113905












113906	static void fts3GetDeltaVarint3(
113907	char **pp,
113908	char *pEnd,
113909	int bDescIdx,
113910	sqlite3_int64 *pVal
113911	){
113912	if( *pp>=pEnd ){
113913	*pp = 0;
113914	}else{
113915	sqlite3_int64 iVal;
	@@ -113920,10 +114350,25 @@
113920	*pVal += iVal;
113921	}
113922	}
113923	}
113924















113925	static void fts3PutDeltaVarint3(
113926	char *pp, / IN/OUT: Output pointer */
113927	int bDescIdx, /* True for descending docids */
113928	sqlite3_int64 piPrev, / IN/OUT: Previous value written to list */
113929	int pbFirst, / IN/OUT: True after first int written */
	@@ -113940,14 +114385,38 @@
113940	pp += sqlite3Fts3PutVarint(pp, iWrite);
113941	*piPrev = iVal;
113942	*pbFirst = 1;
113943	}
113944
113945	#define COMPARE_DOCID(i1, i2) ((bDescIdx?-1:1) * (i1-i2))
113946

























113947	static int fts3DoclistOrMerge(
113948	int bDescIdx, /* True if arguments are desc */
113949	char a1, int n1, / First doclist */
113950	char a2, int n2, / Second doclist */
113951	char *paOut, int pnOut /* OUT: Malloc'd doclist */
113952	){
113953	sqlite3_int64 i1 = 0;
	@@ -113968,35 +114437,47 @@
113968
113969	p = aOut;
113970	fts3GetDeltaVarint3(&p1, pEnd1, 0, &i1);
113971	fts3GetDeltaVarint3(&p2, pEnd2, 0, &i2);
113972	while( p1 \|\| p2 ){
113973	sqlite3_int64 iDiff = COMPARE_DOCID(i1, i2);
113974
113975	if( p2 && p1 && iDiff==0 ){
113976	fts3PutDeltaVarint3(&p, bDescIdx, &iPrev, &bFirstOut, i1);
113977	fts3PoslistMerge(&p, &p1, &p2);
113978	fts3GetDeltaVarint3(&p1, pEnd1, bDescIdx, &i1);
113979	fts3GetDeltaVarint3(&p2, pEnd2, bDescIdx, &i2);
113980	}else if( !p2 \|\| (p1 && iDiff<0) ){
113981	fts3PutDeltaVarint3(&p, bDescIdx, &iPrev, &bFirstOut, i1);
113982	fts3PoslistCopy(&p, &p1);
113983	fts3GetDeltaVarint3(&p1, pEnd1, bDescIdx, &i1);
113984	}else{
113985	fts3PutDeltaVarint3(&p, bDescIdx, &iPrev, &bFirstOut, i2);
113986	fts3PoslistCopy(&p, &p2);
113987	fts3GetDeltaVarint3(&p2, pEnd2, bDescIdx, &i2);
113988	}
113989	}
113990
113991	*paOut = aOut;
113992	*pnOut = (p-aOut);
113993	return SQLITE_OK;
113994	}
113995












113996	static void fts3DoclistPhraseMerge(
113997	int bDescIdx, /* True if arguments are desc */
113998	int nDist, /* Distance from left to right (1=adjacent) */
113999	char aLeft, int nLeft, / Left doclist */
114000	char aRight, int pnRight /* IN/OUT: Right/output doclist */
114001	){
114002	sqlite3_int64 i1 = 0;
	@@ -114015,30 +114496,30 @@
114015	p = aOut;
114016	fts3GetDeltaVarint3(&p1, pEnd1, 0, &i1);
114017	fts3GetDeltaVarint3(&p2, pEnd2, 0, &i2);
114018
114019	while( p1 && p2 ){
114020	sqlite3_int64 iDiff = COMPARE_DOCID(i1, i2);
114021	if( iDiff==0 ){
114022	char *pSave = p;
114023	sqlite3_int64 iPrevSave = iPrev;
114024	int bFirstOutSave = bFirstOut;
114025
114026	fts3PutDeltaVarint3(&p, bDescIdx, &iPrev, &bFirstOut, i1);
114027	if( 0==fts3PoslistPhraseMerge(&p, nDist, 0, 1, &p1, &p2) ){
114028	p = pSave;
114029	iPrev = iPrevSave;
114030	bFirstOut = bFirstOutSave;
114031	}
114032	fts3GetDeltaVarint3(&p1, pEnd1, bDescIdx, &i1);
114033	fts3GetDeltaVarint3(&p2, pEnd2, bDescIdx, &i2);
114034	}else if( iDiff<0 ){
114035	fts3PoslistCopy(0, &p1);
114036	fts3GetDeltaVarint3(&p1, pEnd1, bDescIdx, &i1);
114037	}else{
114038	fts3PoslistCopy(0, &p2);
114039	fts3GetDeltaVarint3(&p2, pEnd2, bDescIdx, &i2);
114040	}
114041	}
114042
114043	*pnRight = p - aOut;
114044	}
	@@ -114051,11 +114532,11 @@
114051	**
114052	** If an OOM error occurs, return SQLITE_NOMEM. In this case it is
114053	** the responsibility of the caller to free any doclists left in the
114054	** TermSelect.aaOutput[] array.
114055	*/
114056	static int fts3TermSelectMerge(Fts3Table p, TermSelect pTS){
114057	char *aOut = 0;
114058	int nOut = 0;
114059	int i;
114060
114061	/* Loop through the doclists in the aaOutput[] array. Merge them all
	@@ -114092,28 +114573,29 @@
114092	pTS->anOutput[0] = nOut;
114093	return SQLITE_OK;
114094	}
114095
114096	/*
114097	** This function is used as the sqlite3Fts3SegReaderIterate() callback when
114098	** querying the full-text index for a doclist associated with a term or
114099	** term-prefix.









114100	*/
114101	static int fts3TermSelectCb(
114102	Fts3Table p, / Virtual table object */
114103	void pContext, / Pointer to TermSelect structure */
114104	char *zTerm,
114105	int nTerm,
114106	char *aDoclist,
114107	int nDoclist
114108	){
114109	TermSelect pTS = (TermSelect )pContext;
114110
114111	UNUSED_PARAMETER(p);
114112	UNUSED_PARAMETER(zTerm);
114113	UNUSED_PARAMETER(nTerm);
114114
114115	if( pTS->aaOutput[0]==0 ){
114116	/* If this is the first term selected, copy the doclist to the output
114117	** buffer using memcpy(). */
114118	pTS->aaOutput[0] = sqlite3_malloc(nDoclist);
114119	pTS->anOutput[0] = nDoclist;
	@@ -114180,23 +114662,30 @@
114180	}
114181	pCsr->apSegment[pCsr->nSegment++] = pNew;
114182	return SQLITE_OK;
114183	}
114184







114185	static int fts3SegReaderCursor(
114186	Fts3Table p, / FTS3 table handle */
114187	int iIndex, /* Index to search (from 0 to p->nIndex-1) */
114188	int iLevel, /* Level of segments to scan */
114189	const char zTerm, / Term to query for */
114190	int nTerm, /* Size of zTerm in bytes */
114191	int isPrefix, /* True for a prefix search */
114192	int isScan, /* True to scan from zTerm to EOF */
114193	Fts3MultiSegReader pCsr / Cursor object to populate */
114194	){
114195	int rc = SQLITE_OK;
114196	int rc2;
114197	sqlite3_stmt *pStmt = 0;
114198
114199	/* If iLevel is less than 0 and this is not a scan, include a seg-reader
114200	** for the pending-terms. If this is a scan, then this call must be being
114201	** made by an fts4aux module, not an FTS table. In this case calling
114202	** Fts3SegReaderPending might segfault, as the data structures used by
	@@ -114281,28 +114770,46 @@
114281	return fts3SegReaderCursor(
114282	p, iIndex, iLevel, zTerm, nTerm, isPrefix, isScan, pCsr
114283	);
114284	}
114285






114286	static int fts3SegReaderCursorAddZero(
114287	Fts3Table *p,
114288	const char *zTerm,
114289	int nTerm,
114290	Fts3MultiSegReader *pCsr
114291	){
114292	return fts3SegReaderCursor(p, 0, FTS3_SEGCURSOR_ALL, zTerm, nTerm, 0, 0,pCsr);
114293	}
114294
114295
114296	SQLITE_PRIVATE int sqlite3Fts3TermSegReaderCursor(












114297	Fts3Cursor pCsr, / Virtual table cursor handle */
114298	const char zTerm, / Term to query for */
114299	int nTerm, /* Size of zTerm in bytes */
114300	int isPrefix, /* True for a prefix search */
114301	Fts3MultiSegReader *ppSegcsr / OUT: Allocated seg-reader cursor */
114302	){
114303	Fts3MultiSegReader pSegcsr; / Object to allocate and return */
114304	int rc = SQLITE_NOMEM; /* Return code */
114305
114306	pSegcsr = sqlite3_malloc(sizeof(Fts3MultiSegReader));
114307	if( pSegcsr ){
114308	int i;
	@@ -114342,62 +114849,53 @@
114342
114343	*ppSegcsr = pSegcsr;
114344	return rc;
114345	}
114346



114347	static void fts3SegReaderCursorFree(Fts3MultiSegReader *pSegcsr){
114348	sqlite3Fts3SegReaderFinish(pSegcsr);
114349	sqlite3_free(pSegcsr);
114350	}
114351
114352	/*
114353	** This function retreives the doclist for the specified term (or term
114354	** prefix) from the database.
114355	**
114356	** The returned doclist may be in one of two formats, depending on the
114357	** value of parameter isReqPos. If isReqPos is zero, then the doclist is
114358	** a sorted list of delta-compressed docids (a bare doclist). If isReqPos
114359	** is non-zero, then the returned list is in the same format as is stored
114360	** in the database without the found length specifier at the start of on-disk
114361	** doclists.
114362	*/
114363	static int fts3TermSelect(
114364	Fts3Table p, / Virtual table handle */
114365	Fts3PhraseToken pTok, / Token to query for */
114366	int iColumn, /* Column to query (or -ve for all columns) */
114367	int isReqPos, /* True to include position lists in output */
114368	int pnOut, / OUT: Size of buffer at ppOut /
114369	char *ppOut / OUT: Malloced result buffer */
114370	){
114371	int rc; /* Return code */
114372	Fts3MultiSegReader pSegcsr; / Seg-reader cursor for this term */
114373	TermSelect tsc; /* Context object for fts3TermSelectCb() */
114374	Fts3SegFilter filter; /* Segment term filter configuration */
114375
114376	pSegcsr = pTok->pSegcsr;
114377	memset(&tsc, 0, sizeof(TermSelect));
114378	tsc.isReqPos = isReqPos;
114379
114380	filter.flags = FTS3_SEGMENT_IGNORE_EMPTY
114381	\| (pTok->isPrefix ? FTS3_SEGMENT_PREFIX : 0)
114382	\| (isReqPos ? FTS3_SEGMENT_REQUIRE_POS : 0)
114383	\| (iColumn<p->nColumn ? FTS3_SEGMENT_COLUMN_FILTER : 0);
114384	filter.iCol = iColumn;
114385	filter.zTerm = pTok->z;
114386	filter.nTerm = pTok->n;
114387
114388	rc = sqlite3Fts3SegReaderStart(p, pSegcsr, &filter);
114389	while( SQLITE_OK==rc
114390	&& SQLITE_ROW==(rc = sqlite3Fts3SegReaderStep(p, pSegcsr))
114391	){
114392	rc = fts3TermSelectCb(p, (void *)&tsc,
114393	pSegcsr->zTerm, pSegcsr->nTerm, pSegcsr->aDoclist, pSegcsr->nDoclist
114394	);
114395	}
114396
114397	if( rc==SQLITE_OK ){
114398	rc = fts3TermSelectMerge(p, &tsc);
114399	}
114400	if( rc==SQLITE_OK ){
114401	*ppOut = tsc.aaOutput[0];
114402	*pnOut = tsc.anOutput[0];
114403	}else{
	@@ -114419,28 +114917,19 @@
114419	** If the isPoslist argument is true, then it is assumed that the doclist
114420	** contains a position-list following each docid. Otherwise, it is assumed
114421	** that the doclist is simply a list of docids stored as delta encoded
114422	** varints.
114423	*/
114424	static int fts3DoclistCountDocids(int isPoslist, char *aList, int nList){
114425	int nDoc = 0; /* Return value */
114426	if( aList ){
114427	char aEnd = &aList[nList]; / Pointer to one byte after EOF */
114428	char p = aList; / Cursor */
114429	if( !isPoslist ){
114430	/* The number of docids in the list is the same as the number of
114431	** varints. In FTS3 a varint consists of a single byte with the 0x80
114432	** bit cleared and zero or more bytes with the 0x80 bit set. So to
114433	** count the varints in the buffer, just count the number of bytes
114434	** with the 0x80 bit clear. */
114435	while( p<aEnd ) nDoc += (((*p++)&0x80)==0);
114436	}else{
114437	while( p<aEnd ){
114438	nDoc++;
114439	while( (p++)&0x80 ); / Skip docid varint */
114440	fts3PoslistCopy(0, &p); /* Skip over position list */
114441	}
114442	}
114443	}
114444
114445	return nDoc;
114446	}
	@@ -114466,11 +114955,11 @@
114466	}else{
114467	pCsr->iPrevId = sqlite3_column_int64(pCsr->pStmt, 0);
114468	rc = SQLITE_OK;
114469	}
114470	}else{
114471	rc = sqlite3Fts3EvalNext((Fts3Cursor *)pCursor);
114472	}
114473	assert( ((Fts3Table *)pCsr->base.pVtab)->pSegments==0 );
114474	return rc;
114475	}
114476
	@@ -114543,11 +115032,11 @@
114543	}
114544
114545	rc = sqlite3Fts3ReadLock(p);
114546	if( rc!=SQLITE_OK ) return rc;
114547
114548	rc = sqlite3Fts3EvalStart(pCsr, pCsr->pExpr, 1);
114549
114550	sqlite3Fts3SegmentsClose(p);
114551	if( rc!=SQLITE_OK ) return rc;
114552	pCsr->pNextId = pCsr->aDoclist;
114553	pCsr->iPrevId = 0;
	@@ -114950,26 +115439,43 @@
114950	p->zDb, p->zName, zName
114951	);
114952	return rc;
114953	}
114954





114955	static int fts3SavepointMethod(sqlite3_vtab *pVtab, int iSavepoint){
114956	UNUSED_PARAMETER(iSavepoint);
114957	assert( ((Fts3Table *)pVtab)->inTransaction );
114958	assert( ((Fts3Table *)pVtab)->mxSavepoint < iSavepoint );
114959	TESTONLY( ((Fts3Table *)pVtab)->mxSavepoint = iSavepoint );
114960	return fts3SyncMethod(pVtab);
114961	}






114962	static int fts3ReleaseMethod(sqlite3_vtab *pVtab, int iSavepoint){
114963	TESTONLY( Fts3Table p = (Fts3Table)pVtab );
114964	UNUSED_PARAMETER(iSavepoint);
114965	UNUSED_PARAMETER(pVtab);
114966	assert( p->inTransaction );
114967	assert( p->mxSavepoint >= iSavepoint );
114968	TESTONLY( p->mxSavepoint = iSavepoint-1 );
114969	return SQLITE_OK;
114970	}






114971	static int fts3RollbackToMethod(sqlite3_vtab *pVtab, int iSavepoint){
114972	Fts3Table p = (Fts3Table)pVtab;
114973	UNUSED_PARAMETER(iSavepoint);
114974	assert( p->inTransaction );
114975	assert( p->mxSavepoint >= iSavepoint );
	@@ -115114,22 +115620,10 @@
115114	sqlite3Fts3HashClear(pHash);
115115	sqlite3_free(pHash);
115116	}
115117	return rc;
115118	}
115119
115120	#if !SQLITE_CORE
115121	SQLITE_API int sqlite3_extension_init(
115122	sqlite3 *db,
115123	char **pzErrMsg,
115124	const sqlite3_api_routines *pApi
115125	){
115126	SQLITE_EXTENSION_INIT2(pApi)
115127	return sqlite3Fts3Init(db);
115128	}
115129	#endif
115130
115131
115132	/*
115133	** Allocate an Fts3MultiSegReader for each token in the expression headed
115134	** by pExpr.
115135	**
	@@ -115143,24 +115637,24 @@
115143	** there exists prefix b-tree of the right length) then it may be traversed
115144	** and merged incrementally. Otherwise, it has to be merged into an in-memory
115145	** doclist and then traversed.
115146	*/
115147	static void fts3EvalAllocateReaders(
115148	Fts3Cursor *pCsr,
115149	Fts3Expr *pExpr,
115150	int pnToken, / OUT: Total number of tokens in phrase. */
115151	int pnOr, / OUT: Total number of OR nodes in expr. */
115152	int *pRc
115153	){
115154	if( pExpr && SQLITE_OK==*pRc ){
115155	if( pExpr->eType==FTSQUERY_PHRASE ){
115156	int i;
115157	int nToken = pExpr->pPhrase->nToken;
115158	*pnToken += nToken;
115159	for(i=0; i<nToken; i++){
115160	Fts3PhraseToken *pToken = &pExpr->pPhrase->aToken[i];
115161	int rc = sqlite3Fts3TermSegReaderCursor(pCsr,
115162	pToken->z, pToken->n, pToken->isPrefix, &pToken->pSegcsr
115163	);
115164	if( rc!=SQLITE_OK ){
115165	*pRc = rc;
115166	return;
	@@ -115174,16 +115668,24 @@
115174	fts3EvalAllocateReaders(pCsr, pExpr->pRight, pnToken, pnOr, pRc);
115175	}
115176	}
115177	}
115178








115179	static void fts3EvalPhraseMergeToken(
115180	Fts3Table *pTab,
115181	Fts3Phrase *p,
115182	int iToken,
115183	char *pList,
115184	int nList
115185	){
115186	assert( iToken!=p->iDoclistToken );
115187
115188	if( pList==0 ){
115189	sqlite3_free(p->doclist.aAll);
	@@ -115228,13 +115730,19 @@
115228	}
115229
115230	if( iToken>p->iDoclistToken ) p->iDoclistToken = iToken;
115231	}
115232






115233	static int fts3EvalPhraseLoad(
115234	Fts3Cursor *pCsr,
115235	Fts3Phrase *p
115236	){
115237	Fts3Table pTab = (Fts3Table )pCsr->base.pVtab;
115238	int iToken;
115239	int rc = SQLITE_OK;
115240
	@@ -115243,11 +115751,11 @@
115243	assert( pToken->pDeferred==0 \|\| pToken->pSegcsr==0 );
115244
115245	if( pToken->pSegcsr ){
115246	int nThis = 0;
115247	char *pThis = 0;
115248	rc = fts3TermSelect(pTab, pToken, p->iColumn, 1, &nThis, &pThis);
115249	if( rc==SQLITE_OK ){
115250	fts3EvalPhraseMergeToken(pTab, p, iToken, pThis, nThis);
115251	}
115252	}
115253	assert( pToken->pSegcsr==0 );
	@@ -115254,18 +115762,26 @@
115254	}
115255
115256	return rc;
115257	}
115258










115259	static int fts3EvalDeferredPhrase(Fts3Cursor pCsr, Fts3Phrase pPhrase){
115260	int iToken;
115261	int rc = SQLITE_OK;
115262
115263	int nMaxUndeferred = pPhrase->iDoclistToken;
115264	char *aPoslist = 0;
115265	int nPoslist = 0;
115266	int iPrev = -1;
115267
115268	assert( pPhrase->doclist.bFreeList==0 );
115269
115270	for(iToken=0; rc==SQLITE_OK && iToken<pPhrase->nToken; iToken++){
115271	Fts3PhraseToken *pToken = &pPhrase->aToken[iToken];
	@@ -115307,10 +115823,11 @@
115307	iPrev = iToken;
115308	}
115309	}
115310
115311	if( iPrev>=0 ){

115312	if( nMaxUndeferred<0 ){
115313	pPhrase->doclist.pList = aPoslist;
115314	pPhrase->doclist.nList = nPoslist;
115315	pPhrase->doclist.iDocid = pCsr->iPrevId;
115316	pPhrase->doclist.bFreeList = 1;
	@@ -115355,13 +115872,19 @@
115355	/*
115356	** This function is called for each Fts3Phrase in a full-text query
115357	** expression to initialize the mechanism for returning rows. Once this
115358	** function has been called successfully on an Fts3Phrase, it may be
115359	** used with fts3EvalPhraseNext() to iterate through the matching docids.






115360	*/
115361	static int fts3EvalPhraseStart(Fts3Cursor pCsr, int bOptOk, Fts3Phrase p){
115362	int rc;
115363	Fts3PhraseToken *pFirst = &p->aToken[0];
115364	Fts3Table pTab = (Fts3Table )pCsr->base.pVtab;
115365
115366	if( pCsr->bDesc==pTab->bDescIdx
115367	&& bOptOk==1
	@@ -115385,11 +115908,17 @@
115385	return rc;
115386	}
115387
115388	/*
115389	** This function is used to iterate backwards (from the end to start)
115390	** through doclists.






115391	*/
115392	SQLITE_PRIVATE void sqlite3Fts3DoclistPrev(
115393	int bDescIdx, /* True if the doclist is desc */
115394	char aDoclist, / Pointer to entire doclist */
115395	int nDoclist, /* Length of aDoclist in bytes */
	@@ -115450,13 +115979,13 @@
115450	** If there is no "next" entry and no error occurs, then *pbEof is set to
115451	** 1 before returning. Otherwise, if no error occurs and the iterator is
115452	** successfully advanced, *pbEof is set to 0.
115453	*/
115454	static int fts3EvalPhraseNext(
115455	Fts3Cursor *pCsr,
115456	Fts3Phrase *p,
115457	u8 *pbEof
115458	){
115459	int rc = SQLITE_OK;
115460	Fts3Doclist *pDL = &p->doclist;
115461	Fts3Table pTab = (Fts3Table )pCsr->base.pVtab;
115462
	@@ -115498,14 +116027,14 @@
115498	fts3PoslistCopy(0, &pIter);
115499	pDL->nList = (pIter - pDL->pList);
115500
115501	/* pIter now points just past the 0x00 that terminates the position-
115502	** list for document pDL->iDocid. However, if this position-list was
115503	** edited in place by fts3EvalNearTrim2(), then pIter may not actually
115504	** point to the start of the next docid value. The following line deals
115505	** with this case by advancing pIter past the zero-padding added by
115506	** fts3EvalNearTrim2(). */
115507	while( pIter<pEnd && *pIter==0 ) pIter++;
115508
115509	pDL->pNextDocid = pIter;
115510	assert( pIter>=&pDL->aAll[pDL->nAll] \|\| *pIter );
115511	*pbEof = 0;
	@@ -115513,15 +116042,31 @@
115513	}
115514
115515	return rc;
115516	}
115517
















115518	static void fts3EvalStartReaders(
115519	Fts3Cursor *pCsr,
115520	Fts3Expr *pExpr,
115521	int bOptOk,
115522	int *pRc
115523	){
115524	if( pExpr && SQLITE_OK==*pRc ){
115525	if( pExpr->eType==FTSQUERY_PHRASE ){
115526	int i;
115527	int nToken = pExpr->pPhrase->nToken;
	@@ -115536,27 +116081,46 @@
115536	pExpr->bDeferred = (pExpr->pLeft->bDeferred && pExpr->pRight->bDeferred);
115537	}
115538	}
115539	}
115540












115541	typedef struct Fts3TokenAndCost Fts3TokenAndCost;
115542	struct Fts3TokenAndCost {
115543	Fts3Phrase pPhrase; / The phrase the token belongs to */
115544	int iToken; /* Position of token in phrase */
115545	Fts3PhraseToken pToken; / The token itself */
115546	Fts3Expr *pRoot;
115547	int nOvfl;
115548	int iCol; /* The column the token must match */
115549	};
115550







115551	static void fts3EvalTokenCosts(
115552	Fts3Cursor *pCsr,
115553	Fts3Expr *pRoot,
115554	Fts3Expr *pExpr,
115555	Fts3TokenAndCost **ppTC,
115556	Fts3Expr ***ppOr,
115557	int *pRc
115558	){
115559	if( *pRc==SQLITE_OK && pExpr ){
115560	if( pExpr->eType==FTSQUERY_PHRASE ){
115561	Fts3Phrase *pPhrase = pExpr->pPhrase;
115562	int i;
	@@ -115584,23 +116148,34 @@
115584	fts3EvalTokenCosts(pCsr, pRoot, pExpr->pRight, ppTC, ppOr, pRc);
115585	}
115586	}
115587	}
115588











115589	static int fts3EvalAverageDocsize(Fts3Cursor pCsr, int pnPage){
115590	if( pCsr->nRowAvg==0 ){
115591	/* The average document size, which is required to calculate the cost
115592	** of each doclist, has not yet been determined. Read the required
115593	** data from the %_stat table to calculate it.
115594	**
115595	** Entry 0 of the %_stat table is a blob containing (nCol+1) FTS3
115596	** varints, where nCol is the number of columns in the FTS3 table.
115597	** The first varint is the number of documents currently stored in
115598	** the table. The following nCol varints contain the total amount of
115599	** data stored in all rows of each column of the table, from left
115600	** to right.
115601	*/
115602	int rc;
115603	Fts3Table p = (Fts3Table)pCsr->base.pVtab;
115604	sqlite3_stmt *pStmt;
115605	sqlite3_int64 nDoc = 0;
115606	sqlite3_int64 nByte = 0;
	@@ -115631,109 +116206,151 @@
115631
115632	*pnPage = pCsr->nRowAvg;
115633	return SQLITE_OK;
115634	}
115635














115636	static int fts3EvalSelectDeferred(
115637	Fts3Cursor *pCsr,
115638	Fts3Expr *pRoot,
115639	Fts3TokenAndCost *aTC,
115640	int nTC
115641	){
115642	int nDocSize = 0;
115643	int nDocEst = 0;
115644	int rc = SQLITE_OK;
115645	Fts3Table pTab = (Fts3Table )pCsr->base.pVtab;
115646	int ii;




115647
115648	int nOvfl = 0;
115649	int nTerm = 0;
115650



115651	for(ii=0; ii<nTC; ii++){
115652	if( aTC[ii].pRoot==pRoot ){
115653	nOvfl += aTC[ii].nOvfl;
115654	nTerm++;
115655	}
115656	}
115657	if( nOvfl==0 \|\| nTerm<2 ) return SQLITE_OK;
115658

115659	rc = fts3EvalAverageDocsize(pCsr, &nDocSize);
115660
115661	for(ii=0; ii<nTerm && rc==SQLITE_OK; ii++){
115662	int jj;
115663	Fts3TokenAndCost *pTC = 0;
115664
115665	for(jj=0; jj<nTC; jj++){
115666	if( aTC[jj].pToken && aTC[jj].pRoot==pRoot
115667	&& (!pTC \|\| aTC[jj].nOvfl<pTC->nOvfl)

























115668	){
115669	pTC = &aTC[jj];
115670	}
115671	}
115672	assert( pTC );
115673
115674	/* At this point pTC points to the cheapest remaining token. */
115675	if( ii==0 ){
115676	if( pTC->nOvfl ){
115677	nDocEst = (pTC->nOvfl * pTab->nPgsz + pTab->nPgsz) / 10;
115678	}else{










115679	Fts3PhraseToken *pToken = pTC->pToken;
115680	int nList = 0;
115681	char *pList = 0;
115682	rc = fts3TermSelect(pTab, pToken, pTC->iCol, 1, &nList, &pList);
115683	assert( rc==SQLITE_OK \|\| pList==0 );
115684
115685	if( rc==SQLITE_OK ){
115686	nDocEst = fts3DoclistCountDocids(1, pList, nList);
115687	fts3EvalPhraseMergeToken(pTab, pTC->pPhrase, pTC->iToken,pList,nList);




115688	}
115689	}
115690	}else{
115691	if( pTC->nOvfl>=(nDocEst*nDocSize) ){
115692	Fts3PhraseToken *pToken = pTC->pToken;
115693	rc = sqlite3Fts3DeferToken(pCsr, pToken, pTC->iCol);
115694	fts3SegReaderCursorFree(pToken->pSegcsr);
115695	pToken->pSegcsr = 0;
115696	}
115697	nDocEst = 1 + (nDocEst/4);
115698	}
115699	pTC->pToken = 0;
115700	}
115701
115702	return rc;
115703	}
115704
115705	SQLITE_PRIVATE int sqlite3Fts3EvalStart(Fts3Cursor pCsr, Fts3Expr pExpr, int bOptOk){












115706	Fts3Table pTab = (Fts3Table )pCsr->base.pVtab;
115707	int rc = SQLITE_OK;
115708	int nToken = 0;
115709	int nOr = 0;
115710
115711	/* Allocate a MultiSegReader for each token in the expression. */
115712	fts3EvalAllocateReaders(pCsr, pExpr, &nToken, &nOr, &rc);
115713
115714	/* Call fts3EvalPhraseStart() on all phrases in the expression. TODO:
115715	** This call will eventually also be responsible for determining which
115716	** tokens are 'deferred' until the document text is loaded into memory.
115717	**
115718	** Each token in each phrase is dealt with using one of the following
115719	** three strategies:
115720	**
115721	** 1. Entire doclist loaded into memory as part of the
115722	** fts3EvalStartReaders() call.
115723	**
115724	** 2. Doclist loaded into memory incrementally, as part of each
115725	** sqlite3Fts3EvalNext() call.
115726	**
115727	** 3. Token doclist is never loaded. Instead, documents are loaded into
115728	** memory and scanned for the token as part of the sqlite3Fts3EvalNext()
115729	** call. This is known as a "deferred" token.
115730	*/
115731
115732	/* If bOptOk is true, check if there are any tokens that should be deferred.
115733	*/
115734	if( rc==SQLITE_OK && bOptOk && nToken>1 && pTab->bHasStat ){
115735	Fts3TokenAndCost *aTC;
115736	Fts3Expr **apOr;
115737	aTC = (Fts3TokenAndCost *)sqlite3_malloc(
115738	sizeof(Fts3TokenAndCost) * nToken
115739	+ sizeof(Fts3Expr ) nOr * 2
	@@ -115745,11 +116362,11 @@
115745	}else{
115746	int ii;
115747	Fts3TokenAndCost *pTC = aTC;
115748	Fts3Expr **ppOr = apOr;
115749
115750	fts3EvalTokenCosts(pCsr, 0, pExpr, &pTC, &ppOr, &rc);
115751	nToken = pTC-aTC;
115752	nOr = ppOr-apOr;
115753
115754	if( rc==SQLITE_OK ){
115755	rc = fts3EvalSelectDeferred(pCsr, 0, aTC, nToken);
	@@ -115760,25 +116377,50 @@
115760
115761	sqlite3_free(aTC);
115762	}
115763	}
115764
115765	fts3EvalStartReaders(pCsr, pExpr, bOptOk, &rc);
115766	return rc;
115767	}
115768
115769	static void fts3EvalZeroPoslist(Fts3Phrase *pPhrase){



115770	if( pPhrase->doclist.bFreeList ){
115771	sqlite3_free(pPhrase->doclist.pList);
115772	}
115773	pPhrase->doclist.pList = 0;
115774	pPhrase->doclist.nList = 0;
115775	pPhrase->doclist.bFreeList = 0;
115776	}
115777
115778	static int fts3EvalNearTrim2(
115779	int nNear,






















115780	char aTmp, / Temporary space to use */
115781	char *paPoslist, / IN/OUT: Position list */
115782	int pnToken, / IN/OUT: Tokens in phrase of paPoslist /
115783	Fts3Phrase pPhrase / The phrase object to trim the doclist of */
115784	){
	@@ -115806,10 +116448,176 @@
115806	}
115807
115808	return res;
115809	}
115810






































































































































































115811	static int fts3EvalNearTest(Fts3Expr pExpr, int pRc){
115812	int res = 1;
115813
115814	/* The following block runs if pExpr is the root of a NEAR query.
115815	** For example, the query:
	@@ -115827,11 +116635,11 @@
115827	** \| \|
115828	** "w" "x"
115829	**
115830	** The right-hand child of a NEAR node is always a phrase. The
115831	** left-hand child may be either a phrase or a NEAR node. There are
115832	** no exceptions to this.
115833	*/
115834	if( *pRc==SQLITE_OK
115835	&& pExpr->eType==FTSQUERY_NEAR
115836	&& pExpr->bEof==0
115837	&& (pExpr->pParent==0 \|\| pExpr->pParent->eType!=FTSQUERY_NEAR)
	@@ -115854,21 +116662,21 @@
115854	int nToken = p->pPhrase->nToken;
115855
115856	for(p=p->pParent;res && p && p->eType==FTSQUERY_NEAR; p=p->pParent){
115857	Fts3Phrase *pPhrase = p->pRight->pPhrase;
115858	int nNear = p->nNear;
115859	res = fts3EvalNearTrim2(nNear, aTmp, &aPoslist, &nToken, pPhrase);
115860	}
115861
115862	aPoslist = pExpr->pRight->pPhrase->doclist.pList;
115863	nToken = pExpr->pRight->pPhrase->nToken;
115864	for(p=pExpr->pLeft; p && res; p=p->pLeft){
115865	int nNear = p->pParent->nNear;
115866	Fts3Phrase *pPhrase = (
115867	p->eType==FTSQUERY_NEAR ? p->pRight->pPhrase : p->pPhrase
115868	);
115869	res = fts3EvalNearTrim2(nNear, aTmp, &aPoslist, &nToken, pPhrase);
115870	}
115871	}
115872
115873	sqlite3_free(aTmp);
115874	}
	@@ -115875,132 +116683,33 @@
115875
115876	return res;
115877	}
115878
115879	/*
115880	** This macro is used by the fts3EvalNext() function. The two arguments are
115881	** 64-bit docid values. If the current query is "ORDER BY docid ASC", then
115882	** the macro returns (i1 - i2). Or if it is "ORDER BY docid DESC", then
115883	** it returns (i2 - i1). This allows the same code to be used for merging
115884	** doclists in ascending or descending order.




115885	*/
115886	#define DOCID_CMP(i1, i2) ((pCsr->bDesc?-1:1) * (i1-i2))
115887
115888	static void fts3EvalNext(
115889	Fts3Cursor *pCsr,
115890	Fts3Expr *pExpr,
115891	int *pRc
115892	){
115893	if( *pRc==SQLITE_OK ){
115894	assert( pExpr->bEof==0 );
115895	pExpr->bStart = 1;
115896
115897	switch( pExpr->eType ){
115898	case FTSQUERY_NEAR:
115899	case FTSQUERY_AND: {
115900	Fts3Expr *pLeft = pExpr->pLeft;
115901	Fts3Expr *pRight = pExpr->pRight;
115902	assert( !pLeft->bDeferred \|\| !pRight->bDeferred );
115903	if( pLeft->bDeferred ){
115904	fts3EvalNext(pCsr, pRight, pRc);
115905	pExpr->iDocid = pRight->iDocid;
115906	pExpr->bEof = pRight->bEof;
115907	}else if( pRight->bDeferred ){
115908	fts3EvalNext(pCsr, pLeft, pRc);
115909	pExpr->iDocid = pLeft->iDocid;
115910	pExpr->bEof = pLeft->bEof;
115911	}else{
115912	fts3EvalNext(pCsr, pLeft, pRc);
115913	fts3EvalNext(pCsr, pRight, pRc);
115914
115915	while( !pLeft->bEof && !pRight->bEof && *pRc==SQLITE_OK ){
115916	sqlite3_int64 iDiff = DOCID_CMP(pLeft->iDocid, pRight->iDocid);
115917	if( iDiff==0 ) break;
115918	if( iDiff<0 ){
115919	fts3EvalNext(pCsr, pLeft, pRc);
115920	}else{
115921	fts3EvalNext(pCsr, pRight, pRc);
115922	}
115923	}
115924
115925	pExpr->iDocid = pLeft->iDocid;
115926	pExpr->bEof = (pLeft->bEof \|\| pRight->bEof);
115927	}
115928	break;
115929	}
115930
115931	case FTSQUERY_OR: {
115932	Fts3Expr *pLeft = pExpr->pLeft;
115933	Fts3Expr *pRight = pExpr->pRight;
115934	sqlite3_int64 iCmp = DOCID_CMP(pLeft->iDocid, pRight->iDocid);
115935
115936	assert( pLeft->bStart \|\| pLeft->iDocid==pRight->iDocid );
115937	assert( pRight->bStart \|\| pLeft->iDocid==pRight->iDocid );
115938
115939	if( pRight->bEof \|\| (pLeft->bEof==0 && iCmp<0) ){
115940	fts3EvalNext(pCsr, pLeft, pRc);
115941	}else if( pLeft->bEof \|\| (pRight->bEof==0 && iCmp>0) ){
115942	fts3EvalNext(pCsr, pRight, pRc);
115943	}else{
115944	fts3EvalNext(pCsr, pLeft, pRc);
115945	fts3EvalNext(pCsr, pRight, pRc);
115946	}
115947
115948	pExpr->bEof = (pLeft->bEof && pRight->bEof);
115949	iCmp = DOCID_CMP(pLeft->iDocid, pRight->iDocid);
115950	if( pRight->bEof \|\| (pLeft->bEof==0 && iCmp<0) ){
115951	pExpr->iDocid = pLeft->iDocid;
115952	}else{
115953	pExpr->iDocid = pRight->iDocid;
115954	}
115955
115956	break;
115957	}
115958
115959	case FTSQUERY_NOT: {
115960	Fts3Expr *pLeft = pExpr->pLeft;
115961	Fts3Expr *pRight = pExpr->pRight;
115962
115963	if( pRight->bStart==0 ){
115964	fts3EvalNext(pCsr, pRight, pRc);
115965	assert( *pRc!=SQLITE_OK \|\| pRight->bStart );
115966	}
115967
115968	fts3EvalNext(pCsr, pLeft, pRc);
115969	if( pLeft->bEof==0 ){
115970	while( !*pRc
115971	&& !pRight->bEof
115972	&& DOCID_CMP(pLeft->iDocid, pRight->iDocid)>0
115973	){
115974	fts3EvalNext(pCsr, pRight, pRc);
115975	}
115976	}
115977	pExpr->iDocid = pLeft->iDocid;
115978	pExpr->bEof = pLeft->bEof;
115979	break;
115980	}
115981
115982	default: {
115983	Fts3Phrase *pPhrase = pExpr->pPhrase;
115984	fts3EvalZeroPoslist(pPhrase);
115985	*pRc = fts3EvalPhraseNext(pCsr, pPhrase, &pExpr->bEof);
115986	pExpr->iDocid = pPhrase->doclist.iDocid;
115987	break;
115988	}
115989	}
115990	}
115991	}
115992
115993	static int fts3EvalDeferredTest(Fts3Cursor pCsr, Fts3Expr pExpr, int *pRc){
115994	int bHit = 1;
115995	if( *pRc==SQLITE_OK ){
115996	switch( pExpr->eType ){
115997	case FTSQUERY_NEAR:
115998	case FTSQUERY_AND:
115999	bHit = (
116000	fts3EvalDeferredTest(pCsr, pExpr->pLeft, pRc)
116001	&& fts3EvalDeferredTest(pCsr, pExpr->pRight, pRc)
116002	&& fts3EvalNearTest(pExpr, pRc)
116003	);
116004
116005	/* If the NEAR expression does not match any rows, zero the doclist for
116006	** all phrases involved in the NEAR. This is because the snippet(),
	@@ -116022,31 +116731,31 @@
116022	&& (pExpr->pParent==0 \|\| pExpr->pParent->eType!=FTSQUERY_NEAR)
116023	){
116024	Fts3Expr *p;
116025	for(p=pExpr; p->pPhrase==0; p=p->pLeft){
116026	if( p->pRight->iDocid==pCsr->iPrevId ){
116027	fts3EvalZeroPoslist(p->pRight->pPhrase);
116028	}
116029	}
116030	if( p->iDocid==pCsr->iPrevId ){
116031	fts3EvalZeroPoslist(p->pPhrase);
116032	}
116033	}
116034
116035	break;
116036
116037	case FTSQUERY_OR: {
116038	int bHit1 = fts3EvalDeferredTest(pCsr, pExpr->pLeft, pRc);
116039	int bHit2 = fts3EvalDeferredTest(pCsr, pExpr->pRight, pRc);
116040	bHit = bHit1 \|\| bHit2;
116041	break;
116042	}
116043
116044	case FTSQUERY_NOT:
116045	bHit = (
116046	fts3EvalDeferredTest(pCsr, pExpr->pLeft, pRc)
116047	&& !fts3EvalDeferredTest(pCsr, pExpr->pRight, pRc)
116048	);
116049	break;
116050
116051	default: {
116052	if( pCsr->pDeferred
	@@ -116053,11 +116762,11 @@
116053	&& (pExpr->iDocid==pCsr->iPrevId \|\| pExpr->bDeferred)
116054	){
116055	Fts3Phrase *pPhrase = pExpr->pPhrase;
116056	assert( pExpr->bDeferred \|\| pPhrase->doclist.bFreeList==0 );
116057	if( pExpr->bDeferred ){
116058	fts3EvalZeroPoslist(pPhrase);
116059	}
116060	*pRc = fts3EvalDeferredPhrase(pCsr, pPhrase);
116061	bHit = (pPhrase->doclist.pList!=0);
116062	pExpr->iDocid = pCsr->iPrevId;
116063	}else{
	@@ -116069,31 +116778,53 @@
116069	}
116070	return bHit;
116071	}
116072
116073	/*
116074	** Return 1 if both of the following are true:













116075	**
116076	** 1. *pRc is SQLITE_OK when this function returns, and
116077	**
116078	** 2. After scanning the current FTS table row for the deferred tokens,
116079	** it is determined that the row does not match the query.
116080	**
116081	** Or, if no error occurs and it seems the current row does match the FTS
116082	** query, return 0.
116083	*/
116084	static int fts3EvalLoadDeferred(Fts3Cursor pCsr, int pRc){
116085	int rc = *pRc;
116086	int bMiss = 0;
116087	if( rc==SQLITE_OK ){







116088	if( pCsr->pDeferred ){
116089	rc = fts3CursorSeek(0, pCsr);
116090	if( rc==SQLITE_OK ){
116091	rc = sqlite3Fts3CacheDeferredDoclists(pCsr);
116092	}
116093	}
116094	bMiss = (0==fts3EvalDeferredTest(pCsr, pCsr->pExpr, &rc));


116095	sqlite3Fts3FreeDeferredDoclists(pCsr);
116096	*pRc = rc;
116097	}
116098	return (rc==SQLITE_OK && bMiss);
116099	}
	@@ -116100,11 +116831,11 @@
116100
116101	/*
116102	** Advance to the next document that matches the FTS expression in
116103	** Fts3Cursor.pExpr.
116104	*/
116105	SQLITE_PRIVATE int sqlite3Fts3EvalNext(Fts3Cursor *pCsr){
116106	int rc = SQLITE_OK; /* Return Code */
116107	Fts3Expr *pExpr = pCsr->pExpr;
116108	assert( pCsr->isEof==0 );
116109	if( pExpr==0 ){
116110	pCsr->isEof = 1;
	@@ -116112,23 +116843,23 @@
116112	do {
116113	if( pCsr->isRequireSeek==0 ){
116114	sqlite3_reset(pCsr->pStmt);
116115	}
116116	assert( sqlite3_data_count(pCsr->pStmt)==0 );
116117	fts3EvalNext(pCsr, pExpr, &rc);
116118	pCsr->isEof = pExpr->bEof;
116119	pCsr->isRequireSeek = 1;
116120	pCsr->isMatchinfoNeeded = 1;
116121	pCsr->iPrevId = pExpr->iDocid;
116122	}while( pCsr->isEof==0 && fts3EvalLoadDeferred(pCsr, &rc) );
116123	}
116124	return rc;
116125	}
116126
116127	/*
116128	** Restart interation for expression pExpr so that the next call to
116129	** sqlite3Fts3EvalNext() visits the first row. Do not allow incremental
116130	** loading or merging of phrase doclists for this iteration.
116131	**
116132	** If *pRc is other than SQLITE_OK when this function is called, it is
116133	** a no-op. If an error occurs within this function, *pRc is set to an
116134	** SQLite error code before returning.
	@@ -116140,11 +116871,11 @@
116140	){
116141	if( pExpr && *pRc==SQLITE_OK ){
116142	Fts3Phrase *pPhrase = pExpr->pPhrase;
116143
116144	if( pPhrase ){
116145	fts3EvalZeroPoslist(pPhrase);
116146	if( pPhrase->bIncr ){
116147	assert( pPhrase->nToken==1 );
116148	assert( pPhrase->aToken[0].pSegcsr );
116149	sqlite3Fts3MsrIncrRestart(pPhrase->aToken[0].pSegcsr);
116150	*pRc = fts3EvalPhraseStart(pCsr, 0, pPhrase);
	@@ -116256,18 +116987,18 @@
116256	/* Ensure the %_content statement is reset. */
116257	if( pCsr->isRequireSeek==0 ) sqlite3_reset(pCsr->pStmt);
116258	assert( sqlite3_data_count(pCsr->pStmt)==0 );
116259
116260	/* Advance to the next document */
116261	fts3EvalNext(pCsr, pRoot, &rc);
116262	pCsr->isEof = pRoot->bEof;
116263	pCsr->isRequireSeek = 1;
116264	pCsr->isMatchinfoNeeded = 1;
116265	pCsr->iPrevId = pRoot->iDocid;
116266	}while( pCsr->isEof==0
116267	&& pRoot->eType==FTSQUERY_NEAR
116268	&& fts3EvalLoadDeferred(pCsr, &rc)
116269	);
116270
116271	if( rc==SQLITE_OK && pCsr->isEof==0 ){
116272	fts3EvalUpdateCounts(pRoot);
116273	}
	@@ -116285,14 +117016,14 @@
116285	**
116286	** do {...} while( pRoot->iDocid<iDocid && rc==SQLITE_OK );
116287	*/
116288	fts3EvalRestart(pCsr, pRoot, &rc);
116289	do {
116290	fts3EvalNext(pCsr, pRoot, &rc);
116291	assert( pRoot->bEof==0 );
116292	}while( pRoot->iDocid!=iDocid && rc==SQLITE_OK );
116293	fts3EvalLoadDeferred(pCsr, &rc);
116294	}
116295	}
116296	return rc;
116297	}
116298
	@@ -116419,18 +117150,32 @@
116419	*/
116420	SQLITE_PRIVATE void sqlite3Fts3EvalPhraseCleanup(Fts3Phrase *pPhrase){
116421	if( pPhrase ){
116422	int i;
116423	sqlite3_free(pPhrase->doclist.aAll);
116424	fts3EvalZeroPoslist(pPhrase);
116425	memset(&pPhrase->doclist, 0, sizeof(Fts3Doclist));
116426	for(i=0; i<pPhrase->nToken; i++){
116427	fts3SegReaderCursorFree(pPhrase->aToken[i].pSegcsr);
116428	pPhrase->aToken[i].pSegcsr = 0;
116429	}
116430	}
116431	}














116432
116433	#endif
116434
116435	/************ End of fts3.c **********************************************/
116436	/************ Begin file fts3_aux.c **************************************/
	@@ -118917,14 +119662,10 @@
118917	** (in which case SQLITE_CORE is not defined), or
118918	**
118919	** * The FTS3 module is being built into the core of
118920	** SQLite (in which case SQLITE_ENABLE_FTS3 is defined).
118921	*/
118922	#ifndef SQLITE_CORE
118923	SQLITE_EXTENSION_INIT1
118924	#endif
118925
118926	#if !defined(SQLITE_CORE) \|\| defined(SQLITE_ENABLE_FTS3)
118927
118928
118929	/*
118930	** Implementation of the SQL scalar function for accessing the underlying
118931

	--- src/sqlite3.c
	+++ src/sqlite3.c
	@@ -1,8 +1,8 @@
1	/******************************************************************************
2	** This file is an amalgamation of many separate C source files from SQLite
3	** version 3.7.8. By combining all the individual C code files into this
4	** single large file, the entire code can be compiled as a single translation
5	** unit. This allows many compilers to do optimizations that would not be
6	** possible if the files were compiled separately. Performance improvements
7	** of 5% or more are commonly seen when SQLite is compiled as a single
8	** translation unit.
	@@ -648,13 +648,13 @@
648	**
649	** See also: [sqlite3_libversion()],
650	** [sqlite3_libversion_number()], [sqlite3_sourceid()],
651	** [sqlite_version()] and [sqlite_source_id()].
652	*/
653	#define SQLITE_VERSION "3.7.8"
654	#define SQLITE_VERSION_NUMBER 3007008
655	#define SQLITE_SOURCE_ID "2011-07-19 18:29:00 ed5f0aad6b21066bacd01521e82c22e96991f400"
656
657	/*
658	** CAPI3REF: Run-Time Library Version Numbers
659	** KEYWORDS: sqlite3_version, sqlite3_sourceid
660	**
	@@ -1282,20 +1282,37 @@
1282	** when [PRAGMA synchronous \| PRAGMA synchronous=OFF] is set, but most
1283	** VFSes do not need this signal and should silently ignore this opcode.
1284	** Applications should not call [sqlite3_file_control()] with this
1285	** opcode as doing so may disrupt the operation of the specialized VFSes
1286	** that do require it.
1287	**
1288	** ^The [SQLITE_FCNTL_WIN32_AV_RETRY] opcode is used to configure automatic
1289	** retry counts and intervals for certain disk I/O operations for the
1290	** windows [VFS] in order to work to provide robustness against
1291	** anti-virus programs. By default, the windows VFS will retry file read,
1292	** file write, and file delete opertions up to 10 times, with a delay
1293	** of 25 milliseconds before the first retry and with the delay increasing
1294	** by an additional 25 milliseconds with each subsequent retry. This
1295	** opcode allows those to values (10 retries and 25 milliseconds of delay)
1296	** to be adjusted. The values are changed for all database connections
1297	** within the same process. The argument is a pointer to an array of two
1298	** integers where the first integer i the new retry count and the second
1299	** integer is the delay. If either integer is negative, then the setting
1300	** is not changed but instead the prior value of that setting is written
1301	** into the array entry, allowing the current retry settings to be
1302	** interrogated. The zDbName parameter is ignored.
1303	**
1304	*/
1305	#define SQLITE_FCNTL_LOCKSTATE 1
1306	#define SQLITE_GET_LOCKPROXYFILE 2
1307	#define SQLITE_SET_LOCKPROXYFILE 3
1308	#define SQLITE_LAST_ERRNO 4
1309	#define SQLITE_FCNTL_SIZE_HINT 5
1310	#define SQLITE_FCNTL_CHUNK_SIZE 6
1311	#define SQLITE_FCNTL_FILE_POINTER 7
1312	#define SQLITE_FCNTL_SYNC_OMITTED 8
1313	#define SQLITE_FCNTL_WIN32_AV_RETRY 9
1314
1315	/*
1316	** CAPI3REF: Mutex Handle
1317	**
1318	** The mutex module within SQLite defines [sqlite3_mutex] to be an
	@@ -9558,10 +9575,11 @@
9575	#define SQLITE_IndexSearch 0x08 /* Disable indexes for searching */
9576	#define SQLITE_IndexCover 0x10 /* Disable index covering table */
9577	#define SQLITE_GroupByOrder 0x20 /* Disable GROUPBY cover of ORDERBY */
9578	#define SQLITE_FactorOutConst 0x40 /* Disable factoring out constants */
9579	#define SQLITE_IdxRealAsInt 0x80 /* Store REAL as INT in indices */
9580	#define SQLITE_DistinctOpt 0x80 /* DISTINCT using indexes */
9581	#define SQLITE_OptMask 0xff /* Mask of all disablable opts */
9582
9583	/*
9584	** Possible values for the sqlite.magic field.
9585	** The numbers are obtained at random and have no special meaning, other
	@@ -10449,10 +10467,11 @@
10467	Table pTab; / An SQL table corresponding to zName */
10468	Select pSelect; / A SELECT statement used in place of a table name */
10469	u8 isPopulated; /* Temporary table associated with SELECT is populated */
10470	u8 jointype; /* Type of join between this able and the previous */
10471	u8 notIndexed; /* True if there is a NOT INDEXED clause */
10472	u8 isCorrelated; /* True if sub-query is correlated */
10473	#ifndef SQLITE_OMIT_EXPLAIN
10474	u8 iSelectId; /* If pSelect!=0, the id of the sub-select in EQP */
10475	#endif
10476	int iCursor; /* The VDBE cursor number used to access this table */
10477	Expr pOn; / The ON clause of a join */
	@@ -10568,10 +10587,11 @@
10587	struct WhereInfo {
10588	Parse pParse; / Parsing and code generating context */
10589	u16 wctrlFlags; /* Flags originally passed to sqlite3WhereBegin() */
10590	u8 okOnePass; /* Ok to use one-pass algorithm for UPDATE or DELETE */
10591	u8 untestedTerms; /* Not all WHERE terms resolved by outer loop */
10592	u8 eDistinct;
10593	SrcList pTabList; / List of tables in the join */
10594	int iTop; /* The very beginning of the WHERE loop */
10595	int iContinue; /* Jump here to continue with next record */
10596	int iBreak; /* Jump here to break out of the loop */
10597	int nLevel; /* Number of nested loop */
	@@ -10579,10 +10599,13 @@
10599	double savedNQueryLoop; /* pParse->nQueryLoop outside the WHERE loop */
10600	double nRowOut; /* Estimated number of output rows */
10601	WhereLevel a[1]; /* Information about each nest loop in WHERE */
10602	};
10603
10604	#define WHERE_DISTINCT_UNIQUE 1
10605	#define WHERE_DISTINCT_ORDERED 2
10606
10607	/*
10608	** A NameContext defines a context in which to resolve table and column
10609	** names. The context consists of a list of tables (the pSrcList) field and
10610	** a list of named expression (pEList). The named expression list may
10611	** be NULL. The pSrc corresponds to the FROM clause of a SELECT or
	@@ -11340,11 +11363,11 @@
11363	#if defined(SQLITE_ENABLE_UPDATE_DELETE_LIMIT) && !defined(SQLITE_OMIT_SUBQUERY)
11364	SQLITE_PRIVATE Expr sqlite3LimitWhere(Parse , SrcList , Expr , ExprList , Expr , Expr , char );
11365	#endif
11366	SQLITE_PRIVATE void sqlite3DeleteFrom(Parse, SrcList, Expr*);
11367	SQLITE_PRIVATE void sqlite3Update(Parse, SrcList, ExprList, Expr, int);
11368	SQLITE_PRIVATE WhereInfo sqlite3WhereBegin(Parse, SrcList, Expr, ExprList*,ExprList,u16);
11369	SQLITE_PRIVATE void sqlite3WhereEnd(WhereInfo*);
11370	SQLITE_PRIVATE int sqlite3ExprCodeGetColumn(Parse, Table, int, int, int);
11371	SQLITE_PRIVATE void sqlite3ExprCodeGetColumnOfTable(Vdbe, Table, int, int, int);
11372	SQLITE_PRIVATE void sqlite3ExprCodeMove(Parse*, int, int, int);
11373	SQLITE_PRIVATE void sqlite3ExprCodeCopy(Parse*, int, int, int);
	@@ -15841,11 +15864,11 @@
15864	** Free an outstanding memory allocation.
15865	**
15866	** This function assumes that the necessary mutexes, if any, are
15867	** already held by the caller. Hence "Unsafe".
15868	*/
15869	static void memsys3FreeUnsafe(void *pOld){
15870	Mem3Block p = (Mem3Block)pOld;
15871	int i;
15872	u32 size, x;
15873	assert( sqlite3_mutex_held(mem3.mutex) );
15874	assert( p>mem3.aPool && p<&mem3.aPool[mem3.nPool] );
	@@ -15916,21 +15939,21 @@
15939	}
15940
15941	/*
15942	** Free memory.
15943	*/
15944	static void memsys3Free(void *pPrior){
15945	assert( pPrior );
15946	memsys3Enter();
15947	memsys3FreeUnsafe(pPrior);
15948	memsys3Leave();
15949	}
15950
15951	/*
15952	** Change the size of an existing memory allocation
15953	*/
15954	static void memsys3Realloc(void pPrior, int nBytes){
15955	int nOld;
15956	void *p;
15957	if( pPrior==0 ){
15958	return sqlite3_malloc(nBytes);
15959	}
	@@ -25143,11 +25166,13 @@
25166	case EINVAL:
25167	case ENOTCONN:
25168	case ENODEV:
25169	case ENXIO:
25170	case ENOENT:
25171	#ifdef ESTALE /* ESTALE is not defined on Interix systems */
25172	case ESTALE:
25173	#endif
25174	case ENOSYS:
25175	/* these should force the client to close the file and reconnect */
25176
25177	default:
25178	return sqliteIOErr;
	@@ -31820,10 +31845,58 @@
31845	iLine, iErrno, zFunc, zPath, zMsg
31846	);
31847
31848	return errcode;
31849	}
31850
31851	/*
31852	** The number of times that a ReadFile(), WriteFile(), and DeleteFile()
31853	** will be retried following a locking error - probably caused by
31854	** antivirus software. Also the initial delay before the first retry.
31855	** The delay increases linearly with each retry.
31856	*/
31857	#ifndef SQLITE_WIN32_IOERR_RETRY
31858	# define SQLITE_WIN32_IOERR_RETRY 10
31859	#endif
31860	#ifndef SQLITE_WIN32_IOERR_RETRY_DELAY
31861	# define SQLITE_WIN32_IOERR_RETRY_DELAY 25
31862	#endif
31863	static int win32IoerrRetry = SQLITE_WIN32_IOERR_RETRY;
31864	static int win32IoerrRetryDelay = SQLITE_WIN32_IOERR_RETRY_DELAY;
31865
31866	/*
31867	** If a ReadFile() or WriteFile() error occurs, invoke this routine
31868	** to see if it should be retried. Return TRUE to retry. Return FALSE
31869	** to give up with an error.
31870	*/
31871	static int retryIoerr(int *pnRetry){
31872	DWORD e;
31873	if( *pnRetry>=win32IoerrRetry ){
31874	return 0;
31875	}
31876	e = GetLastError();
31877	if( e==ERROR_ACCESS_DENIED \|\|
31878	e==ERROR_LOCK_VIOLATION \|\|
31879	e==ERROR_SHARING_VIOLATION ){
31880	Sleep(win32IoerrRetryDelay(1+pnRetry));
31881	++*pnRetry;
31882	return 1;
31883	}
31884	return 0;
31885	}
31886
31887	/*
31888	** Log a I/O error retry episode.
31889	*/
31890	static void logIoerr(int nRetry){
31891	if( nRetry ){
31892	sqlite3_log(SQLITE_IOERR,
31893	"delayed %dms for lock/sharing conflict",
31894	win32IoerrRetryDelaynRetry(nRetry+1)/2
31895	);
31896	}
31897	}
31898
31899	#if SQLITE_OS_WINCE
31900	/*************************************************************************
31901	** This section contains code for WinCE only.
31902	*/
	@@ -32238,22 +32311,25 @@
32311	int amt, /* Number of bytes to read */
32312	sqlite3_int64 offset /* Begin reading at this offset */
32313	){
32314	winFile pFile = (winFile)id; /* file handle */
32315	DWORD nRead; /* Number of bytes actually read from file */
32316	int nRetry = 0; /* Number of retrys */
32317
32318	assert( id!=0 );
32319	SimulateIOError(return SQLITE_IOERR_READ);
32320	OSTRACE(("READ %d lock=%d\n", pFile->h, pFile->locktype));
32321
32322	if( seekWinFile(pFile, offset) ){
32323	return SQLITE_FULL;
32324	}
32325	while( !ReadFile(pFile->h, pBuf, amt, &nRead, 0) ){
32326	if( retryIoerr(&nRetry) ) continue;
32327	pFile->lastErrno = GetLastError();
32328	return winLogError(SQLITE_IOERR_READ, "winRead", pFile->zPath);
32329	}
32330	logIoerr(nRetry);
32331	if( nRead<(DWORD)amt ){
32332	/* Unread parts of the buffer must be zero-filled */
32333	memset(&((char*)pBuf)[nRead], 0, amt-nRead);
32334	return SQLITE_IOERR_SHORT_READ;
32335	}
	@@ -32271,10 +32347,11 @@
32347	int amt, /* Number of bytes to write */
32348	sqlite3_int64 offset /* Offset into the file to begin writing at */
32349	){
32350	int rc; /* True if error has occured, else false */
32351	winFile pFile = (winFile)id; /* File handle */
32352	int nRetry = 0; /* Number of retries */
32353
32354	assert( amt>0 );
32355	assert( pFile );
32356	SimulateIOError(return SQLITE_IOERR_WRITE);
32357	SimulateDiskfullError(return SQLITE_FULL);
	@@ -32285,11 +32362,16 @@
32362	if( rc==0 ){
32363	u8 aRem = (u8 )pBuf; /* Data yet to be written */
32364	int nRem = amt; /* Number of bytes yet to be written */
32365	DWORD nWrite; /* Bytes written by each WriteFile() call */
32366
32367	while( nRem>0 ){
32368	if( !WriteFile(pFile->h, aRem, nRem, &nWrite, 0) ){
32369	if( retryIoerr(&nRetry) ) continue;
32370	break;
32371	}
32372	if( nWrite<=0 ) break;
32373	aRem += nWrite;
32374	nRem -= nWrite;
32375	}
32376	if( nRem>0 ){
32377	pFile->lastErrno = GetLastError();
	@@ -32301,10 +32383,12 @@
32383	if( ( pFile->lastErrno==ERROR_HANDLE_DISK_FULL )
32384	\|\| ( pFile->lastErrno==ERROR_DISK_FULL )){
32385	return SQLITE_FULL;
32386	}
32387	return winLogError(SQLITE_IOERR_WRITE, "winWrite", pFile->zPath);
32388	}else{
32389	logIoerr(nRetry);
32390	}
32391	return SQLITE_OK;
32392	}
32393
32394	/*
	@@ -32716,10 +32800,24 @@
32800	SimulateIOErrorBenign(0);
32801	return SQLITE_OK;
32802	}
32803	case SQLITE_FCNTL_SYNC_OMITTED: {
32804	return SQLITE_OK;
32805	}
32806	case SQLITE_FCNTL_WIN32_AV_RETRY: {
32807	int a = (int)pArg;
32808	if( a[0]>0 ){
32809	win32IoerrRetry = a[0];
32810	}else{
32811	a[0] = win32IoerrRetry;
32812	}
32813	if( a[1]>0 ){
32814	win32IoerrRetryDelay = a[1];
32815	}else{
32816	a[1] = win32IoerrRetryDelay;
32817	}
32818	return SQLITE_OK;
32819	}
32820	}
32821	return SQLITE_NOTFOUND;
32822	}
32823
	@@ -33734,19 +33832,17 @@
33832	** file open, we will be unable to delete it. To work around this
33833	** problem, we delay 100 milliseconds and try to delete again. Up
33834	** to MX_DELETION_ATTEMPTs deletion attempts are run before giving
33835	** up and returning an error.
33836	*/

33837	static int winDelete(
33838	sqlite3_vfs pVfs, / Not used on win32 */
33839	const char zFilename, / Name of file to delete */
33840	int syncDir /* Not used on win32 */
33841	){
33842	int cnt = 0;
33843	int rc;

33844	void *zConverted;
33845	UNUSED_PARAMETER(pVfs);
33846	UNUSED_PARAMETER(syncDir);
33847
33848	SimulateIOError(return SQLITE_IOERR_DELETE);
	@@ -33753,38 +33849,34 @@
33849	zConverted = convertUtf8Filename(zFilename);
33850	if( zConverted==0 ){
33851	return SQLITE_NOMEM;
33852	}
33853	if( isNT() ){
33854	rc = 1;
33855	while( GetFileAttributesW(zConverted)!=INVALID_FILE_ATTRIBUTES &&
33856	(rc = DeleteFileW(zConverted))==0 && retryIoerr(&cnt) ){}
33857	rc = rc ? SQLITE_OK : SQLITE_ERROR;


33858	/* isNT() is 1 if SQLITE_OS_WINCE==1, so this else is never executed.
33859	** Since the ASCII version of these Windows API do not exist for WINCE,
33860	** it's important to not reference them for WINCE builds.
33861	*/
33862	#if SQLITE_OS_WINCE==0
33863	}else{
33864	rc = 1;
33865	while( GetFileAttributesA(zConverted)!=INVALID_FILE_ATTRIBUTES &&
33866	(rc = DeleteFileA(zConverted))==0 && retryIoerr(&cnt) ){}
33867	rc = rc ? SQLITE_OK : SQLITE_ERROR;


33868	#endif
33869	}
33870	if( rc ){
33871	rc = winLogError(SQLITE_IOERR_DELETE, "winDelete", zFilename);
33872	}else{
33873	logIoerr(cnt);
33874	}
33875	free(zConverted);
33876	OSTRACE(("DELETE \"%s\" %s\n", zFilename, (rc ? "failed" : "ok" )));
33877	return rc;





33878	}
33879
33880	/*
33881	** Check the existance and status of a file.
33882	*/
	@@ -59049,10 +59141,11 @@
59141	nMem = 10;
59142	}
59143	memset(zCsr, 0, zEnd-zCsr);
59144	zCsr += (zCsr - (u8*)0)&7;
59145	assert( EIGHT_BYTE_ALIGNMENT(zCsr) );
59146	p->expired = 0;
59147
59148	/* Memory for registers, parameters, cursor, etc, is allocated in two
59149	** passes. On the first pass, we try to reuse unused space at the
59150	** end of the opcode array. If we are unable to satisfy all memory
59151	** requirements by reusing the opcode array tail, then the second
	@@ -61277,11 +61370,11 @@
61370	sqlite3_mutex_enter(db->mutex);
61371	while( (rc = sqlite3Step(v))==SQLITE_SCHEMA
61372	&& cnt++ < SQLITE_MAX_SCHEMA_RETRY
61373	&& (rc2 = rc = sqlite3Reprepare(v))==SQLITE_OK ){
61374	sqlite3_reset(pStmt);
61375	assert( v->expired==0 );
61376	}
61377	if( rc2!=SQLITE_OK && ALWAYS(v->isPrepareV2) && ALWAYS(db->pErr) ){
61378	/* This case occurs after failing to recompile an sql statement.
61379	** The error message from the SQL compiler has already been loaded
61380	** into the database handle. This block copies the error message
	@@ -65968,11 +66061,11 @@
66061	** automatically created table with root-page 1 (an BLOB_INTKEY table).
66062	*/
66063	if( pOp->p4.pKeyInfo ){
66064	int pgno;
66065	assert( pOp->p4type==P4_KEYINFO );
66066	rc = sqlite3BtreeCreateTable(u.ax.pCx->pBt, &pgno, BTREE_BLOBKEY \| pOp->p5);
66067	if( rc==SQLITE_OK ){
66068	assert( pgno==MASTER_ROOT+1 );
66069	rc = sqlite3BtreeCursor(u.ax.pCx->pBt, pgno, 1,
66070	(KeyInfo*)pOp->p4.z, u.ax.pCx->pCursor);
66071	u.ax.pCx->pKeyInfo = pOp->p4.pKeyInfo;
	@@ -71007,15 +71100,29 @@
71100	/* Recursively resolve names in all subqueries
71101	*/
71102	for(i=0; i<p->pSrc->nSrc; i++){
71103	struct SrcList_item *pItem = &p->pSrc->a[i];
71104	if( pItem->pSelect ){
71105	NameContext pNC; / Used to iterate name contexts */
71106	int nRef = 0; /* Refcount for pOuterNC and outer contexts */
71107	const char *zSavedContext = pParse->zAuthContext;
71108
71109	/* Count the total number of references to pOuterNC and all of its
71110	** parent contexts. After resolving references to expressions in
71111	** pItem->pSelect, check if this value has changed. If so, then
71112	** SELECT statement pItem->pSelect must be correlated. Set the
71113	** pItem->isCorrelated flag if this is the case. */
71114	for(pNC=pOuterNC; pNC; pNC=pNC->pNext) nRef += pNC->nRef;
71115
71116	if( pItem->zName ) pParse->zAuthContext = pItem->zName;
71117	sqlite3ResolveSelectNames(pParse, pItem->pSelect, pOuterNC);
71118	pParse->zAuthContext = zSavedContext;
71119	if( pParse->nErr \|\| db->mallocFailed ) return WRC_Abort;
71120
71121	for(pNC=pOuterNC; pNC; pNC=pNC->pNext) nRef -= pNC->nRef;
71122	assert( pItem->isCorrelated==0 && nRef<=0 );
71123	pItem->isCorrelated = (nRef!=0);
71124	}
71125	}
71126
71127	/* If there are no aggregate functions in the result-set, and no GROUP BY
71128	** expression, do not allow aggregates in any of the other expressions.
	@@ -72119,10 +72226,11 @@
72226	pNewItem->zName = sqlite3DbStrDup(db, pOldItem->zName);
72227	pNewItem->zAlias = sqlite3DbStrDup(db, pOldItem->zAlias);
72228	pNewItem->jointype = pOldItem->jointype;
72229	pNewItem->iCursor = pOldItem->iCursor;
72230	pNewItem->isPopulated = pOldItem->isPopulated;
72231	pNewItem->isCorrelated = pOldItem->isCorrelated;
72232	pNewItem->zIndex = sqlite3DbStrDup(db, pOldItem->zIndex);
72233	pNewItem->notIndexed = pOldItem->notIndexed;
72234	pNewItem->pIndex = pOldItem->pIndex;
72235	pTab = pNewItem->pTab = pOldItem->pTab;
72236	if( pTab ){
	@@ -80126,11 +80234,11 @@
80234	if( pStart ){
80235	assert( pEnd!=0 );
80236	/* A named index with an explicit CREATE INDEX statement */
80237	zStmt = sqlite3MPrintf(db, "CREATE%s INDEX %.*s",
80238	onError==OE_None ? "" : " UNIQUE",
80239	(int)(pEnd->z - pName->z) + 1,
80240	pName->z);
80241	}else{
80242	/* An automatic index created by a PRIMARY KEY or UNIQUE constraint */
80243	/* zStmt = sqlite3MPrintf(""); */
80244	zStmt = 0;
	@@ -81921,11 +82029,13 @@
82029	int regRowid; /* Actual register containing rowids */
82030
82031	/* Collect rowids of every row to be deleted.
82032	*/
82033	sqlite3VdbeAddOp2(v, OP_Null, 0, iRowSet);
82034	pWInfo = sqlite3WhereBegin(
82035	pParse, pTabList, pWhere, 0, 0, WHERE_DUPLICATES_OK
82036	);
82037	if( pWInfo==0 ) goto delete_from_cleanup;
82038	regRowid = sqlite3ExprCodeGetColumn(pParse, pTab, -1, iCur, iRowid);
82039	sqlite3VdbeAddOp2(v, OP_RowSetAdd, iRowSet, regRowid);
82040	if( db->flags & SQLITE_CountRows ){
82041	sqlite3VdbeAddOp2(v, OP_AddImm, memCnt, 1);
	@@ -84368,11 +84478,11 @@
84478
84479	/* Create VDBE to loop through the entries in pSrc that match the WHERE
84480	** clause. If the constraint is not deferred, throw an exception for
84481	** each row found. Otherwise, for deferred constraints, increment the
84482	** deferred constraint counter by nIncr for each row selected. */
84483	pWInfo = sqlite3WhereBegin(pParse, pSrc, pWhere, 0, 0, 0);
84484	if( nIncr>0 && pFKey->isDeferred==0 ){
84485	sqlite3ParseToplevel(pParse)->mayAbort = 1;
84486	}
84487	sqlite3VdbeAddOp2(v, OP_FkCounter, pFKey->isDeferred, nIncr);
84488	if( pWInfo ){
	@@ -87235,10 +87345,13 @@
87345	int (strnicmp)(const char,const char*,int);
87346	int (unlock_notify)(sqlite3,void()(void,int),void);
87347	int (wal_autocheckpoint)(sqlite3,int);
87348	int (wal_checkpoint)(sqlite3,const char*);
87349	void (wal_hook)(sqlite3,int()(void,sqlite3,const char,int),void);
87350	int (blob_reopen)(sqlite3_blob,sqlite3_int64);
87351	int (vtab_config)(sqlite3,int op,...);
87352	int (vtab_on_conflict)(sqlite3);
87353	};
87354
87355	/*
87356	** The following macros redefine the API routines so that they are
87357	** redirected throught the global sqlite3_api structure.
	@@ -87435,10 +87548,13 @@
87548	#define sqlite3_strnicmp sqlite3_api->strnicmp
87549	#define sqlite3_unlock_notify sqlite3_api->unlock_notify
87550	#define sqlite3_wal_autocheckpoint sqlite3_api->wal_autocheckpoint
87551	#define sqlite3_wal_checkpoint sqlite3_api->wal_checkpoint
87552	#define sqlite3_wal_hook sqlite3_api->wal_hook
87553	#define sqlite3_blob_reopen sqlite3_api->blob_reopen
87554	#define sqlite3_vtab_config sqlite3_api->vtab_config
87555	#define sqlite3_vtab_on_conflict sqlite3_api->vtab_on_conflict
87556	#endif /* SQLITE_CORE */
87557
87558	#define SQLITE_EXTENSION_INIT1 const sqlite3_api_routines *sqlite3_api = 0;
87559	#define SQLITE_EXTENSION_INIT2(v) sqlite3_api = v;
87560
	@@ -87509,10 +87625,12 @@
87625
87626	#ifdef SQLITE_OMIT_VIRTUALTABLE
87627	# define sqlite3_create_module 0
87628	# define sqlite3_create_module_v2 0
87629	# define sqlite3_declare_vtab 0
87630	# define sqlite3_vtab_config 0
87631	# define sqlite3_vtab_on_conflict 0
87632	#endif
87633
87634	#ifdef SQLITE_OMIT_SHARED_CACHE
87635	# define sqlite3_enable_shared_cache 0
87636	#endif
	@@ -87532,10 +87650,11 @@
87650	#define sqlite3_blob_bytes 0
87651	#define sqlite3_blob_close 0
87652	#define sqlite3_blob_open 0
87653	#define sqlite3_blob_read 0
87654	#define sqlite3_blob_write 0
87655	#define sqlite3_blob_reopen 0
87656	#endif
87657
87658	/*
87659	** The following structure contains pointers to all SQLite API routines.
87660	** A pointer to this structure is passed into extensions when they are
	@@ -87797,10 +87916,13 @@
87916	#else
87917	0,
87918	0,
87919	0,
87920	#endif
87921	sqlite3_blob_reopen,
87922	sqlite3_vtab_config,
87923	sqlite3_vtab_on_conflict,
87924	};
87925
87926	/*
87927	** Attempt to load an SQLite extension library contained in the file
87928	** zFile. The entry point is zProc. zProc may be 0 in which case a
	@@ -94186,10 +94308,11 @@
94308	Expr pHaving; / The HAVING clause. May be NULL */
94309	int isDistinct; /* True if the DISTINCT keyword is present */
94310	int distinct; /* Table to use for the distinct set */
94311	int rc = 1; /* Value to return from this function */
94312	int addrSortIndex; /* Address of an OP_OpenEphemeral instruction */
94313	int addrDistinctIndex; /* Address of an OP_OpenEphemeral instruction */
94314	AggInfo sAggInfo; /* Information used by aggregate queries */
94315	int iEnd; /* Address of the end of the query */
94316	sqlite3 db; / The database connection */
94317
94318	#ifndef SQLITE_OMIT_EXPLAIN
	@@ -94312,20 +94435,10 @@
94435	explainSetInteger(pParse->iSelectId, iRestoreSelectId);
94436	return rc;
94437	}
94438	#endif
94439










94440	/* If there is both a GROUP BY and an ORDER BY clause and they are
94441	** identical, then disable the ORDER BY clause since the GROUP BY
94442	** will cause elements to come out in the correct order. This is
94443	** an optimization - the correct answer should result regardless.
94444	** Use the SQLITE_GroupByOrder flag with SQLITE_TESTCTRL_OPTIMIZER
	@@ -94333,10 +94446,34 @@
94446	*/
94447	if( sqlite3ExprListCompare(p->pGroupBy, pOrderBy)==0
94448	&& (db->flags & SQLITE_GroupByOrder)==0 ){
94449	pOrderBy = 0;
94450	}
94451
94452	/* If the query is DISTINCT with an ORDER BY but is not an aggregate, and
94453	** if the select-list is the same as the ORDER BY list, then this query
94454	** can be rewritten as a GROUP BY. In other words, this:
94455	**
94456	** SELECT DISTINCT xyz FROM ... ORDER BY xyz
94457	**
94458	** is transformed to:
94459	**
94460	** SELECT xyz FROM ... GROUP BY xyz
94461	**
94462	** The second form is preferred as a single index (or temp-table) may be
94463	** used for both the ORDER BY and DISTINCT processing. As originally
94464	** written the query must use a temp-table for at least one of the ORDER
94465	** BY and DISTINCT, and an index or separate temp-table for the other.
94466	*/
94467	if( (p->selFlags & (SF_Distinct\|SF_Aggregate))==SF_Distinct
94468	&& sqlite3ExprListCompare(pOrderBy, p->pEList)==0
94469	){
94470	p->selFlags &= ~SF_Distinct;
94471	p->pGroupBy = sqlite3ExprListDup(db, p->pEList, 0);
94472	pGroupBy = p->pGroupBy;
94473	pOrderBy = 0;
94474	}
94475
94476	/* If there is an ORDER BY clause, then this sorting
94477	** index might end up being unused if the data can be
94478	** extracted in pre-sorted order. If that is the case, then the
94479	** OP_OpenEphemeral instruction will be changed to an OP_Noop once
	@@ -94369,26 +94506,25 @@
94506
94507	/* Open a virtual index to use for the distinct set.
94508	*/
94509	if( p->selFlags & SF_Distinct ){
94510	KeyInfo *pKeyInfo;

94511	distinct = pParse->nTab++;
94512	pKeyInfo = keyInfoFromExprList(pParse, p->pEList);
94513	addrDistinctIndex = sqlite3VdbeAddOp4(v, OP_OpenEphemeral, distinct, 0, 0,
94514	(char*)pKeyInfo, P4_KEYINFO_HANDOFF);
94515	sqlite3VdbeChangeP5(v, BTREE_UNORDERED);
94516	}else{
94517	distinct = addrDistinctIndex = -1;
94518	}
94519
94520	/* Aggregate and non-aggregate queries are handled differently */
94521	if( !isAgg && pGroupBy==0 ){
94522	ExprList *pDist = (isDistinct ? p->pEList : 0);
94523
94524	/* Begin the database scan. */
94525	pWInfo = sqlite3WhereBegin(pParse, pTabList, pWhere, &pOrderBy, pDist, 0);
94526	if( pWInfo==0 ) goto select_end;
94527	if( pWInfo->nRowOut < p->nSelectRow ) p->nSelectRow = pWInfo->nRowOut;
94528
94529	/* If sorting index that was created by a prior OP_OpenEphemeral
94530	** instruction ended up not being needed, then change the OP_OpenEphemeral
	@@ -94397,14 +94533,56 @@
94533	if( addrSortIndex>=0 && pOrderBy==0 ){
94534	sqlite3VdbeChangeToNoop(v, addrSortIndex, 1);
94535	p->addrOpenEphm[2] = -1;
94536	}
94537
94538	if( pWInfo->eDistinct ){
94539	VdbeOp pOp; / No longer required OpenEphemeral instr. */
94540
94541	assert( addrDistinctIndex>0 );
94542	pOp = sqlite3VdbeGetOp(v, addrDistinctIndex);
94543
94544	assert( isDistinct );
94545	assert( pWInfo->eDistinct==WHERE_DISTINCT_ORDERED
94546	\|\| pWInfo->eDistinct==WHERE_DISTINCT_UNIQUE
94547	);
94548	distinct = -1;
94549	if( pWInfo->eDistinct==WHERE_DISTINCT_ORDERED ){
94550	int iJump;
94551	int iExpr;
94552	int iFlag = ++pParse->nMem;
94553	int iBase = pParse->nMem+1;
94554	int iBase2 = iBase + pEList->nExpr;
94555	pParse->nMem += (pEList->nExpr*2);
94556
94557	/* Change the OP_OpenEphemeral coded earlier to an OP_Integer. The
94558	** OP_Integer initializes the "first row" flag. */
94559	pOp->opcode = OP_Integer;
94560	pOp->p1 = 1;
94561	pOp->p2 = iFlag;
94562
94563	sqlite3ExprCodeExprList(pParse, pEList, iBase, 1);
94564	iJump = sqlite3VdbeCurrentAddr(v) + 1 + pEList->nExpr + 1 + 1;
94565	sqlite3VdbeAddOp2(v, OP_If, iFlag, iJump-1);
94566	for(iExpr=0; iExpr<pEList->nExpr; iExpr++){
94567	CollSeq *pColl = sqlite3ExprCollSeq(pParse, pEList->a[iExpr].pExpr);
94568	sqlite3VdbeAddOp3(v, OP_Ne, iBase+iExpr, iJump, iBase2+iExpr);
94569	sqlite3VdbeChangeP4(v, -1, (const char *)pColl, P4_COLLSEQ);
94570	sqlite3VdbeChangeP5(v, SQLITE_NULLEQ);
94571	}
94572	sqlite3VdbeAddOp2(v, OP_Goto, 0, pWInfo->iContinue);
94573
94574	sqlite3VdbeAddOp2(v, OP_Integer, 0, iFlag);
94575	assert( sqlite3VdbeCurrentAddr(v)==iJump );
94576	sqlite3VdbeAddOp3(v, OP_Move, iBase, iBase2, pEList->nExpr);
94577	}else{
94578	pOp->opcode = OP_Noop;
94579	}
94580	}
94581
94582	/* Use the standard inner loop. */
94583	selectInnerLoop(pParse, p, pEList, 0, 0, pOrderBy, distinct, pDest,
94584	pWInfo->iContinue, pWInfo->iBreak);
94585
94586	/* End the database scan loop.
94587	*/
94588	sqlite3WhereEnd(pWInfo);
	@@ -94510,11 +94688,11 @@
94688	** This might involve two separate loops with an OP_Sort in between, or
94689	** it might be a single loop that uses an index to extract information
94690	** in the right order to begin with.
94691	*/
94692	sqlite3VdbeAddOp2(v, OP_Gosub, regReset, addrReset);
94693	pWInfo = sqlite3WhereBegin(pParse, pTabList, pWhere, &pGroupBy, 0, 0);
94694	if( pWInfo==0 ) goto select_end;
94695	if( pGroupBy==0 ){
94696	/* The optimizer is able to deliver rows in group by order so
94697	** we do not have to sort. The OP_OpenEphemeral table will be
94698	** cancelled later because we still need to use the pKeyInfo
	@@ -94772,11 +94950,11 @@
94950	/* This case runs if the aggregate has no GROUP BY clause. The
94951	** processing is much simpler since there is only a single row
94952	** of output.
94953	*/
94954	resetAccumulator(pParse, &sAggInfo);
94955	pWInfo = sqlite3WhereBegin(pParse, pTabList, pWhere, &pMinMax, 0, flag);
94956	if( pWInfo==0 ){
94957	sqlite3ExprListDelete(db, pDel);
94958	goto select_end;
94959	}
94960	updateAccumulator(pParse, &sAggInfo);
	@@ -95248,19 +95426,32 @@
95426	iDb = sqlite3TwoPartName(pParse, pName1, pName2, &pName);
95427	if( iDb<0 ){
95428	goto trigger_cleanup;
95429	}
95430	}
95431	if( !pTableName \|\| db->mallocFailed ){
95432	goto trigger_cleanup;
95433	}
95434
95435	/* A long-standing parser bug is that this syntax was allowed:
95436	**
95437	** CREATE TRIGGER attached.demo AFTER INSERT ON attached.tab ....
95438	** ^^^^^^^^
95439	**
95440	** To maintain backwards compatibility, ignore the database
95441	** name on pTableName if we are reparsing our of SQLITE_MASTER.
95442	*/
95443	if( db->init.busy && iDb!=1 ){
95444	sqlite3DbFree(db, pTableName->a[0].zDatabase);
95445	pTableName->a[0].zDatabase = 0;
95446	}
95447
95448	/* If the trigger name was unqualified, and the table is a temp table,
95449	** then set iDb to 1 to create the trigger in the temporary database.
95450	** If sqlite3SrcListLookup() returns 0, indicating the table does not
95451	** exist, the error is caught by the block below.
95452	*/



95453	pTab = sqlite3SrcListLookup(pParse, pTableName);
95454	if( db->init.busy==0 && pName2->n==0 && pTab
95455	&& pTab->pSchema==db->aDb[1].pSchema ){
95456	iDb = 1;
95457	}
	@@ -96554,11 +96745,13 @@
96745	}
96746
96747	/* Begin the database scan
96748	*/
96749	sqlite3VdbeAddOp2(v, OP_Null, 0, regOldRowid);
96750	pWInfo = sqlite3WhereBegin(
96751	pParse, pTabList, pWhere, 0, 0, WHERE_ONEPASS_DESIRED
96752	);
96753	if( pWInfo==0 ) goto update_cleanup;
96754	okOnePass = pWInfo->okOnePass;
96755
96756	/* Remember the rowid of every item to be updated.
96757	*/
	@@ -98580,10 +98773,11 @@
98773	#define WHERE_REVERSE 0x02000000 /* Scan in reverse order */
98774	#define WHERE_UNIQUE 0x04000000 /* Selects no more than one row */
98775	#define WHERE_VIRTUALTABLE 0x08000000 /* Use virtual-table processing */
98776	#define WHERE_MULTI_OR 0x10000000 /* OR using multiple indices */
98777	#define WHERE_TEMP_INDEX 0x20000000 /* Uses an ephemeral index */
98778	#define WHERE_DISTINCT 0x40000000 /* Correct order for DISTINCT */
98779
98780	/*
98781	** Initialize a preallocated WhereClause structure.
98782	*/
98783	static void whereClauseInit(
	@@ -99724,10 +99918,166 @@
99918	}
99919	}
99920	return 0;
99921	}
99922
99923	/*
99924	** This function searches the expression list passed as the second argument
99925	** for an expression of type TK_COLUMN that refers to the same column and
99926	** uses the same collation sequence as the iCol'th column of index pIdx.
99927	** Argument iBase is the cursor number used for the table that pIdx refers
99928	** to.
99929	**
99930	** If such an expression is found, its index in pList->a[] is returned. If
99931	** no expression is found, -1 is returned.
99932	*/
99933	static int findIndexCol(
99934	Parse pParse, / Parse context */
99935	ExprList pList, / Expression list to search */
99936	int iBase, /* Cursor for table associated with pIdx */
99937	Index pIdx, / Index to match column of */
99938	int iCol /* Column of index to match */
99939	){
99940	int i;
99941	const char *zColl = pIdx->azColl[iCol];
99942
99943	for(i=0; i<pList->nExpr; i++){
99944	Expr *p = pList->a[i].pExpr;
99945	if( p->op==TK_COLUMN
99946	&& p->iColumn==pIdx->aiColumn[iCol]
99947	&& p->iTable==iBase
99948	){
99949	CollSeq *pColl = sqlite3ExprCollSeq(pParse, p);
99950	if( ALWAYS(pColl) && 0==sqlite3StrICmp(pColl->zName, zColl) ){
99951	return i;
99952	}
99953	}
99954	}
99955
99956	return -1;
99957	}
99958
99959	/*
99960	** This routine determines if pIdx can be used to assist in processing a
99961	** DISTINCT qualifier. In other words, it tests whether or not using this
99962	** index for the outer loop guarantees that rows with equal values for
99963	** all expressions in the pDistinct list are delivered grouped together.
99964	**
99965	** For example, the query
99966	**
99967	** SELECT DISTINCT a, b, c FROM tbl WHERE a = ?
99968	**
99969	** can benefit from any index on columns "b" and "c".
99970	*/
99971	static int isDistinctIndex(
99972	Parse pParse, / Parsing context */
99973	WhereClause pWC, / The WHERE clause */
99974	Index pIdx, / The index being considered */
99975	int base, /* Cursor number for the table pIdx is on */
99976	ExprList pDistinct, / The DISTINCT expressions */
99977	int nEqCol /* Number of index columns with == */
99978	){
99979	Bitmask mask = 0; /* Mask of unaccounted for pDistinct exprs */
99980	int i; /* Iterator variable */
99981
99982	if( pIdx->zName==0 \|\| pDistinct==0 \|\| pDistinct->nExpr>=BMS ) return 0;
99983	testcase( pDistinct->nExpr==BMS-1 );
99984
99985	/* Loop through all the expressions in the distinct list. If any of them
99986	** are not simple column references, return early. Otherwise, test if the
99987	** WHERE clause contains a "col=X" clause. If it does, the expression
99988	** can be ignored. If it does not, and the column does not belong to the
99989	** same table as index pIdx, return early. Finally, if there is no
99990	** matching "col=X" expression and the column is on the same table as pIdx,
99991	** set the corresponding bit in variable mask.
99992	*/
99993	for(i=0; i<pDistinct->nExpr; i++){
99994	WhereTerm *pTerm;
99995	Expr *p = pDistinct->a[i].pExpr;
99996	if( p->op!=TK_COLUMN ) return 0;
99997	pTerm = findTerm(pWC, p->iTable, p->iColumn, ~(Bitmask)0, WO_EQ, 0);
99998	if( pTerm ){
99999	Expr *pX = pTerm->pExpr;
100000	CollSeq *p1 = sqlite3BinaryCompareCollSeq(pParse, pX->pLeft, pX->pRight);
100001	CollSeq *p2 = sqlite3ExprCollSeq(pParse, p);
100002	if( p1==p2 ) continue;
100003	}
100004	if( p->iTable!=base ) return 0;
100005	mask \|= (((Bitmask)1) << i);
100006	}
100007
100008	for(i=nEqCol; mask && i<pIdx->nColumn; i++){
100009	int iExpr = findIndexCol(pParse, pDistinct, base, pIdx, i);
100010	if( iExpr<0 ) break;
100011	mask &= ~(((Bitmask)1) << iExpr);
100012	}
100013
100014	return (mask==0);
100015	}
100016
100017
100018	/*
100019	** Return true if the DISTINCT expression-list passed as the third argument
100020	** is redundant. A DISTINCT list is redundant if the database contains a
100021	** UNIQUE index that guarantees that the result of the query will be distinct
100022	** anyway.
100023	*/
100024	static int isDistinctRedundant(
100025	Parse *pParse,
100026	SrcList *pTabList,
100027	WhereClause *pWC,
100028	ExprList *pDistinct
100029	){
100030	Table *pTab;
100031	Index *pIdx;
100032	int i;
100033	int iBase;
100034
100035	/* If there is more than one table or sub-select in the FROM clause of
100036	** this query, then it will not be possible to show that the DISTINCT
100037	** clause is redundant. */
100038	if( pTabList->nSrc!=1 ) return 0;
100039	iBase = pTabList->a[0].iCursor;
100040	pTab = pTabList->a[0].pTab;
100041
100042	/* If any of the expressions is an IPK column on table iBase, then return
100043	** true. Note: The (p->iTable==iBase) part of this test may be false if the
100044	** current SELECT is a correlated sub-query.
100045	*/
100046	for(i=0; i<pDistinct->nExpr; i++){
100047	Expr *p = pDistinct->a[i].pExpr;
100048	if( p->op==TK_COLUMN && p->iTable==iBase && p->iColumn<0 ) return 1;
100049	}
100050
100051	/* Loop through all indices on the table, checking each to see if it makes
100052	** the DISTINCT qualifier redundant. It does so if:
100053	**
100054	** 1. The index is itself UNIQUE, and
100055	**
100056	** 2. All of the columns in the index are either part of the pDistinct
100057	** list, or else the WHERE clause contains a term of the form "col=X",
100058	** where X is a constant value. The collation sequences of the
100059	** comparison and select-list expressions must match those of the index.
100060	*/
100061	for(pIdx=pTab->pIndex; pIdx; pIdx=pIdx->pNext){
100062	if( pIdx->onError==OE_None ) continue;
100063	for(i=0; i<pIdx->nColumn; i++){
100064	int iCol = pIdx->aiColumn[i];
100065	if( 0==findTerm(pWC, iBase, iCol, ~(Bitmask)0, WO_EQ, pIdx)
100066	&& 0>findIndexCol(pParse, pDistinct, iBase, pIdx, i)
100067	){
100068	break;
100069	}
100070	}
100071	if( i==pIdx->nColumn ){
100072	/* This index implies that the DISTINCT qualifier is redundant. */
100073	return 1;
100074	}
100075	}
100076
100077	return 0;
100078	}
100079
100080	/*
100081	** This routine decides if pIdx can be used to satisfy the ORDER BY
100082	** clause. If it can, it returns 1. If pIdx cannot satisfy the
100083	** ORDER BY clause, this routine returns 0.
	@@ -99760,11 +100110,14 @@
100110	int sortOrder = 0; /* XOR of index and ORDER BY sort direction */
100111	int nTerm; /* Number of ORDER BY terms */
100112	struct ExprList_item pTerm; / A term of the ORDER BY clause */
100113	sqlite3 *db = pParse->db;
100114
100115	if( !pOrderBy ) return 0;
100116	if( wsFlags & WHERE_COLUMN_IN ) return 0;
100117	if( pIdx->bUnordered ) return 0;
100118
100119	nTerm = pOrderBy->nExpr;
100120	assert( nTerm>0 );
100121
100122	/* Argument pIdx must either point to a 'real' named index structure,
100123	** or an index structure allocated on the stack by bestBtreeIndex() to
	@@ -100073,10 +100426,14 @@
100426	double costTempIdx; /* per-query cost of the transient index */
100427	WhereTerm pTerm; / A single term of the WHERE clause */
100428	WhereTerm pWCEnd; / End of pWC->a[] */
100429	Table pTable; / Table tht might be indexed */
100430
100431	if( pParse->nQueryLoop<=(double)1 ){
100432	/* There is no point in building an automatic index for a single scan */
100433	return;
100434	}
100435	if( (pParse->db->flags & SQLITE_AutoIndex)==0 ){
100436	/* Automatic indices are disabled at run-time */
100437	return;
100438	}
100439	if( (pCost->plan.wsFlags & WHERE_NOT_FULLSCAN)!=0 ){
	@@ -100084,10 +100441,14 @@
100441	return;
100442	}
100443	if( pSrc->notIndexed ){
100444	/* The NOT INDEXED clause appears in the SQL. */
100445	return;
100446	}
100447	if( pSrc->isCorrelated ){
100448	/* The source is a correlated sub-query. No point in indexing it. */
100449	return;
100450	}
100451
100452	assert( pParse->nQueryLoop >= (double)1 );
100453	pTable = pSrc->pTab;
100454	nTableRow = pTable->nRowEst;
	@@ -101016,10 +101377,11 @@
101377	WhereClause pWC, / The WHERE clause */
101378	struct SrcList_item pSrc, / The FROM clause term to search */
101379	Bitmask notReady, /* Mask of cursors not available for indexing */
101380	Bitmask notValid, /* Cursors not available for any purpose */
101381	ExprList pOrderBy, / The ORDER BY clause */
101382	ExprList pDistinct, / The select-list if query is DISTINCT */
101383	WhereCost pCost / Lowest cost query plan */
101384	){
101385	int iCur = pSrc->iCursor; /* The cursor of the table to be accessed */
101386	Index pProbe; / An index we are evaluating */
101387	Index pIdx; / Copy of pProbe, or zero for IPK index */
	@@ -101156,11 +101518,12 @@
101518	int nEq; /* Number of == or IN terms matching index */
101519	int bInEst = 0; /* True if "x IN (SELECT...)" seen */
101520	int nInMul = 1; /* Number of distinct equalities to lookup */
101521	int estBound = 100; /* Estimated reduction in search space */
101522	int nBound = 0; /* Number of range constraints seen */
101523	int bSort = !!pOrderBy; /* True if external sort required */
101524	int bDist = !!pDistinct; /* True if index cannot help with DISTINCT */
101525	int bLookup = 0; /* True if not a covering index */
101526	WhereTerm pTerm; / A single term of the WHERE clause */
101527	#ifdef SQLITE_ENABLE_STAT2
101528	WhereTerm pFirstTerm = 0; / First term matching the index */
101529	#endif
	@@ -101220,21 +101583,24 @@
101583
101584	/* If there is an ORDER BY clause and the index being considered will
101585	** naturally scan rows in the required order, set the appropriate flags
101586	** in wsFlags. Otherwise, if there is an ORDER BY clause but the index
101587	** will scan rows in a different order, set the bSort variable. */
101588	if( isSortingIndex(
101589	pParse, pWC->pMaskSet, pProbe, iCur, pOrderBy, nEq, wsFlags, &rev)
101590	){
101591	bSort = 0;
101592	wsFlags \|= WHERE_ROWID_RANGE\|WHERE_COLUMN_RANGE\|WHERE_ORDERBY;
101593	wsFlags \|= (rev ? WHERE_REVERSE : 0);
101594	}
101595
101596	/* If there is a DISTINCT qualifier and this index will scan rows in
101597	** order of the DISTINCT expressions, clear bDist and set the appropriate
101598	** flags in wsFlags. */
101599	if( isDistinctIndex(pParse, pWC, pProbe, iCur, pDistinct, nEq) ){
101600	bDist = 0;
101601	wsFlags \|= WHERE_ROWID_RANGE\|WHERE_COLUMN_RANGE\|WHERE_DISTINCT;
101602	}
101603
101604	/* If currently calculating the cost of using an index (not the IPK
101605	** index), determine if all required column data may be obtained without
101606	** using the main table (i.e. if the index is a covering
	@@ -101265,16 +101631,17 @@
101631	nRow = aiRowEst[0]/2;
101632	nInMul = (int)(nRow / aiRowEst[nEq]);
101633	}
101634
101635	#ifdef SQLITE_ENABLE_STAT2
101636	/* If the constraint is of the form x=VALUE or x IN (E1,E2,...)
101637	** and we do not think that values of x are unique and if histogram
101638	** data is available for column x, then it might be possible
101639	** to get a better estimate on the number of rows based on
101640	** VALUE and how common that value is according to the histogram.
101641	*/
101642	if( nRow>(double)1 && nEq==1 && pFirstTerm!=0 && aiRowEst[1]>1 ){
101643	if( pFirstTerm->eOperator & (WO_EQ\|WO_ISNULL) ){
101644	testcase( pFirstTerm->eOperator==WO_EQ );
101645	testcase( pFirstTerm->eOperator==WO_ISNULL );
101646	whereEqualScanEst(pParse, pProbe, pFirstTerm->pExpr->pRight, &nRow);
101647	}else if( pFirstTerm->eOperator==WO_IN && bInEst==0 ){
	@@ -101347,10 +101714,13 @@
101714	** difference and select C of 3.0.
101715	*/
101716	if( bSort ){
101717	cost += nRowestLog(nRow)3;
101718	}
101719	if( bDist ){
101720	cost += nRowestLog(nRow)3;
101721	}
101722
101723	/** Cost of using this index has now been computed **/
101724
101725	/* If there are additional constraints on this table that cannot
101726	** be used with the current index, but which might lower the number
	@@ -101492,11 +101862,11 @@
101862	}
101863	sqlite3DbFree(pParse->db, p);
101864	}else
101865	#endif
101866	{
101867	bestBtreeIndex(pParse, pWC, pSrc, notReady, notValid, pOrderBy, 0, pCost);
101868	}
101869	}
101870
101871	/*
101872	** Disable a term in the WHERE clause. Except, do not disable the term
	@@ -102454,11 +102824,11 @@
102824	for(ii=0; ii<pOrWc->nTerm; ii++){
102825	WhereTerm *pOrTerm = &pOrWc->a[ii];
102826	if( pOrTerm->leftCursor==iCur \|\| pOrTerm->eOperator==WO_AND ){
102827	WhereInfo pSubWInfo; / Info for single OR-term scan */
102828	/* Loop through table entries that match term pOrTerm. */
102829	pSubWInfo = sqlite3WhereBegin(pParse, pOrTab, pOrTerm->pExpr, 0, 0,
102830	WHERE_OMIT_OPEN \| WHERE_OMIT_CLOSE \|
102831	WHERE_FORCE_TABLE \| WHERE_ONETABLE_ONLY);
102832	if( pSubWInfo ){
102833	explainOneScan(
102834	pParse, pOrTab, &pSubWInfo->a[0], iLevel, pLevel->iFrom, 0
	@@ -102695,10 +103065,11 @@
103065	SQLITE_PRIVATE WhereInfo *sqlite3WhereBegin(
103066	Parse pParse, / The parser context */
103067	SrcList pTabList, / A list of all tables to be scanned */
103068	Expr pWhere, / The WHERE clause */
103069	ExprList *ppOrderBy, / An ORDER BY clause, or NULL */
103070	ExprList pDistinct, / The select-list for DISTINCT queries - or NULL */
103071	u16 wctrlFlags /* One of the WHERE_* flags defined in sqliteInt.h */
103072	){
103073	int i; /* Loop counter */
103074	int nByteWInfo; /* Num. bytes allocated for WhereInfo struct */
103075	int nTabList; /* Number of elements in pTabList */
	@@ -102754,10 +103125,14 @@
103125	pWInfo->iBreak = sqlite3VdbeMakeLabel(v);
103126	pWInfo->pWC = pWC = (WhereClause )&((u8 )pWInfo)[nByteWInfo];
103127	pWInfo->wctrlFlags = wctrlFlags;
103128	pWInfo->savedNQueryLoop = pParse->nQueryLoop;
103129	pMaskSet = (WhereMaskSet*)&pWC[1];
103130
103131	/* Disable the DISTINCT optimization if SQLITE_DistinctOpt is set via
103132	** sqlite3_test_ctrl(SQLITE_TESTCTRL_OPTIMIZATIONS,...) */
103133	if( db->flags & SQLITE_DistinctOpt ) pDistinct = 0;
103134
103135	/* Split the WHERE clause into separate subexpressions where each
103136	** subexpression is separated by an AND operator.
103137	*/
103138	initMaskSet(pMaskSet);
	@@ -102821,10 +103196,19 @@
103196	*/
103197	exprAnalyzeAll(pTabList, pWC);
103198	if( db->mallocFailed ){
103199	goto whereBeginError;
103200	}
103201
103202	/* Check if the DISTINCT qualifier, if there is one, is redundant.
103203	** If it is, then set pDistinct to NULL and WhereInfo.eDistinct to
103204	** WHERE_DISTINCT_UNIQUE to tell the caller to ignore the DISTINCT.
103205	*/
103206	if( pDistinct && isDistinctRedundant(pParse, pTabList, pWC, pDistinct) ){
103207	pDistinct = 0;
103208	pWInfo->eDistinct = WHERE_DISTINCT_UNIQUE;
103209	}
103210
103211	/* Chose the best index to use for each table in the FROM clause.
103212	**
103213	** This loop fills in the following fields:
103214	**
	@@ -102905,10 +103289,11 @@
103289	Bitmask mask; /* Mask of tables not yet ready */
103290	for(j=iFrom, pTabItem=&pTabList->a[j]; j<nTabList; j++, pTabItem++){
103291	int doNotReorder; /* True if this table should not be reordered */
103292	WhereCost sCost; /* Cost information from best[Virtual]Index() */
103293	ExprList pOrderBy; / ORDER BY clause for index to optimize */
103294	ExprList pDist; / DISTINCT clause for index to optimize */
103295
103296	doNotReorder = (pTabItem->jointype & (JT_LEFT\|JT_CROSS))!=0;
103297	if( j!=iFrom && doNotReorder ) break;
103298	m = getMask(pMaskSet, pTabItem->iCursor);
103299	if( (m & notReady)==0 ){
	@@ -102915,10 +103300,11 @@
103300	if( j==iFrom ) iFrom++;
103301	continue;
103302	}
103303	mask = (isOptimal ? m : notReady);
103304	pOrderBy = ((i==0 && ppOrderBy )?*ppOrderBy:0);
103305	pDist = (i==0 ? pDistinct : 0);
103306	if( pTabItem->pIndex==0 ) nUnconstrained++;
103307
103308	WHERETRACE(("=== trying table %d with isOptimal=%d ===\n",
103309	j, isOptimal));
103310	assert( pTabItem->pTab );
	@@ -102929,11 +103315,11 @@
103315	&sCost, pp);
103316	}else
103317	#endif
103318	{
103319	bestBtreeIndex(pParse, pWC, pTabItem, mask, notReady, pOrderBy,
103320	pDist, &sCost);
103321	}
103322	assert( isOptimal \|\| (sCost.used&notReady)==0 );
103323
103324	/* If an INDEXED BY clause is present, then the plan must use that
103325	** index if it uses any index at all */
	@@ -102989,10 +103375,14 @@
103375	WHERETRACE(("*** Optimizer selects table %d for loop %d"
103376	" with cost=%g and nRow=%g\n",
103377	bestJ, pLevel-pWInfo->a, bestPlan.rCost, bestPlan.plan.nRow));
103378	if( (bestPlan.plan.wsFlags & WHERE_ORDERBY)!=0 ){
103379	*ppOrderBy = 0;
103380	}
103381	if( (bestPlan.plan.wsFlags & WHERE_DISTINCT)!=0 ){
103382	assert( pWInfo->eDistinct==0 );
103383	pWInfo->eDistinct = WHERE_DISTINCT_ORDERED;
103384	}
103385	andFlags &= bestPlan.plan.wsFlags;
103386	pLevel->plan = bestPlan.plan;
103387	testcase( bestPlan.plan.wsFlags & WHERE_INDEXED );
103388	testcase( bestPlan.plan.wsFlags & WHERE_TEMP_INDEX );
	@@ -111519,11 +111909,17 @@
111909	*/
111910	#if defined(SQLITE_ENABLE_FTS4) && !defined(SQLITE_ENABLE_FTS3)
111911	# define SQLITE_ENABLE_FTS3
111912	#endif
111913
111914	#if !defined(SQLITE_CORE) \|\| defined(SQLITE_ENABLE_FTS3)
111915
111916	/* If not building as part of the core, include sqlite3ext.h. */
111917	#ifndef SQLITE_CORE
111918	SQLITE_API extern const sqlite3_api_routines *sqlite3_api;
111919	#endif
111920
111921	/************ Include fts3_tokenizer.h in the middle of fts3Int.h ********/
111922	/************ Begin file fts3_tokenizer.h ********************************/
111923	/*
111924	** 2006 July 10
111925	**
	@@ -112052,11 +112448,11 @@
112448
112449	sqlite3_int64 iDocid; /* Current docid (if pList!=0) */
112450	int bFreeList; /* True if pList should be sqlite3_free()d */
112451	char pList; / Pointer to position list following iDocid */
112452	int nList; /* Length of position list */
112453	};
112454
112455	/*
112456	** A "phrase" is a sequence of one or more tokens that must match in
112457	** sequence. A single token is the base case and the most common case.
112458	** For a sequence of tokens contained in double-quotes (i.e. "one two three")
	@@ -112252,23 +112648,12 @@
112648	#endif
112649
112650	/* fts3_aux.c */
112651	SQLITE_PRIVATE int sqlite3Fts3InitAux(sqlite3 *db);
112652








112653	SQLITE_PRIVATE void sqlite3Fts3EvalPhraseCleanup(Fts3Phrase *);
112654



112655	SQLITE_PRIVATE int sqlite3Fts3MsrIncrStart(
112656	Fts3Table, Fts3MultiSegReader, int, const char*, int);
112657	SQLITE_PRIVATE int sqlite3Fts3MsrIncrNext(
112658	Fts3Table , Fts3MultiSegReader , sqlite3_int64 , char , int );
112659	SQLITE_PRIVATE char sqlite3Fts3EvalPhrasePoslist(Fts3Cursor , Fts3Expr *, int iCol);
	@@ -112275,11 +112660,11 @@
112660	SQLITE_PRIVATE int sqlite3Fts3MsrOvfl(Fts3Cursor , Fts3MultiSegReader , int *);
112661	SQLITE_PRIVATE int sqlite3Fts3MsrIncrRestart(Fts3MultiSegReader *pCsr);
112662
112663	SQLITE_PRIVATE int sqlite3Fts3DeferredTokenList(Fts3DeferredToken , char , int );
112664
112665	#endif /* !SQLITE_CORE \|\| SQLITE_ENABLE_FTS3 */
112666	#endif /* _FTSINT_H */
112667
112668	/************ End of fts3Int.h *******************************************/
112669	/************ Continuing where we left off in fts3.c *********************/
112670	#if !defined(SQLITE_CORE) \|\| defined(SQLITE_ENABLE_FTS3)
	@@ -112291,10 +112676,15 @@
112676
112677	#ifndef SQLITE_CORE
112678	SQLITE_EXTENSION_INIT1
112679	#endif
112680
112681	static int fts3EvalNext(Fts3Cursor *pCsr);
112682	static int fts3EvalStart(Fts3Cursor *pCsr);
112683	static int fts3TermSegReaderCursor(
112684	Fts3Cursor , const char , int, int, Fts3MultiSegReader **);
112685
112686	/*
112687	** Write a 64-bit variable-length integer to memory starting at p[0].
112688	** The length of data written will be between 1 and FTS3_VARINT_MAX bytes.
112689	** The number of bytes written is returned.
112690	*/
	@@ -112799,31 +113189,60 @@
113189	}
113190	sqlite3_free(zFree);
113191	return zRet;
113192	}
113193
113194	/*
113195	** This function interprets the string at (*pp) as a non-negative integer
113196	** value. It reads the integer and sets *pnOut to the value read, then
113197	** sets *pp to point to the byte immediately following the last byte of
113198	** the integer value.
113199	**
113200	** Only decimal digits ('0'..'9') may be part of an integer value.
113201	**
113202	** If *pp does not being with a decimal digit SQLITE_ERROR is returned and
113203	** the output value undefined. Otherwise SQLITE_OK is returned.
113204	**
113205	** This function is used when parsing the "prefix=" FTS4 parameter.
113206	*/
113207	static int fts3GobbleInt(const char *pp, int pnOut){
113208	const char p = pp; /* Iterator pointer */
113209	int nInt = 0; /* Output value */
113210
113211	for(p=*pp; p[0]>='0' && p[0]<='9'; p++){
113212	nInt = nInt * 10 + (p[0] - '0');
113213	}
113214	if( p==*pp ) return SQLITE_ERROR;
113215	*pnOut = nInt;
113216	*pp = p;
113217	return SQLITE_OK;
113218	}
113219
113220	/*
113221	** This function is called to allocate an array of Fts3Index structures
113222	** representing the indexes maintained by the current FTS table. FTS tables
113223	** always maintain the main "terms" index, but may also maintain one or
113224	** more "prefix" indexes, depending on the value of the "prefix=" parameter
113225	** (if any) specified as part of the CREATE VIRTUAL TABLE statement.
113226	**
113227	** Argument zParam is passed the value of the "prefix=" option if one was
113228	** specified, or NULL otherwise.
113229	**
113230	** If no error occurs, SQLITE_OK is returned and *apIndex set to point to
113231	** the allocated array. *pnIndex is set to the number of elements in the
113232	** array. If an error does occur, an SQLite error code is returned.
113233	**
113234	** Regardless of whether or not an error is returned, it is the responsibility
113235	** of the caller to call sqlite3_free() on the output array to free it.
113236	*/
113237	static int fts3PrefixParameter(
113238	const char zParam, / ABC in prefix=ABC parameter to parse */
113239	int pnIndex, / OUT: size of apIndex[] array /
113240	struct Fts3Index *apIndex / OUT: Array of indexes for this table */

113241	){
113242	struct Fts3Index aIndex; / Allocated array */
113243	int nIndex = 1; /* Number of entries in array */
113244
113245	if( zParam && zParam[0] ){
113246	const char *p;
113247	nIndex++;
113248	for(p=zParam; *p; p++){
	@@ -112830,11 +113249,11 @@
113249	if( *p==',' ) nIndex++;
113250	}
113251	}
113252
113253	aIndex = sqlite3_malloc(sizeof(struct Fts3Index) * nIndex);
113254	*apIndex = aIndex;
113255	*pnIndex = nIndex;
113256	if( !aIndex ){
113257	return SQLITE_NOMEM;
113258	}
113259
	@@ -112887,12 +113306,11 @@
113306	int isFts4 = (argv[0][3]=='4'); /* True for FTS4, false for FTS3 */
113307	const char *aCol; / Array of column names */
113308	sqlite3_tokenizer pTokenizer = 0; / Tokenizer for this table */
113309
113310	int nIndex; /* Size of aIndex[] array */
113311	struct Fts3Index aIndex = 0; / Array of indexes for this table */

113312
113313	/* The results of parsing supported FTS4 key=value options: */
113314	int bNoDocsize = 0; /* True to omit %_docsize table */
113315	int bDescIdx = 0; /* True to store descending indexes */
113316	char zPrefix = 0; / Prefix parameter value (or NULL) */
	@@ -113025,11 +113443,11 @@
113443	rc = sqlite3Fts3InitTokenizer(pHash, "simple", &pTokenizer, pzErr);
113444	if( rc!=SQLITE_OK ) goto fts3_init_out;
113445	}
113446	assert( pTokenizer );
113447
113448	rc = fts3PrefixParameter(zPrefix, &nIndex, &aIndex);
113449	if( rc==SQLITE_ERROR ){
113450	assert( zPrefix );
113451	*pzErr = sqlite3_mprintf("error parsing prefix parameter: %s", zPrefix);
113452	}
113453	if( rc!=SQLITE_OK ) goto fts3_init_out;
	@@ -113112,11 +113530,11 @@
113530	/* Declare the table schema to SQLite. */
113531	fts3DeclareVtab(&rc, p);
113532
113533	fts3_init_out:
113534	sqlite3_free(zPrefix);
113535	sqlite3_free(aIndex);
113536	sqlite3_free(zCompress);
113537	sqlite3_free(zUncompress);
113538	sqlite3_free((void *)aCol);
113539	if( rc!=SQLITE_OK ){
113540	if( p ){
	@@ -113703,12 +114121,10 @@
114121	*pp1 = p1 + 1;
114122	*pp2 = p2 + 1;
114123	}
114124
114125	/*


114126	** This function is used to merge two position lists into one. When it is
114127	** called, pp1 and pp2 must both point to position lists. A position-list is
114128	** the part of a doclist that follows each document id. For example, if a row
114129	** contains:
114130	**
	@@ -113724,10 +114140,12 @@
114140	** If isSaveLeft is 0, an entry is added to the output position list for
114141	** each position in *pp2 for which there exists one or more positions in
114142	** pp1 so that (pos(pp2)>pos(pp1) && pos(pp2)-pos(*pp1)<=nToken). i.e.
114143	** when the pp1 token appears before the pp2 token, but not more than nToken
114144	** slots before it.
114145	**
114146	** e.g. nToken==1 searches for adjacent positions.
114147	*/
114148	static int fts3PoslistPhraseMerge(
114149	char *pp, / IN/OUT: Preallocated output buffer */
114150	int nToken, /* Maximum difference in token positions */
114151	int isSaveLeft, /* Save the left position */
	@@ -113890,26 +114308,38 @@
114308
114309	return res;
114310	}
114311
114312	/*
114313	** An instance of this function is used to merge together the (potentially
114314	** large number of) doclists for each term that matches a prefix query.
114315	** See function fts3TermSelectMerge() for details.
114316	*/
114317	typedef struct TermSelect TermSelect;
114318	struct TermSelect {
114319	char aaOutput[16]; / Malloc'd output buffers */
114320	int anOutput[16]; /* Size each output buffer in bytes */

114321	};
114322
114323	/*
114324	** This function is used to read a single varint from a buffer. Parameter
114325	** pEnd points 1 byte past the end of the buffer. When this function is
114326	** called, if *pp points to pEnd or greater, then the end of the buffer
114327	** has been reached. In this case *pp is set to 0 and the function returns.
114328	**
114329	** If *pp does not point to or past pEnd, then a single varint is read
114330	** from pp. pp is then set to point 1 byte past the end of the read varint.
114331	**
114332	** If bDescIdx is false, the value read is added to *pVal before returning.
114333	** If it is true, the value read is subtracted from *pVal before this
114334	** function returns.
114335	*/
114336	static void fts3GetDeltaVarint3(
114337	char *pp, / IN/OUT: Point to read varint from */
114338	char pEnd, / End of buffer */
114339	int bDescIdx, /* True if docids are descending */
114340	sqlite3_int64 pVal / IN/OUT: Integer value */
114341	){
114342	if( *pp>=pEnd ){
114343	*pp = 0;
114344	}else{
114345	sqlite3_int64 iVal;
	@@ -113920,10 +114350,25 @@
114350	*pVal += iVal;
114351	}
114352	}
114353	}
114354
114355	/*
114356	** This function is used to write a single varint to a buffer. The varint
114357	** is written to pp. Before returning, pp is set to point 1 byte past the
114358	** end of the value written.
114359	**
114360	** If *pbFirst is zero when this function is called, the value written to
114361	** the buffer is that of parameter iVal.
114362	**
114363	** If *pbFirst is non-zero when this function is called, then the value
114364	** written is either (iVal-piPrev) (if bDescIdx is zero) or (piPrev-iVal)
114365	** (if bDescIdx is non-zero).
114366	**
114367	** Before returning, this function always sets pbFirst to 1 and piPrev
114368	** to the value of parameter iVal.
114369	*/
114370	static void fts3PutDeltaVarint3(
114371	char *pp, / IN/OUT: Output pointer */
114372	int bDescIdx, /* True for descending docids */
114373	sqlite3_int64 piPrev, / IN/OUT: Previous value written to list */
114374	int pbFirst, / IN/OUT: True after first int written */
	@@ -113940,14 +114385,38 @@
114385	pp += sqlite3Fts3PutVarint(pp, iWrite);
114386	*piPrev = iVal;
114387	*pbFirst = 1;
114388	}
114389

114390
114391	/*
114392	** This macro is used by various functions that merge doclists. The two
114393	** arguments are 64-bit docid values. If the value of the stack variable
114394	** bDescDoclist is 0 when this macro is invoked, then it returns (i1-i2).
114395	** Otherwise, (i2-i1).
114396	**
114397	** Using this makes it easier to write code that can merge doclists that are
114398	** sorted in either ascending or descending order.
114399	*/
114400	#define DOCID_CMP(i1, i2) ((bDescDoclist?-1:1) * (i1-i2))
114401
114402	/*
114403	** This function does an "OR" merge of two doclists (output contains all
114404	** positions contained in either argument doclist). If the docids in the
114405	** input doclists are sorted in ascending order, parameter bDescDoclist
114406	** should be false. If they are sorted in ascending order, it should be
114407	** passed a non-zero value.
114408	**
114409	** If no error occurs, *paOut is set to point at an sqlite3_malloc'd buffer
114410	** containing the output doclist and SQLITE_OK is returned. In this case
114411	** *pnOut is set to the number of bytes in the output doclist.
114412	**
114413	** If an error occurs, an SQLite error code is returned. The output values
114414	** are undefined in this case.
114415	*/
114416	static int fts3DoclistOrMerge(
114417	int bDescDoclist, /* True if arguments are desc */
114418	char a1, int n1, / First doclist */
114419	char a2, int n2, / Second doclist */
114420	char *paOut, int pnOut /* OUT: Malloc'd doclist */
114421	){
114422	sqlite3_int64 i1 = 0;
	@@ -113968,35 +114437,47 @@
114437
114438	p = aOut;
114439	fts3GetDeltaVarint3(&p1, pEnd1, 0, &i1);
114440	fts3GetDeltaVarint3(&p2, pEnd2, 0, &i2);
114441	while( p1 \|\| p2 ){
114442	sqlite3_int64 iDiff = DOCID_CMP(i1, i2);
114443
114444	if( p2 && p1 && iDiff==0 ){
114445	fts3PutDeltaVarint3(&p, bDescDoclist, &iPrev, &bFirstOut, i1);
114446	fts3PoslistMerge(&p, &p1, &p2);
114447	fts3GetDeltaVarint3(&p1, pEnd1, bDescDoclist, &i1);
114448	fts3GetDeltaVarint3(&p2, pEnd2, bDescDoclist, &i2);
114449	}else if( !p2 \|\| (p1 && iDiff<0) ){
114450	fts3PutDeltaVarint3(&p, bDescDoclist, &iPrev, &bFirstOut, i1);
114451	fts3PoslistCopy(&p, &p1);
114452	fts3GetDeltaVarint3(&p1, pEnd1, bDescDoclist, &i1);
114453	}else{
114454	fts3PutDeltaVarint3(&p, bDescDoclist, &iPrev, &bFirstOut, i2);
114455	fts3PoslistCopy(&p, &p2);
114456	fts3GetDeltaVarint3(&p2, pEnd2, bDescDoclist, &i2);
114457	}
114458	}
114459
114460	*paOut = aOut;
114461	*pnOut = (p-aOut);
114462	return SQLITE_OK;
114463	}
114464
114465	/*
114466	** This function does a "phrase" merge of two doclists. In a phrase merge,
114467	** the output contains a copy of each position from the right-hand input
114468	** doclist for which there is a position in the left-hand input doclist
114469	** exactly nDist tokens before it.
114470	**
114471	** If the docids in the input doclists are sorted in ascending order,
114472	** parameter bDescDoclist should be false. If they are sorted in ascending
114473	** order, it should be passed a non-zero value.
114474	**
114475	** The right-hand input doclist is overwritten by this function.
114476	*/
114477	static void fts3DoclistPhraseMerge(
114478	int bDescDoclist, /* True if arguments are desc */
114479	int nDist, /* Distance from left to right (1=adjacent) */
114480	char aLeft, int nLeft, / Left doclist */
114481	char aRight, int pnRight /* IN/OUT: Right/output doclist */
114482	){
114483	sqlite3_int64 i1 = 0;
	@@ -114015,30 +114496,30 @@
114496	p = aOut;
114497	fts3GetDeltaVarint3(&p1, pEnd1, 0, &i1);
114498	fts3GetDeltaVarint3(&p2, pEnd2, 0, &i2);
114499
114500	while( p1 && p2 ){
114501	sqlite3_int64 iDiff = DOCID_CMP(i1, i2);
114502	if( iDiff==0 ){
114503	char *pSave = p;
114504	sqlite3_int64 iPrevSave = iPrev;
114505	int bFirstOutSave = bFirstOut;
114506
114507	fts3PutDeltaVarint3(&p, bDescDoclist, &iPrev, &bFirstOut, i1);
114508	if( 0==fts3PoslistPhraseMerge(&p, nDist, 0, 1, &p1, &p2) ){
114509	p = pSave;
114510	iPrev = iPrevSave;
114511	bFirstOut = bFirstOutSave;
114512	}
114513	fts3GetDeltaVarint3(&p1, pEnd1, bDescDoclist, &i1);
114514	fts3GetDeltaVarint3(&p2, pEnd2, bDescDoclist, &i2);
114515	}else if( iDiff<0 ){
114516	fts3PoslistCopy(0, &p1);
114517	fts3GetDeltaVarint3(&p1, pEnd1, bDescDoclist, &i1);
114518	}else{
114519	fts3PoslistCopy(0, &p2);
114520	fts3GetDeltaVarint3(&p2, pEnd2, bDescDoclist, &i2);
114521	}
114522	}
114523
114524	*pnRight = p - aOut;
114525	}
	@@ -114051,11 +114532,11 @@
114532	**
114533	** If an OOM error occurs, return SQLITE_NOMEM. In this case it is
114534	** the responsibility of the caller to free any doclists left in the
114535	** TermSelect.aaOutput[] array.
114536	*/
114537	static int fts3TermSelectFinishMerge(Fts3Table p, TermSelect pTS){
114538	char *aOut = 0;
114539	int nOut = 0;
114540	int i;
114541
114542	/* Loop through the doclists in the aaOutput[] array. Merge them all
	@@ -114092,28 +114573,29 @@
114573	pTS->anOutput[0] = nOut;
114574	return SQLITE_OK;
114575	}
114576
114577	/*
114578	** Merge the doclist aDoclist/nDoclist into the TermSelect object passed
114579	** as the first argument. The merge is an "OR" merge (see function
114580	** fts3DoclistOrMerge() for details).
114581	**
114582	** This function is called with the doclist for each term that matches
114583	** a queried prefix. It merges all these doclists into one, the doclist
114584	** for the specified prefix. Since there can be a very large number of
114585	** doclists to merge, the merging is done pair-wise using the TermSelect
114586	** object.
114587	**
114588	** This function returns SQLITE_OK if the merge is successful, or an
114589	** SQLite error code (SQLITE_NOMEM) if an error occurs.
114590	*/
114591	static int fts3TermSelectMerge(
114592	Fts3Table p, / FTS table handle */
114593	TermSelect pTS, / TermSelect object to merge into */
114594	char aDoclist, / Pointer to doclist */
114595	int nDoclist /* Size of aDoclist in bytes */


114596	){






114597	if( pTS->aaOutput[0]==0 ){
114598	/* If this is the first term selected, copy the doclist to the output
114599	** buffer using memcpy(). */
114600	pTS->aaOutput[0] = sqlite3_malloc(nDoclist);
114601	pTS->anOutput[0] = nDoclist;
	@@ -114180,23 +114662,30 @@
114662	}
114663	pCsr->apSegment[pCsr->nSegment++] = pNew;
114664	return SQLITE_OK;
114665	}
114666
114667	/*
114668	** Add seg-reader objects to the Fts3MultiSegReader object passed as the
114669	** 8th argument.
114670	**
114671	** This function returns SQLITE_OK if successful, or an SQLite error code
114672	** otherwise.
114673	*/
114674	static int fts3SegReaderCursor(
114675	Fts3Table p, / FTS3 table handle */
114676	int iIndex, /* Index to search (from 0 to p->nIndex-1) */
114677	int iLevel, /* Level of segments to scan */
114678	const char zTerm, / Term to query for */
114679	int nTerm, /* Size of zTerm in bytes */
114680	int isPrefix, /* True for a prefix search */
114681	int isScan, /* True to scan from zTerm to EOF */
114682	Fts3MultiSegReader pCsr / Cursor object to populate */
114683	){
114684	int rc = SQLITE_OK; /* Error code */
114685	sqlite3_stmt pStmt = 0; / Statement to iterate through segments */
114686	int rc2; /* Result of sqlite3_reset() */
114687
114688	/* If iLevel is less than 0 and this is not a scan, include a seg-reader
114689	** for the pending-terms. If this is a scan, then this call must be being
114690	** made by an fts4aux module, not an FTS table. In this case calling
114691	** Fts3SegReaderPending might segfault, as the data structures used by
	@@ -114281,28 +114770,46 @@
114770	return fts3SegReaderCursor(
114771	p, iIndex, iLevel, zTerm, nTerm, isPrefix, isScan, pCsr
114772	);
114773	}
114774
114775	/*
114776	** In addition to its current configuration, have the Fts3MultiSegReader
114777	** passed as the 4th argument also scan the doclist for term zTerm/nTerm.
114778	**
114779	** SQLITE_OK is returned if no error occurs, otherwise an SQLite error code.
114780	*/
114781	static int fts3SegReaderCursorAddZero(
114782	Fts3Table p, / FTS virtual table handle */
114783	const char zTerm, / Term to scan doclist of */
114784	int nTerm, /* Number of bytes in zTerm */
114785	Fts3MultiSegReader pCsr / Fts3MultiSegReader to modify */
114786	){
114787	return fts3SegReaderCursor(p, 0, FTS3_SEGCURSOR_ALL, zTerm, nTerm, 0, 0,pCsr);
114788	}
114789
114790	/*
114791	** Open an Fts3MultiSegReader to scan the doclist for term zTerm/nTerm. Or,
114792	** if isPrefix is true, to scan the doclist for all terms for which
114793	** zTerm/nTerm is a prefix. If successful, return SQLITE_OK and write
114794	** a pointer to the new Fts3MultiSegReader to *ppSegcsr. Otherwise, return
114795	** an SQLite error code.
114796	**
114797	** It is the responsibility of the caller to free this object by eventually
114798	** passing it to fts3SegReaderCursorFree()
114799	**
114800	** SQLITE_OK is returned if no error occurs, otherwise an SQLite error code.
114801	** Output parameter *ppSegcsr is set to 0 if an error occurs.
114802	*/
114803	static int fts3TermSegReaderCursor(
114804	Fts3Cursor pCsr, / Virtual table cursor handle */
114805	const char zTerm, / Term to query for */
114806	int nTerm, /* Size of zTerm in bytes */
114807	int isPrefix, /* True for a prefix search */
114808	Fts3MultiSegReader *ppSegcsr / OUT: Allocated seg-reader cursor */
114809	){
114810	Fts3MultiSegReader pSegcsr; / Object to allocate and return */
114811	int rc = SQLITE_NOMEM; /* Return code */
114812
114813	pSegcsr = sqlite3_malloc(sizeof(Fts3MultiSegReader));
114814	if( pSegcsr ){
114815	int i;
	@@ -114342,62 +114849,53 @@
114849
114850	*ppSegcsr = pSegcsr;
114851	return rc;
114852	}
114853
114854	/*
114855	** Free an Fts3MultiSegReader allocated by fts3TermSegReaderCursor().
114856	*/
114857	static void fts3SegReaderCursorFree(Fts3MultiSegReader *pSegcsr){
114858	sqlite3Fts3SegReaderFinish(pSegcsr);
114859	sqlite3_free(pSegcsr);
114860	}
114861
114862	/*
114863	** This function retreives the doclist for the specified term (or term
114864	** prefix) from the database.







114865	*/
114866	static int fts3TermSelect(
114867	Fts3Table p, / Virtual table handle */
114868	Fts3PhraseToken pTok, / Token to query for */
114869	int iColumn, /* Column to query (or -ve for all columns) */

114870	int pnOut, / OUT: Size of buffer at ppOut /
114871	char *ppOut / OUT: Malloced result buffer */
114872	){
114873	int rc; /* Return code */
114874	Fts3MultiSegReader pSegcsr; / Seg-reader cursor for this term */
114875	TermSelect tsc; /* Object for pair-wise doclist merging */
114876	Fts3SegFilter filter; /* Segment term filter configuration */
114877
114878	pSegcsr = pTok->pSegcsr;
114879	memset(&tsc, 0, sizeof(TermSelect));

114880
114881	filter.flags = FTS3_SEGMENT_IGNORE_EMPTY \| FTS3_SEGMENT_REQUIRE_POS
114882	\| (pTok->isPrefix ? FTS3_SEGMENT_PREFIX : 0)

114883	\| (iColumn<p->nColumn ? FTS3_SEGMENT_COLUMN_FILTER : 0);
114884	filter.iCol = iColumn;
114885	filter.zTerm = pTok->z;
114886	filter.nTerm = pTok->n;
114887
114888	rc = sqlite3Fts3SegReaderStart(p, pSegcsr, &filter);
114889	while( SQLITE_OK==rc
114890	&& SQLITE_ROW==(rc = sqlite3Fts3SegReaderStep(p, pSegcsr))
114891	){
114892	rc = fts3TermSelectMerge(p, &tsc, pSegcsr->aDoclist, pSegcsr->nDoclist);


114893	}
114894
114895	if( rc==SQLITE_OK ){
114896	rc = fts3TermSelectFinishMerge(p, &tsc);
114897	}
114898	if( rc==SQLITE_OK ){
114899	*ppOut = tsc.aaOutput[0];
114900	*pnOut = tsc.anOutput[0];
114901	}else{
	@@ -114419,28 +114917,19 @@
114917	** If the isPoslist argument is true, then it is assumed that the doclist
114918	** contains a position-list following each docid. Otherwise, it is assumed
114919	** that the doclist is simply a list of docids stored as delta encoded
114920	** varints.
114921	*/
114922	static int fts3DoclistCountDocids(char *aList, int nList){
114923	int nDoc = 0; /* Return value */
114924	if( aList ){
114925	char aEnd = &aList[nList]; / Pointer to one byte after EOF */
114926	char p = aList; / Cursor */
114927	while( p<aEnd ){
114928	nDoc++;
114929	while( (p++)&0x80 ); / Skip docid varint */
114930	fts3PoslistCopy(0, &p); /* Skip over position list */









114931	}
114932	}
114933
114934	return nDoc;
114935	}
	@@ -114466,11 +114955,11 @@
114955	}else{
114956	pCsr->iPrevId = sqlite3_column_int64(pCsr->pStmt, 0);
114957	rc = SQLITE_OK;
114958	}
114959	}else{
114960	rc = fts3EvalNext((Fts3Cursor *)pCursor);
114961	}
114962	assert( ((Fts3Table *)pCsr->base.pVtab)->pSegments==0 );
114963	return rc;
114964	}
114965
	@@ -114543,11 +115032,11 @@
115032	}
115033
115034	rc = sqlite3Fts3ReadLock(p);
115035	if( rc!=SQLITE_OK ) return rc;
115036
115037	rc = fts3EvalStart(pCsr);
115038
115039	sqlite3Fts3SegmentsClose(p);
115040	if( rc!=SQLITE_OK ) return rc;
115041	pCsr->pNextId = pCsr->aDoclist;
115042	pCsr->iPrevId = 0;
	@@ -114950,26 +115439,43 @@
115439	p->zDb, p->zName, zName
115440	);
115441	return rc;
115442	}
115443
115444	/*
115445	** The xSavepoint() method.
115446	**
115447	** Flush the contents of the pending-terms table to disk.
115448	*/
115449	static int fts3SavepointMethod(sqlite3_vtab *pVtab, int iSavepoint){
115450	UNUSED_PARAMETER(iSavepoint);
115451	assert( ((Fts3Table *)pVtab)->inTransaction );
115452	assert( ((Fts3Table *)pVtab)->mxSavepoint < iSavepoint );
115453	TESTONLY( ((Fts3Table *)pVtab)->mxSavepoint = iSavepoint );
115454	return fts3SyncMethod(pVtab);
115455	}
115456
115457	/*
115458	** The xRelease() method.
115459	**
115460	** This is a no-op.
115461	*/
115462	static int fts3ReleaseMethod(sqlite3_vtab *pVtab, int iSavepoint){
115463	TESTONLY( Fts3Table p = (Fts3Table)pVtab );
115464	UNUSED_PARAMETER(iSavepoint);
115465	UNUSED_PARAMETER(pVtab);
115466	assert( p->inTransaction );
115467	assert( p->mxSavepoint >= iSavepoint );
115468	TESTONLY( p->mxSavepoint = iSavepoint-1 );
115469	return SQLITE_OK;
115470	}
115471
115472	/*
115473	** The xRollbackTo() method.
115474	**
115475	** Discard the contents of the pending terms table.
115476	*/
115477	static int fts3RollbackToMethod(sqlite3_vtab *pVtab, int iSavepoint){
115478	Fts3Table p = (Fts3Table)pVtab;
115479	UNUSED_PARAMETER(iSavepoint);
115480	assert( p->inTransaction );
115481	assert( p->mxSavepoint >= iSavepoint );
	@@ -115114,22 +115620,10 @@
115620	sqlite3Fts3HashClear(pHash);
115621	sqlite3_free(pHash);
115622	}
115623	return rc;
115624	}












115625
115626	/*
115627	** Allocate an Fts3MultiSegReader for each token in the expression headed
115628	** by pExpr.
115629	**
	@@ -115143,24 +115637,24 @@
115637	** there exists prefix b-tree of the right length) then it may be traversed
115638	** and merged incrementally. Otherwise, it has to be merged into an in-memory
115639	** doclist and then traversed.
115640	*/
115641	static void fts3EvalAllocateReaders(
115642	Fts3Cursor pCsr, / FTS cursor handle */
115643	Fts3Expr pExpr, / Allocate readers for this expression */
115644	int pnToken, / OUT: Total number of tokens in phrase. */
115645	int pnOr, / OUT: Total number of OR nodes in expr. */
115646	int pRc / IN/OUT: Error code */
115647	){
115648	if( pExpr && SQLITE_OK==*pRc ){
115649	if( pExpr->eType==FTSQUERY_PHRASE ){
115650	int i;
115651	int nToken = pExpr->pPhrase->nToken;
115652	*pnToken += nToken;
115653	for(i=0; i<nToken; i++){
115654	Fts3PhraseToken *pToken = &pExpr->pPhrase->aToken[i];
115655	int rc = fts3TermSegReaderCursor(pCsr,
115656	pToken->z, pToken->n, pToken->isPrefix, &pToken->pSegcsr
115657	);
115658	if( rc!=SQLITE_OK ){
115659	*pRc = rc;
115660	return;
	@@ -115174,16 +115668,24 @@
115668	fts3EvalAllocateReaders(pCsr, pExpr->pRight, pnToken, pnOr, pRc);
115669	}
115670	}
115671	}
115672
115673	/*
115674	** Arguments pList/nList contain the doclist for token iToken of phrase p.
115675	** It is merged into the main doclist stored in p->doclist.aAll/nAll.
115676	**
115677	** This function assumes that pList points to a buffer allocated using
115678	** sqlite3_malloc(). This function takes responsibility for eventually
115679	** freeing the buffer.
115680	*/
115681	static void fts3EvalPhraseMergeToken(
115682	Fts3Table pTab, / FTS Table pointer */
115683	Fts3Phrase p, / Phrase to merge pList/nList into */
115684	int iToken, /* Token pList/nList corresponds to */
115685	char pList, / Pointer to doclist */
115686	int nList /* Number of bytes in pList */
115687	){
115688	assert( iToken!=p->iDoclistToken );
115689
115690	if( pList==0 ){
115691	sqlite3_free(p->doclist.aAll);
	@@ -115228,13 +115730,19 @@
115730	}
115731
115732	if( iToken>p->iDoclistToken ) p->iDoclistToken = iToken;
115733	}
115734
115735	/*
115736	** Load the doclist for phrase p into p->doclist.aAll/nAll. The loaded doclist
115737	** does not take deferred tokens into account.
115738	**
115739	** SQLITE_OK is returned if no error occurs, otherwise an SQLite error code.
115740	*/
115741	static int fts3EvalPhraseLoad(
115742	Fts3Cursor pCsr, / FTS Cursor handle */
115743	Fts3Phrase p / Phrase object */
115744	){
115745	Fts3Table pTab = (Fts3Table )pCsr->base.pVtab;
115746	int iToken;
115747	int rc = SQLITE_OK;
115748
	@@ -115243,11 +115751,11 @@
115751	assert( pToken->pDeferred==0 \|\| pToken->pSegcsr==0 );
115752
115753	if( pToken->pSegcsr ){
115754	int nThis = 0;
115755	char *pThis = 0;
115756	rc = fts3TermSelect(pTab, pToken, p->iColumn, &nThis, &pThis);
115757	if( rc==SQLITE_OK ){
115758	fts3EvalPhraseMergeToken(pTab, p, iToken, pThis, nThis);
115759	}
115760	}
115761	assert( pToken->pSegcsr==0 );
	@@ -115254,18 +115762,26 @@
115762	}
115763
115764	return rc;
115765	}
115766
115767	/*
115768	** This function is called on each phrase after the position lists for
115769	** any deferred tokens have been loaded into memory. It updates the phrases
115770	** current position list to include only those positions that are really
115771	** instances of the phrase (after considering deferred tokens). If this
115772	** means that the phrase does not appear in the current row, doclist.pList
115773	** and doclist.nList are both zeroed.
115774	**
115775	** SQLITE_OK is returned if no error occurs, otherwise an SQLite error code.
115776	*/
115777	static int fts3EvalDeferredPhrase(Fts3Cursor pCsr, Fts3Phrase pPhrase){
115778	int iToken; /* Used to iterate through phrase tokens */
115779	int rc = SQLITE_OK; /* Return code */
115780	char aPoslist = 0; / Position list for deferred tokens */
115781	int nPoslist = 0; /* Number of bytes in aPoslist */
115782	int iPrev = -1; /* Token number of previous deferred token */


115783
115784	assert( pPhrase->doclist.bFreeList==0 );
115785
115786	for(iToken=0; rc==SQLITE_OK && iToken<pPhrase->nToken; iToken++){
115787	Fts3PhraseToken *pToken = &pPhrase->aToken[iToken];
	@@ -115307,10 +115823,11 @@
115823	iPrev = iToken;
115824	}
115825	}
115826
115827	if( iPrev>=0 ){
115828	int nMaxUndeferred = pPhrase->iDoclistToken;
115829	if( nMaxUndeferred<0 ){
115830	pPhrase->doclist.pList = aPoslist;
115831	pPhrase->doclist.nList = nPoslist;
115832	pPhrase->doclist.iDocid = pCsr->iPrevId;
115833	pPhrase->doclist.bFreeList = 1;
	@@ -115355,13 +115872,19 @@
115872	/*
115873	** This function is called for each Fts3Phrase in a full-text query
115874	** expression to initialize the mechanism for returning rows. Once this
115875	** function has been called successfully on an Fts3Phrase, it may be
115876	** used with fts3EvalPhraseNext() to iterate through the matching docids.
115877	**
115878	** If parameter bOptOk is true, then the phrase may (or may not) use the
115879	** incremental loading strategy. Otherwise, the entire doclist is loaded into
115880	** memory within this call.
115881	**
115882	** SQLITE_OK is returned if no error occurs, otherwise an SQLite error code.
115883	*/
115884	static int fts3EvalPhraseStart(Fts3Cursor pCsr, int bOptOk, Fts3Phrase p){
115885	int rc; /* Error code */
115886	Fts3PhraseToken *pFirst = &p->aToken[0];
115887	Fts3Table pTab = (Fts3Table )pCsr->base.pVtab;
115888
115889	if( pCsr->bDesc==pTab->bDescIdx
115890	&& bOptOk==1
	@@ -115385,11 +115908,17 @@
115908	return rc;
115909	}
115910
115911	/*
115912	** This function is used to iterate backwards (from the end to start)
115913	** through doclists. It is used by this module to iterate through phrase
115914	** doclists in reverse and by the fts3_write.c module to iterate through
115915	** pending-terms lists when writing to databases with "order=desc".
115916	**
115917	** The doclist may be sorted in ascending (parameter bDescIdx==0) or
115918	** descending (parameter bDescIdx==1) order of docid. Regardless, this
115919	** function iterates from the end of the doclist to the beginning.
115920	*/
115921	SQLITE_PRIVATE void sqlite3Fts3DoclistPrev(
115922	int bDescIdx, /* True if the doclist is desc */
115923	char aDoclist, / Pointer to entire doclist */
115924	int nDoclist, /* Length of aDoclist in bytes */
	@@ -115450,13 +115979,13 @@
115979	** If there is no "next" entry and no error occurs, then *pbEof is set to
115980	** 1 before returning. Otherwise, if no error occurs and the iterator is
115981	** successfully advanced, *pbEof is set to 0.
115982	*/
115983	static int fts3EvalPhraseNext(
115984	Fts3Cursor pCsr, / FTS Cursor handle */
115985	Fts3Phrase p, / Phrase object to advance to next docid */
115986	u8 pbEof / OUT: Set to 1 if EOF */
115987	){
115988	int rc = SQLITE_OK;
115989	Fts3Doclist *pDL = &p->doclist;
115990	Fts3Table pTab = (Fts3Table )pCsr->base.pVtab;
115991
	@@ -115498,14 +116027,14 @@
116027	fts3PoslistCopy(0, &pIter);
116028	pDL->nList = (pIter - pDL->pList);
116029
116030	/* pIter now points just past the 0x00 that terminates the position-
116031	** list for document pDL->iDocid. However, if this position-list was
116032	** edited in place by fts3EvalNearTrim(), then pIter may not actually
116033	** point to the start of the next docid value. The following line deals
116034	** with this case by advancing pIter past the zero-padding added by
116035	** fts3EvalNearTrim(). */
116036	while( pIter<pEnd && *pIter==0 ) pIter++;
116037
116038	pDL->pNextDocid = pIter;
116039	assert( pIter>=&pDL->aAll[pDL->nAll] \|\| *pIter );
116040	*pbEof = 0;
	@@ -115513,15 +116042,31 @@
116042	}
116043
116044	return rc;
116045	}
116046
116047	/*
116048	**
116049	** If *pRc is not SQLITE_OK when this function is called, it is a no-op.
116050	** Otherwise, fts3EvalPhraseStart() is called on all phrases within the
116051	** expression. Also the Fts3Expr.bDeferred variable is set to true for any
116052	** expressions for which all descendent tokens are deferred.
116053	**
116054	** If parameter bOptOk is zero, then it is guaranteed that the
116055	** Fts3Phrase.doclist.aAll/nAll variables contain the entire doclist for
116056	** each phrase in the expression (subject to deferred token processing).
116057	** Or, if bOptOk is non-zero, then one or more tokens within the expression
116058	** may be loaded incrementally, meaning doclist.aAll/nAll is not available.
116059	**
116060	** If an error occurs within this function, *pRc is set to an SQLite error
116061	** code before returning.
116062	*/
116063	static void fts3EvalStartReaders(
116064	Fts3Cursor pCsr, / FTS Cursor handle */
116065	Fts3Expr pExpr, / Expression to initialize phrases in */
116066	int bOptOk, /* True to enable incremental loading */
116067	int pRc / IN/OUT: Error code */
116068	){
116069	if( pExpr && SQLITE_OK==*pRc ){
116070	if( pExpr->eType==FTSQUERY_PHRASE ){
116071	int i;
116072	int nToken = pExpr->pPhrase->nToken;
	@@ -115536,27 +116081,46 @@
116081	pExpr->bDeferred = (pExpr->pLeft->bDeferred && pExpr->pRight->bDeferred);
116082	}
116083	}
116084	}
116085
116086	/*
116087	** An array of the following structures is assembled as part of the process
116088	** of selecting tokens to defer before the query starts executing (as part
116089	** of the xFilter() method). There is one element in the array for each
116090	** token in the FTS expression.
116091	**
116092	** Tokens are divided into AND/NEAR clusters. All tokens in a cluster belong
116093	** to phrases that are connected only by AND and NEAR operators (not OR or
116094	** NOT). When determining tokens to defer, each AND/NEAR cluster is considered
116095	** separately. The root of a tokens AND/NEAR cluster is stored in
116096	** Fts3TokenAndCost.pRoot.
116097	*/
116098	typedef struct Fts3TokenAndCost Fts3TokenAndCost;
116099	struct Fts3TokenAndCost {
116100	Fts3Phrase pPhrase; / The phrase the token belongs to */
116101	int iToken; /* Position of token in phrase */
116102	Fts3PhraseToken pToken; / The token itself */
116103	Fts3Expr pRoot; / Root of NEAR/AND cluster */
116104	int nOvfl; /* Number of overflow pages to load doclist */
116105	int iCol; /* The column the token must match */
116106	};
116107
116108	/*
116109	** This function is used to populate an allocated Fts3TokenAndCost array.
116110	**
116111	** If *pRc is not SQLITE_OK when this function is called, it is a no-op.
116112	** Otherwise, if an error occurs during execution, *pRc is set to an
116113	** SQLite error code.
116114	*/
116115	static void fts3EvalTokenCosts(
116116	Fts3Cursor pCsr, / FTS Cursor handle */
116117	Fts3Expr pRoot, / Root of current AND/NEAR cluster */
116118	Fts3Expr pExpr, / Expression to consider */
116119	Fts3TokenAndCost *ppTC, / Write new entries to (ppTC)++ */
116120	Fts3Expr **ppOr, / Write new OR root to (ppOr)++ */
116121	int pRc / IN/OUT: Error code */
116122	){
116123	if( *pRc==SQLITE_OK && pExpr ){
116124	if( pExpr->eType==FTSQUERY_PHRASE ){
116125	Fts3Phrase *pPhrase = pExpr->pPhrase;
116126	int i;
	@@ -115584,23 +116148,34 @@
116148	fts3EvalTokenCosts(pCsr, pRoot, pExpr->pRight, ppTC, ppOr, pRc);
116149	}
116150	}
116151	}
116152
116153	/*
116154	** Determine the average document (row) size in pages. If successful,
116155	** write this value to *pnPage and return SQLITE_OK. Otherwise, return
116156	** an SQLite error code.
116157	**
116158	** The average document size in pages is calculated by first calculating
116159	** determining the average size in bytes, B. If B is less than the amount
116160	** of data that will fit on a single leaf page of an intkey table in
116161	** this database, then the average docsize is 1. Otherwise, it is 1 plus
116162	** the number of overflow pages consumed by a record B bytes in size.
116163	*/
116164	static int fts3EvalAverageDocsize(Fts3Cursor pCsr, int pnPage){
116165	if( pCsr->nRowAvg==0 ){
116166	/* The average document size, which is required to calculate the cost
116167	** of each doclist, has not yet been determined. Read the required
116168	** data from the %_stat table to calculate it.
116169	**
116170	** Entry 0 of the %_stat table is a blob containing (nCol+1) FTS3
116171	** varints, where nCol is the number of columns in the FTS3 table.
116172	** The first varint is the number of documents currently stored in
116173	** the table. The following nCol varints contain the total amount of
116174	** data stored in all rows of each column of the table, from left
116175	** to right.
116176	*/
116177	int rc;
116178	Fts3Table p = (Fts3Table)pCsr->base.pVtab;
116179	sqlite3_stmt *pStmt;
116180	sqlite3_int64 nDoc = 0;
116181	sqlite3_int64 nByte = 0;
	@@ -115631,109 +116206,151 @@
116206
116207	*pnPage = pCsr->nRowAvg;
116208	return SQLITE_OK;
116209	}
116210
116211	/*
116212	** This function is called to select the tokens (if any) that will be
116213	** deferred. The array aTC[] has already been populated when this is
116214	** called.
116215	**
116216	** This function is called once for each AND/NEAR cluster in the
116217	** expression. Each invocation determines which tokens to defer within
116218	** the cluster with root node pRoot. See comments above the definition
116219	** of struct Fts3TokenAndCost for more details.
116220	**
116221	** If no error occurs, SQLITE_OK is returned and sqlite3Fts3DeferToken()
116222	** called on each token to defer. Otherwise, an SQLite error code is
116223	** returned.
116224	*/
116225	static int fts3EvalSelectDeferred(
116226	Fts3Cursor pCsr, / FTS Cursor handle */
116227	Fts3Expr pRoot, / Consider tokens with this root node */
116228	Fts3TokenAndCost aTC, / Array of expression tokens and costs */
116229	int nTC /* Number of entries in aTC[] */
116230	){



116231	Fts3Table pTab = (Fts3Table )pCsr->base.pVtab;
116232	int nDocSize = 0; /* Number of pages per doc loaded */
116233	int rc = SQLITE_OK; /* Return code */
116234	int ii; /* Iterator variable for various purposes */
116235	int nOvfl = 0; /* Total overflow pages used by doclists */
116236	int nToken = 0; /* Total number of tokens in cluster */
116237
116238	int nMinEst = 0; /* The minimum count for any phrase so far. */
116239	int nLoad4 = 1; /* (Phrases that will be loaded)^4. */
116240
116241	/* Count the tokens in this AND/NEAR cluster. If none of the doclists
116242	** associated with the tokens spill onto overflow pages, or if there is
116243	** only 1 token, exit early. No tokens to defer in this case. */
116244	for(ii=0; ii<nTC; ii++){
116245	if( aTC[ii].pRoot==pRoot ){
116246	nOvfl += aTC[ii].nOvfl;
116247	nToken++;
116248	}
116249	}
116250	if( nOvfl==0 \|\| nToken<2 ) return SQLITE_OK;
116251
116252	/* Obtain the average docsize (in pages). */
116253	rc = fts3EvalAverageDocsize(pCsr, &nDocSize);
116254	assert( rc!=SQLITE_OK \|\| nDocSize>0 );
116255
116256
116257	/* Iterate through all tokens in this AND/NEAR cluster, in ascending order
116258	** of the number of overflow pages that will be loaded by the pager layer
116259	** to retrieve the entire doclist for the token from the full-text index.
116260	** Load the doclists for tokens that are either:
116261	**
116262	** a. The cheapest token in the entire query (i.e. the one visited by the
116263	** first iteration of this loop), or
116264	**
116265	** b. Part of a multi-token phrase.
116266	**
116267	** After each token doclist is loaded, merge it with the others from the
116268	** same phrase and count the number of documents that the merged doclist
116269	** contains. Set variable "nMinEst" to the smallest number of documents in
116270	** any phrase doclist for which 1 or more token doclists have been loaded.
116271	** Let nOther be the number of other phrases for which it is certain that
116272	** one or more tokens will not be deferred.
116273	**
116274	** Then, for each token, defer it if loading the doclist would result in
116275	** loading N or more overflow pages into memory, where N is computed as:
116276	**
116277	** (nMinEst + 4^nOther - 1) / (4^nOther)
116278	*/
116279	for(ii=0; ii<nToken && rc==SQLITE_OK; ii++){
116280	int iTC; /* Used to iterate through aTC[] array. */
116281	Fts3TokenAndCost pTC = 0; / Set to cheapest remaining token. */
116282
116283	/* Set pTC to point to the cheapest remaining token. */
116284	for(iTC=0; iTC<nTC; iTC++){
116285	if( aTC[iTC].pToken && aTC[iTC].pRoot==pRoot
116286	&& (!pTC \|\| aTC[iTC].nOvfl<pTC->nOvfl)
116287	){
116288	pTC = &aTC[iTC];
116289	}
116290	}
116291	assert( pTC );
116292
116293	if( ii && pTC->nOvfl>=((nMinEst+(nLoad4/4)-1)/(nLoad4/4))*nDocSize ){
116294	/* The number of overflow pages to load for this (and therefore all
116295	** subsequent) tokens is greater than the estimated number of pages
116296	** that will be loaded if all subsequent tokens are deferred.
116297	*/
116298	Fts3PhraseToken *pToken = pTC->pToken;
116299	rc = sqlite3Fts3DeferToken(pCsr, pToken, pTC->iCol);
116300	fts3SegReaderCursorFree(pToken->pSegcsr);
116301	pToken->pSegcsr = 0;
116302	}else{
116303	nLoad4 = nLoad4*4;
116304	if( ii==0 \|\| pTC->pPhrase->nToken>1 ){
116305	/* Either this is the cheapest token in the entire query, or it is
116306	** part of a multi-token phrase. Either way, the entire doclist will
116307	** (eventually) be loaded into memory. It may as well be now. */
116308	Fts3PhraseToken *pToken = pTC->pToken;
116309	int nList = 0;
116310	char *pList = 0;
116311	rc = fts3TermSelect(pTab, pToken, pTC->iCol, &nList, &pList);
116312	assert( rc==SQLITE_OK \|\| pList==0 );

116313	if( rc==SQLITE_OK ){
116314	int nCount;
116315	fts3EvalPhraseMergeToken(pTab, pTC->pPhrase, pTC->iToken,pList,nList);
116316	nCount = fts3DoclistCountDocids(
116317	pTC->pPhrase->doclist.aAll, pTC->pPhrase->doclist.nAll
116318	);
116319	if( ii==0 \|\| nCount<nMinEst ) nMinEst = nCount;
116320	}
116321	}








116322	}
116323	pTC->pToken = 0;
116324	}
116325
116326	return rc;
116327	}
116328
116329	/*
116330	** This function is called from within the xFilter method. It initializes
116331	** the full-text query currently stored in pCsr->pExpr. To iterate through
116332	** the results of a query, the caller does:
116333	**
116334	** fts3EvalStart(pCsr);
116335	** while( 1 ){
116336	** fts3EvalNext(pCsr);
116337	** if( pCsr->bEof ) break;
116338	** ... return row pCsr->iPrevId to the caller ...
116339	** }
116340	*/
116341	static int fts3EvalStart(Fts3Cursor *pCsr){
116342	Fts3Table pTab = (Fts3Table )pCsr->base.pVtab;
116343	int rc = SQLITE_OK;
116344	int nToken = 0;
116345	int nOr = 0;
116346
116347	/* Allocate a MultiSegReader for each token in the expression. */
116348	fts3EvalAllocateReaders(pCsr, pCsr->pExpr, &nToken, &nOr, &rc);
116349
116350	/* Determine which, if any, tokens in the expression should be deferred. */
116351	if( rc==SQLITE_OK && nToken>1 && pTab->bHasStat ){



















116352	Fts3TokenAndCost *aTC;
116353	Fts3Expr **apOr;
116354	aTC = (Fts3TokenAndCost *)sqlite3_malloc(
116355	sizeof(Fts3TokenAndCost) * nToken
116356	+ sizeof(Fts3Expr ) nOr * 2
	@@ -115745,11 +116362,11 @@
116362	}else{
116363	int ii;
116364	Fts3TokenAndCost *pTC = aTC;
116365	Fts3Expr **ppOr = apOr;
116366
116367	fts3EvalTokenCosts(pCsr, 0, pCsr->pExpr, &pTC, &ppOr, &rc);
116368	nToken = pTC-aTC;
116369	nOr = ppOr-apOr;
116370
116371	if( rc==SQLITE_OK ){
116372	rc = fts3EvalSelectDeferred(pCsr, 0, aTC, nToken);
	@@ -115760,25 +116377,50 @@
116377
116378	sqlite3_free(aTC);
116379	}
116380	}
116381
116382	fts3EvalStartReaders(pCsr, pCsr->pExpr, 1, &rc);
116383	return rc;
116384	}
116385
116386	/*
116387	** Invalidate the current position list for phrase pPhrase.
116388	*/
116389	static void fts3EvalInvalidatePoslist(Fts3Phrase *pPhrase){
116390	if( pPhrase->doclist.bFreeList ){
116391	sqlite3_free(pPhrase->doclist.pList);
116392	}
116393	pPhrase->doclist.pList = 0;
116394	pPhrase->doclist.nList = 0;
116395	pPhrase->doclist.bFreeList = 0;
116396	}
116397
116398	/*
116399	** This function is called to edit the position list associated with
116400	** the phrase object passed as the fifth argument according to a NEAR
116401	** condition. For example:
116402	**
116403	** abc NEAR/5 "def ghi"
116404	**
116405	** Parameter nNear is passed the NEAR distance of the expression (5 in
116406	** the example above). When this function is called, *paPoslist points to
116407	** the position list, and *pnToken is the number of phrase tokens in, the
116408	** phrase on the other side of the NEAR operator to pPhrase. For example,
116409	** if pPhrase refers to the "def ghi" phrase, then *paPoslist points to
116410	** the position list associated with phrase "abc".
116411	**
116412	** All positions in the pPhrase position list that are not sufficiently
116413	** close to a position in the *paPoslist position list are removed. If this
116414	** leaves 0 positions, zero is returned. Otherwise, non-zero.
116415	**
116416	** Before returning, *paPoslist is set to point to the position lsit
116417	** associated with pPhrase. And *pnToken is set to the number of tokens in
116418	** pPhrase.
116419	*/
116420	static int fts3EvalNearTrim(
116421	int nNear, /* NEAR distance. As in "NEAR/nNear". */
116422	char aTmp, / Temporary space to use */
116423	char *paPoslist, / IN/OUT: Position list */
116424	int pnToken, / IN/OUT: Tokens in phrase of paPoslist /
116425	Fts3Phrase pPhrase / The phrase object to trim the doclist of */
116426	){
	@@ -115806,10 +116448,176 @@
116448	}
116449
116450	return res;
116451	}
116452
116453	/*
116454	** This function is a no-op if *pRc is other than SQLITE_OK when it is called.
116455	** Otherwise, it advances the expression passed as the second argument to
116456	** point to the next matching row in the database. Expressions iterate through
116457	** matching rows in docid order. Ascending order if Fts3Cursor.bDesc is zero,
116458	** or descending if it is non-zero.
116459	**
116460	** If an error occurs, *pRc is set to an SQLite error code. Otherwise, if
116461	** successful, the following variables in pExpr are set:
116462	**
116463	** Fts3Expr.bEof (non-zero if EOF - there is no next row)
116464	** Fts3Expr.iDocid (valid if bEof==0. The docid of the next row)
116465	**
116466	** If the expression is of type FTSQUERY_PHRASE, and the expression is not
116467	** at EOF, then the following variables are populated with the position list
116468	** for the phrase for the visited row:
116469	**
116470	** FTs3Expr.pPhrase->doclist.nList (length of pList in bytes)
116471	** FTs3Expr.pPhrase->doclist.pList (pointer to position list)
116472	**
116473	** It says above that this function advances the expression to the next
116474	** matching row. This is usually true, but there are the following exceptions:
116475	**
116476	** 1. Deferred tokens are not taken into account. If a phrase consists
116477	** entirely of deferred tokens, it is assumed to match every row in
116478	** the db. In this case the position-list is not populated at all.
116479	**
116480	** Or, if a phrase contains one or more deferred tokens and one or
116481	** more non-deferred tokens, then the expression is advanced to the
116482	** next possible match, considering only non-deferred tokens. In other
116483	** words, if the phrase is "A B C", and "B" is deferred, the expression
116484	** is advanced to the next row that contains an instance of "A * C",
116485	** where "*" may match any single token. The position list in this case
116486	** is populated as for "A * C" before returning.
116487	**
116488	** 2. NEAR is treated as AND. If the expression is "x NEAR y", it is
116489	** advanced to point to the next row that matches "x AND y".
116490	**
116491	** See fts3EvalTestDeferredAndNear() for details on testing if a row is
116492	** really a match, taking into account deferred tokens and NEAR operators.
116493	*/
116494	static void fts3EvalNextRow(
116495	Fts3Cursor pCsr, / FTS Cursor handle */
116496	Fts3Expr pExpr, / Expr. to advance to next matching row */
116497	int pRc / IN/OUT: Error code */
116498	){
116499	if( *pRc==SQLITE_OK ){
116500	int bDescDoclist = pCsr->bDesc; /* Used by DOCID_CMP() macro */
116501	assert( pExpr->bEof==0 );
116502	pExpr->bStart = 1;
116503
116504	switch( pExpr->eType ){
116505	case FTSQUERY_NEAR:
116506	case FTSQUERY_AND: {
116507	Fts3Expr *pLeft = pExpr->pLeft;
116508	Fts3Expr *pRight = pExpr->pRight;
116509	assert( !pLeft->bDeferred \|\| !pRight->bDeferred );
116510
116511	if( pLeft->bDeferred ){
116512	/* LHS is entirely deferred. So we assume it matches every row.
116513	** Advance the RHS iterator to find the next row visited. */
116514	fts3EvalNextRow(pCsr, pRight, pRc);
116515	pExpr->iDocid = pRight->iDocid;
116516	pExpr->bEof = pRight->bEof;
116517	}else if( pRight->bDeferred ){
116518	/* RHS is entirely deferred. So we assume it matches every row.
116519	** Advance the LHS iterator to find the next row visited. */
116520	fts3EvalNextRow(pCsr, pLeft, pRc);
116521	pExpr->iDocid = pLeft->iDocid;
116522	pExpr->bEof = pLeft->bEof;
116523	}else{
116524	/* Neither the RHS or LHS are deferred. */
116525	fts3EvalNextRow(pCsr, pLeft, pRc);
116526	fts3EvalNextRow(pCsr, pRight, pRc);
116527	while( !pLeft->bEof && !pRight->bEof && *pRc==SQLITE_OK ){
116528	sqlite3_int64 iDiff = DOCID_CMP(pLeft->iDocid, pRight->iDocid);
116529	if( iDiff==0 ) break;
116530	if( iDiff<0 ){
116531	fts3EvalNextRow(pCsr, pLeft, pRc);
116532	}else{
116533	fts3EvalNextRow(pCsr, pRight, pRc);
116534	}
116535	}
116536	pExpr->iDocid = pLeft->iDocid;
116537	pExpr->bEof = (pLeft->bEof \|\| pRight->bEof);
116538	}
116539	break;
116540	}
116541
116542	case FTSQUERY_OR: {
116543	Fts3Expr *pLeft = pExpr->pLeft;
116544	Fts3Expr *pRight = pExpr->pRight;
116545	sqlite3_int64 iCmp = DOCID_CMP(pLeft->iDocid, pRight->iDocid);
116546
116547	assert( pLeft->bStart \|\| pLeft->iDocid==pRight->iDocid );
116548	assert( pRight->bStart \|\| pLeft->iDocid==pRight->iDocid );
116549
116550	if( pRight->bEof \|\| (pLeft->bEof==0 && iCmp<0) ){
116551	fts3EvalNextRow(pCsr, pLeft, pRc);
116552	}else if( pLeft->bEof \|\| (pRight->bEof==0 && iCmp>0) ){
116553	fts3EvalNextRow(pCsr, pRight, pRc);
116554	}else{
116555	fts3EvalNextRow(pCsr, pLeft, pRc);
116556	fts3EvalNextRow(pCsr, pRight, pRc);
116557	}
116558
116559	pExpr->bEof = (pLeft->bEof && pRight->bEof);
116560	iCmp = DOCID_CMP(pLeft->iDocid, pRight->iDocid);
116561	if( pRight->bEof \|\| (pLeft->bEof==0 && iCmp<0) ){
116562	pExpr->iDocid = pLeft->iDocid;
116563	}else{
116564	pExpr->iDocid = pRight->iDocid;
116565	}
116566
116567	break;
116568	}
116569
116570	case FTSQUERY_NOT: {
116571	Fts3Expr *pLeft = pExpr->pLeft;
116572	Fts3Expr *pRight = pExpr->pRight;
116573
116574	if( pRight->bStart==0 ){
116575	fts3EvalNextRow(pCsr, pRight, pRc);
116576	assert( *pRc!=SQLITE_OK \|\| pRight->bStart );
116577	}
116578
116579	fts3EvalNextRow(pCsr, pLeft, pRc);
116580	if( pLeft->bEof==0 ){
116581	while( !*pRc
116582	&& !pRight->bEof
116583	&& DOCID_CMP(pLeft->iDocid, pRight->iDocid)>0
116584	){
116585	fts3EvalNextRow(pCsr, pRight, pRc);
116586	}
116587	}
116588	pExpr->iDocid = pLeft->iDocid;
116589	pExpr->bEof = pLeft->bEof;
116590	break;
116591	}
116592
116593	default: {
116594	Fts3Phrase *pPhrase = pExpr->pPhrase;
116595	fts3EvalInvalidatePoslist(pPhrase);
116596	*pRc = fts3EvalPhraseNext(pCsr, pPhrase, &pExpr->bEof);
116597	pExpr->iDocid = pPhrase->doclist.iDocid;
116598	break;
116599	}
116600	}
116601	}
116602	}
116603
116604	/*
116605	** If *pRc is not SQLITE_OK, or if pExpr is not the root node of a NEAR
116606	** cluster, then this function returns 1 immediately.
116607	**
116608	** Otherwise, it checks if the current row really does match the NEAR
116609	** expression, using the data currently stored in the position lists
116610	** (Fts3Expr->pPhrase.doclist.pList/nList) for each phrase in the expression.
116611	**
116612	** If the current row is a match, the position list associated with each
116613	** phrase in the NEAR expression is edited in place to contain only those
116614	** phrase instances sufficiently close to their peers to satisfy all NEAR
116615	** constraints. In this case it returns 1. If the NEAR expression does not
116616	** match the current row, 0 is returned. The position lists may or may not
116617	** be edited if 0 is returned.
116618	*/
116619	static int fts3EvalNearTest(Fts3Expr pExpr, int pRc){
116620	int res = 1;
116621
116622	/* The following block runs if pExpr is the root of a NEAR query.
116623	** For example, the query:
	@@ -115827,11 +116635,11 @@
116635	** \| \|
116636	** "w" "x"
116637	**
116638	** The right-hand child of a NEAR node is always a phrase. The
116639	** left-hand child may be either a phrase or a NEAR node. There are
116640	** no exceptions to this - it's the way the parser in fts3_expr.c works.
116641	*/
116642	if( *pRc==SQLITE_OK
116643	&& pExpr->eType==FTSQUERY_NEAR
116644	&& pExpr->bEof==0
116645	&& (pExpr->pParent==0 \|\| pExpr->pParent->eType!=FTSQUERY_NEAR)
	@@ -115854,21 +116662,21 @@
116662	int nToken = p->pPhrase->nToken;
116663
116664	for(p=p->pParent;res && p && p->eType==FTSQUERY_NEAR; p=p->pParent){
116665	Fts3Phrase *pPhrase = p->pRight->pPhrase;
116666	int nNear = p->nNear;
116667	res = fts3EvalNearTrim(nNear, aTmp, &aPoslist, &nToken, pPhrase);
116668	}
116669
116670	aPoslist = pExpr->pRight->pPhrase->doclist.pList;
116671	nToken = pExpr->pRight->pPhrase->nToken;
116672	for(p=pExpr->pLeft; p && res; p=p->pLeft){
116673	int nNear = p->pParent->nNear;
116674	Fts3Phrase *pPhrase = (
116675	p->eType==FTSQUERY_NEAR ? p->pRight->pPhrase : p->pPhrase
116676	);
116677	res = fts3EvalNearTrim(nNear, aTmp, &aPoslist, &nToken, pPhrase);
116678	}
116679	}
116680
116681	sqlite3_free(aTmp);
116682	}
	@@ -115875,132 +116683,33 @@
116683
116684	return res;
116685	}
116686
116687	/*
116688	** This function is a helper function for fts3EvalTestDeferredAndNear().
116689	** Assuming no error occurs or has occurred, It returns non-zero if the
116690	** expression passed as the second argument matches the row that pCsr
116691	** currently points to, or zero if it does not.
116692	**
116693	** If *pRc is not SQLITE_OK when this function is called, it is a no-op.
116694	** If an error occurs during execution of this function, *pRc is set to
116695	** the appropriate SQLite error code. In this case the returned value is
116696	** undefined.
116697	*/
116698	static int fts3EvalTestExpr(
116699	Fts3Cursor pCsr, / FTS cursor handle */
116700	Fts3Expr pExpr, / Expr to test. May or may not be root. */
116701	int pRc / IN/OUT: Error code */
116702	){
116703	int bHit = 1; /* Return value */







































































































116704	if( *pRc==SQLITE_OK ){
116705	switch( pExpr->eType ){
116706	case FTSQUERY_NEAR:
116707	case FTSQUERY_AND:
116708	bHit = (
116709	fts3EvalTestExpr(pCsr, pExpr->pLeft, pRc)
116710	&& fts3EvalTestExpr(pCsr, pExpr->pRight, pRc)
116711	&& fts3EvalNearTest(pExpr, pRc)
116712	);
116713
116714	/* If the NEAR expression does not match any rows, zero the doclist for
116715	** all phrases involved in the NEAR. This is because the snippet(),
	@@ -116022,31 +116731,31 @@
116731	&& (pExpr->pParent==0 \|\| pExpr->pParent->eType!=FTSQUERY_NEAR)
116732	){
116733	Fts3Expr *p;
116734	for(p=pExpr; p->pPhrase==0; p=p->pLeft){
116735	if( p->pRight->iDocid==pCsr->iPrevId ){
116736	fts3EvalInvalidatePoslist(p->pRight->pPhrase);
116737	}
116738	}
116739	if( p->iDocid==pCsr->iPrevId ){
116740	fts3EvalInvalidatePoslist(p->pPhrase);
116741	}
116742	}
116743
116744	break;
116745
116746	case FTSQUERY_OR: {
116747	int bHit1 = fts3EvalTestExpr(pCsr, pExpr->pLeft, pRc);
116748	int bHit2 = fts3EvalTestExpr(pCsr, pExpr->pRight, pRc);
116749	bHit = bHit1 \|\| bHit2;
116750	break;
116751	}
116752
116753	case FTSQUERY_NOT:
116754	bHit = (
116755	fts3EvalTestExpr(pCsr, pExpr->pLeft, pRc)
116756	&& !fts3EvalTestExpr(pCsr, pExpr->pRight, pRc)
116757	);
116758	break;
116759
116760	default: {
116761	if( pCsr->pDeferred
	@@ -116053,11 +116762,11 @@
116762	&& (pExpr->iDocid==pCsr->iPrevId \|\| pExpr->bDeferred)
116763	){
116764	Fts3Phrase *pPhrase = pExpr->pPhrase;
116765	assert( pExpr->bDeferred \|\| pPhrase->doclist.bFreeList==0 );
116766	if( pExpr->bDeferred ){
116767	fts3EvalInvalidatePoslist(pPhrase);
116768	}
116769	*pRc = fts3EvalDeferredPhrase(pCsr, pPhrase);
116770	bHit = (pPhrase->doclist.pList!=0);
116771	pExpr->iDocid = pCsr->iPrevId;
116772	}else{
	@@ -116069,31 +116778,53 @@
116778	}
116779	return bHit;
116780	}
116781
116782	/*
116783	** This function is called as the second part of each xNext operation when
116784	** iterating through the results of a full-text query. At this point the
116785	** cursor points to a row that matches the query expression, with the
116786	** following caveats:
116787	**
116788	** * Up until this point, "NEAR" operators in the expression have been
116789	** treated as "AND".
116790	**
116791	** * Deferred tokens have not yet been considered.
116792	**
116793	** If *pRc is not SQLITE_OK when this function is called, it immediately
116794	** returns 0. Otherwise, it tests whether or not after considering NEAR
116795	** operators and deferred tokens the current row is still a match for the
116796	** expression. It returns 1 if both of the following are true:
116797	**
116798	** 1. *pRc is SQLITE_OK when this function returns, and
116799	**
116800	** 2. After scanning the current FTS table row for the deferred tokens,
116801	** it is determined that the row does not match the query.
116802	**
116803	** Or, if no error occurs and it seems the current row does match the FTS
116804	** query, return 0.
116805	*/
116806	static int fts3EvalTestDeferredAndNear(Fts3Cursor pCsr, int pRc){
116807	int rc = *pRc;
116808	int bMiss = 0;
116809	if( rc==SQLITE_OK ){
116810
116811	/* If there are one or more deferred tokens, load the current row into
116812	** memory and scan it to determine the position list for each deferred
116813	** token. Then, see if this row is really a match, considering deferred
116814	** tokens and NEAR operators (neither of which were taken into account
116815	** earlier, by fts3EvalNextRow()).
116816	*/
116817	if( pCsr->pDeferred ){
116818	rc = fts3CursorSeek(0, pCsr);
116819	if( rc==SQLITE_OK ){
116820	rc = sqlite3Fts3CacheDeferredDoclists(pCsr);
116821	}
116822	}
116823	bMiss = (0==fts3EvalTestExpr(pCsr, pCsr->pExpr, &rc));
116824
116825	/* Free the position-lists accumulated for each deferred token above. */
116826	sqlite3Fts3FreeDeferredDoclists(pCsr);
116827	*pRc = rc;
116828	}
116829	return (rc==SQLITE_OK && bMiss);
116830	}
	@@ -116100,11 +116831,11 @@
116831
116832	/*
116833	** Advance to the next document that matches the FTS expression in
116834	** Fts3Cursor.pExpr.
116835	*/
116836	static int fts3EvalNext(Fts3Cursor *pCsr){
116837	int rc = SQLITE_OK; /* Return Code */
116838	Fts3Expr *pExpr = pCsr->pExpr;
116839	assert( pCsr->isEof==0 );
116840	if( pExpr==0 ){
116841	pCsr->isEof = 1;
	@@ -116112,23 +116843,23 @@
116843	do {
116844	if( pCsr->isRequireSeek==0 ){
116845	sqlite3_reset(pCsr->pStmt);
116846	}
116847	assert( sqlite3_data_count(pCsr->pStmt)==0 );
116848	fts3EvalNextRow(pCsr, pExpr, &rc);
116849	pCsr->isEof = pExpr->bEof;
116850	pCsr->isRequireSeek = 1;
116851	pCsr->isMatchinfoNeeded = 1;
116852	pCsr->iPrevId = pExpr->iDocid;
116853	}while( pCsr->isEof==0 && fts3EvalTestDeferredAndNear(pCsr, &rc) );
116854	}
116855	return rc;
116856	}
116857
116858	/*
116859	** Restart interation for expression pExpr so that the next call to
116860	** fts3EvalNext() visits the first row. Do not allow incremental
116861	** loading or merging of phrase doclists for this iteration.
116862	**
116863	** If *pRc is other than SQLITE_OK when this function is called, it is
116864	** a no-op. If an error occurs within this function, *pRc is set to an
116865	** SQLite error code before returning.
	@@ -116140,11 +116871,11 @@
116871	){
116872	if( pExpr && *pRc==SQLITE_OK ){
116873	Fts3Phrase *pPhrase = pExpr->pPhrase;
116874
116875	if( pPhrase ){
116876	fts3EvalInvalidatePoslist(pPhrase);
116877	if( pPhrase->bIncr ){
116878	assert( pPhrase->nToken==1 );
116879	assert( pPhrase->aToken[0].pSegcsr );
116880	sqlite3Fts3MsrIncrRestart(pPhrase->aToken[0].pSegcsr);
116881	*pRc = fts3EvalPhraseStart(pCsr, 0, pPhrase);
	@@ -116256,18 +116987,18 @@
116987	/* Ensure the %_content statement is reset. */
116988	if( pCsr->isRequireSeek==0 ) sqlite3_reset(pCsr->pStmt);
116989	assert( sqlite3_data_count(pCsr->pStmt)==0 );
116990
116991	/* Advance to the next document */
116992	fts3EvalNextRow(pCsr, pRoot, &rc);
116993	pCsr->isEof = pRoot->bEof;
116994	pCsr->isRequireSeek = 1;
116995	pCsr->isMatchinfoNeeded = 1;
116996	pCsr->iPrevId = pRoot->iDocid;
116997	}while( pCsr->isEof==0
116998	&& pRoot->eType==FTSQUERY_NEAR
116999	&& fts3EvalTestDeferredAndNear(pCsr, &rc)
117000	);
117001
117002	if( rc==SQLITE_OK && pCsr->isEof==0 ){
117003	fts3EvalUpdateCounts(pRoot);
117004	}
	@@ -116285,14 +117016,14 @@
117016	**
117017	** do {...} while( pRoot->iDocid<iDocid && rc==SQLITE_OK );
117018	*/
117019	fts3EvalRestart(pCsr, pRoot, &rc);
117020	do {
117021	fts3EvalNextRow(pCsr, pRoot, &rc);
117022	assert( pRoot->bEof==0 );
117023	}while( pRoot->iDocid!=iDocid && rc==SQLITE_OK );
117024	fts3EvalTestDeferredAndNear(pCsr, &rc);
117025	}
117026	}
117027	return rc;
117028	}
117029
	@@ -116419,18 +117150,32 @@
117150	*/
117151	SQLITE_PRIVATE void sqlite3Fts3EvalPhraseCleanup(Fts3Phrase *pPhrase){
117152	if( pPhrase ){
117153	int i;
117154	sqlite3_free(pPhrase->doclist.aAll);
117155	fts3EvalInvalidatePoslist(pPhrase);
117156	memset(&pPhrase->doclist, 0, sizeof(Fts3Doclist));
117157	for(i=0; i<pPhrase->nToken; i++){
117158	fts3SegReaderCursorFree(pPhrase->aToken[i].pSegcsr);
117159	pPhrase->aToken[i].pSegcsr = 0;
117160	}
117161	}
117162	}
117163
117164	#if !SQLITE_CORE
117165	/*
117166	** Initialize API pointer table, if required.
117167	*/
117168	SQLITE_API int sqlite3_extension_init(
117169	sqlite3 *db,
117170	char **pzErrMsg,
117171	const sqlite3_api_routines *pApi
117172	){
117173	SQLITE_EXTENSION_INIT2(pApi)
117174	return sqlite3Fts3Init(db);
117175	}
117176	#endif
117177
117178	#endif
117179
117180	/************ End of fts3.c **********************************************/
117181	/************ Begin file fts3_aux.c **************************************/
	@@ -118917,14 +119662,10 @@
119662	** (in which case SQLITE_CORE is not defined), or
119663	**
119664	** * The FTS3 module is being built into the core of
119665	** SQLite (in which case SQLITE_ENABLE_FTS3 is defined).
119666	*/




119667	#if !defined(SQLITE_CORE) \|\| defined(SQLITE_ENABLE_FTS3)
119668
119669
119670	/*
119671	** Implementation of the SQL scalar function for accessing the underlying
119672

M src/sqlite3.h

+21 -4

		--- src/sqlite3.h
		+++ src/sqlite3.h
		@@ -105,13 +105,13 @@
105	105	**
106	106	** See also: [sqlite3_libversion()],
107	107	** [sqlite3_libversion_number()], [sqlite3_sourceid()],
108	108	** [sqlite_version()] and [sqlite_source_id()].
109	109	*/
110		-#define SQLITE_VERSION "3.7.7"
111		-#define SQLITE_VERSION_NUMBER 3007007
112		-#define SQLITE_SOURCE_ID "2011-06-24 11:29:51 9b191bb4c7c1e1b12b188c0b3eee1f8f587887c8"
	110	+#define SQLITE_VERSION "3.7.8"
	111	+#define SQLITE_VERSION_NUMBER 3007008
	112	+#define SQLITE_SOURCE_ID "2011-07-19 18:29:00 ed5f0aad6b21066bacd01521e82c22e96991f400"
113	113
114	114	/*
115	115	** CAPI3REF: Run-Time Library Version Numbers
116	116	** KEYWORDS: sqlite3_version, sqlite3_sourceid
117	117	**
		@@ -739,20 +739,37 @@
739	739	** when [PRAGMA synchronous \| PRAGMA synchronous=OFF] is set, but most
740	740	** VFSes do not need this signal and should silently ignore this opcode.
741	741	** Applications should not call [sqlite3_file_control()] with this
742	742	** opcode as doing so may disrupt the operation of the specialized VFSes
743	743	** that do require it.
	744	+**
	745	+** ^The [SQLITE_FCNTL_WIN32_AV_RETRY] opcode is used to configure automatic
	746	+** retry counts and intervals for certain disk I/O operations for the
	747	+** windows [VFS] in order to work to provide robustness against
	748	+** anti-virus programs. By default, the windows VFS will retry file read,
	749	+** file write, and file delete opertions up to 10 times, with a delay
	750	+** of 25 milliseconds before the first retry and with the delay increasing
	751	+** by an additional 25 milliseconds with each subsequent retry. This
	752	+** opcode allows those to values (10 retries and 25 milliseconds of delay)
	753	+** to be adjusted. The values are changed for all database connections
	754	+** within the same process. The argument is a pointer to an array of two
	755	+** integers where the first integer i the new retry count and the second
	756	+** integer is the delay. If either integer is negative, then the setting
	757	+** is not changed but instead the prior value of that setting is written
	758	+** into the array entry, allowing the current retry settings to be
	759	+** interrogated. The zDbName parameter is ignored.
	760	+**
744	761	*/
745	762	#define SQLITE_FCNTL_LOCKSTATE 1
746	763	#define SQLITE_GET_LOCKPROXYFILE 2
747	764	#define SQLITE_SET_LOCKPROXYFILE 3
748	765	#define SQLITE_LAST_ERRNO 4
749	766	#define SQLITE_FCNTL_SIZE_HINT 5
750	767	#define SQLITE_FCNTL_CHUNK_SIZE 6
751	768	#define SQLITE_FCNTL_FILE_POINTER 7
752	769	#define SQLITE_FCNTL_SYNC_OMITTED 8
753		-
	770	+#define SQLITE_FCNTL_WIN32_AV_RETRY 9
754	771
755	772	/*
756	773	** CAPI3REF: Mutex Handle
757	774	**
758	775	** The mutex module within SQLite defines [sqlite3_mutex] to be an
759	776

	--- src/sqlite3.h
	+++ src/sqlite3.h
	@@ -105,13 +105,13 @@
105	**
106	** See also: [sqlite3_libversion()],
107	** [sqlite3_libversion_number()], [sqlite3_sourceid()],
108	** [sqlite_version()] and [sqlite_source_id()].
109	*/
110	#define SQLITE_VERSION "3.7.7"
111	#define SQLITE_VERSION_NUMBER 3007007
112	#define SQLITE_SOURCE_ID "2011-06-24 11:29:51 9b191bb4c7c1e1b12b188c0b3eee1f8f587887c8"
113
114	/*
115	** CAPI3REF: Run-Time Library Version Numbers
116	** KEYWORDS: sqlite3_version, sqlite3_sourceid
117	**
	@@ -739,20 +739,37 @@
739	** when [PRAGMA synchronous \| PRAGMA synchronous=OFF] is set, but most
740	** VFSes do not need this signal and should silently ignore this opcode.
741	** Applications should not call [sqlite3_file_control()] with this
742	** opcode as doing so may disrupt the operation of the specialized VFSes
743	** that do require it.

















744	*/
745	#define SQLITE_FCNTL_LOCKSTATE 1
746	#define SQLITE_GET_LOCKPROXYFILE 2
747	#define SQLITE_SET_LOCKPROXYFILE 3
748	#define SQLITE_LAST_ERRNO 4
749	#define SQLITE_FCNTL_SIZE_HINT 5
750	#define SQLITE_FCNTL_CHUNK_SIZE 6
751	#define SQLITE_FCNTL_FILE_POINTER 7
752	#define SQLITE_FCNTL_SYNC_OMITTED 8
753
754
755	/*
756	** CAPI3REF: Mutex Handle
757	**
758	** The mutex module within SQLite defines [sqlite3_mutex] to be an
759

	--- src/sqlite3.h
	+++ src/sqlite3.h
	@@ -105,13 +105,13 @@
105	**
106	** See also: [sqlite3_libversion()],
107	** [sqlite3_libversion_number()], [sqlite3_sourceid()],
108	** [sqlite_version()] and [sqlite_source_id()].
109	*/
110	#define SQLITE_VERSION "3.7.8"
111	#define SQLITE_VERSION_NUMBER 3007008
112	#define SQLITE_SOURCE_ID "2011-07-19 18:29:00 ed5f0aad6b21066bacd01521e82c22e96991f400"
113
114	/*
115	** CAPI3REF: Run-Time Library Version Numbers
116	** KEYWORDS: sqlite3_version, sqlite3_sourceid
117	**
	@@ -739,20 +739,37 @@
739	** when [PRAGMA synchronous \| PRAGMA synchronous=OFF] is set, but most
740	** VFSes do not need this signal and should silently ignore this opcode.
741	** Applications should not call [sqlite3_file_control()] with this
742	** opcode as doing so may disrupt the operation of the specialized VFSes
743	** that do require it.
744	**
745	** ^The [SQLITE_FCNTL_WIN32_AV_RETRY] opcode is used to configure automatic
746	** retry counts and intervals for certain disk I/O operations for the
747	** windows [VFS] in order to work to provide robustness against
748	** anti-virus programs. By default, the windows VFS will retry file read,
749	** file write, and file delete opertions up to 10 times, with a delay
750	** of 25 milliseconds before the first retry and with the delay increasing
751	** by an additional 25 milliseconds with each subsequent retry. This
752	** opcode allows those to values (10 retries and 25 milliseconds of delay)
753	** to be adjusted. The values are changed for all database connections
754	** within the same process. The argument is a pointer to an array of two
755	** integers where the first integer i the new retry count and the second
756	** integer is the delay. If either integer is negative, then the setting
757	** is not changed but instead the prior value of that setting is written
758	** into the array entry, allowing the current retry settings to be
759	** interrogated. The zDbName parameter is ignored.
760	**
761	*/
762	#define SQLITE_FCNTL_LOCKSTATE 1
763	#define SQLITE_GET_LOCKPROXYFILE 2
764	#define SQLITE_SET_LOCKPROXYFILE 3
765	#define SQLITE_LAST_ERRNO 4
766	#define SQLITE_FCNTL_SIZE_HINT 5
767	#define SQLITE_FCNTL_CHUNK_SIZE 6
768	#define SQLITE_FCNTL_FILE_POINTER 7
769	#define SQLITE_FCNTL_SYNC_OMITTED 8
770	#define SQLITE_FCNTL_WIN32_AV_RETRY 9
771
772	/*
773	** CAPI3REF: Mutex Handle
774	**
775	** The mutex module within SQLite defines [sqlite3_mutex] to be an
776

M src/tar.c

+278 -29

		--- src/tar.c
		+++ src/tar.c
		@@ -27,31 +27,253 @@
27	27	*/
28	28	static struct tarball_t {
29	29	unsigned char aHdr; / Space for building headers */
30	30	char zSpaces; / Spaces for padding */
31	31	char zPrevDir; / Name of directory for previous entry */
	32	+ int nPrevDirAlloc; /* size of zPrevDir */
	33	+ Blob pax; /* PAX data */
32	34	} tball;
	35	+
	36	+
	37	+/*
	38	+** field lengths of 'ustar' name and prefix fields.
	39	+*/
	40	+#define USTAR_NAME_LEN 100
	41	+#define USTAR_PREFIX_LEN 155
	42	+
33	43
34	44	/*
35	45	** Begin the process of generating a tarball.
36	46	**
37	47	** Initialize the GZIP compressor and the table of directory names.
38	48	*/
39	49	static void tar_begin(void){
40	50	assert( tball.aHdr==0 );
41		- tball.aHdr = fossil_malloc(512+512+256);
42		- memset(tball.aHdr, 0, 512+512+256);
	51	+ tball.aHdr = fossil_malloc(512+512);
	52	+ memset(tball.aHdr, 0, 512+512);
43	53	tball.zSpaces = (char*)&tball.aHdr[512];
44		- tball.zPrevDir = (char*)&tball.zSpaces[512];
	54	+ /* zPrevDir init */
	55	+ tball.zPrevDir = NULL;
	56	+ tball.nPrevDirAlloc = 0;
	57	+ /* scratch buffer init */
	58	+ blob_zero(&tball.pax);
	59	+
45	60	memcpy(&tball.aHdr[108], "0000000", 8); /* Owner ID */
46	61	memcpy(&tball.aHdr[116], "0000000", 8); /* Group ID */
47		- memcpy(&tball.aHdr[257], "ustar ", 7); /* Format */
	62	+ memcpy(&tball.aHdr[257], "ustar\00000", 8); /* POSIX.1 format */
	63	+ memcpy(&tball.aHdr[265], "nobody", 7); /* Owner name */
	64	+ memcpy(&tball.aHdr[297], "nobody", 7); /* Group name */
48	65	gzip_begin();
49	66	db_multi_exec(
50	67	"CREATE TEMP TABLE dir(name UNIQUE);"
51	68	);
52	69	}
	70	+
	71	+
	72	+/*
	73	+** verify that lla characters in 'zName' are in the
	74	+** ISO646 (=ASCII) character set.
	75	+*/
	76	+static int is_iso646_name(
	77	+ const char zName, / file path */
	78	+ int nName /* path length */
	79	+){
	80	+ int i;
	81	+ for(i = 0; i < nName; i++){
	82	+ unsigned char c = (unsigned char)zName[i];
	83	+ if( c>0x7e ) return 0;
	84	+ }
	85	+ return 1;
	86	+}
	87	+
	88	+
	89	+/*
	90	+** copy string pSrc into pDst, truncating or padding with 0 if necessary
	91	+*/
	92	+static void padded_copy(
	93	+ char *pDest,
	94	+ int nDest,
	95	+ const char *pSrc,
	96	+ int nSrc
	97	+){
	98	+ if(nSrc >= nDest){
	99	+ memcpy(pDest, pSrc, nDest);
	100	+ }else{
	101	+ memcpy(pDest, pSrc, nSrc);
	102	+ memset(&pDest[nSrc], 0, nDest - nSrc);
	103	+ }
	104	+}
	105	+
	106	+
	107	+
	108	+/******************************************************************************
	109	+**
	110	+** The 'tar' format has evolved over time. Initially the name was stored
	111	+** in a 100 byte null-terminated field 'name'. File path names were
	112	+** limited to 99 bytes.
	113	+**
	114	+** The Posix.1 'ustar' format added a 155 byte field 'prefix', allowing
	115	+** for up to 255 characters to be stored. The full file path is formed by
	116	+** concatenating the field 'prefix', a slash, and the field 'name'. This
	117	+** gives some measure of compatibility with programs that only understand
	118	+** the oldest format.
	119	+**
	120	+** The latest Posix extension is called the 'pax Interchange Format'.
	121	+** It removes all the limitations of the previous two formats by allowing
	122	+** the storage of arbitrary-length attributes in a separate object that looks
	123	+** like a file to programs that do not understand this extension. So the
	124	+** contents of the 'name' and 'prefix' fields should contain values that allow
	125	+** versions of tar that do not understand this extension to still do
	126	+** something useful.
	127	+**
	128	+******************************************************************************/
	129	+
	130	+/*
	131	+** The position we use to split a file path into the 'name' and 'prefix'
	132	+** fields needs to meet the following criteria:
	133	+**
	134	+** - not at the beginning or end of the string
	135	+** - the position must contain a slash
	136	+** - no more than 100 characters follow the slash
	137	+** - no more than 155 characters precede it
	138	+**
	139	+** The routine 'find_split_pos' finds a split position. It will meet the
	140	+** criteria of listed above if such a position exists. If no such
	141	+** position exists it generates one that useful for generating the
	142	+** values used for backward compatibility.
	143	+*/
	144	+static int find_split_pos(
	145	+ const char zName, / file path */
	146	+ int nName /* path length */
	147	+){
	148	+ int i, split = 0;
	149	+ /* only search if the string needs splitting */
	150	+ if(nName > USTAR_NAME_LEN){
	151	+ for(i = 1; i+1 < nName; i++)
	152	+ if(zName[i] == '/'){
	153	+ split = i+1;
	154	+ /* if the split position is within USTAR_NAME_LEN bytes from
	155	+ * the end we can quit */
	156	+ if(nName - split <= USTAR_NAME_LEN) break;
	157	+ }
	158	+ }
	159	+ return split;
	160	+}
	161	+
	162	+
	163	+/*
	164	+** attempt to split the file name path to meet 'ustar' header
	165	+** criteria.
	166	+*/
	167	+static int tar_split_path(
	168	+ const char zName, / path */
	169	+ int nName, /* path length */
	170	+ char pName, / name field */
	171	+ char pPrefix / prefix field */
	172	+){
	173	+ int split = find_split_pos(zName, nName);
	174	+ /* check whether both pieces fit */
	175	+ if(nName - split > USTAR_NAME_LEN \|\| split > USTAR_PREFIX_LEN+1){
	176	+ return 0; /* no */
	177	+ }
	178	+
	179	+ /* extract name */
	180	+ padded_copy(pName, USTAR_NAME_LEN, &zName[split], nName - split);
	181	+
	182	+ /* extract prefix */
	183	+ padded_copy(pPrefix, USTAR_PREFIX_LEN, zName, (split > 0 ? split - 1 : 0));
	184	+
	185	+ return 1; /* success */
	186	+}
	187	+
	188	+
	189	+/*
	190	+** When using an extension header we still need to put something
	191	+** reasonable in the name and prefix fields. This is probably as
	192	+** good as it gets.
	193	+*/
	194	+static void approximate_split_path(
	195	+ const char zName, / path */
	196	+ int nName, /* path length */
	197	+ char pName, / name field */
	198	+ char pPrefix, / prefix field */
	199	+ int bHeader /* is this a 'x' type tar header? */
	200	+){
	201	+ int split;
	202	+
	203	+ /* if this is a Pax Interchange header prepend "PaxHeader/"
	204	+ ** so we can tell files apart from metadata */
	205	+ if( bHeader ){
	206	+ int n;
	207	+ blob_reset(&tball.pax);
	208	+ blob_appendf(&tball.pax, "PaxHeader/%.s", nName, nName, zName);
	209	+ zName = blob_buffer(&tball.pax);
	210	+ nName = blob_size(&tball.pax);
	211	+ }
	212	+
	213	+ /* find the split position */
	214	+ split = find_split_pos(zName, nName);
	215	+
	216	+ /* extract a name, truncate if needed */
	217	+ padded_copy(pName, USTAR_NAME_LEN, &zName[split], nName - split);
	218	+
	219	+ /* extract a prefix field, truncate when needed */
	220	+ padded_copy(pPrefix, USTAR_PREFIX_LEN, zName, (split > 0 ? split-1 : 0));
	221	+}
	222	+
	223	+
	224	+/*
	225	+** add a Pax Interchange header to the scratch buffer
	226	+**
	227	+** format: <length> <key>=<value>\n
	228	+** the tricky part is that each header contains its own
	229	+** size in decimal, counting that length.
	230	+*/
	231	+static void add_pax_header(
	232	+ const char *zField,
	233	+ const char *zValue,
	234	+ int nValue
	235	+){
	236	+ /* calculate length without length field */
	237	+ int blen = strlen(zField) + nValue + 3;
	238	+ /* calculate the length of the length field */
	239	+ int next10 = 1;
	240	+ int n;
	241	+ for(n = blen; n > 0; ){
	242	+ blen++; next10 *= 10;
	243	+ n /= 10;
	244	+ }
	245	+ /* adding the length extended the length field? */
	246	+ if(blen > next10){
	247	+ blen++;
	248	+ }
	249	+ /* build the string */
	250	+ blob_appendf(&tball.pax, "%d %s=%.s\n", blen, zField, nValue, nValue, zValue);
	251	+ /* this _must_ be right */
	252	+ if(blob_size(&tball.pax) != blen){
	253	+ fossil_fatal("internal error: PAX tar header has bad length");
	254	+ }
	255	+}
	256	+
	257	+
	258	+/*
	259	+** set the header type, calculate the checksum and output
	260	+** the header
	261	+*/
	262	+static void cksum_and_write_header(
	263	+ char cType
	264	+){
	265	+ unsigned int cksum = 0;
	266	+ int i;
	267	+ memset(&tball.aHdr[148], ' ', 8);
	268	+ tball.aHdr[156] = cType;
	269	+ for(i=0; i<512; i++) cksum += tball.aHdr[i];
	270	+ sqlite3_snprintf(8, (char*)&tball.aHdr[148], "%07o", cksum);
	271	+ tball.aHdr[155] = 0;
	272	+ gzip_step((char*)tball.aHdr, 512);
	273	+}
	274	+
53	275
54	276	/*
55	277	** Build a header for a file or directory and write that header
56	278	** into the growing tarball.
57	279	*/
		@@ -59,33 +281,49 @@
59	281	const char zName, / Name of the object */
60	282	int nName, /* Number of characters in zName */
61	283	int iMode, /* Mode. 0644 or 0755 */
62	284	unsigned int mTime, /* File modification time */
63	285	int iSize, /* Size of the object in bytes */
64		- int iType /* Type of object. 0==file. 5==directory */
	286	+ char cType /* Type of object. '0'==file. '5'==directory */
65	287	){
66		- unsigned int cksum = 0;
67		- int i;
68		- if( nName>100 ){
69		- memcpy(&tball.aHdr[345], zName, nName-100);
70		- memcpy(tball.aHdr, &zName[nName-100], 100);
71		- memset(&tball.aHdr[245+nName], 0, 267-nName);
72		- }else{
73		- memcpy(tball.aHdr, zName, nName);
74		- memset(&tball.aHdr[nName], 0, 100-nName);
75		- memset(&tball.aHdr[345], 0, 167);
76		- }
	288	+ /* set mode and modification time */
77	289	sqlite3_snprintf(8, (char*)&tball.aHdr[100], "%07o", iMode);
78		- sqlite3_snprintf(12, (char*)&tball.aHdr[124], "%011o", iSize);
79	290	sqlite3_snprintf(12, (char*)&tball.aHdr[136], "%011o", mTime);
80		- memset(&tball.aHdr[148], ' ', 8);
81		- tball.aHdr[156] = iType + '0';
82		- for(i=0; i<512; i++) cksum += tball.aHdr[i];
83		- sqlite3_snprintf(7, (char*)&tball.aHdr[148], "%06o", cksum);
84		- tball.aHdr[154] = 0;
85		- gzip_step((char*)tball.aHdr, 512);
	291	+
	292	+ /* see if we need to output a Pax Interchange Header */
	293	+ if( !is_iso646_name(zName, nName) \|\|
	294	+ !tar_split_path(zName, nName, tball.aHdr, &tball.aHdr[345]) ){
	295	+ int lastPage;
	296	+ /* add a file name for interoperability with older programs */
	297	+ approximate_split_path(zName, nName, tball.aHdr, &tball.aHdr[345], 1);
	298	+
	299	+ /* generate the Pax Interchange path header */
	300	+ blob_reset(&tball.pax);
	301	+ add_pax_header("path", zName, nName);
	302	+
	303	+ /* set the header length, and write the header */
	304	+ sqlite3_snprintf(12, (char*)&tball.aHdr[124], "%011o",
	305	+ blob_size(&tball.pax));
	306	+ cksum_and_write_header('x');
	307	+
	308	+ /* write the Pax Interchange data */
	309	+ gzip_step(blob_buffer(&tball.pax), blob_size(&tball.pax));
	310	+ lastPage = blob_size(&tball.pax) % 512;
	311	+ if( lastPage!=0 ){
	312	+ gzip_step(tball.zSpaces, 512 - lastPage);
	313	+ }
	314	+
	315	+ /* generate an approximate path for the regular header */
	316	+ approximate_split_path(zName, nName, tball.aHdr, &tball.aHdr[345], 0);
	317	+ }
	318	+ /* set the size */
	319	+ sqlite3_snprintf(12, (char*)&tball.aHdr[124], "%011o", iSize);
	320	+
	321	+ /* write the regular header */
	322	+ cksum_and_write_header(cType);
86	323	}
	324	+
87	325
88	326	/*
89	327	** Recursively add an directory entry for the given file if those
90	328	** directories have not previously been seen.
91	329	*/
		@@ -95,18 +333,27 @@
95	333	unsigned int mTime /* Modification time */
96	334	){
97	335	int i;
98	336	for(i=nName-1; i>0 && zName[i]!='/'; i--){}
99	337	if( i<=0 ) return;
100		- if( tball.zPrevDir[i]==0 && memcmp(tball.zPrevDir, zName, i)==0 ) return;
	338	+ if( i < tball.nPrevDirAlloc && tball.zPrevDir[i]==0 &&
	339	+ memcmp(tball.zPrevDir, zName, i)==0 ) return;
101	340	db_multi_exec("INSERT OR IGNORE INTO dir VALUES('%#q')", i, zName);
102	341	if( sqlite3_changes(g.db)==0 ) return;
103	342	tar_add_directory_of(zName, i-1, mTime);
104		- tar_add_header(zName, i, 0755, mTime, 0, 5);
	343	+ tar_add_header(zName, i, 0755, mTime, 0, '5');
	344	+ if( i >= tball.nPrevDirAlloc ){
	345	+ int nsize = tball.nPrevDirAlloc * 2;
	346	+ if(i+1 > nsize)
	347	+ nsize = i+1;
	348	+ tball.zPrevDir = fossil_realloc(tball.zPrevDir, nsize);
	349	+ tball.nPrevDirAlloc = nsize;
	350	+ }
105	351	memcpy(tball.zPrevDir, zName, i);
106	352	tball.zPrevDir[i] = 0;
107	353	}
	354	+
108	355
109	356	/*
110	357	** Add a single file to the growing tarball.
111	358	*/
112	359	static void tar_add_file(
		@@ -117,15 +364,13 @@
117	364	){
118	365	int nName = strlen(zName);
119	366	int n = blob_size(pContent);
120	367	int lastPage;
121	368
122		- if( nName>=250 ){
123		- fossil_fatal("name too long for ustar format: \"%s\"", zName);
124		- }
	369	+ /* length check moved to tar_split_path */
125	370	tar_add_directory_of(zName, nName, mTime);
126		- tar_add_header(zName, nName, isExe ? 0755 : 0644, mTime, n, 0);
	371	+ tar_add_header(zName, nName, isExe ? 0755 : 0644, mTime, n, '0');
127	372	if( n ){
128	373	gzip_step(blob_buffer(pContent), n);
129	374	lastPage = n % 512;
130	375	if( lastPage!=0 ){
131	376	gzip_step(tball.zSpaces, 512 - lastPage);
		@@ -142,10 +387,14 @@
142	387	gzip_step(tball.zSpaces, 512);
143	388	gzip_step(tball.zSpaces, 512);
144	389	gzip_finish(pOut);
145	390	fossil_free(tball.aHdr);
146	391	tball.aHdr = 0;
	392	+ fossil_free(tball.zPrevDir);
	393	+ tball.zPrevDir = NULL;
	394	+ tball.nPrevDirAlloc = 0;
	395	+ blob_reset(&tball.pax);
147	396	}
148	397
149	398
150	399	/*
151	400	** COMMAND: test-tarball
152	401

	--- src/tar.c
	+++ src/tar.c
	@@ -27,31 +27,253 @@
27	*/
28	static struct tarball_t {
29	unsigned char aHdr; / Space for building headers */
30	char zSpaces; / Spaces for padding */
31	char zPrevDir; / Name of directory for previous entry */


32	} tball;








33
34	/*
35	** Begin the process of generating a tarball.
36	**
37	** Initialize the GZIP compressor and the table of directory names.
38	*/
39	static void tar_begin(void){
40	assert( tball.aHdr==0 );
41	tball.aHdr = fossil_malloc(512+512+256);
42	memset(tball.aHdr, 0, 512+512+256);
43	tball.zSpaces = (char*)&tball.aHdr[512];
44	tball.zPrevDir = (char*)&tball.zSpaces[512];





45	memcpy(&tball.aHdr[108], "0000000", 8); /* Owner ID */
46	memcpy(&tball.aHdr[116], "0000000", 8); /* Group ID */
47	memcpy(&tball.aHdr[257], "ustar ", 7); /* Format */


48	gzip_begin();
49	db_multi_exec(
50	"CREATE TEMP TABLE dir(name UNIQUE);"
51	);
52	}













































































































































































































53
54	/*
55	** Build a header for a file or directory and write that header
56	** into the growing tarball.
57	*/
	@@ -59,33 +281,49 @@
59	const char zName, / Name of the object */
60	int nName, /* Number of characters in zName */
61	int iMode, /* Mode. 0644 or 0755 */
62	unsigned int mTime, /* File modification time */
63	int iSize, /* Size of the object in bytes */
64	int iType /* Type of object. 0==file. 5==directory */
65	){
66	unsigned int cksum = 0;
67	int i;
68	if( nName>100 ){
69	memcpy(&tball.aHdr[345], zName, nName-100);
70	memcpy(tball.aHdr, &zName[nName-100], 100);
71	memset(&tball.aHdr[245+nName], 0, 267-nName);
72	}else{
73	memcpy(tball.aHdr, zName, nName);
74	memset(&tball.aHdr[nName], 0, 100-nName);
75	memset(&tball.aHdr[345], 0, 167);
76	}
77	sqlite3_snprintf(8, (char*)&tball.aHdr[100], "%07o", iMode);
78	sqlite3_snprintf(12, (char*)&tball.aHdr[124], "%011o", iSize);
79	sqlite3_snprintf(12, (char*)&tball.aHdr[136], "%011o", mTime);
80	memset(&tball.aHdr[148], ' ', 8);
81	tball.aHdr[156] = iType + '0';
82	for(i=0; i<512; i++) cksum += tball.aHdr[i];
83	sqlite3_snprintf(7, (char*)&tball.aHdr[148], "%06o", cksum);
84	tball.aHdr[154] = 0;
85	gzip_step((char*)tball.aHdr, 512);


























86	}

87
88	/*
89	** Recursively add an directory entry for the given file if those
90	** directories have not previously been seen.
91	*/
	@@ -95,18 +333,27 @@
95	unsigned int mTime /* Modification time */
96	){
97	int i;
98	for(i=nName-1; i>0 && zName[i]!='/'; i--){}
99	if( i<=0 ) return;
100	if( tball.zPrevDir[i]==0 && memcmp(tball.zPrevDir, zName, i)==0 ) return;

101	db_multi_exec("INSERT OR IGNORE INTO dir VALUES('%#q')", i, zName);
102	if( sqlite3_changes(g.db)==0 ) return;
103	tar_add_directory_of(zName, i-1, mTime);
104	tar_add_header(zName, i, 0755, mTime, 0, 5);







105	memcpy(tball.zPrevDir, zName, i);
106	tball.zPrevDir[i] = 0;
107	}

108
109	/*
110	** Add a single file to the growing tarball.
111	*/
112	static void tar_add_file(
	@@ -117,15 +364,13 @@
117	){
118	int nName = strlen(zName);
119	int n = blob_size(pContent);
120	int lastPage;
121
122	if( nName>=250 ){
123	fossil_fatal("name too long for ustar format: \"%s\"", zName);
124	}
125	tar_add_directory_of(zName, nName, mTime);
126	tar_add_header(zName, nName, isExe ? 0755 : 0644, mTime, n, 0);
127	if( n ){
128	gzip_step(blob_buffer(pContent), n);
129	lastPage = n % 512;
130	if( lastPage!=0 ){
131	gzip_step(tball.zSpaces, 512 - lastPage);
	@@ -142,10 +387,14 @@
142	gzip_step(tball.zSpaces, 512);
143	gzip_step(tball.zSpaces, 512);
144	gzip_finish(pOut);
145	fossil_free(tball.aHdr);
146	tball.aHdr = 0;




147	}
148
149
150	/*
151	** COMMAND: test-tarball
152

	--- src/tar.c
	+++ src/tar.c
	@@ -27,31 +27,253 @@
27	*/
28	static struct tarball_t {
29	unsigned char aHdr; / Space for building headers */
30	char zSpaces; / Spaces for padding */
31	char zPrevDir; / Name of directory for previous entry */
32	int nPrevDirAlloc; /* size of zPrevDir */
33	Blob pax; /* PAX data */
34	} tball;
35
36
37	/*
38	** field lengths of 'ustar' name and prefix fields.
39	*/
40	#define USTAR_NAME_LEN 100
41	#define USTAR_PREFIX_LEN 155
42
43
44	/*
45	** Begin the process of generating a tarball.
46	**
47	** Initialize the GZIP compressor and the table of directory names.
48	*/
49	static void tar_begin(void){
50	assert( tball.aHdr==0 );
51	tball.aHdr = fossil_malloc(512+512);
52	memset(tball.aHdr, 0, 512+512);
53	tball.zSpaces = (char*)&tball.aHdr[512];
54	/* zPrevDir init */
55	tball.zPrevDir = NULL;
56	tball.nPrevDirAlloc = 0;
57	/* scratch buffer init */
58	blob_zero(&tball.pax);
59
60	memcpy(&tball.aHdr[108], "0000000", 8); /* Owner ID */
61	memcpy(&tball.aHdr[116], "0000000", 8); /* Group ID */
62	memcpy(&tball.aHdr[257], "ustar\00000", 8); /* POSIX.1 format */
63	memcpy(&tball.aHdr[265], "nobody", 7); /* Owner name */
64	memcpy(&tball.aHdr[297], "nobody", 7); /* Group name */
65	gzip_begin();
66	db_multi_exec(
67	"CREATE TEMP TABLE dir(name UNIQUE);"
68	);
69	}
70
71
72	/*
73	** verify that lla characters in 'zName' are in the
74	** ISO646 (=ASCII) character set.
75	*/
76	static int is_iso646_name(
77	const char zName, / file path */
78	int nName /* path length */
79	){
80	int i;
81	for(i = 0; i < nName; i++){
82	unsigned char c = (unsigned char)zName[i];
83	if( c>0x7e ) return 0;
84	}
85	return 1;
86	}
87
88
89	/*
90	** copy string pSrc into pDst, truncating or padding with 0 if necessary
91	*/
92	static void padded_copy(
93	char *pDest,
94	int nDest,
95	const char *pSrc,
96	int nSrc
97	){
98	if(nSrc >= nDest){
99	memcpy(pDest, pSrc, nDest);
100	}else{
101	memcpy(pDest, pSrc, nSrc);
102	memset(&pDest[nSrc], 0, nDest - nSrc);
103	}
104	}
105
106
107
108	/******************************************************************************
109	**
110	** The 'tar' format has evolved over time. Initially the name was stored
111	** in a 100 byte null-terminated field 'name'. File path names were
112	** limited to 99 bytes.
113	**
114	** The Posix.1 'ustar' format added a 155 byte field 'prefix', allowing
115	** for up to 255 characters to be stored. The full file path is formed by
116	** concatenating the field 'prefix', a slash, and the field 'name'. This
117	** gives some measure of compatibility with programs that only understand
118	** the oldest format.
119	**
120	** The latest Posix extension is called the 'pax Interchange Format'.
121	** It removes all the limitations of the previous two formats by allowing
122	** the storage of arbitrary-length attributes in a separate object that looks
123	** like a file to programs that do not understand this extension. So the
124	** contents of the 'name' and 'prefix' fields should contain values that allow
125	** versions of tar that do not understand this extension to still do
126	** something useful.
127	**
128	******************************************************************************/
129
130	/*
131	** The position we use to split a file path into the 'name' and 'prefix'
132	** fields needs to meet the following criteria:
133	**
134	** - not at the beginning or end of the string
135	** - the position must contain a slash
136	** - no more than 100 characters follow the slash
137	** - no more than 155 characters precede it
138	**
139	** The routine 'find_split_pos' finds a split position. It will meet the
140	** criteria of listed above if such a position exists. If no such
141	** position exists it generates one that useful for generating the
142	** values used for backward compatibility.
143	*/
144	static int find_split_pos(
145	const char zName, / file path */
146	int nName /* path length */
147	){
148	int i, split = 0;
149	/* only search if the string needs splitting */
150	if(nName > USTAR_NAME_LEN){
151	for(i = 1; i+1 < nName; i++)
152	if(zName[i] == '/'){
153	split = i+1;
154	/* if the split position is within USTAR_NAME_LEN bytes from
155	* the end we can quit */
156	if(nName - split <= USTAR_NAME_LEN) break;
157	}
158	}
159	return split;
160	}
161
162
163	/*
164	** attempt to split the file name path to meet 'ustar' header
165	** criteria.
166	*/
167	static int tar_split_path(
168	const char zName, / path */
169	int nName, /* path length */
170	char pName, / name field */
171	char pPrefix / prefix field */
172	){
173	int split = find_split_pos(zName, nName);
174	/* check whether both pieces fit */
175	if(nName - split > USTAR_NAME_LEN \|\| split > USTAR_PREFIX_LEN+1){
176	return 0; /* no */
177	}
178
179	/* extract name */
180	padded_copy(pName, USTAR_NAME_LEN, &zName[split], nName - split);
181
182	/* extract prefix */
183	padded_copy(pPrefix, USTAR_PREFIX_LEN, zName, (split > 0 ? split - 1 : 0));
184
185	return 1; /* success */
186	}
187
188
189	/*
190	** When using an extension header we still need to put something
191	** reasonable in the name and prefix fields. This is probably as
192	** good as it gets.
193	*/
194	static void approximate_split_path(
195	const char zName, / path */
196	int nName, /* path length */
197	char pName, / name field */
198	char pPrefix, / prefix field */
199	int bHeader /* is this a 'x' type tar header? */
200	){
201	int split;
202
203	/* if this is a Pax Interchange header prepend "PaxHeader/"
204	** so we can tell files apart from metadata */
205	if( bHeader ){
206	int n;
207	blob_reset(&tball.pax);
208	blob_appendf(&tball.pax, "PaxHeader/%.s", nName, nName, zName);
209	zName = blob_buffer(&tball.pax);
210	nName = blob_size(&tball.pax);
211	}
212
213	/* find the split position */
214	split = find_split_pos(zName, nName);
215
216	/* extract a name, truncate if needed */
217	padded_copy(pName, USTAR_NAME_LEN, &zName[split], nName - split);
218
219	/* extract a prefix field, truncate when needed */
220	padded_copy(pPrefix, USTAR_PREFIX_LEN, zName, (split > 0 ? split-1 : 0));
221	}
222
223
224	/*
225	** add a Pax Interchange header to the scratch buffer
226	**
227	** format: <length> <key>=<value>\n
228	** the tricky part is that each header contains its own
229	** size in decimal, counting that length.
230	*/
231	static void add_pax_header(
232	const char *zField,
233	const char *zValue,
234	int nValue
235	){
236	/* calculate length without length field */
237	int blen = strlen(zField) + nValue + 3;
238	/* calculate the length of the length field */
239	int next10 = 1;
240	int n;
241	for(n = blen; n > 0; ){
242	blen++; next10 *= 10;
243	n /= 10;
244	}
245	/* adding the length extended the length field? */
246	if(blen > next10){
247	blen++;
248	}
249	/* build the string */
250	blob_appendf(&tball.pax, "%d %s=%.s\n", blen, zField, nValue, nValue, zValue);
251	/* this _must_ be right */
252	if(blob_size(&tball.pax) != blen){
253	fossil_fatal("internal error: PAX tar header has bad length");
254	}
255	}
256
257
258	/*
259	** set the header type, calculate the checksum and output
260	** the header
261	*/
262	static void cksum_and_write_header(
263	char cType
264	){
265	unsigned int cksum = 0;
266	int i;
267	memset(&tball.aHdr[148], ' ', 8);
268	tball.aHdr[156] = cType;
269	for(i=0; i<512; i++) cksum += tball.aHdr[i];
270	sqlite3_snprintf(8, (char*)&tball.aHdr[148], "%07o", cksum);
271	tball.aHdr[155] = 0;
272	gzip_step((char*)tball.aHdr, 512);
273	}
274
275
276	/*
277	** Build a header for a file or directory and write that header
278	** into the growing tarball.
279	*/
	@@ -59,33 +281,49 @@
281	const char zName, / Name of the object */
282	int nName, /* Number of characters in zName */
283	int iMode, /* Mode. 0644 or 0755 */
284	unsigned int mTime, /* File modification time */
285	int iSize, /* Size of the object in bytes */
286	char cType /* Type of object. '0'==file. '5'==directory */
287	){
288	/* set mode and modification time */










289	sqlite3_snprintf(8, (char*)&tball.aHdr[100], "%07o", iMode);

290	sqlite3_snprintf(12, (char*)&tball.aHdr[136], "%011o", mTime);
291
292	/* see if we need to output a Pax Interchange Header */
293	if( !is_iso646_name(zName, nName) \|\|
294	!tar_split_path(zName, nName, tball.aHdr, &tball.aHdr[345]) ){
295	int lastPage;
296	/* add a file name for interoperability with older programs */
297	approximate_split_path(zName, nName, tball.aHdr, &tball.aHdr[345], 1);
298
299	/* generate the Pax Interchange path header */
300	blob_reset(&tball.pax);
301	add_pax_header("path", zName, nName);
302
303	/* set the header length, and write the header */
304	sqlite3_snprintf(12, (char*)&tball.aHdr[124], "%011o",
305	blob_size(&tball.pax));
306	cksum_and_write_header('x');
307
308	/* write the Pax Interchange data */
309	gzip_step(blob_buffer(&tball.pax), blob_size(&tball.pax));
310	lastPage = blob_size(&tball.pax) % 512;
311	if( lastPage!=0 ){
312	gzip_step(tball.zSpaces, 512 - lastPage);
313	}
314
315	/* generate an approximate path for the regular header */
316	approximate_split_path(zName, nName, tball.aHdr, &tball.aHdr[345], 0);
317	}
318	/* set the size */
319	sqlite3_snprintf(12, (char*)&tball.aHdr[124], "%011o", iSize);
320
321	/* write the regular header */
322	cksum_and_write_header(cType);
323	}
324
325
326	/*
327	** Recursively add an directory entry for the given file if those
328	** directories have not previously been seen.
329	*/
	@@ -95,18 +333,27 @@
333	unsigned int mTime /* Modification time */
334	){
335	int i;
336	for(i=nName-1; i>0 && zName[i]!='/'; i--){}
337	if( i<=0 ) return;
338	if( i < tball.nPrevDirAlloc && tball.zPrevDir[i]==0 &&
339	memcmp(tball.zPrevDir, zName, i)==0 ) return;
340	db_multi_exec("INSERT OR IGNORE INTO dir VALUES('%#q')", i, zName);
341	if( sqlite3_changes(g.db)==0 ) return;
342	tar_add_directory_of(zName, i-1, mTime);
343	tar_add_header(zName, i, 0755, mTime, 0, '5');
344	if( i >= tball.nPrevDirAlloc ){
345	int nsize = tball.nPrevDirAlloc * 2;
346	if(i+1 > nsize)
347	nsize = i+1;
348	tball.zPrevDir = fossil_realloc(tball.zPrevDir, nsize);
349	tball.nPrevDirAlloc = nsize;
350	}
351	memcpy(tball.zPrevDir, zName, i);
352	tball.zPrevDir[i] = 0;
353	}
354
355
356	/*
357	** Add a single file to the growing tarball.
358	*/
359	static void tar_add_file(
	@@ -117,15 +364,13 @@
364	){
365	int nName = strlen(zName);
366	int n = blob_size(pContent);
367	int lastPage;
368
369	/* length check moved to tar_split_path */


370	tar_add_directory_of(zName, nName, mTime);
371	tar_add_header(zName, nName, isExe ? 0755 : 0644, mTime, n, '0');
372	if( n ){
373	gzip_step(blob_buffer(pContent), n);
374	lastPage = n % 512;
375	if( lastPage!=0 ){
376	gzip_step(tball.zSpaces, 512 - lastPage);
	@@ -142,10 +387,14 @@
387	gzip_step(tball.zSpaces, 512);
388	gzip_step(tball.zSpaces, 512);
389	gzip_finish(pOut);
390	fossil_free(tball.aHdr);
391	tball.aHdr = 0;
392	fossil_free(tball.zPrevDir);
393	tball.zPrevDir = NULL;
394	tball.nPrevDirAlloc = 0;
395	blob_reset(&tball.pax);
396	}
397
398
399	/*
400	** COMMAND: test-tarball
401

M src/tar.c

+278 -29

		--- src/tar.c
		+++ src/tar.c
		@@ -27,31 +27,253 @@
27	27	*/
28	28	static struct tarball_t {
29	29	unsigned char aHdr; / Space for building headers */
30	30	char zSpaces; / Spaces for padding */
31	31	char zPrevDir; / Name of directory for previous entry */
	32	+ int nPrevDirAlloc; /* size of zPrevDir */
	33	+ Blob pax; /* PAX data */
32	34	} tball;
	35	+
	36	+
	37	+/*
	38	+** field lengths of 'ustar' name and prefix fields.
	39	+*/
	40	+#define USTAR_NAME_LEN 100
	41	+#define USTAR_PREFIX_LEN 155
	42	+
33	43
34	44	/*
35	45	** Begin the process of generating a tarball.
36	46	**
37	47	** Initialize the GZIP compressor and the table of directory names.
38	48	*/
39	49	static void tar_begin(void){
40	50	assert( tball.aHdr==0 );
41		- tball.aHdr = fossil_malloc(512+512+256);
42		- memset(tball.aHdr, 0, 512+512+256);
	51	+ tball.aHdr = fossil_malloc(512+512);
	52	+ memset(tball.aHdr, 0, 512+512);
43	53	tball.zSpaces = (char*)&tball.aHdr[512];
44		- tball.zPrevDir = (char*)&tball.zSpaces[512];
	54	+ /* zPrevDir init */
	55	+ tball.zPrevDir = NULL;
	56	+ tball.nPrevDirAlloc = 0;
	57	+ /* scratch buffer init */
	58	+ blob_zero(&tball.pax);
	59	+
45	60	memcpy(&tball.aHdr[108], "0000000", 8); /* Owner ID */
46	61	memcpy(&tball.aHdr[116], "0000000", 8); /* Group ID */
47		- memcpy(&tball.aHdr[257], "ustar ", 7); /* Format */
	62	+ memcpy(&tball.aHdr[257], "ustar\00000", 8); /* POSIX.1 format */
	63	+ memcpy(&tball.aHdr[265], "nobody", 7); /* Owner name */
	64	+ memcpy(&tball.aHdr[297], "nobody", 7); /* Group name */
48	65	gzip_begin();
49	66	db_multi_exec(
50	67	"CREATE TEMP TABLE dir(name UNIQUE);"
51	68	);
52	69	}
	70	+
	71	+
	72	+/*
	73	+** verify that lla characters in 'zName' are in the
	74	+** ISO646 (=ASCII) character set.
	75	+*/
	76	+static int is_iso646_name(
	77	+ const char zName, / file path */
	78	+ int nName /* path length */
	79	+){
	80	+ int i;
	81	+ for(i = 0; i < nName; i++){
	82	+ unsigned char c = (unsigned char)zName[i];
	83	+ if( c>0x7e ) return 0;
	84	+ }
	85	+ return 1;
	86	+}
	87	+
	88	+
	89	+/*
	90	+** copy string pSrc into pDst, truncating or padding with 0 if necessary
	91	+*/
	92	+static void padded_copy(
	93	+ char *pDest,
	94	+ int nDest,
	95	+ const char *pSrc,
	96	+ int nSrc
	97	+){
	98	+ if(nSrc >= nDest){
	99	+ memcpy(pDest, pSrc, nDest);
	100	+ }else{
	101	+ memcpy(pDest, pSrc, nSrc);
	102	+ memset(&pDest[nSrc], 0, nDest - nSrc);
	103	+ }
	104	+}
	105	+
	106	+
	107	+
	108	+/******************************************************************************
	109	+**
	110	+** The 'tar' format has evolved over time. Initially the name was stored
	111	+** in a 100 byte null-terminated field 'name'. File path names were
	112	+** limited to 99 bytes.
	113	+**
	114	+** The Posix.1 'ustar' format added a 155 byte field 'prefix', allowing
	115	+** for up to 255 characters to be stored. The full file path is formed by
	116	+** concatenating the field 'prefix', a slash, and the field 'name'. This
	117	+** gives some measure of compatibility with programs that only understand
	118	+** the oldest format.
	119	+**
	120	+** The latest Posix extension is called the 'pax Interchange Format'.
	121	+** It removes all the limitations of the previous two formats by allowing
	122	+** the storage of arbitrary-length attributes in a separate object that looks
	123	+** like a file to programs that do not understand this extension. So the
	124	+** contents of the 'name' and 'prefix' fields should contain values that allow
	125	+** versions of tar that do not understand this extension to still do
	126	+** something useful.
	127	+**
	128	+******************************************************************************/
	129	+
	130	+/*
	131	+** The position we use to split a file path into the 'name' and 'prefix'
	132	+** fields needs to meet the following criteria:
	133	+**
	134	+** - not at the beginning or end of the string
	135	+** - the position must contain a slash
	136	+** - no more than 100 characters follow the slash
	137	+** - no more than 155 characters precede it
	138	+**
	139	+** The routine 'find_split_pos' finds a split position. It will meet the
	140	+** criteria of listed above if such a position exists. If no such
	141	+** position exists it generates one that useful for generating the
	142	+** values used for backward compatibility.
	143	+*/
	144	+static int find_split_pos(
	145	+ const char zName, / file path */
	146	+ int nName /* path length */
	147	+){
	148	+ int i, split = 0;
	149	+ /* only search if the string needs splitting */
	150	+ if(nName > USTAR_NAME_LEN){
	151	+ for(i = 1; i+1 < nName; i++)
	152	+ if(zName[i] == '/'){
	153	+ split = i+1;
	154	+ /* if the split position is within USTAR_NAME_LEN bytes from
	155	+ * the end we can quit */
	156	+ if(nName - split <= USTAR_NAME_LEN) break;
	157	+ }
	158	+ }
	159	+ return split;
	160	+}
	161	+
	162	+
	163	+/*
	164	+** attempt to split the file name path to meet 'ustar' header
	165	+** criteria.
	166	+*/
	167	+static int tar_split_path(
	168	+ const char zName, / path */
	169	+ int nName, /* path length */
	170	+ char pName, / name field */
	171	+ char pPrefix / prefix field */
	172	+){
	173	+ int split = find_split_pos(zName, nName);
	174	+ /* check whether both pieces fit */
	175	+ if(nName - split > USTAR_NAME_LEN \|\| split > USTAR_PREFIX_LEN+1){
	176	+ return 0; /* no */
	177	+ }
	178	+
	179	+ /* extract name */
	180	+ padded_copy(pName, USTAR_NAME_LEN, &zName[split], nName - split);
	181	+
	182	+ /* extract prefix */
	183	+ padded_copy(pPrefix, USTAR_PREFIX_LEN, zName, (split > 0 ? split - 1 : 0));
	184	+
	185	+ return 1; /* success */
	186	+}
	187	+
	188	+
	189	+/*
	190	+** When using an extension header we still need to put something
	191	+** reasonable in the name and prefix fields. This is probably as
	192	+** good as it gets.
	193	+*/
	194	+static void approximate_split_path(
	195	+ const char zName, / path */
	196	+ int nName, /* path length */
	197	+ char pName, / name field */
	198	+ char pPrefix, / prefix field */
	199	+ int bHeader /* is this a 'x' type tar header? */
	200	+){
	201	+ int split;
	202	+
	203	+ /* if this is a Pax Interchange header prepend "PaxHeader/"
	204	+ ** so we can tell files apart from metadata */
	205	+ if( bHeader ){
	206	+ int n;
	207	+ blob_reset(&tball.pax);
	208	+ blob_appendf(&tball.pax, "PaxHeader/%.s", nName, nName, zName);
	209	+ zName = blob_buffer(&tball.pax);
	210	+ nName = blob_size(&tball.pax);
	211	+ }
	212	+
	213	+ /* find the split position */
	214	+ split = find_split_pos(zName, nName);
	215	+
	216	+ /* extract a name, truncate if needed */
	217	+ padded_copy(pName, USTAR_NAME_LEN, &zName[split], nName - split);
	218	+
	219	+ /* extract a prefix field, truncate when needed */
	220	+ padded_copy(pPrefix, USTAR_PREFIX_LEN, zName, (split > 0 ? split-1 : 0));
	221	+}
	222	+
	223	+
	224	+/*
	225	+** add a Pax Interchange header to the scratch buffer
	226	+**
	227	+** format: <length> <key>=<value>\n
	228	+** the tricky part is that each header contains its own
	229	+** size in decimal, counting that length.
	230	+*/
	231	+static void add_pax_header(
	232	+ const char *zField,
	233	+ const char *zValue,
	234	+ int nValue
	235	+){
	236	+ /* calculate length without length field */
	237	+ int blen = strlen(zField) + nValue + 3;
	238	+ /* calculate the length of the length field */
	239	+ int next10 = 1;
	240	+ int n;
	241	+ for(n = blen; n > 0; ){
	242	+ blen++; next10 *= 10;
	243	+ n /= 10;
	244	+ }
	245	+ /* adding the length extended the length field? */
	246	+ if(blen > next10){
	247	+ blen++;
	248	+ }
	249	+ /* build the string */
	250	+ blob_appendf(&tball.pax, "%d %s=%.s\n", blen, zField, nValue, nValue, zValue);
	251	+ /* this _must_ be right */
	252	+ if(blob_size(&tball.pax) != blen){
	253	+ fossil_fatal("internal error: PAX tar header has bad length");
	254	+ }
	255	+}
	256	+
	257	+
	258	+/*
	259	+** set the header type, calculate the checksum and output
	260	+** the header
	261	+*/
	262	+static void cksum_and_write_header(
	263	+ char cType
	264	+){
	265	+ unsigned int cksum = 0;
	266	+ int i;
	267	+ memset(&tball.aHdr[148], ' ', 8);
	268	+ tball.aHdr[156] = cType;
	269	+ for(i=0; i<512; i++) cksum += tball.aHdr[i];
	270	+ sqlite3_snprintf(8, (char*)&tball.aHdr[148], "%07o", cksum);
	271	+ tball.aHdr[155] = 0;
	272	+ gzip_step((char*)tball.aHdr, 512);
	273	+}
	274	+
53	275
54	276	/*
55	277	** Build a header for a file or directory and write that header
56	278	** into the growing tarball.
57	279	*/
		@@ -59,33 +281,49 @@
59	281	const char zName, / Name of the object */
60	282	int nName, /* Number of characters in zName */
61	283	int iMode, /* Mode. 0644 or 0755 */
62	284	unsigned int mTime, /* File modification time */
63	285	int iSize, /* Size of the object in bytes */
64		- int iType /* Type of object. 0==file. 5==directory */
	286	+ char cType /* Type of object. '0'==file. '5'==directory */
65	287	){
66		- unsigned int cksum = 0;
67		- int i;
68		- if( nName>100 ){
69		- memcpy(&tball.aHdr[345], zName, nName-100);
70		- memcpy(tball.aHdr, &zName[nName-100], 100);
71		- memset(&tball.aHdr[245+nName], 0, 267-nName);
72		- }else{
73		- memcpy(tball.aHdr, zName, nName);
74		- memset(&tball.aHdr[nName], 0, 100-nName);
75		- memset(&tball.aHdr[345], 0, 167);
76		- }
	288	+ /* set mode and modification time */
77	289	sqlite3_snprintf(8, (char*)&tball.aHdr[100], "%07o", iMode);
78		- sqlite3_snprintf(12, (char*)&tball.aHdr[124], "%011o", iSize);
79	290	sqlite3_snprintf(12, (char*)&tball.aHdr[136], "%011o", mTime);
80		- memset(&tball.aHdr[148], ' ', 8);
81		- tball.aHdr[156] = iType + '0';
82		- for(i=0; i<512; i++) cksum += tball.aHdr[i];
83		- sqlite3_snprintf(7, (char*)&tball.aHdr[148], "%06o", cksum);
84		- tball.aHdr[154] = 0;
85		- gzip_step((char*)tball.aHdr, 512);
	291	+
	292	+ /* see if we need to output a Pax Interchange Header */
	293	+ if( !is_iso646_name(zName, nName) \|\|
	294	+ !tar_split_path(zName, nName, tball.aHdr, &tball.aHdr[345]) ){
	295	+ int lastPage;
	296	+ /* add a file name for interoperability with older programs */
	297	+ approximate_split_path(zName, nName, tball.aHdr, &tball.aHdr[345], 1);
	298	+
	299	+ /* generate the Pax Interchange path header */
	300	+ blob_reset(&tball.pax);
	301	+ add_pax_header("path", zName, nName);
	302	+
	303	+ /* set the header length, and write the header */
	304	+ sqlite3_snprintf(12, (char*)&tball.aHdr[124], "%011o",
	305	+ blob_size(&tball.pax));
	306	+ cksum_and_write_header('x');
	307	+
	308	+ /* write the Pax Interchange data */
	309	+ gzip_step(blob_buffer(&tball.pax), blob_size(&tball.pax));
	310	+ lastPage = blob_size(&tball.pax) % 512;
	311	+ if( lastPage!=0 ){
	312	+ gzip_step(tball.zSpaces, 512 - lastPage);
	313	+ }
	314	+
	315	+ /* generate an approximate path for the regular header */
	316	+ approximate_split_path(zName, nName, tball.aHdr, &tball.aHdr[345], 0);
	317	+ }
	318	+ /* set the size */
	319	+ sqlite3_snprintf(12, (char*)&tball.aHdr[124], "%011o", iSize);
	320	+
	321	+ /* write the regular header */
	322	+ cksum_and_write_header(cType);
86	323	}
	324	+
87	325
88	326	/*
89	327	** Recursively add an directory entry for the given file if those
90	328	** directories have not previously been seen.
91	329	*/
		@@ -95,18 +333,27 @@
95	333	unsigned int mTime /* Modification time */
96	334	){
97	335	int i;
98	336	for(i=nName-1; i>0 && zName[i]!='/'; i--){}
99	337	if( i<=0 ) return;
100		- if( tball.zPrevDir[i]==0 && memcmp(tball.zPrevDir, zName, i)==0 ) return;
	338	+ if( i < tball.nPrevDirAlloc && tball.zPrevDir[i]==0 &&
	339	+ memcmp(tball.zPrevDir, zName, i)==0 ) return;
101	340	db_multi_exec("INSERT OR IGNORE INTO dir VALUES('%#q')", i, zName);
102	341	if( sqlite3_changes(g.db)==0 ) return;
103	342	tar_add_directory_of(zName, i-1, mTime);
104		- tar_add_header(zName, i, 0755, mTime, 0, 5);
	343	+ tar_add_header(zName, i, 0755, mTime, 0, '5');
	344	+ if( i >= tball.nPrevDirAlloc ){
	345	+ int nsize = tball.nPrevDirAlloc * 2;
	346	+ if(i+1 > nsize)
	347	+ nsize = i+1;
	348	+ tball.zPrevDir = fossil_realloc(tball.zPrevDir, nsize);
	349	+ tball.nPrevDirAlloc = nsize;
	350	+ }
105	351	memcpy(tball.zPrevDir, zName, i);
106	352	tball.zPrevDir[i] = 0;
107	353	}
	354	+
108	355
109	356	/*
110	357	** Add a single file to the growing tarball.
111	358	*/
112	359	static void tar_add_file(
		@@ -117,15 +364,13 @@
117	364	){
118	365	int nName = strlen(zName);
119	366	int n = blob_size(pContent);
120	367	int lastPage;
121	368
122		- if( nName>=250 ){
123		- fossil_fatal("name too long for ustar format: \"%s\"", zName);
124		- }
	369	+ /* length check moved to tar_split_path */
125	370	tar_add_directory_of(zName, nName, mTime);
126		- tar_add_header(zName, nName, isExe ? 0755 : 0644, mTime, n, 0);
	371	+ tar_add_header(zName, nName, isExe ? 0755 : 0644, mTime, n, '0');
127	372	if( n ){
128	373	gzip_step(blob_buffer(pContent), n);
129	374	lastPage = n % 512;
130	375	if( lastPage!=0 ){
131	376	gzip_step(tball.zSpaces, 512 - lastPage);
		@@ -142,10 +387,14 @@
142	387	gzip_step(tball.zSpaces, 512);
143	388	gzip_step(tball.zSpaces, 512);
144	389	gzip_finish(pOut);
145	390	fossil_free(tball.aHdr);
146	391	tball.aHdr = 0;
	392	+ fossil_free(tball.zPrevDir);
	393	+ tball.zPrevDir = NULL;
	394	+ tball.nPrevDirAlloc = 0;
	395	+ blob_reset(&tball.pax);
147	396	}
148	397
149	398
150	399	/*
151	400	** COMMAND: test-tarball
152	401

	--- src/tar.c
	+++ src/tar.c
	@@ -27,31 +27,253 @@
27	*/
28	static struct tarball_t {
29	unsigned char aHdr; / Space for building headers */
30	char zSpaces; / Spaces for padding */
31	char zPrevDir; / Name of directory for previous entry */


32	} tball;








33
34	/*
35	** Begin the process of generating a tarball.
36	**
37	** Initialize the GZIP compressor and the table of directory names.
38	*/
39	static void tar_begin(void){
40	assert( tball.aHdr==0 );
41	tball.aHdr = fossil_malloc(512+512+256);
42	memset(tball.aHdr, 0, 512+512+256);
43	tball.zSpaces = (char*)&tball.aHdr[512];
44	tball.zPrevDir = (char*)&tball.zSpaces[512];





45	memcpy(&tball.aHdr[108], "0000000", 8); /* Owner ID */
46	memcpy(&tball.aHdr[116], "0000000", 8); /* Group ID */
47	memcpy(&tball.aHdr[257], "ustar ", 7); /* Format */


48	gzip_begin();
49	db_multi_exec(
50	"CREATE TEMP TABLE dir(name UNIQUE);"
51	);
52	}













































































































































































































53
54	/*
55	** Build a header for a file or directory and write that header
56	** into the growing tarball.
57	*/
	@@ -59,33 +281,49 @@
59	const char zName, / Name of the object */
60	int nName, /* Number of characters in zName */
61	int iMode, /* Mode. 0644 or 0755 */
62	unsigned int mTime, /* File modification time */
63	int iSize, /* Size of the object in bytes */
64	int iType /* Type of object. 0==file. 5==directory */
65	){
66	unsigned int cksum = 0;
67	int i;
68	if( nName>100 ){
69	memcpy(&tball.aHdr[345], zName, nName-100);
70	memcpy(tball.aHdr, &zName[nName-100], 100);
71	memset(&tball.aHdr[245+nName], 0, 267-nName);
72	}else{
73	memcpy(tball.aHdr, zName, nName);
74	memset(&tball.aHdr[nName], 0, 100-nName);
75	memset(&tball.aHdr[345], 0, 167);
76	}
77	sqlite3_snprintf(8, (char*)&tball.aHdr[100], "%07o", iMode);
78	sqlite3_snprintf(12, (char*)&tball.aHdr[124], "%011o", iSize);
79	sqlite3_snprintf(12, (char*)&tball.aHdr[136], "%011o", mTime);
80	memset(&tball.aHdr[148], ' ', 8);
81	tball.aHdr[156] = iType + '0';
82	for(i=0; i<512; i++) cksum += tball.aHdr[i];
83	sqlite3_snprintf(7, (char*)&tball.aHdr[148], "%06o", cksum);
84	tball.aHdr[154] = 0;
85	gzip_step((char*)tball.aHdr, 512);


























86	}

87
88	/*
89	** Recursively add an directory entry for the given file if those
90	** directories have not previously been seen.
91	*/
	@@ -95,18 +333,27 @@
95	unsigned int mTime /* Modification time */
96	){
97	int i;
98	for(i=nName-1; i>0 && zName[i]!='/'; i--){}
99	if( i<=0 ) return;
100	if( tball.zPrevDir[i]==0 && memcmp(tball.zPrevDir, zName, i)==0 ) return;

101	db_multi_exec("INSERT OR IGNORE INTO dir VALUES('%#q')", i, zName);
102	if( sqlite3_changes(g.db)==0 ) return;
103	tar_add_directory_of(zName, i-1, mTime);
104	tar_add_header(zName, i, 0755, mTime, 0, 5);







105	memcpy(tball.zPrevDir, zName, i);
106	tball.zPrevDir[i] = 0;
107	}

108
109	/*
110	** Add a single file to the growing tarball.
111	*/
112	static void tar_add_file(
	@@ -117,15 +364,13 @@
117	){
118	int nName = strlen(zName);
119	int n = blob_size(pContent);
120	int lastPage;
121
122	if( nName>=250 ){
123	fossil_fatal("name too long for ustar format: \"%s\"", zName);
124	}
125	tar_add_directory_of(zName, nName, mTime);
126	tar_add_header(zName, nName, isExe ? 0755 : 0644, mTime, n, 0);
127	if( n ){
128	gzip_step(blob_buffer(pContent), n);
129	lastPage = n % 512;
130	if( lastPage!=0 ){
131	gzip_step(tball.zSpaces, 512 - lastPage);
	@@ -142,10 +387,14 @@
142	gzip_step(tball.zSpaces, 512);
143	gzip_step(tball.zSpaces, 512);
144	gzip_finish(pOut);
145	fossil_free(tball.aHdr);
146	tball.aHdr = 0;




147	}
148
149
150	/*
151	** COMMAND: test-tarball
152

	--- src/tar.c
	+++ src/tar.c
	@@ -27,31 +27,253 @@
27	*/
28	static struct tarball_t {
29	unsigned char aHdr; / Space for building headers */
30	char zSpaces; / Spaces for padding */
31	char zPrevDir; / Name of directory for previous entry */
32	int nPrevDirAlloc; /* size of zPrevDir */
33	Blob pax; /* PAX data */
34	} tball;
35
36
37	/*
38	** field lengths of 'ustar' name and prefix fields.
39	*/
40	#define USTAR_NAME_LEN 100
41	#define USTAR_PREFIX_LEN 155
42
43
44	/*
45	** Begin the process of generating a tarball.
46	**
47	** Initialize the GZIP compressor and the table of directory names.
48	*/
49	static void tar_begin(void){
50	assert( tball.aHdr==0 );
51	tball.aHdr = fossil_malloc(512+512);
52	memset(tball.aHdr, 0, 512+512);
53	tball.zSpaces = (char*)&tball.aHdr[512];
54	/* zPrevDir init */
55	tball.zPrevDir = NULL;
56	tball.nPrevDirAlloc = 0;
57	/* scratch buffer init */
58	blob_zero(&tball.pax);
59
60	memcpy(&tball.aHdr[108], "0000000", 8); /* Owner ID */
61	memcpy(&tball.aHdr[116], "0000000", 8); /* Group ID */
62	memcpy(&tball.aHdr[257], "ustar\00000", 8); /* POSIX.1 format */
63	memcpy(&tball.aHdr[265], "nobody", 7); /* Owner name */
64	memcpy(&tball.aHdr[297], "nobody", 7); /* Group name */
65	gzip_begin();
66	db_multi_exec(
67	"CREATE TEMP TABLE dir(name UNIQUE);"
68	);
69	}
70
71
72	/*
73	** verify that lla characters in 'zName' are in the
74	** ISO646 (=ASCII) character set.
75	*/
76	static int is_iso646_name(
77	const char zName, / file path */
78	int nName /* path length */
79	){
80	int i;
81	for(i = 0; i < nName; i++){
82	unsigned char c = (unsigned char)zName[i];
83	if( c>0x7e ) return 0;
84	}
85	return 1;
86	}
87
88
89	/*
90	** copy string pSrc into pDst, truncating or padding with 0 if necessary
91	*/
92	static void padded_copy(
93	char *pDest,
94	int nDest,
95	const char *pSrc,
96	int nSrc
97	){
98	if(nSrc >= nDest){
99	memcpy(pDest, pSrc, nDest);
100	}else{
101	memcpy(pDest, pSrc, nSrc);
102	memset(&pDest[nSrc], 0, nDest - nSrc);
103	}
104	}
105
106
107
108	/******************************************************************************
109	**
110	** The 'tar' format has evolved over time. Initially the name was stored
111	** in a 100 byte null-terminated field 'name'. File path names were
112	** limited to 99 bytes.
113	**
114	** The Posix.1 'ustar' format added a 155 byte field 'prefix', allowing
115	** for up to 255 characters to be stored. The full file path is formed by
116	** concatenating the field 'prefix', a slash, and the field 'name'. This
117	** gives some measure of compatibility with programs that only understand
118	** the oldest format.
119	**
120	** The latest Posix extension is called the 'pax Interchange Format'.
121	** It removes all the limitations of the previous two formats by allowing
122	** the storage of arbitrary-length attributes in a separate object that looks
123	** like a file to programs that do not understand this extension. So the
124	** contents of the 'name' and 'prefix' fields should contain values that allow
125	** versions of tar that do not understand this extension to still do
126	** something useful.
127	**
128	******************************************************************************/
129
130	/*
131	** The position we use to split a file path into the 'name' and 'prefix'
132	** fields needs to meet the following criteria:
133	**
134	** - not at the beginning or end of the string
135	** - the position must contain a slash
136	** - no more than 100 characters follow the slash
137	** - no more than 155 characters precede it
138	**
139	** The routine 'find_split_pos' finds a split position. It will meet the
140	** criteria of listed above if such a position exists. If no such
141	** position exists it generates one that useful for generating the
142	** values used for backward compatibility.
143	*/
144	static int find_split_pos(
145	const char zName, / file path */
146	int nName /* path length */
147	){
148	int i, split = 0;
149	/* only search if the string needs splitting */
150	if(nName > USTAR_NAME_LEN){
151	for(i = 1; i+1 < nName; i++)
152	if(zName[i] == '/'){
153	split = i+1;
154	/* if the split position is within USTAR_NAME_LEN bytes from
155	* the end we can quit */
156	if(nName - split <= USTAR_NAME_LEN) break;
157	}
158	}
159	return split;
160	}
161
162
163	/*
164	** attempt to split the file name path to meet 'ustar' header
165	** criteria.
166	*/
167	static int tar_split_path(
168	const char zName, / path */
169	int nName, /* path length */
170	char pName, / name field */
171	char pPrefix / prefix field */
172	){
173	int split = find_split_pos(zName, nName);
174	/* check whether both pieces fit */
175	if(nName - split > USTAR_NAME_LEN \|\| split > USTAR_PREFIX_LEN+1){
176	return 0; /* no */
177	}
178
179	/* extract name */
180	padded_copy(pName, USTAR_NAME_LEN, &zName[split], nName - split);
181
182	/* extract prefix */
183	padded_copy(pPrefix, USTAR_PREFIX_LEN, zName, (split > 0 ? split - 1 : 0));
184
185	return 1; /* success */
186	}
187
188
189	/*
190	** When using an extension header we still need to put something
191	** reasonable in the name and prefix fields. This is probably as
192	** good as it gets.
193	*/
194	static void approximate_split_path(
195	const char zName, / path */
196	int nName, /* path length */
197	char pName, / name field */
198	char pPrefix, / prefix field */
199	int bHeader /* is this a 'x' type tar header? */
200	){
201	int split;
202
203	/* if this is a Pax Interchange header prepend "PaxHeader/"
204	** so we can tell files apart from metadata */
205	if( bHeader ){
206	int n;
207	blob_reset(&tball.pax);
208	blob_appendf(&tball.pax, "PaxHeader/%.s", nName, nName, zName);
209	zName = blob_buffer(&tball.pax);
210	nName = blob_size(&tball.pax);
211	}
212
213	/* find the split position */
214	split = find_split_pos(zName, nName);
215
216	/* extract a name, truncate if needed */
217	padded_copy(pName, USTAR_NAME_LEN, &zName[split], nName - split);
218
219	/* extract a prefix field, truncate when needed */
220	padded_copy(pPrefix, USTAR_PREFIX_LEN, zName, (split > 0 ? split-1 : 0));
221	}
222
223
224	/*
225	** add a Pax Interchange header to the scratch buffer
226	**
227	** format: <length> <key>=<value>\n
228	** the tricky part is that each header contains its own
229	** size in decimal, counting that length.
230	*/
231	static void add_pax_header(
232	const char *zField,
233	const char *zValue,
234	int nValue
235	){
236	/* calculate length without length field */
237	int blen = strlen(zField) + nValue + 3;
238	/* calculate the length of the length field */
239	int next10 = 1;
240	int n;
241	for(n = blen; n > 0; ){
242	blen++; next10 *= 10;
243	n /= 10;
244	}
245	/* adding the length extended the length field? */
246	if(blen > next10){
247	blen++;
248	}
249	/* build the string */
250	blob_appendf(&tball.pax, "%d %s=%.s\n", blen, zField, nValue, nValue, zValue);
251	/* this _must_ be right */
252	if(blob_size(&tball.pax) != blen){
253	fossil_fatal("internal error: PAX tar header has bad length");
254	}
255	}
256
257
258	/*
259	** set the header type, calculate the checksum and output
260	** the header
261	*/
262	static void cksum_and_write_header(
263	char cType
264	){
265	unsigned int cksum = 0;
266	int i;
267	memset(&tball.aHdr[148], ' ', 8);
268	tball.aHdr[156] = cType;
269	for(i=0; i<512; i++) cksum += tball.aHdr[i];
270	sqlite3_snprintf(8, (char*)&tball.aHdr[148], "%07o", cksum);
271	tball.aHdr[155] = 0;
272	gzip_step((char*)tball.aHdr, 512);
273	}
274
275
276	/*
277	** Build a header for a file or directory and write that header
278	** into the growing tarball.
279	*/
	@@ -59,33 +281,49 @@
281	const char zName, / Name of the object */
282	int nName, /* Number of characters in zName */
283	int iMode, /* Mode. 0644 or 0755 */
284	unsigned int mTime, /* File modification time */
285	int iSize, /* Size of the object in bytes */
286	char cType /* Type of object. '0'==file. '5'==directory */
287	){
288	/* set mode and modification time */










289	sqlite3_snprintf(8, (char*)&tball.aHdr[100], "%07o", iMode);

290	sqlite3_snprintf(12, (char*)&tball.aHdr[136], "%011o", mTime);
291
292	/* see if we need to output a Pax Interchange Header */
293	if( !is_iso646_name(zName, nName) \|\|
294	!tar_split_path(zName, nName, tball.aHdr, &tball.aHdr[345]) ){
295	int lastPage;
296	/* add a file name for interoperability with older programs */
297	approximate_split_path(zName, nName, tball.aHdr, &tball.aHdr[345], 1);
298
299	/* generate the Pax Interchange path header */
300	blob_reset(&tball.pax);
301	add_pax_header("path", zName, nName);
302
303	/* set the header length, and write the header */
304	sqlite3_snprintf(12, (char*)&tball.aHdr[124], "%011o",
305	blob_size(&tball.pax));
306	cksum_and_write_header('x');
307
308	/* write the Pax Interchange data */
309	gzip_step(blob_buffer(&tball.pax), blob_size(&tball.pax));
310	lastPage = blob_size(&tball.pax) % 512;
311	if( lastPage!=0 ){
312	gzip_step(tball.zSpaces, 512 - lastPage);
313	}
314
315	/* generate an approximate path for the regular header */
316	approximate_split_path(zName, nName, tball.aHdr, &tball.aHdr[345], 0);
317	}
318	/* set the size */
319	sqlite3_snprintf(12, (char*)&tball.aHdr[124], "%011o", iSize);
320
321	/* write the regular header */
322	cksum_and_write_header(cType);
323	}
324
325
326	/*
327	** Recursively add an directory entry for the given file if those
328	** directories have not previously been seen.
329	*/
	@@ -95,18 +333,27 @@
333	unsigned int mTime /* Modification time */
334	){
335	int i;
336	for(i=nName-1; i>0 && zName[i]!='/'; i--){}
337	if( i<=0 ) return;
338	if( i < tball.nPrevDirAlloc && tball.zPrevDir[i]==0 &&
339	memcmp(tball.zPrevDir, zName, i)==0 ) return;
340	db_multi_exec("INSERT OR IGNORE INTO dir VALUES('%#q')", i, zName);
341	if( sqlite3_changes(g.db)==0 ) return;
342	tar_add_directory_of(zName, i-1, mTime);
343	tar_add_header(zName, i, 0755, mTime, 0, '5');
344	if( i >= tball.nPrevDirAlloc ){
345	int nsize = tball.nPrevDirAlloc * 2;
346	if(i+1 > nsize)
347	nsize = i+1;
348	tball.zPrevDir = fossil_realloc(tball.zPrevDir, nsize);
349	tball.nPrevDirAlloc = nsize;
350	}
351	memcpy(tball.zPrevDir, zName, i);
352	tball.zPrevDir[i] = 0;
353	}
354
355
356	/*
357	** Add a single file to the growing tarball.
358	*/
359	static void tar_add_file(
	@@ -117,15 +364,13 @@
364	){
365	int nName = strlen(zName);
366	int n = blob_size(pContent);
367	int lastPage;
368
369	/* length check moved to tar_split_path */


370	tar_add_directory_of(zName, nName, mTime);
371	tar_add_header(zName, nName, isExe ? 0755 : 0644, mTime, n, '0');
372	if( n ){
373	gzip_step(blob_buffer(pContent), n);
374	lastPage = n % 512;
375	if( lastPage!=0 ){
376	gzip_step(tball.zSpaces, 512 - lastPage);
	@@ -142,10 +387,14 @@
387	gzip_step(tball.zSpaces, 512);
388	gzip_step(tball.zSpaces, 512);
389	gzip_finish(pOut);
390	fossil_free(tball.aHdr);
391	tball.aHdr = 0;
392	fossil_free(tball.zPrevDir);
393	tball.zPrevDir = NULL;
394	tball.nPrevDirAlloc = 0;
395	blob_reset(&tball.pax);
396	}
397
398
399	/*
400	** COMMAND: test-tarball
401

M src/timeline.c

+74 -13

		--- src/timeline.c
		+++ src/timeline.c
		@@ -107,11 +107,55 @@
107	107	#define TIMELINE_LEAFONLY 0x0002 /* Show "Leaf", but not "Merge", "Fork" etc */
108	108	#define TIMELINE_BRIEF 0x0004 /* Combine adjacent elements of same object */
109	109	#define TIMELINE_GRAPH 0x0008 /* Compute a graph */
110	110	#define TIMELINE_DISJOINT 0x0010 /* Elements are not contiguous */
111	111	#define TIMELINE_FCHANGES 0x0020 /* Detail file changes */
	112	+#define TIMELINE_BRCOLOR 0x0040 /* Background color by branch name */
	113	+#define TIMELINE_UCOLOR 0x0080 /* Background color by user */
112	114	#endif
	115	+
	116	+/*
	117	+** Hash a string and use the hash to determine a background color.
	118	+*/
	119	+char hash_color(const char z){
	120	+ int i; /* Loop counter */
	121	+ unsigned int h = 0; /* Hash on the branch name */
	122	+ int r, g, b; /* Values for red, green, and blue */
	123	+ int h1, h2, h3, h4; /* Elements of the hash value */
	124	+ int mx, mn; /* Components of HSV */
	125	+ static char zColor[10]; /* The resulting color */
	126	+ static int ix[2] = {0,0}; /* Color chooser parameters */
	127	+
	128	+ if( ix[0]==0 ){
	129	+ if( db_get_boolean("white-foreground", 0) ){
	130	+ ix[0] = 140;
	131	+ ix[1] = 40;
	132	+ }else{
	133	+ ix[0] = 216;
	134	+ ix[1] = 16;
	135	+ }
	136	+ }
	137	+ for(i=0; z[i]; i++ ){
	138	+ h = (h<<11) ^ (h<<1) ^ (h>>3) ^ z[i];
	139	+ }
	140	+ h1 = h % 6; h /= 6;
	141	+ h3 = h % 30; h /= 30;
	142	+ h4 = h % 40; h /= 40;
	143	+ mx = ix[0] - h3;
	144	+ mn = mx - h4 - ix[1];
	145	+ h2 = (h%(mx - mn)) + mn;
	146	+ switch( h1 ){
	147	+ case 0: r = mx; g = h2, b = mn; break;
	148	+ case 1: r = h2; g = mx, b = mn; break;
	149	+ case 2: r = mn; g = mx, b = h2; break;
	150	+ case 3: r = mn; g = h2, b = mx; break;
	151	+ case 4: r = h2; g = mn, b = mx; break;
	152	+ default: r = mx; g = mn, b = h2; break;
	153	+ }
	154	+ sqlite3_snprintf(8, zColor, "#%02x%02x%02x", r,g,b);
	155	+ return zColor;
	156	+}
113	157
114	158	/*
115	159	** Output a timeline in the web format given a query. The query
116	160	** should return these columns:
117	161	**
		@@ -142,10 +186,11 @@
142	186	char zPrevDate[20];
143	187	GraphContext *pGraph = 0;
144	188	int prevWasDivider = 0; /* True if previous output row was <hr> */
145	189	int fchngQueryInit = 0; /* True if fchngQuery is initialized */
146	190	Stmt fchngQuery; /* Query for file changes on check-ins */
	191	+ static Stmt qbranch;
147	192
148	193	zPrevDate[0] = 0;
149	194	mxWikiLen = db_get_int("timeline-max-comment", 0);
150	195	if( db_get_boolean("timeline-block-markup", 0) ){
151	196	wikiFlags = WIKI_INLINE;
		@@ -157,10 +202,14 @@
157	202	/* style is not moved to css, because this is
158	203	** a technical div for the timeline graph
159	204	*/
160	205	@ <div id="canvas" style="position:relative;width:1px;height:1px;"></div>
161	206	}
	207	+ db_static_prepare(&qbranch,
	208	+ "SELECT value FROM tagxref WHERE tagid=%d AND tagtype>0 AND rid=:rid",
	209	+ TAG_BRANCH
	210	+ );
162	211
163	212	@ <table id="timelineTable" class="timelineTable">
164	213	blob_zero(&comment);
165	214	while( db_step(pQuery)==SQLITE_ROW ){
166	215	int rid = db_column_int(pQuery, 0);
		@@ -170,10 +219,11 @@
170	219	const char *zDate = db_column_text(pQuery, 2);
171	220	const char *zType = db_column_text(pQuery, 7);
172	221	const char *zUser = db_column_text(pQuery, 4);
173	222	const char *zTagList = db_column_text(pQuery, 8);
174	223	int tagid = db_column_int(pQuery, 9);
	224	+ const char zBr = 0; / Branch */
175	225	int commentColumn = 3; /* Column containing comment text */
176	226	char zTime[8];
177	227	if( tagid ){
178	228	if( tagid==prevTagid ){
179	229	if( tmFlags & TIMELINE_BRIEF ){
		@@ -208,37 +258,44 @@
208	258	memcpy(zTime, &zDate[11], 5);
209	259	zTime[5] = 0;
210	260	@ <tr>
211	261	@ <td class="timelineTime">%s(zTime)</td>
212	262	@ <td class="timelineGraph">
213		- if( pGraph && zType[0]=='c' ){
	263	+ if( tmFlags & TIMELINE_UCOLOR ) zBgClr = zUser ? hash_color(zUser) : 0;
	264	+ if( zType[0]=='c'
	265	+ && (pGraph \|\| zBgClr==0 \|\| (tmFlags & TIMELINE_BRCOLOR)!=0)
	266	+ ){
	267	+ db_reset(&qbranch);
	268	+ db_bind_int(&qbranch, ":rid", rid);
	269	+ if( db_step(&qbranch)==SQLITE_ROW ){
	270	+ zBr = db_column_text(&qbranch, 0);
	271	+ }else{
	272	+ zBr = "trunk";
	273	+ }
	274	+ if( zBgClr==0 \|\| (tmFlags & TIMELINE_BRCOLOR)!=0 ){
	275	+ if( zBr==0 \|\| strcmp(zBr,"trunk")==0 ){
	276	+ zBgClr = 0;
	277	+ }else{
	278	+ zBgClr = hash_color(zBr);
	279	+ }
	280	+ }
	281	+ }
	282	+ if( zType[0]=='c' && (pGraph \|\| (tmFlags & TIMELINE_BRCOLOR)!=0) ){
214	283	int nParent = 0;
215	284	int aParent[32];
216		- const char *zBr;
217	285	int gidx;
218	286	static Stmt qparent;
219		- static Stmt qbranch;
220	287	db_static_prepare(&qparent,
221	288	"SELECT pid FROM plink"
222	289	" WHERE cid=:rid AND pid NOT IN phantom"
223	290	" ORDER BY isprim DESC /sort/"
224	291	);
225		- db_static_prepare(&qbranch,
226		- "SELECT value FROM tagxref WHERE tagid=%d AND tagtype>0 AND rid=:rid",
227		- TAG_BRANCH
228		- );
229	292	db_bind_int(&qparent, ":rid", rid);
230	293	while( db_step(&qparent)==SQLITE_ROW && nParent<32 ){
231	294	aParent[nParent++] = db_column_int(&qparent, 0);
232	295	}
233	296	db_reset(&qparent);
234		- db_bind_int(&qbranch, ":rid", rid);
235		- if( db_step(&qbranch)==SQLITE_ROW ){
236		- zBr = db_column_text(&qbranch, 0);
237		- }else{
238		- zBr = "trunk";
239		- }
240	297	gidx = graph_add_row(pGraph, rid, nParent, aParent, zBr, zBgClr, isLeaf);
241	298	db_reset(&qbranch);
242	299	@ <div id="m%d(gidx)"></div>
243	300	}
244	301	@</td>
		@@ -768,10 +825,12 @@
768	825	** fc Show details of files changed
769	826	** f=RID Show family (immediate parents and children) of RID
770	827	** from=RID Path from...
771	828	** to=RID ... to this
772	829	** nomerge ... avoid merge links on the path
	830	+** brbg Background color from branch name
	831	+** ubg Background color from user
773	832	**
774	833	** p= and d= can appear individually or together. If either p= or d=
775	834	** appear, then u=, y=, a=, and b= are ignored.
776	835	**
777	836	** If a= and b= appear, only a= is used. If neither appear, the most
		@@ -826,10 +885,12 @@
826	885	tmFlags = TIMELINE_GRAPH;
827	886	}
828	887	if( P("ng")!=0 \|\| zSearch!=0 ){
829	888	tmFlags &= ~TIMELINE_GRAPH;
830	889	}
	890	+ if( P("brbg")!=0 ) tmFlags \|= TIMELINE_BRCOLOR;
	891	+ if( P("ubg")!=0 ) tmFlags \|= TIMELINE_UCOLOR;
831	892
832	893	style_header("Timeline");
833	894	login_anonymous_available();
834	895	timeline_temp_table();
835	896	blob_zero(&sql);
836	897

	--- src/timeline.c
	+++ src/timeline.c
	@@ -107,11 +107,55 @@
107	#define TIMELINE_LEAFONLY 0x0002 /* Show "Leaf", but not "Merge", "Fork" etc */
108	#define TIMELINE_BRIEF 0x0004 /* Combine adjacent elements of same object */
109	#define TIMELINE_GRAPH 0x0008 /* Compute a graph */
110	#define TIMELINE_DISJOINT 0x0010 /* Elements are not contiguous */
111	#define TIMELINE_FCHANGES 0x0020 /* Detail file changes */


112	#endif










































113
114	/*
115	** Output a timeline in the web format given a query. The query
116	** should return these columns:
117	**
	@@ -142,10 +186,11 @@
142	char zPrevDate[20];
143	GraphContext *pGraph = 0;
144	int prevWasDivider = 0; /* True if previous output row was <hr> */
145	int fchngQueryInit = 0; /* True if fchngQuery is initialized */
146	Stmt fchngQuery; /* Query for file changes on check-ins */

147
148	zPrevDate[0] = 0;
149	mxWikiLen = db_get_int("timeline-max-comment", 0);
150	if( db_get_boolean("timeline-block-markup", 0) ){
151	wikiFlags = WIKI_INLINE;
	@@ -157,10 +202,14 @@
157	/* style is not moved to css, because this is
158	** a technical div for the timeline graph
159	*/
160	@ <div id="canvas" style="position:relative;width:1px;height:1px;"></div>
161	}




162
163	@ <table id="timelineTable" class="timelineTable">
164	blob_zero(&comment);
165	while( db_step(pQuery)==SQLITE_ROW ){
166	int rid = db_column_int(pQuery, 0);
	@@ -170,10 +219,11 @@
170	const char *zDate = db_column_text(pQuery, 2);
171	const char *zType = db_column_text(pQuery, 7);
172	const char *zUser = db_column_text(pQuery, 4);
173	const char *zTagList = db_column_text(pQuery, 8);
174	int tagid = db_column_int(pQuery, 9);

175	int commentColumn = 3; /* Column containing comment text */
176	char zTime[8];
177	if( tagid ){
178	if( tagid==prevTagid ){
179	if( tmFlags & TIMELINE_BRIEF ){
	@@ -208,37 +258,44 @@
208	memcpy(zTime, &zDate[11], 5);
209	zTime[5] = 0;
210	@ <tr>
211	@ <td class="timelineTime">%s(zTime)</td>
212	@ <td class="timelineGraph">
213	if( pGraph && zType[0]=='c' ){



















214	int nParent = 0;
215	int aParent[32];
216	const char *zBr;
217	int gidx;
218	static Stmt qparent;
219	static Stmt qbranch;
220	db_static_prepare(&qparent,
221	"SELECT pid FROM plink"
222	" WHERE cid=:rid AND pid NOT IN phantom"
223	" ORDER BY isprim DESC /sort/"
224	);
225	db_static_prepare(&qbranch,
226	"SELECT value FROM tagxref WHERE tagid=%d AND tagtype>0 AND rid=:rid",
227	TAG_BRANCH
228	);
229	db_bind_int(&qparent, ":rid", rid);
230	while( db_step(&qparent)==SQLITE_ROW && nParent<32 ){
231	aParent[nParent++] = db_column_int(&qparent, 0);
232	}
233	db_reset(&qparent);
234	db_bind_int(&qbranch, ":rid", rid);
235	if( db_step(&qbranch)==SQLITE_ROW ){
236	zBr = db_column_text(&qbranch, 0);
237	}else{
238	zBr = "trunk";
239	}
240	gidx = graph_add_row(pGraph, rid, nParent, aParent, zBr, zBgClr, isLeaf);
241	db_reset(&qbranch);
242	@ <div id="m%d(gidx)"></div>
243	}
244	@</td>
	@@ -768,10 +825,12 @@
768	** fc Show details of files changed
769	** f=RID Show family (immediate parents and children) of RID
770	** from=RID Path from...
771	** to=RID ... to this
772	** nomerge ... avoid merge links on the path


773	**
774	** p= and d= can appear individually or together. If either p= or d=
775	** appear, then u=, y=, a=, and b= are ignored.
776	**
777	** If a= and b= appear, only a= is used. If neither appear, the most
	@@ -826,10 +885,12 @@
826	tmFlags = TIMELINE_GRAPH;
827	}
828	if( P("ng")!=0 \|\| zSearch!=0 ){
829	tmFlags &= ~TIMELINE_GRAPH;
830	}


831
832	style_header("Timeline");
833	login_anonymous_available();
834	timeline_temp_table();
835	blob_zero(&sql);
836

	--- src/timeline.c
	+++ src/timeline.c
	@@ -107,11 +107,55 @@
107	#define TIMELINE_LEAFONLY 0x0002 /* Show "Leaf", but not "Merge", "Fork" etc */
108	#define TIMELINE_BRIEF 0x0004 /* Combine adjacent elements of same object */
109	#define TIMELINE_GRAPH 0x0008 /* Compute a graph */
110	#define TIMELINE_DISJOINT 0x0010 /* Elements are not contiguous */
111	#define TIMELINE_FCHANGES 0x0020 /* Detail file changes */
112	#define TIMELINE_BRCOLOR 0x0040 /* Background color by branch name */
113	#define TIMELINE_UCOLOR 0x0080 /* Background color by user */
114	#endif
115
116	/*
117	** Hash a string and use the hash to determine a background color.
118	*/
119	char hash_color(const char z){
120	int i; /* Loop counter */
121	unsigned int h = 0; /* Hash on the branch name */
122	int r, g, b; /* Values for red, green, and blue */
123	int h1, h2, h3, h4; /* Elements of the hash value */
124	int mx, mn; /* Components of HSV */
125	static char zColor[10]; /* The resulting color */
126	static int ix[2] = {0,0}; /* Color chooser parameters */
127
128	if( ix[0]==0 ){
129	if( db_get_boolean("white-foreground", 0) ){
130	ix[0] = 140;
131	ix[1] = 40;
132	}else{
133	ix[0] = 216;
134	ix[1] = 16;
135	}
136	}
137	for(i=0; z[i]; i++ ){
138	h = (h<<11) ^ (h<<1) ^ (h>>3) ^ z[i];
139	}
140	h1 = h % 6; h /= 6;
141	h3 = h % 30; h /= 30;
142	h4 = h % 40; h /= 40;
143	mx = ix[0] - h3;
144	mn = mx - h4 - ix[1];
145	h2 = (h%(mx - mn)) + mn;
146	switch( h1 ){
147	case 0: r = mx; g = h2, b = mn; break;
148	case 1: r = h2; g = mx, b = mn; break;
149	case 2: r = mn; g = mx, b = h2; break;
150	case 3: r = mn; g = h2, b = mx; break;
151	case 4: r = h2; g = mn, b = mx; break;
152	default: r = mx; g = mn, b = h2; break;
153	}
154	sqlite3_snprintf(8, zColor, "#%02x%02x%02x", r,g,b);
155	return zColor;
156	}
157
158	/*
159	** Output a timeline in the web format given a query. The query
160	** should return these columns:
161	**
	@@ -142,10 +186,11 @@
186	char zPrevDate[20];
187	GraphContext *pGraph = 0;
188	int prevWasDivider = 0; /* True if previous output row was <hr> */
189	int fchngQueryInit = 0; /* True if fchngQuery is initialized */
190	Stmt fchngQuery; /* Query for file changes on check-ins */
191	static Stmt qbranch;
192
193	zPrevDate[0] = 0;
194	mxWikiLen = db_get_int("timeline-max-comment", 0);
195	if( db_get_boolean("timeline-block-markup", 0) ){
196	wikiFlags = WIKI_INLINE;
	@@ -157,10 +202,14 @@
202	/* style is not moved to css, because this is
203	** a technical div for the timeline graph
204	*/
205	@ <div id="canvas" style="position:relative;width:1px;height:1px;"></div>
206	}
207	db_static_prepare(&qbranch,
208	"SELECT value FROM tagxref WHERE tagid=%d AND tagtype>0 AND rid=:rid",
209	TAG_BRANCH
210	);
211
212	@ <table id="timelineTable" class="timelineTable">
213	blob_zero(&comment);
214	while( db_step(pQuery)==SQLITE_ROW ){
215	int rid = db_column_int(pQuery, 0);
	@@ -170,10 +219,11 @@
219	const char *zDate = db_column_text(pQuery, 2);
220	const char *zType = db_column_text(pQuery, 7);
221	const char *zUser = db_column_text(pQuery, 4);
222	const char *zTagList = db_column_text(pQuery, 8);
223	int tagid = db_column_int(pQuery, 9);
224	const char zBr = 0; / Branch */
225	int commentColumn = 3; /* Column containing comment text */
226	char zTime[8];
227	if( tagid ){
228	if( tagid==prevTagid ){
229	if( tmFlags & TIMELINE_BRIEF ){
	@@ -208,37 +258,44 @@
258	memcpy(zTime, &zDate[11], 5);
259	zTime[5] = 0;
260	@ <tr>
261	@ <td class="timelineTime">%s(zTime)</td>
262	@ <td class="timelineGraph">
263	if( tmFlags & TIMELINE_UCOLOR ) zBgClr = zUser ? hash_color(zUser) : 0;
264	if( zType[0]=='c'
265	&& (pGraph \|\| zBgClr==0 \|\| (tmFlags & TIMELINE_BRCOLOR)!=0)
266	){
267	db_reset(&qbranch);
268	db_bind_int(&qbranch, ":rid", rid);
269	if( db_step(&qbranch)==SQLITE_ROW ){
270	zBr = db_column_text(&qbranch, 0);
271	}else{
272	zBr = "trunk";
273	}
274	if( zBgClr==0 \|\| (tmFlags & TIMELINE_BRCOLOR)!=0 ){
275	if( zBr==0 \|\| strcmp(zBr,"trunk")==0 ){
276	zBgClr = 0;
277	}else{
278	zBgClr = hash_color(zBr);
279	}
280	}
281	}
282	if( zType[0]=='c' && (pGraph \|\| (tmFlags & TIMELINE_BRCOLOR)!=0) ){
283	int nParent = 0;
284	int aParent[32];

285	int gidx;
286	static Stmt qparent;

287	db_static_prepare(&qparent,
288	"SELECT pid FROM plink"
289	" WHERE cid=:rid AND pid NOT IN phantom"
290	" ORDER BY isprim DESC /sort/"
291	);




292	db_bind_int(&qparent, ":rid", rid);
293	while( db_step(&qparent)==SQLITE_ROW && nParent<32 ){
294	aParent[nParent++] = db_column_int(&qparent, 0);
295	}
296	db_reset(&qparent);






297	gidx = graph_add_row(pGraph, rid, nParent, aParent, zBr, zBgClr, isLeaf);
298	db_reset(&qbranch);
299	@ <div id="m%d(gidx)"></div>
300	}
301	@</td>
	@@ -768,10 +825,12 @@
825	** fc Show details of files changed
826	** f=RID Show family (immediate parents and children) of RID
827	** from=RID Path from...
828	** to=RID ... to this
829	** nomerge ... avoid merge links on the path
830	** brbg Background color from branch name
831	** ubg Background color from user
832	**
833	** p= and d= can appear individually or together. If either p= or d=
834	** appear, then u=, y=, a=, and b= are ignored.
835	**
836	** If a= and b= appear, only a= is used. If neither appear, the most
	@@ -826,10 +885,12 @@
885	tmFlags = TIMELINE_GRAPH;
886	}
887	if( P("ng")!=0 \|\| zSearch!=0 ){
888	tmFlags &= ~TIMELINE_GRAPH;
889	}
890	if( P("brbg")!=0 ) tmFlags \|= TIMELINE_BRCOLOR;
891	if( P("ubg")!=0 ) tmFlags \|= TIMELINE_UCOLOR;
892
893	style_header("Timeline");
894	login_anonymous_available();
895	timeline_temp_table();
896	blob_zero(&sql);
897

M src/winhttp.c

+7 -7

		--- src/winhttp.c
		+++ src/winhttp.c
		@@ -444,11 +444,11 @@
444	444	** is logged in.
445	445	**
446	446	** In the following description of the methods, "Fossil-DSCM" will be
447	447	** used as the default SERVICE-NAME:
448	448	**
449		-** fossil service create ?SERVICE-NAME? ?OPTIONS?
	449	+** fossil winsrv create ?SERVICE-NAME? ?OPTIONS?
450	450	**
451	451	** Creates a service. Available options include:
452	452	**
453	453	** -D\|--display DISPLAY-NAME
454	454	**
		@@ -458,11 +458,11 @@
458	458	**
459	459	** -S\|--start TYPE
460	460	**
461	461	** Sets the start type of the service. TYPE can be "manual",
462	462	** which means you need to start the service yourself with the
463		-** 'fossil service start' command or with the "net start" command
	463	+** 'fossil winsrv start' command or with the "net start" command
464	464	** from the operating system. If TYPE is set to "auto", the service
465	465	** will be started automatically by the system during startup.
466	466	**
467	467	** -U\|--username USERNAME
468	468	**
		@@ -509,27 +509,27 @@
509	509	** Enables automatic login if the --localauth option is present
510	510	** and the "localauth" setting is off and the connection is from
511	511	** localhost.
512	512	**
513	513	**
514		-** fossil service delete ?SERVICE-NAME?
	514	+** fossil winsrv delete ?SERVICE-NAME?
515	515	**
516	516	** Deletes a service. If the service is currently running, it will be
517	517	** stopped first and then deleted.
518	518	**
519	519	**
520		-** fossil service show ?SERVICE-NAME?
	520	+** fossil winsrv show ?SERVICE-NAME?
521	521	**
522	522	** Shows how the service is configured and its current state.
523	523	**
524	524	**
525		-** fossil service start ?SERVICE-NAME?
	525	+** fossil winsrv start ?SERVICE-NAME?
526	526	**
527	527	** Start the service.
528	528	**
529	529	**
530		-** fossil service stop ?SERVICE-NAME?
	530	+** fossil winsrv stop ?SERVICE-NAME?
531	531	**
532	532	** Stop the service.
533	533	**
534	534	**
535	535	** NOTE: This command is available on Windows operating systems only and
		@@ -852,11 +852,11 @@
852	852
853	853	#else /* _WIN32 -- This code is for win32 only */
854	854	#include "winhttp.h"
855	855
856	856	void cmd_win32_service(void){
857		- fossil_fatal("The service command is platform specific "
	857	+ fossil_fatal("The winsrv command is platform specific "
858	858	"and not available on this platform.");
859	859	return;
860	860	}
861	861
862	862	#endif /* _WIN32 -- This code is for win32 only */
863	863

	--- src/winhttp.c
	+++ src/winhttp.c
	@@ -444,11 +444,11 @@
444	** is logged in.
445	**
446	** In the following description of the methods, "Fossil-DSCM" will be
447	** used as the default SERVICE-NAME:
448	**
449	** fossil service create ?SERVICE-NAME? ?OPTIONS?
450	**
451	** Creates a service. Available options include:
452	**
453	** -D\|--display DISPLAY-NAME
454	**
	@@ -458,11 +458,11 @@
458	**
459	** -S\|--start TYPE
460	**
461	** Sets the start type of the service. TYPE can be "manual",
462	** which means you need to start the service yourself with the
463	** 'fossil service start' command or with the "net start" command
464	** from the operating system. If TYPE is set to "auto", the service
465	** will be started automatically by the system during startup.
466	**
467	** -U\|--username USERNAME
468	**
	@@ -509,27 +509,27 @@
509	** Enables automatic login if the --localauth option is present
510	** and the "localauth" setting is off and the connection is from
511	** localhost.
512	**
513	**
514	** fossil service delete ?SERVICE-NAME?
515	**
516	** Deletes a service. If the service is currently running, it will be
517	** stopped first and then deleted.
518	**
519	**
520	** fossil service show ?SERVICE-NAME?
521	**
522	** Shows how the service is configured and its current state.
523	**
524	**
525	** fossil service start ?SERVICE-NAME?
526	**
527	** Start the service.
528	**
529	**
530	** fossil service stop ?SERVICE-NAME?
531	**
532	** Stop the service.
533	**
534	**
535	** NOTE: This command is available on Windows operating systems only and
	@@ -852,11 +852,11 @@
852
853	#else /* _WIN32 -- This code is for win32 only */
854	#include "winhttp.h"
855
856	void cmd_win32_service(void){
857	fossil_fatal("The service command is platform specific "
858	"and not available on this platform.");
859	return;
860	}
861
862	#endif /* _WIN32 -- This code is for win32 only */
863

	--- src/winhttp.c
	+++ src/winhttp.c
	@@ -444,11 +444,11 @@
444	** is logged in.
445	**
446	** In the following description of the methods, "Fossil-DSCM" will be
447	** used as the default SERVICE-NAME:
448	**
449	** fossil winsrv create ?SERVICE-NAME? ?OPTIONS?
450	**
451	** Creates a service. Available options include:
452	**
453	** -D\|--display DISPLAY-NAME
454	**
	@@ -458,11 +458,11 @@
458	**
459	** -S\|--start TYPE
460	**
461	** Sets the start type of the service. TYPE can be "manual",
462	** which means you need to start the service yourself with the
463	** 'fossil winsrv start' command or with the "net start" command
464	** from the operating system. If TYPE is set to "auto", the service
465	** will be started automatically by the system during startup.
466	**
467	** -U\|--username USERNAME
468	**
	@@ -509,27 +509,27 @@
509	** Enables automatic login if the --localauth option is present
510	** and the "localauth" setting is off and the connection is from
511	** localhost.
512	**
513	**
514	** fossil winsrv delete ?SERVICE-NAME?
515	**
516	** Deletes a service. If the service is currently running, it will be
517	** stopped first and then deleted.
518	**
519	**
520	** fossil winsrv show ?SERVICE-NAME?
521	**
522	** Shows how the service is configured and its current state.
523	**
524	**
525	** fossil winsrv start ?SERVICE-NAME?
526	**
527	** Start the service.
528	**
529	**
530	** fossil winsrv stop ?SERVICE-NAME?
531	**
532	** Stop the service.
533	**
534	**
535	** NOTE: This command is available on Windows operating systems only and
	@@ -852,11 +852,11 @@
852
853	#else /* _WIN32 -- This code is for win32 only */
854	#include "winhttp.h"
855
856	void cmd_win32_service(void){
857	fossil_fatal("The winsrv command is platform specific "
858	"and not available on this platform.");
859	return;
860	}
861
862	#endif /* _WIN32 -- This code is for win32 only */
863

M win/Makefile.dmc

+1 -1

		--- win/Makefile.dmc
		+++ win/Makefile.dmc
		@@ -84,11 +84,11 @@
84	84	clean:
85	85	-del $(OBJDIR)\*.obj
86	86	-del .obj _.c .h .map
87	87
88	88	realclean:
89		- -del $(APPNAME) translate$E mkindex$E makeheaders$E version$E
	89	+ -del $(APPNAME) translate$E mkindex$E makeheaders$E mkversion$E
90	90
91	91
92	92	$(OBJDIR)\add$O : add_.c add.h
93	93	$(TCC) -o$@ -c add_.c
94	94
95	95

	--- win/Makefile.dmc
	+++ win/Makefile.dmc
	@@ -84,11 +84,11 @@
84	clean:
85	-del $(OBJDIR)\*.obj
86	-del .obj _.c .h .map
87
88	realclean:
89	-del $(APPNAME) translate$E mkindex$E makeheaders$E version$E
90
91
92	$(OBJDIR)\add$O : add_.c add.h
93	$(TCC) -o$@ -c add_.c
94
95

	--- win/Makefile.dmc
	+++ win/Makefile.dmc
	@@ -84,11 +84,11 @@
84	clean:
85	-del $(OBJDIR)\*.obj
86	-del .obj _.c .h .map
87
88	realclean:
89	-del $(APPNAME) translate$E mkindex$E makeheaders$E mkversion$E
90
91
92	$(OBJDIR)\add$O : add_.c add.h
93	$(TCC) -o$@ -c add_.c
94
95

M win/Makefile.msc

+1 -1

		--- win/Makefile.msc
		+++ win/Makefile.msc
		@@ -180,11 +180,11 @@
180	180	-del $(OX)\*.obj
181	181	-del .obj _.c .h .map
182	182	-del headers linkopts
183	183
184	184	realclean:
185		- -del $(APPNAME) translate$E mkindex$E makeheaders$E version$E
	185	+ -del $(APPNAME) translate$E mkindex$E makeheaders$E mkversion$E
186	186
187	187
188	188	$(OX)\add$O : add_.c add.h
189	189	$(TCC) /Fo$@ -c add_.c
190	190
191	191

	--- win/Makefile.msc
	+++ win/Makefile.msc
	@@ -180,11 +180,11 @@
180	-del $(OX)\*.obj
181	-del .obj _.c .h .map
182	-del headers linkopts
183
184	realclean:
185	-del $(APPNAME) translate$E mkindex$E makeheaders$E version$E
186
187
188	$(OX)\add$O : add_.c add.h
189	$(TCC) /Fo$@ -c add_.c
190
191

	--- win/Makefile.msc
	+++ win/Makefile.msc
	@@ -180,11 +180,11 @@
180	-del $(OX)\*.obj
181	-del .obj _.c .h .map
182	-del headers linkopts
183
184	realclean:
185	-del $(APPNAME) translate$E mkindex$E makeheaders$E mkversion$E
186
187
188	$(OX)\add$O : add_.c add.h
189	$(TCC) /Fo$@ -c add_.c
190
191

Fossil SCM

Keyboard Shortcuts