Add special fast paths to sqlite3FtsUnicodeTolower() and Isalnum() for codepoints...

FossilOrigin-Name: cf7b25d47687635a04f4347d45f135c686b9d758

diff --git a/ext/fts3/fts3_unicode.c b/ext/fts3/fts3_unicode.c
index 2c3e4580e38ff3d0b26e465b88f5dbc97b8d54dd..cbdc1099b3ab2648d54ea40d6934fa2e8a237d22 100644 (file)
--- a/ext/fts3/fts3_unicode.c
+++ b/ext/fts3/fts3_unicode.c
@@ -201,6 +201,7 @@ static int unicodeNext(
       if( !zNew ) return SQLITE_NOMEM;
       zOut = &zNew[zOut - pCsr->zToken];
       pCsr->zToken = zNew;
+      pCsr->nAlloc += 64;
     }
 
     /* Write the folded case of the last character read to the output */

diff --git a/ext/fts3/fts3_unicode2.c b/ext/fts3/fts3_unicode2.c
index 51d60549b7b64bdd1a417cca7dc484ea04d1d141..fb04656d191a752e41f6e6dfbd4c7db64ad210cc 100644 (file)
--- a/ext/fts3/fts3_unicode2.c
+++ b/ext/fts3/fts3_unicode2.c
@@ -121,8 +121,13 @@ int sqlite3FtsUnicodeIsalnum(int c){
     0x38008060, 0x380400F0, 0x3C000001, 0x3FFFF401, 0x40000001,
     0x43FFF401,
   };
+  static const unsigned int aAscii[4] = {
+    0xFFFFFFFF, 0xFC00FFFF, 0xF8000001, 0xF8000001,
+  };
 
-  if( c<(1<<22) ){
+  if( c<128 ){
+    return ( (aAscii[c >> 5] & (1 << (c & 0x001F)))==0 );
+  }else if( c<(1<<22) ){
     unsigned int key = (((unsigned int)c)<<10) | 0x000003FF;
     int iRes;
     int iHi = sizeof(aEntry)/sizeof(aEntry[0]) - 1;
@@ -236,7 +241,9 @@ int sqlite3FtsUnicodeTolower(int c){
   assert( c>=0 );
   assert( sizeof(unsigned short)==2 && sizeof(unsigned char)==1 );
 
-  if( c<65536 ){
+  if( c<128 ){
+    if( c>='A' && c<='Z' ) ret = c + ('a' - 'A');
+  }else if( c<65536 ){
     int iHi = sizeof(aEntry)/sizeof(aEntry[0]) - 1;
     int iLo = 0;
     int iRes = -1;

diff --git a/ext/fts3/unicode/mkunicode.tcl b/ext/fts3/unicode/mkunicode.tcl
index 7e93a531b25246f4aa942c94d542ee7e81f04acf..7ba73482419db759e760d9e31cc13d4bcf419341 100644 (file)
--- a/ext/fts3/unicode/mkunicode.tcl
+++ b/ext/fts3/unicode/mkunicode.tcl
@@ -105,6 +105,27 @@ proc an_print_range_array {lRange} {
   puts "  \};"
 }
 
+proc an_print_ascii_bitmap {lRange} {
+  foreach range $lRange {
+    foreach {iFirst nRange} $range {}
+    for {set i $iFirst} {$i < ($iFirst+$nRange)} {incr i} {
+      if {$i<=127} { set a($i) 1 }
+    }
+  }
+
+  set aAscii [list 0 0 0 0]
+  foreach key [array names a] {
+    set idx [expr $key >> 5]
+    lset aAscii $idx [expr [lindex $aAscii $idx] | (1 << ($key&0x001F))]
+  }
+
+  puts "  static const unsigned int aAscii\[4\] = \{"
+  puts -nonewline "   "
+  foreach v $aAscii { puts -nonewline [format " 0x%08X," $v] }
+  puts ""
+  puts "  \};"
+}
+
 proc print_isalnum {zFunc lRange} {
   puts "/*"
   puts "** Return true if the argument corresponds to a unicode codepoint"
@@ -115,8 +136,11 @@ proc print_isalnum {zFunc lRange} {
   puts "*/"
   puts "int ${zFunc}\(int c)\{"
   an_print_range_array $lRange
+  an_print_ascii_bitmap $lRange
   puts {
-  if( c<(1<<22) ){
+  if( c<128 ){
+    return ( (aAscii[c >> 5] & (1 << (c & 0x001F)))==0 );
+  }else if( c<(1<<22) ){
     unsigned int key = (((unsigned int)c)<<10) | 0x000003FF;
     int iRes;
     int iHi = sizeof(aEntry)/sizeof(aEntry[0]) - 1;
@@ -365,7 +389,9 @@ proc print_tolower {zFunc} {
   assert( c>=0 );
   assert( sizeof(unsigned short)==2 && sizeof(unsigned char)==1 );
 
-  if( c<65536 ){
+  if( c<128 ){
+    if( c>='A' && c<='Z' ) ret = c + ('a' - 'A');
+  }else if( c<65536 ){
     int iHi = sizeof(aEntry)/sizeof(aEntry[0]) - 1;
     int iLo = 0;
     int iRes = -1;

diff --git a/manifest b/manifest
index c55da8cba4e6fe035ad01d457ecd853aeb9eb8d8..575fce7aa7a5dd340f8998bc1a99557ad70bcc0a 100644 (file)
--- a/manifest
+++ b/manifest
@@ -1,5 +1,5 @@
-C Fix\scomments\sin\sgenerated\sfile\sfts3_unicode2.c.
-D 2012-05-25T18:48:48.456
+C Add\sspecial\sfast\spaths\sto\ssqlite3FtsUnicodeTolower()\sand\sIsalnum()\sfor\scodepoints\sin\sthe\sASCII\srange.
+D 2012-05-25T19:50:12.903
 F Makefile.arm-wince-mingw32ce-gcc d6df77f1f48d690bd73162294bbba7f59507c72f
 F Makefile.in 2f37e468503dbe79d35c9f6dffcf3fae1ae9ec20
 F Makefile.linux-gcc 91d710bdc4998cb015f39edf3cb314ec4f4d7e23
@@ -70,15 +70,15 @@ F ext/fts3/fts3_test.c 348f7d08cae05285794e23dc4fe8b8fdf66e264a
 F ext/fts3/fts3_tokenizer.c 3da7254a9881f7e270ab28e2004e0d22b3212bce
 F ext/fts3/fts3_tokenizer.h 66dec98e365854b6cd2d54f1a96bb6d428fc5a68
 F ext/fts3/fts3_tokenizer1.c 5c98225a53705e5ee34824087478cf477bdb7004
-F ext/fts3/fts3_unicode.c 033ee5d10d1a69613890d892829e6d3cf7177e40
-F ext/fts3/fts3_unicode2.c 6989db92aff500ae9795c1b16720ff5a17bfbf0f
+F ext/fts3/fts3_unicode.c f487c6ef0ac2cc35aec3c3636c0a35483da6453c
+F ext/fts3/fts3_unicode2.c 75fa8f249a5c17d324969e90d7066e7021a90874
 F ext/fts3/fts3_write.c cd4af00b3b0512b4d76177a267fcaafab44cbce4
 F ext/fts3/fts3speed.tcl b54caf6a18d38174f1a6e84219950d85e98bb1e9
 F ext/fts3/mkfts3amal.tcl 252ecb7fe6467854f2aa237bf2c390b74e71f100
 F ext/fts3/tool/fts3view.c 6cfc5b67a5f0e09c0d698f9fd012c784bfaa9197
 F ext/fts3/unicode/CaseFolding.txt 8c678ca52ecc95e16bc7afc2dbf6fc9ffa05db8c
 F ext/fts3/unicode/UnicodeData.txt cd07314edb62d49fde34debdaf92fa2aa69011e7
-F ext/fts3/unicode/mkunicode.tcl 3ff244e41222fa5e43c60739c501131a2395b310
+F ext/fts3/unicode/mkunicode.tcl a7214d1705cb57ff56fb828002fa811192a25524
 F ext/icu/README.txt bf8461d8cdc6b8f514c080e4e10dc3b2bbdfefa9
 F ext/icu/icu.c eb9ae1d79046bd7871aa97ee6da51eb770134b5a
 F ext/icu/sqliteicu.h 728867a802baa5a96de7495e9689a8e01715ef37
@@ -1004,7 +1004,7 @@ F tool/tostr.awk e75472c2f98dd76e06b8c9c1367f4ab07e122d06
 F tool/vdbe-compress.tcl d70ea6d8a19e3571d7ab8c9b75cba86d1173ff0f
 F tool/warnings-clang.sh a8a0a3babda96dfb1ff51adda3cbbf3dfb7266c2
 F tool/warnings.sh fbc018d67fd7395f440c28f33ef0f94420226381
-P 0c13570ec78c6887103dc99b81b470829fa28385
-R 9b33e64eb6dd16074aa14c4bbd40c221
+P 3dc567ef4702d9a63d78d11ff705cb7f7359f7a6
+R a4a319ece993f7d02e2811bac8d8bd2b
 U dan
-Z a6323715198787ea389cc785d187a551
+Z e7946b1fde61e6ce4870009a94d8c133

diff --git a/manifest.uuid b/manifest.uuid
index 327ad0f2db0d0e6a5fb804464921902c498a1082..35c2e0876820aa9404df74014b289f36dbb3c52b 100644 (file)
--- a/manifest.uuid
+++ b/manifest.uuid
@@ -1 +1 @@
-3dc567ef4702d9a63d78d11ff705cb7f7359f7a6
\ No newline at end of file
+cf7b25d47687635a04f4347d45f135c686b9d758
\ No newline at end of file