Add the "categories" option to the unicode61 tokenizer in fts5.

FossilOrigin-Name: 80d2b9e635e3100f90cffdcffa5b5038da6fbbfccc9f5777c59a4ae760d4cb62

diff --git a/ext/fts3/unicode/mkunicode.tcl b/ext/fts3/unicode/mkunicode.tcl
index de89099122ca597fcca86407efac44a761625fb8..84b8ddc80b5e863811ef37133febe5efb200db44 100644 (file)
--- a/ext/fts3/unicode/mkunicode.tcl
+++ b/ext/fts3/unicode/mkunicode.tcl
@@ -529,6 +529,263 @@ proc print_fold {zFunc} {
   puts "\}"
 }
 
+proc code {txt} {
+  set txt [string trimright $txt]
+  set txt [string trimleft $txt "\n"]
+  set n [expr {[string length $txt] - [string length [string trim $txt]]}]
+  set ret ""
+  foreach L [split $txt "\n"] {
+    append ret "[string range $L $n end]\n"
+  }
+  return [uplevel "subst -nocommands {$ret}"]
+}
+
+proc intarray {lInt} {
+  set ret ""
+  set n [llength $lInt]
+  for {set i 0} {$i < $n} {incr i 10} {
+    append ret "\n    "
+    foreach int [lrange $lInt $i [expr $i+9]] {
+      append ret [format "%-7s" "$int, "]
+    }
+  }
+  append ret "\n  "
+  set ret
+}
+
+proc categories_switch {Cvar first lSecond} {
+  upvar $Cvar C
+  set ret ""
+  append ret "case '$first':\n"
+  append ret "          switch( zCat\[1\] ){\n"
+  foreach s $lSecond {
+    append ret "            case '$s': aArray\[$C($first$s)\] = 1; break;\n"
+  }
+  append ret "            case '*': \n"
+  foreach s $lSecond {
+    append ret "              aArray\[$C($first$s)\] = 1;\n"
+  }
+  append ret "              break;\n"
+  append ret "            default: return 1;"
+  append ret "          }\n"
+  append ret "          break;\n"
+}
+
+# Argument is a list. Each element of which is itself a list of two elements:
+#
+#   * the codepoint
+#   * the category
+#
+# List elements are sorted in order of codepoint.
+#
+proc print_categories {lMap} {
+  set categories {
+    Cc Cf Cn Cs
+    Ll Lm Lo Lt Lu
+    Mc Me Mn
+    Nd Nl No
+    Pc Pd Pe Pf Pi Po Ps
+    Sc Sk Sm So
+    Zl Zp Zs
+
+    LC Co
+  }
+
+  for {set i 0} {$i < [llength $categories]} {incr i} {
+    set C([lindex $categories $i]) [expr 1+$i]
+  }
+
+  set caseC [categories_switch C C {c f n s o}]
+  set caseL [categories_switch C L {l m o t u C}]
+  set caseM [categories_switch C M {c e n}]
+  set caseN [categories_switch C N {d l o}]
+  set caseP [categories_switch C P {c d e f i o s}]
+  set caseS [categories_switch C S {c k m o}]
+  set caseZ [categories_switch C Z {l p s}]
+
+  set nCat [expr [llength [array names C]] + 1]
+  puts [code {
+    int sqlite3Fts5UnicodeNCat(void) { 
+      return $nCat;
+    }
+
+    int sqlite3Fts5UnicodeCatParse(const char *zCat, u8 *aArray){ 
+      aArray[0] = 1;
+      switch( zCat[0] ){
+        $caseC
+        $caseL
+        $caseM
+        $caseN
+        $caseP
+        $caseS
+        $caseZ
+      }
+      return 0;
+    }
+  }]
+
+  set nRepeat 0
+  set first   [lindex $lMap 0 0]
+  set class   [lindex $lMap 0 1]
+  set prev -1
+
+  set CASE(0) "Lu"
+  set CASE(1) "Ll"
+
+  foreach m $lMap {
+    foreach {codepoint cl} $m {}
+    set codepoint [expr "0x$codepoint"]
+    if {$codepoint>=(1<<20)} continue
+
+    set bNew 0
+    if {$codepoint!=($prev+1)} {
+      set bNew 1
+    } elseif {
+      $cl==$class || ($class=="LC" && $cl==$CASE([expr $nRepeat & 0x01]))
+    } {
+      incr nRepeat
+    } elseif {$class=="Lu" && $nRepeat==1 && $cl=="Ll"} {
+      set class LC
+      incr nRepeat
+    } else {
+      set bNew 1
+    }
+    if {$bNew} {
+      lappend lEntries [list $first $class $nRepeat]
+      set nRepeat 1
+      set first $codepoint
+      set class $cl
+    }
+    set prev $codepoint
+  }
+  if {$nRepeat>0} {
+    lappend lEntries [list $first $class $nRepeat]
+  }
+
+  set aBlock [list 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]
+  set aMap [list]
+  foreach e $lEntries {
+    foreach {cp class nRepeat} $e {}
+    set block [expr ($cp>>16)]
+    if {$block>0 && [lindex $aBlock $block]==0} {
+      for {set i 1} {$i<=$block} {incr i} {
+        if {[lindex $aBlock $i]==0} {
+          lset aBlock $i [llength $aMap]
+        }
+      }
+    }
+    lappend aMap [expr {$cp & 0xFFFF}]
+    lappend aData [expr {($nRepeat << 5) + $C($class)}]
+  }
+  for {set i 1} {$i<[llength $aBlock]} {incr i} {
+    if {[lindex $aBlock $i]==0} {
+      lset aBlock $i [llength $aMap]
+    }
+  }
+
+  set aBlockArray [intarray $aBlock]
+  set aMapArray [intarray $aMap]
+  set aDataArray [intarray $aData]
+  puts [code {
+    static u16 aFts5UnicodeBlock[] = {$aBlockArray};
+    static u16 aFts5UnicodeMap[] = {$aMapArray};
+    static u16 aFts5UnicodeData[] = {$aDataArray};
+
+    int sqlite3Fts5UnicodeCategory(int iCode) { 
+      int iRes = -1;
+      int iHi;
+      int iLo;
+      int ret;
+      u16 iKey;
+
+      if( iCode>=(1<<20) ){
+        return 0;
+      }
+      iLo = aFts5UnicodeBlock[(iCode>>16)];
+      iHi = aFts5UnicodeBlock[1+(iCode>>16)];
+      iKey = (iCode & 0xFFFF);
+      while( iHi>iLo ){
+        int iTest = (iHi + iLo) / 2;
+        assert( iTest>=iLo && iTest<iHi );
+        if( iKey>=aFts5UnicodeMap[iTest] ){
+          iRes = iTest;
+          iLo = iTest+1;
+        }else{
+          iHi = iTest;
+        }
+      }
+
+      if( iRes<0 ) return 0;
+      if( iKey>=(aFts5UnicodeMap[iRes]+(aFts5UnicodeData[iRes]>>5)) ) return 0;
+      ret = aFts5UnicodeData[iRes] & 0x1F;
+      if( ret!=$C(LC) ) return ret;
+      return ((iKey - aFts5UnicodeMap[iRes]) & 0x01) ? $C(Ll) : $C(Lu);
+    }
+
+    void sqlite3Fts5UnicodeAscii(u8 *aArray, u8 *aAscii){
+      int i = 0;
+      int iTbl = 0;
+      while( i<128 ){
+        int bToken = aArray[ aFts5UnicodeData[iTbl] & 0x1F ];
+        int n = (aFts5UnicodeData[iTbl] >> 5) + i;
+        for(; i<128 && i<n; i++){
+          aAscii[i] = bToken;
+        }
+        iTbl++;
+      }
+    }
+  }]
+}
+
+proc print_test_categories {lMap} {
+
+  set lCP [list]
+  foreach e $lMap {
+    foreach {cp cat} $e {}
+    if {[expr 0x$cp] < (1<<20)} {
+      lappend lCP "{0x$cp, \"$cat\"}, "
+    }
+  }
+
+  set aCP "\n"
+  for {set i 0} {$i < [llength $lCP]} {incr i 4} {
+    append aCP "    [join [lrange $lCP $i $i+3]]\n"
+  }
+
+
+  puts [code {
+    static int categories_test (int *piCode){
+      struct Codepoint {
+        int iCode;
+        const char *zCat;
+      } aCP[] = {$aCP};
+      int i;
+      int iCP = 0;
+
+      for(i=0; i<1000000; i++){
+        u8 aArray[40];
+        int cat = 0;
+        int c = 0;
+        memset(aArray, 0, sizeof(aArray));
+        if( aCP[iCP].iCode==i ){
+          sqlite3Fts5UnicodeCatParse(aCP[iCP].zCat, aArray);
+          iCP++;
+        }else{
+          aArray[0] = 1;
+        }
+
+        c = sqlite3Fts5UnicodeCategory(i);
+        if( aArray[c]==0 ){
+          *piCode = i;
+          return 1;
+        }
+      }
+
+      return 0;
+    }
+  }]
+}
+
 proc print_fold_test {zFunc mappings} {
   global tl_lookup_table
 
@@ -605,15 +862,21 @@ proc print_test_main {} {
   puts "#include <stdio.h>"
   puts ""
   puts "int main(int argc, char **argv)\{"
-  puts "  int r1, r2;"
+  puts "  int r1, r2, r3;"
   puts "  int code;"
+  puts "  r3 = 0;"
   puts "  r1 = isalnum_test(&code);"
   puts "  if( r1 ) printf(\"isalnum(): Problem with code %d\\n\",code);"
   puts "  else printf(\"isalnum(): test passed\\n\");"
   puts "  r2 = fold_test(&code);"
   puts "  if( r2 ) printf(\"fold(): Problem with code %d\\n\",code);"
   puts "  else printf(\"fold(): test passed\\n\");"
-  puts "  return (r1 || r2);"
+  if {$::generate_fts5_code} {
+    puts "  r3 = categories_test(&code);"
+    puts "  if( r3 ) printf(\"categories(): Problem with code %d\\n\",code);"
+    puts "  else printf(\"categories(): test passed\\n\");"
+  }
+  puts "  return (r1 || r2 || r3);"
   puts "\}"
 }
 
@@ -651,10 +914,18 @@ for {set i 0} {$i < [llength $argv]-2} {incr i} {
 
 print_fileheader
 
+if {$::generate_test_code} {
+  puts "typedef unsigned short int u16;"
+  puts "typedef unsigned char u8;"
+  puts "#include <string.h>"
+}
+
 # Print the isalnum() function to stdout.
 #
 set lRange [an_load_separator_ranges]
-print_isalnum ${function_prefix}UnicodeIsalnum $lRange
+if {$generate_fts5_code==0} {
+  print_isalnum ${function_prefix}UnicodeIsalnum $lRange
+}
 
 # Leave a gap between the two generated C functions.
 #
@@ -677,12 +948,21 @@ puts ""
 #
 print_fold ${function_prefix}UnicodeFold
 
+if {$generate_fts5_code} {
+  puts ""
+  puts ""
+  print_categories [cc_load_unicodedata_text ${unicodedata.txt}]
+}
+
 # Print the test routines and main() function to stdout, if -test 
 # was specified.
 #
 if {$::generate_test_code} {
-  print_test_isalnum ${function_prefix}UnicodeIsalnum $lRange
+  if {$generate_fts5_code==0} {
+    print_test_isalnum ${function_prefix}UnicodeIsalnum $lRange
+  }
   print_fold_test ${function_prefix}UnicodeFold $mappings
+  print_test_categories [cc_load_unicodedata_text ${unicodedata.txt}]
   print_test_main 
 }
 

diff --git a/ext/fts3/unicode/parseunicode.tcl b/ext/fts3/unicode/parseunicode.tcl
index 0cb2c83a18f76deaf8b75ebe9c5000544b7708e1..966d7bdd3a18d9c629cedeb023822c5ebf7cfd49 100644 (file)
--- a/ext/fts3/unicode/parseunicode.tcl
+++ b/ext/fts3/unicode/parseunicode.tcl
@@ -143,4 +143,40 @@ proc tl_load_casefolding_txt {zName} {
   }
 }
 
+proc cc_load_unicodedata_text {zName} {
+  set fd [open $zName]
+  set lField {
+    code
+    character_name
+    general_category
+    canonical_combining_classes
+    bidirectional_category
+    character_decomposition_mapping
+    decimal_digit_value
+    digit_value
+    numeric_value
+    mirrored
+    unicode_1_name
+    iso10646_comment_field
+    uppercase_mapping
+    lowercase_mapping
+    titlecase_mapping
+  }
+  set lRet [list]
+
+  while { ![eof $fd] } {
+    set line [gets $fd]
+    if {$line == ""} continue
+
+    set fields [split $line ";"]
+    if {[llength $fields] != [llength $lField]} { error "parse error: $line" }
+    foreach $lField $fields {}
+
+    lappend lRet [list $code $general_category]
+  }
+
+  close $fd
+  set lRet
+}
+
 

diff --git a/ext/fts5/fts5Int.h b/ext/fts5/fts5Int.h
index c87583676bb7ce81f47aa942c6404551cd03bfec..d667f3747c77e3438edfb875136309947edafdb4 100644 (file)
--- a/ext/fts5/fts5Int.h
+++ b/ext/fts5/fts5Int.h
@@ -787,6 +787,10 @@ int sqlite3Fts5VocabInit(Fts5Global*, sqlite3*);
 int sqlite3Fts5UnicodeIsalnum(int c);
 int sqlite3Fts5UnicodeIsdiacritic(int c);
 int sqlite3Fts5UnicodeFold(int c, int bRemoveDiacritic);
+
+int sqlite3Fts5UnicodeCatParse(const char*, u8*);
+int sqlite3Fts5UnicodeCategory(int iCode);
+void sqlite3Fts5UnicodeAscii(u8*, u8*);
 /*
 ** End of interface to code in fts5_unicode2.c.
 **************************************************************************/

diff --git a/ext/fts5/fts5_expr.c b/ext/fts5/fts5_expr.c
index c21d69d9ee3a6085e27b3ba540c03d2c101897f7..a09fe7def8cfb347bc7988346d29172b0bd7558f 100644 (file)
--- a/ext/fts5/fts5_expr.c
+++ b/ext/fts5/fts5_expr.c
@@ -2541,14 +2541,19 @@ static void fts5ExprIsAlnum(
   sqlite3_value **apVal           /* Function arguments */
 ){
   int iCode;
+  u8 aArr[32];
   if( nArg!=1 ){
     sqlite3_result_error(pCtx, 
         "wrong number of arguments to function fts5_isalnum", -1
     );
     return;
   }
+  memset(aArr, 0, sizeof(aArr));
+  sqlite3Fts5UnicodeCatParse("L*", aArr);
+  sqlite3Fts5UnicodeCatParse("N*", aArr);
+  sqlite3Fts5UnicodeCatParse("Co", aArr);
   iCode = sqlite3_value_int(apVal[0]);
-  sqlite3_result_int(pCtx, sqlite3Fts5UnicodeIsalnum(iCode));
+  sqlite3_result_int(pCtx, aArr[sqlite3Fts5UnicodeCategory(iCode)]);
 }
 
 static void fts5ExprFold(

diff --git a/ext/fts5/fts5_tokenize.c b/ext/fts5/fts5_tokenize.c
index b72a0c24ab9f4c227c9197443f137d2cf67130cd..af2bc222f2d899687b31cf3590221e7e8d9772e1 100644 (file)
--- a/ext/fts5/fts5_tokenize.c
+++ b/ext/fts5/fts5_tokenize.c
@@ -237,6 +237,8 @@ struct Unicode61Tokenizer {
   int bRemoveDiacritic;           /* True if remove_diacritics=1 is set */
   int nException;
   int *aiException;
+
+  unsigned char aCategory[32];    /* True for token char categories */
 };
 
 static int fts5UnicodeAddExceptions(
@@ -261,7 +263,7 @@ static int fts5UnicodeAddExceptions(
         if( iCode<128 ){
           p->aTokenChar[iCode] = (unsigned char)bTokenChars;
         }else{
-          bToken = sqlite3Fts5UnicodeIsalnum(iCode);
+          bToken = p->aCategory[sqlite3Fts5UnicodeCategory(iCode)];
           assert( (bToken==0 || bToken==1) ); 
           assert( (bTokenChars==0 || bTokenChars==1) );
           if( bToken!=bTokenChars && sqlite3Fts5UnicodeIsdiacritic(iCode)==0 ){
@@ -322,6 +324,21 @@ static void fts5UnicodeDelete(Fts5Tokenizer *pTok){
   return;
 }
 
+static int unicodeSetCategories(Unicode61Tokenizer *p, const char *zCat){
+  const char *z = zCat;
+
+  while( *z ){
+    while( *z==' ' || *z=='\t' ) z++;
+    if( *z && sqlite3Fts5UnicodeCatParse(z, p->aCategory) ){
+      return SQLITE_ERROR;
+    }
+    while( *z!=' ' && *z!='\t' && *z!='\0' ) z++;
+  }
+
+  sqlite3Fts5UnicodeAscii(p->aCategory, p->aTokenChar);
+  return SQLITE_OK;
+}
+
 /*
 ** Create a "unicode61" tokenizer.
 */
@@ -340,15 +357,28 @@ static int fts5UnicodeCreate(
   }else{
     p = (Unicode61Tokenizer*)sqlite3_malloc(sizeof(Unicode61Tokenizer));
     if( p ){
+      const char *zCat = "L* N* Co";
       int i;
       memset(p, 0, sizeof(Unicode61Tokenizer));
-      memcpy(p->aTokenChar, aAsciiTokenChar, sizeof(aAsciiTokenChar));
+
       p->bRemoveDiacritic = 1;
       p->nFold = 64;
       p->aFold = sqlite3_malloc(p->nFold * sizeof(char));
       if( p->aFold==0 ){
         rc = SQLITE_NOMEM;
       }
+
+      /* Search for a "categories" argument */
+      for(i=0; rc==SQLITE_OK && i<nArg; i+=2){
+        if( 0==sqlite3_stricmp(azArg[i], "categories") ){
+          zCat = azArg[i+1];
+        }
+      }
+
+      if( rc==SQLITE_OK ){
+        rc = unicodeSetCategories(p, zCat);
+      }
+
       for(i=0; rc==SQLITE_OK && i<nArg; i+=2){
         const char *zArg = azArg[i+1];
         if( 0==sqlite3_stricmp(azArg[i], "remove_diacritics") ){
@@ -362,10 +392,14 @@ static int fts5UnicodeCreate(
         }else
         if( 0==sqlite3_stricmp(azArg[i], "separators") ){
           rc = fts5UnicodeAddExceptions(p, zArg, 0);
+        }else
+        if( 0==sqlite3_stricmp(azArg[i], "categories") ){
+          /* no-op */
         }else{
           rc = SQLITE_ERROR;
         }
       }
+
     }else{
       rc = SQLITE_NOMEM;
     }
@@ -384,8 +418,10 @@ static int fts5UnicodeCreate(
 ** character (not a separator).
 */
 static int fts5UnicodeIsAlnum(Unicode61Tokenizer *p, int iCode){
-  assert( (sqlite3Fts5UnicodeIsalnum(iCode) & 0xFFFFFFFE)==0 );
-  return sqlite3Fts5UnicodeIsalnum(iCode) ^ fts5UnicodeIsException(p, iCode);
+  return (
+    p->aCategory[sqlite3Fts5UnicodeCategory(iCode)]
+    ^ fts5UnicodeIsException(p, iCode)
+  );
 }
 
 static int fts5UnicodeTokenize(

diff --git a/ext/fts5/fts5_unicode2.c b/ext/fts5/fts5_unicode2.c
index 1ef56f61567cc72934b32113546f4d9f4c2d17d2..e0c5fa07f37bf1890ebe44cb61213885752bec31 100644 (file)
--- a/ext/fts5/fts5_unicode2.c
+++ b/ext/fts5/fts5_unicode2.c
@@ -18,135 +18,6 @@
 
 #include <assert.h>
 
-/*
-** Return true if the argument corresponds to a unicode codepoint
-** classified as either a letter or a number. Otherwise false.
-**
-** The results are undefined if the value passed to this function
-** is less than zero.
-*/
-int sqlite3Fts5UnicodeIsalnum(int c){
-  /* Each unsigned integer in the following array corresponds to a contiguous
-  ** range of unicode codepoints that are not either letters or numbers (i.e.
-  ** codepoints for which this function should return 0).
-  **
-  ** The most significant 22 bits in each 32-bit value contain the first 
-  ** codepoint in the range. The least significant 10 bits are used to store
-  ** the size of the range (always at least 1). In other words, the value 
-  ** ((C<<22) + N) represents a range of N codepoints starting with codepoint 
-  ** C. It is not possible to represent a range larger than 1023 codepoints 
-  ** using this format.
-  */
-  static const unsigned int aEntry[] = {
-    0x00000030, 0x0000E807, 0x00016C06, 0x0001EC2F, 0x0002AC07,
-    0x0002D001, 0x0002D803, 0x0002EC01, 0x0002FC01, 0x00035C01,
-    0x0003DC01, 0x000B0804, 0x000B480E, 0x000B9407, 0x000BB401,
-    0x000BBC81, 0x000DD401, 0x000DF801, 0x000E1002, 0x000E1C01,
-    0x000FD801, 0x00120808, 0x00156806, 0x00162402, 0x00163C01,
-    0x00164437, 0x0017CC02, 0x00180005, 0x00181816, 0x00187802,
-    0x00192C15, 0x0019A804, 0x0019C001, 0x001B5001, 0x001B580F,
-    0x001B9C07, 0x001BF402, 0x001C000E, 0x001C3C01, 0x001C4401,
-    0x001CC01B, 0x001E980B, 0x001FAC09, 0x001FD804, 0x00205804,
-    0x00206C09, 0x00209403, 0x0020A405, 0x0020C00F, 0x00216403,
-    0x00217801, 0x0023901B, 0x00240004, 0x0024E803, 0x0024F812,
-    0x00254407, 0x00258804, 0x0025C001, 0x00260403, 0x0026F001,
-    0x0026F807, 0x00271C02, 0x00272C03, 0x00275C01, 0x00278802,
-    0x0027C802, 0x0027E802, 0x00280403, 0x0028F001, 0x0028F805,
-    0x00291C02, 0x00292C03, 0x00294401, 0x0029C002, 0x0029D401,
-    0x002A0403, 0x002AF001, 0x002AF808, 0x002B1C03, 0x002B2C03,
-    0x002B8802, 0x002BC002, 0x002C0403, 0x002CF001, 0x002CF807,
-    0x002D1C02, 0x002D2C03, 0x002D5802, 0x002D8802, 0x002DC001,
-    0x002E0801, 0x002EF805, 0x002F1803, 0x002F2804, 0x002F5C01,
-    0x002FCC08, 0x00300403, 0x0030F807, 0x00311803, 0x00312804,
-    0x00315402, 0x00318802, 0x0031FC01, 0x00320802, 0x0032F001,
-    0x0032F807, 0x00331803, 0x00332804, 0x00335402, 0x00338802,
-    0x00340802, 0x0034F807, 0x00351803, 0x00352804, 0x00355C01,
-    0x00358802, 0x0035E401, 0x00360802, 0x00372801, 0x00373C06,
-    0x00375801, 0x00376008, 0x0037C803, 0x0038C401, 0x0038D007,
-    0x0038FC01, 0x00391C09, 0x00396802, 0x003AC401, 0x003AD006,
-    0x003AEC02, 0x003B2006, 0x003C041F, 0x003CD00C, 0x003DC417,
-    0x003E340B, 0x003E6424, 0x003EF80F, 0x003F380D, 0x0040AC14,
-    0x00412806, 0x00415804, 0x00417803, 0x00418803, 0x00419C07,
-    0x0041C404, 0x0042080C, 0x00423C01, 0x00426806, 0x0043EC01,
-    0x004D740C, 0x004E400A, 0x00500001, 0x0059B402, 0x005A0001,
-    0x005A6C02, 0x005BAC03, 0x005C4803, 0x005CC805, 0x005D4802,
-    0x005DC802, 0x005ED023, 0x005F6004, 0x005F7401, 0x0060000F,
-    0x0062A401, 0x0064800C, 0x0064C00C, 0x00650001, 0x00651002,
-    0x0066C011, 0x00672002, 0x00677822, 0x00685C05, 0x00687802,
-    0x0069540A, 0x0069801D, 0x0069FC01, 0x006A8007, 0x006AA006,
-    0x006C0005, 0x006CD011, 0x006D6823, 0x006E0003, 0x006E840D,
-    0x006F980E, 0x006FF004, 0x00709014, 0x0070EC05, 0x0071F802,
-    0x00730008, 0x00734019, 0x0073B401, 0x0073C803, 0x00770027,
-    0x0077F004, 0x007EF401, 0x007EFC03, 0x007F3403, 0x007F7403,
-    0x007FB403, 0x007FF402, 0x00800065, 0x0081A806, 0x0081E805,
-    0x00822805, 0x0082801A, 0x00834021, 0x00840002, 0x00840C04,
-    0x00842002, 0x00845001, 0x00845803, 0x00847806, 0x00849401,
-    0x00849C01, 0x0084A401, 0x0084B801, 0x0084E802, 0x00850005,
-    0x00852804, 0x00853C01, 0x00864264, 0x00900027, 0x0091000B,
-    0x0092704E, 0x00940200, 0x009C0475, 0x009E53B9, 0x00AD400A,
-    0x00B39406, 0x00B3BC03, 0x00B3E404, 0x00B3F802, 0x00B5C001,
-    0x00B5FC01, 0x00B7804F, 0x00B8C00C, 0x00BA001A, 0x00BA6C59,
-    0x00BC00D6, 0x00BFC00C, 0x00C00005, 0x00C02019, 0x00C0A807,
-    0x00C0D802, 0x00C0F403, 0x00C26404, 0x00C28001, 0x00C3EC01,
-    0x00C64002, 0x00C6580A, 0x00C70024, 0x00C8001F, 0x00C8A81E,
-    0x00C94001, 0x00C98020, 0x00CA2827, 0x00CB003F, 0x00CC0100,
-    0x01370040, 0x02924037, 0x0293F802, 0x02983403, 0x0299BC10,
-    0x029A7C01, 0x029BC008, 0x029C0017, 0x029C8002, 0x029E2402,
-    0x02A00801, 0x02A01801, 0x02A02C01, 0x02A08C09, 0x02A0D804,
-    0x02A1D004, 0x02A20002, 0x02A2D011, 0x02A33802, 0x02A38012,
-    0x02A3E003, 0x02A4980A, 0x02A51C0D, 0x02A57C01, 0x02A60004,
-    0x02A6CC1B, 0x02A77802, 0x02A8A40E, 0x02A90C01, 0x02A93002,
-    0x02A97004, 0x02A9DC03, 0x02A9EC01, 0x02AAC001, 0x02AAC803,
-    0x02AADC02, 0x02AAF802, 0x02AB0401, 0x02AB7802, 0x02ABAC07,
-    0x02ABD402, 0x02AF8C0B, 0x03600001, 0x036DFC02, 0x036FFC02,
-    0x037FFC01, 0x03EC7801, 0x03ECA401, 0x03EEC810, 0x03F4F802,
-    0x03F7F002, 0x03F8001A, 0x03F88007, 0x03F8C023, 0x03F95013,
-    0x03F9A004, 0x03FBFC01, 0x03FC040F, 0x03FC6807, 0x03FCEC06,
-    0x03FD6C0B, 0x03FF8007, 0x03FFA007, 0x03FFE405, 0x04040003,
-    0x0404DC09, 0x0405E411, 0x0406400C, 0x0407402E, 0x040E7C01,
-    0x040F4001, 0x04215C01, 0x04247C01, 0x0424FC01, 0x04280403,
-    0x04281402, 0x04283004, 0x0428E003, 0x0428FC01, 0x04294009,
-    0x0429FC01, 0x042CE407, 0x04400003, 0x0440E016, 0x04420003,
-    0x0442C012, 0x04440003, 0x04449C0E, 0x04450004, 0x04460003,
-    0x0446CC0E, 0x04471404, 0x045AAC0D, 0x0491C004, 0x05BD442E,
-    0x05BE3C04, 0x074000F6, 0x07440027, 0x0744A4B5, 0x07480046,
-    0x074C0057, 0x075B0401, 0x075B6C01, 0x075BEC01, 0x075C5401,
-    0x075CD401, 0x075D3C01, 0x075DBC01, 0x075E2401, 0x075EA401,
-    0x075F0C01, 0x07BBC002, 0x07C0002C, 0x07C0C064, 0x07C2800F,
-    0x07C2C40E, 0x07C3040F, 0x07C3440F, 0x07C4401F, 0x07C4C03C,
-    0x07C5C02B, 0x07C7981D, 0x07C8402B, 0x07C90009, 0x07C94002,
-    0x07CC0021, 0x07CCC006, 0x07CCDC46, 0x07CE0014, 0x07CE8025,
-    0x07CF1805, 0x07CF8011, 0x07D0003F, 0x07D10001, 0x07D108B6,
-    0x07D3E404, 0x07D4003E, 0x07D50004, 0x07D54018, 0x07D7EC46,
-    0x07D9140B, 0x07DA0046, 0x07DC0074, 0x38000401, 0x38008060,
-    0x380400F0,
-  };
-  static const unsigned int aAscii[4] = {
-    0xFFFFFFFF, 0xFC00FFFF, 0xF8000001, 0xF8000001,
-  };
-
-  if( (unsigned int)c<128 ){
-    return ( (aAscii[c >> 5] & (1 << (c & 0x001F)))==0 );
-  }else if( (unsigned int)c<(1<<22) ){
-    unsigned int key = (((unsigned int)c)<<10) | 0x000003FF;
-    int iRes = 0;
-    int iHi = sizeof(aEntry)/sizeof(aEntry[0]) - 1;
-    int iLo = 0;
-    while( iHi>=iLo ){
-      int iTest = (iHi + iLo) / 2;
-      if( key >= aEntry[iTest] ){
-        iRes = iTest;
-        iLo = iTest+1;
-      }else{
-        iHi = iTest-1;
-      }
-    }
-    assert( aEntry[0]<key );
-    assert( key>=aEntry[iRes] );
-    return (((unsigned int)c) >= ((aEntry[iRes]>>10) + (aEntry[iRes]&0x3FF)));
-  }
-  return 1;
-}
 
 
 /*
@@ -358,3 +229,534 @@ int sqlite3Fts5UnicodeFold(int c, int bRemoveDiacritic){
 
   return ret;
 }
+
+
+int sqlite3Fts5UnicodeNCat(void) { 
+  return 32;
+}
+
+int sqlite3Fts5UnicodeCatParse(const char *zCat, u8 *aArray){ 
+  aArray[0] = 1;
+  switch( zCat[0] ){
+    case 'C':
+          switch( zCat[1] ){
+            case 'c': aArray[1] = 1; break;
+            case 'f': aArray[2] = 1; break;
+            case 'n': aArray[3] = 1; break;
+            case 's': aArray[4] = 1; break;
+            case 'o': aArray[31] = 1; break;
+            case '*': 
+              aArray[1] = 1;
+              aArray[2] = 1;
+              aArray[3] = 1;
+              aArray[4] = 1;
+              aArray[31] = 1;
+              break;
+            default: return 1;          }
+          break;
+
+    case 'L':
+          switch( zCat[1] ){
+            case 'l': aArray[5] = 1; break;
+            case 'm': aArray[6] = 1; break;
+            case 'o': aArray[7] = 1; break;
+            case 't': aArray[8] = 1; break;
+            case 'u': aArray[9] = 1; break;
+            case 'C': aArray[30] = 1; break;
+            case '*': 
+              aArray[5] = 1;
+              aArray[6] = 1;
+              aArray[7] = 1;
+              aArray[8] = 1;
+              aArray[9] = 1;
+              aArray[30] = 1;
+              break;
+            default: return 1;          }
+          break;
+
+    case 'M':
+          switch( zCat[1] ){
+            case 'c': aArray[10] = 1; break;
+            case 'e': aArray[11] = 1; break;
+            case 'n': aArray[12] = 1; break;
+            case '*': 
+              aArray[10] = 1;
+              aArray[11] = 1;
+              aArray[12] = 1;
+              break;
+            default: return 1;          }
+          break;
+
+    case 'N':
+          switch( zCat[1] ){
+            case 'd': aArray[13] = 1; break;
+            case 'l': aArray[14] = 1; break;
+            case 'o': aArray[15] = 1; break;
+            case '*': 
+              aArray[13] = 1;
+              aArray[14] = 1;
+              aArray[15] = 1;
+              break;
+            default: return 1;          }
+          break;
+
+    case 'P':
+          switch( zCat[1] ){
+            case 'c': aArray[16] = 1; break;
+            case 'd': aArray[17] = 1; break;
+            case 'e': aArray[18] = 1; break;
+            case 'f': aArray[19] = 1; break;
+            case 'i': aArray[20] = 1; break;
+            case 'o': aArray[21] = 1; break;
+            case 's': aArray[22] = 1; break;
+            case '*': 
+              aArray[16] = 1;
+              aArray[17] = 1;
+              aArray[18] = 1;
+              aArray[19] = 1;
+              aArray[20] = 1;
+              aArray[21] = 1;
+              aArray[22] = 1;
+              break;
+            default: return 1;          }
+          break;
+
+    case 'S':
+          switch( zCat[1] ){
+            case 'c': aArray[23] = 1; break;
+            case 'k': aArray[24] = 1; break;
+            case 'm': aArray[25] = 1; break;
+            case 'o': aArray[26] = 1; break;
+            case '*': 
+              aArray[23] = 1;
+              aArray[24] = 1;
+              aArray[25] = 1;
+              aArray[26] = 1;
+              break;
+            default: return 1;          }
+          break;
+
+    case 'Z':
+          switch( zCat[1] ){
+            case 'l': aArray[27] = 1; break;
+            case 'p': aArray[28] = 1; break;
+            case 's': aArray[29] = 1; break;
+            case '*': 
+              aArray[27] = 1;
+              aArray[28] = 1;
+              aArray[29] = 1;
+              break;
+            default: return 1;          }
+          break;
+
+  }
+  return 0;
+}
+
+static u16 aFts5UnicodeBlock[] = {
+    0,     1471,  1753,  1760,  1760,  1760,  1760,  1760,  1760,  1760,  
+    1760,  1760,  1760,  1760,  1760,  1763,  1765,  
+  };
+static u16 aFts5UnicodeMap[] = {
+    0,     32,    33,    36,    37,    40,    41,    42,    43,    44,    
+    45,    46,    48,    58,    60,    63,    65,    91,    92,    93,    
+    94,    95,    96,    97,    123,   124,   125,   126,   127,   160,   
+    161,   162,   166,   167,   168,   169,   170,   171,   172,   173,   
+    174,   175,   176,   177,   178,   180,   181,   182,   184,   185,   
+    186,   187,   188,   191,   192,   215,   216,   223,   247,   248,   
+    256,   312,   313,   329,   330,   377,   383,   385,   387,   388,   
+    391,   394,   396,   398,   402,   403,   405,   406,   409,   412,   
+    414,   415,   417,   418,   423,   427,   428,   431,   434,   436,   
+    437,   440,   442,   443,   444,   446,   448,   452,   453,   454,   
+    455,   456,   457,   458,   459,   460,   461,   477,   478,   496,   
+    497,   498,   499,   500,   503,   505,   506,   564,   570,   572,   
+    573,   575,   577,   580,   583,   584,   592,   660,   661,   688,   
+    706,   710,   722,   736,   741,   748,   749,   750,   751,   768,   
+    880,   884,   885,   886,   890,   891,   894,   900,   902,   903,   
+    904,   908,   910,   912,   913,   931,   940,   975,   977,   978,   
+    981,   984,   1008,  1012,  1014,  1015,  1018,  1020,  1021,  1072,  
+    1120,  1154,  1155,  1160,  1162,  1217,  1231,  1232,  1329,  1369,  
+    1370,  1377,  1417,  1418,  1423,  1425,  1470,  1471,  1472,  1473,  
+    1475,  1476,  1478,  1479,  1488,  1520,  1523,  1536,  1542,  1545,  
+    1547,  1548,  1550,  1552,  1563,  1566,  1568,  1600,  1601,  1611,  
+    1632,  1642,  1646,  1648,  1649,  1748,  1749,  1750,  1757,  1758,  
+    1759,  1765,  1767,  1769,  1770,  1774,  1776,  1786,  1789,  1791,  
+    1792,  1807,  1808,  1809,  1810,  1840,  1869,  1958,  1969,  1984,  
+    1994,  2027,  2036,  2038,  2039,  2042,  2048,  2070,  2074,  2075,  
+    2084,  2085,  2088,  2089,  2096,  2112,  2137,  2142,  2208,  2210,  
+    2276,  2304,  2307,  2308,  2362,  2363,  2364,  2365,  2366,  2369,  
+    2377,  2381,  2382,  2384,  2385,  2392,  2402,  2404,  2406,  2416,  
+    2417,  2418,  2425,  2433,  2434,  2437,  2447,  2451,  2474,  2482,  
+    2486,  2492,  2493,  2494,  2497,  2503,  2507,  2509,  2510,  2519,  
+    2524,  2527,  2530,  2534,  2544,  2546,  2548,  2554,  2555,  2561,  
+    2563,  2565,  2575,  2579,  2602,  2610,  2613,  2616,  2620,  2622,  
+    2625,  2631,  2635,  2641,  2649,  2654,  2662,  2672,  2674,  2677,  
+    2689,  2691,  2693,  2703,  2707,  2730,  2738,  2741,  2748,  2749,  
+    2750,  2753,  2759,  2761,  2763,  2765,  2768,  2784,  2786,  2790,  
+    2800,  2801,  2817,  2818,  2821,  2831,  2835,  2858,  2866,  2869,  
+    2876,  2877,  2878,  2879,  2880,  2881,  2887,  2891,  2893,  2902,  
+    2903,  2908,  2911,  2914,  2918,  2928,  2929,  2930,  2946,  2947,  
+    2949,  2958,  2962,  2969,  2972,  2974,  2979,  2984,  2990,  3006,  
+    3008,  3009,  3014,  3018,  3021,  3024,  3031,  3046,  3056,  3059,  
+    3065,  3066,  3073,  3077,  3086,  3090,  3114,  3125,  3133,  3134,  
+    3137,  3142,  3146,  3157,  3160,  3168,  3170,  3174,  3192,  3199,  
+    3202,  3205,  3214,  3218,  3242,  3253,  3260,  3261,  3262,  3263,  
+    3264,  3270,  3271,  3274,  3276,  3285,  3294,  3296,  3298,  3302,  
+    3313,  3330,  3333,  3342,  3346,  3389,  3390,  3393,  3398,  3402,  
+    3405,  3406,  3415,  3424,  3426,  3430,  3440,  3449,  3450,  3458,  
+    3461,  3482,  3507,  3517,  3520,  3530,  3535,  3538,  3542,  3544,  
+    3570,  3572,  3585,  3633,  3634,  3636,  3647,  3648,  3654,  3655,  
+    3663,  3664,  3674,  3713,  3716,  3719,  3722,  3725,  3732,  3737,  
+    3745,  3749,  3751,  3754,  3757,  3761,  3762,  3764,  3771,  3773,  
+    3776,  3782,  3784,  3792,  3804,  3840,  3841,  3844,  3859,  3860,  
+    3861,  3864,  3866,  3872,  3882,  3892,  3893,  3894,  3895,  3896,  
+    3897,  3898,  3899,  3900,  3901,  3902,  3904,  3913,  3953,  3967,  
+    3968,  3973,  3974,  3976,  3981,  3993,  4030,  4038,  4039,  4046,  
+    4048,  4053,  4057,  4096,  4139,  4141,  4145,  4146,  4152,  4153,  
+    4155,  4157,  4159,  4160,  4170,  4176,  4182,  4184,  4186,  4190,  
+    4193,  4194,  4197,  4199,  4206,  4209,  4213,  4226,  4227,  4229,  
+    4231,  4237,  4238,  4239,  4240,  4250,  4253,  4254,  4256,  4295,  
+    4301,  4304,  4347,  4348,  4349,  4682,  4688,  4696,  4698,  4704,  
+    4746,  4752,  4786,  4792,  4800,  4802,  4808,  4824,  4882,  4888,  
+    4957,  4960,  4969,  4992,  5008,  5024,  5120,  5121,  5741,  5743,  
+    5760,  5761,  5787,  5788,  5792,  5867,  5870,  5888,  5902,  5906,  
+    5920,  5938,  5941,  5952,  5970,  5984,  5998,  6002,  6016,  6068,  
+    6070,  6071,  6078,  6086,  6087,  6089,  6100,  6103,  6104,  6107,  
+    6108,  6109,  6112,  6128,  6144,  6150,  6151,  6155,  6158,  6160,  
+    6176,  6211,  6212,  6272,  6313,  6314,  6320,  6400,  6432,  6435,  
+    6439,  6441,  6448,  6450,  6451,  6457,  6464,  6468,  6470,  6480,  
+    6512,  6528,  6576,  6593,  6600,  6608,  6618,  6622,  6656,  6679,  
+    6681,  6686,  6688,  6741,  6742,  6743,  6744,  6752,  6753,  6754,  
+    6755,  6757,  6765,  6771,  6783,  6784,  6800,  6816,  6823,  6824,  
+    6912,  6916,  6917,  6964,  6965,  6966,  6971,  6972,  6973,  6978,  
+    6979,  6981,  6992,  7002,  7009,  7019,  7028,  7040,  7042,  7043,  
+    7073,  7074,  7078,  7080,  7082,  7083,  7084,  7086,  7088,  7098,  
+    7142,  7143,  7144,  7146,  7149,  7150,  7151,  7154,  7164,  7168,  
+    7204,  7212,  7220,  7222,  7227,  7232,  7245,  7248,  7258,  7288,  
+    7294,  7360,  7376,  7379,  7380,  7393,  7394,  7401,  7405,  7406,  
+    7410,  7412,  7413,  7424,  7468,  7531,  7544,  7545,  7579,  7616,  
+    7676,  7680,  7830,  7838,  7936,  7944,  7952,  7960,  7968,  7976,  
+    7984,  7992,  8000,  8008,  8016,  8025,  8027,  8029,  8031,  8033,  
+    8040,  8048,  8064,  8072,  8080,  8088,  8096,  8104,  8112,  8118,  
+    8120,  8124,  8125,  8126,  8127,  8130,  8134,  8136,  8140,  8141,  
+    8144,  8150,  8152,  8157,  8160,  8168,  8173,  8178,  8182,  8184,  
+    8188,  8189,  8192,  8203,  8208,  8214,  8216,  8217,  8218,  8219,  
+    8221,  8222,  8223,  8224,  8232,  8233,  8234,  8239,  8240,  8249,  
+    8250,  8251,  8255,  8257,  8260,  8261,  8262,  8263,  8274,  8275,  
+    8276,  8277,  8287,  8288,  8298,  8304,  8305,  8308,  8314,  8317,  
+    8318,  8319,  8320,  8330,  8333,  8334,  8336,  8352,  8400,  8413,  
+    8417,  8418,  8421,  8448,  8450,  8451,  8455,  8456,  8458,  8459,  
+    8462,  8464,  8467,  8468,  8469,  8470,  8472,  8473,  8478,  8484,  
+    8485,  8486,  8487,  8488,  8489,  8490,  8494,  8495,  8496,  8500,  
+    8501,  8505,  8506,  8508,  8510,  8512,  8517,  8519,  8522,  8523,  
+    8524,  8526,  8527,  8528,  8544,  8579,  8581,  8585,  8592,  8597,  
+    8602,  8604,  8608,  8609,  8611,  8612,  8614,  8615,  8622,  8623,  
+    8654,  8656,  8658,  8659,  8660,  8661,  8692,  8960,  8968,  8972,  
+    8992,  8994,  9001,  9002,  9003,  9084,  9085,  9115,  9140,  9180,  
+    9186,  9216,  9280,  9312,  9372,  9450,  9472,  9655,  9656,  9665,  
+    9666,  9720,  9728,  9839,  9840,  9985,  10088, 10089, 10090, 10091, 
+    10092, 10093, 10094, 10095, 10096, 10097, 10098, 10099, 10100, 10101, 
+    10102, 10132, 10176, 10181, 10182, 10183, 10214, 10215, 10216, 10217, 
+    10218, 10219, 10220, 10221, 10222, 10223, 10224, 10240, 10496, 10627, 
+    10628, 10629, 10630, 10631, 10632, 10633, 10634, 10635, 10636, 10637, 
+    10638, 10639, 10640, 10641, 10642, 10643, 10644, 10645, 10646, 10647, 
+    10648, 10649, 10712, 10713, 10714, 10715, 10716, 10748, 10749, 10750, 
+    11008, 11056, 11077, 11079, 11088, 11264, 11312, 11360, 11363, 11365, 
+    11367, 11374, 11377, 11378, 11380, 11381, 11383, 11388, 11390, 11393, 
+    11394, 11492, 11493, 11499, 11503, 11506, 11513, 11517, 11518, 11520, 
+    11559, 11565, 11568, 11631, 11632, 11647, 11648, 11680, 11688, 11696, 
+    11704, 11712, 11720, 11728, 11736, 11744, 11776, 11778, 11779, 11780, 
+    11781, 11782, 11785, 11786, 11787, 11788, 11789, 11790, 11799, 11800, 
+    11802, 11803, 11804, 11805, 11806, 11808, 11809, 11810, 11811, 11812, 
+    11813, 11814, 11815, 11816, 11817, 11818, 11823, 11824, 11834, 11904, 
+    11931, 12032, 12272, 12288, 12289, 12292, 12293, 12294, 12295, 12296, 
+    12297, 12298, 12299, 12300, 12301, 12302, 12303, 12304, 12305, 12306, 
+    12308, 12309, 12310, 12311, 12312, 12313, 12314, 12315, 12316, 12317, 
+    12318, 12320, 12321, 12330, 12334, 12336, 12337, 12342, 12344, 12347, 
+    12348, 12349, 12350, 12353, 12441, 12443, 12445, 12447, 12448, 12449, 
+    12539, 12540, 12543, 12549, 12593, 12688, 12690, 12694, 12704, 12736, 
+    12784, 12800, 12832, 12842, 12872, 12880, 12881, 12896, 12928, 12938, 
+    12977, 12992, 13056, 13312, 19893, 19904, 19968, 40908, 40960, 40981, 
+    40982, 42128, 42192, 42232, 42238, 42240, 42508, 42509, 42512, 42528, 
+    42538, 42560, 42606, 42607, 42608, 42611, 42612, 42622, 42623, 42624, 
+    42655, 42656, 42726, 42736, 42738, 42752, 42775, 42784, 42786, 42800, 
+    42802, 42864, 42865, 42873, 42878, 42888, 42889, 42891, 42896, 42912, 
+    43000, 43002, 43003, 43010, 43011, 43014, 43015, 43019, 43020, 43043, 
+    43045, 43047, 43048, 43056, 43062, 43064, 43065, 43072, 43124, 43136, 
+    43138, 43188, 43204, 43214, 43216, 43232, 43250, 43256, 43259, 43264, 
+    43274, 43302, 43310, 43312, 43335, 43346, 43359, 43360, 43392, 43395, 
+    43396, 43443, 43444, 43446, 43450, 43452, 43453, 43457, 43471, 43472, 
+    43486, 43520, 43561, 43567, 43569, 43571, 43573, 43584, 43587, 43588, 
+    43596, 43597, 43600, 43612, 43616, 43632, 43633, 43639, 43642, 43643, 
+    43648, 43696, 43697, 43698, 43701, 43703, 43705, 43710, 43712, 43713, 
+    43714, 43739, 43741, 43742, 43744, 43755, 43756, 43758, 43760, 43762, 
+    43763, 43765, 43766, 43777, 43785, 43793, 43808, 43816, 43968, 44003, 
+    44005, 44006, 44008, 44009, 44011, 44012, 44013, 44016, 44032, 55203, 
+    55216, 55243, 55296, 56191, 56319, 57343, 57344, 63743, 63744, 64112, 
+    64256, 64275, 64285, 64286, 64287, 64297, 64298, 64312, 64318, 64320, 
+    64323, 64326, 64434, 64467, 64830, 64831, 64848, 64914, 65008, 65020, 
+    65021, 65024, 65040, 65047, 65048, 65049, 65056, 65072, 65073, 65075, 
+    65077, 65078, 65079, 65080, 65081, 65082, 65083, 65084, 65085, 65086, 
+    65087, 65088, 65089, 65090, 65091, 65092, 65093, 65095, 65096, 65097, 
+    65101, 65104, 65108, 65112, 65113, 65114, 65115, 65116, 65117, 65118, 
+    65119, 65122, 65123, 65124, 65128, 65129, 65130, 65136, 65142, 65279, 
+    65281, 65284, 65285, 65288, 65289, 65290, 65291, 65292, 65293, 65294, 
+    65296, 65306, 65308, 65311, 65313, 65339, 65340, 65341, 65342, 65343, 
+    65344, 65345, 65371, 65372, 65373, 65374, 65375, 65376, 65377, 65378, 
+    65379, 65380, 65382, 65392, 65393, 65438, 65440, 65474, 65482, 65490, 
+    65498, 65504, 65506, 65507, 65508, 65509, 65512, 65513, 65517, 65529, 
+    65532, 0,     13,    40,    60,    63,    80,    128,   256,   263,   
+    311,   320,   373,   377,   394,   400,   464,   509,   640,   672,   
+    768,   800,   816,   833,   834,   842,   896,   927,   928,   968,   
+    976,   977,   1024,  1064,  1104,  1184,  2048,  2056,  2058,  2103,  
+    2108,  2111,  2135,  2136,  2304,  2326,  2335,  2336,  2367,  2432,  
+    2494,  2560,  2561,  2565,  2572,  2576,  2581,  2585,  2616,  2623,  
+    2624,  2640,  2656,  2685,  2687,  2816,  2873,  2880,  2904,  2912,  
+    2936,  3072,  3680,  4096,  4097,  4098,  4099,  4152,  4167,  4178,  
+    4198,  4224,  4226,  4227,  4272,  4275,  4279,  4281,  4283,  4285,  
+    4286,  4304,  4336,  4352,  4355,  4391,  4396,  4397,  4406,  4416,  
+    4480,  4482,  4483,  4531,  4534,  4543,  4545,  4549,  4560,  5760,  
+    5803,  5804,  5805,  5806,  5808,  5814,  5815,  5824,  8192,  9216,  
+    9328,  12288, 26624, 28416, 28496, 28497, 28559, 28563, 45056, 53248, 
+    53504, 53545, 53605, 53607, 53610, 53613, 53619, 53627, 53635, 53637, 
+    53644, 53674, 53678, 53760, 53826, 53829, 54016, 54112, 54272, 54298, 
+    54324, 54350, 54358, 54376, 54402, 54428, 54430, 54434, 54437, 54441, 
+    54446, 54454, 54459, 54461, 54469, 54480, 54506, 54532, 54535, 54541, 
+    54550, 54558, 54584, 54587, 54592, 54598, 54602, 54610, 54636, 54662, 
+    54688, 54714, 54740, 54766, 54792, 54818, 54844, 54870, 54896, 54922, 
+    54952, 54977, 54978, 55003, 55004, 55010, 55035, 55036, 55061, 55062, 
+    55068, 55093, 55094, 55119, 55120, 55126, 55151, 55152, 55177, 55178, 
+    55184, 55209, 55210, 55235, 55236, 55242, 55246, 60928, 60933, 60961, 
+    60964, 60967, 60969, 60980, 60985, 60987, 60994, 60999, 61001, 61003, 
+    61005, 61009, 61012, 61015, 61017, 61019, 61021, 61023, 61025, 61028, 
+    61031, 61036, 61044, 61049, 61054, 61056, 61067, 61089, 61093, 61099, 
+    61168, 61440, 61488, 61600, 61617, 61633, 61649, 61696, 61712, 61744, 
+    61808, 61926, 61968, 62016, 62032, 62208, 62256, 62263, 62336, 62368, 
+    62406, 62432, 62464, 62528, 62530, 62713, 62720, 62784, 62800, 62971, 
+    63045, 63104, 63232, 0,     42710, 42752, 46900, 46912, 47133, 63488, 
+    1,     32,    256,   0,     65533, 
+  };
+static u16 aFts5UnicodeData[] = {
+    1025,  61,    117,   55,    117,   54,    50,    53,    57,    53,    
+    49,    85,    333,   85,    121,   85,    841,   54,    53,    50,    
+    56,    48,    56,    837,   54,    57,    50,    57,    1057,  61,    
+    53,    151,   58,    53,    56,    58,    39,    52,    57,    34,    
+    58,    56,    58,    57,    79,    56,    37,    85,    56,    47,    
+    39,    51,    111,   53,    745,   57,    233,   773,   57,    261,   
+    1822,  37,    542,   37,    1534,  222,   69,    73,    37,    126,   
+    126,   73,    69,    137,   37,    73,    37,    105,   101,   73,    
+    37,    73,    37,    190,   158,   37,    126,   126,   73,    37,    
+    126,   94,    37,    39,    94,    69,    135,   41,    40,    37,    
+    41,    40,    37,    41,    40,    37,    542,   37,    606,   37,    
+    41,    40,    37,    126,   73,    37,    1886,  197,   73,    37,    
+    73,    69,    126,   105,   37,    286,   2181,  39,    869,   582,   
+    152,   390,   472,   166,   248,   38,    56,    38,    568,   3596,  
+    158,   38,    56,    94,    38,    101,   53,    88,    41,    53,    
+    105,   41,    73,    37,    553,   297,   1125,  94,    37,    105,   
+    101,   798,   133,   94,    57,    126,   94,    37,    1641,  1541,  
+    1118,  58,    172,   75,    1790,  478,   37,    2846,  1225,  38,    
+    213,   1253,  53,    49,    55,    1452,  49,    44,    53,    76,    
+    53,    76,    53,    44,    871,   103,   85,    162,   121,   85,    
+    55,    85,    90,    364,   53,    85,    1031,  38,    327,   684,   
+    333,   149,   71,    44,    3175,  53,    39,    236,   34,    58,    
+    204,   70,    76,    58,    140,   71,    333,   103,   90,    39,    
+    469,   34,    39,    44,    967,   876,   2855,  364,   39,    333,   
+    1063,  300,   70,    58,    117,   38,    711,   140,   38,    300,   
+    38,    108,   38,    172,   501,   807,   108,   53,    39,    359,   
+    876,   108,   42,    1735,  44,    42,    44,    39,    106,   268,   
+    138,   44,    74,    39,    236,   327,   76,    85,    333,   53,    
+    38,    199,   231,   44,    74,    263,   71,    711,   231,   39,    
+    135,   44,    39,    106,   140,   74,    74,    44,    39,    42,    
+    71,    103,   76,    333,   71,    87,    207,   58,    55,    76,    
+    42,    199,   71,    711,   231,   71,    71,    71,    44,    106,   
+    76,    76,    108,   44,    135,   39,    333,   76,    103,   44,    
+    76,    42,    295,   103,   711,   231,   71,    167,   44,    39,    
+    106,   172,   76,    42,    74,    44,    39,    71,    76,    333,   
+    53,    55,    44,    74,    263,   71,    711,   231,   71,    167,   
+    44,    39,    42,    44,    42,    140,   74,    74,    44,    44,    
+    42,    71,    103,   76,    333,   58,    39,    207,   44,    39,    
+    199,   103,   135,   71,    39,    71,    71,    103,   391,   74,    
+    44,    74,    106,   106,   44,    39,    42,    333,   111,   218,   
+    55,    58,    106,   263,   103,   743,   327,   167,   39,    108,   
+    138,   108,   140,   76,    71,    71,    76,    333,   239,   58,    
+    74,    263,   103,   743,   327,   167,   44,    39,    42,    44,    
+    170,   44,    74,    74,    76,    74,    39,    71,    76,    333,   
+    71,    74,    263,   103,   1319,  39,    106,   140,   106,   106,   
+    44,    39,    42,    71,    76,    333,   207,   58,    199,   74,    
+    583,   775,   295,   39,    231,   44,    106,   108,   44,    266,   
+    74,    53,    1543,  44,    71,    236,   55,    199,   38,    268,   
+    53,    333,   85,    71,    39,    71,    39,    39,    135,   231,   
+    103,   39,    39,    71,    135,   44,    71,    204,   76,    39,    
+    167,   38,    204,   333,   135,   39,    122,   501,   58,    53,    
+    122,   76,    218,   333,   335,   58,    44,    58,    44,    58,    
+    44,    54,    50,    54,    50,    74,    263,   1159,  460,   42,    
+    172,   53,    76,    167,   364,   1164,  282,   44,    218,   90,    
+    181,   154,   85,    1383,  74,    140,   42,    204,   42,    76,    
+    74,    76,    39,    333,   213,   199,   74,    76,    135,   108,   
+    39,    106,   71,    234,   103,   140,   423,   44,    74,    76,    
+    202,   44,    39,    42,    333,   106,   44,    90,    1225,  41,    
+    41,    1383,  53,    38,    10631, 135,   231,   39,    135,   1319,  
+    135,   1063,  135,   231,   39,    135,   487,   1831,  135,   2151,  
+    108,   309,   655,   519,   346,   2727,  49,    19847, 85,    551,   
+    61,    839,   54,    50,    2407,  117,   110,   423,   135,   108,   
+    583,   108,   85,    583,   76,    423,   103,   76,    1671,  76,    
+    42,    236,   266,   44,    74,    364,   117,   38,    117,   55,    
+    39,    44,    333,   335,   213,   49,    149,   108,   61,    333,   
+    1127,  38,    1671,  1319,  44,    39,    2247,  935,   108,   138,   
+    76,    106,   74,    44,    202,   108,   58,    85,    333,   967,   
+    167,   1415,  554,   231,   74,    333,   47,    1114,  743,   76,    
+    106,   85,    1703,  42,    44,    42,    236,   44,    42,    44,    
+    74,    268,   202,   332,   44,    333,   333,   245,   38,    213,   
+    140,   42,    1511,  44,    42,    172,   42,    44,    170,   44,    
+    74,    231,   333,   245,   346,   300,   314,   76,    42,    967,   
+    42,    140,   74,    76,    42,    44,    74,    71,    333,   1415,  
+    44,    42,    76,    106,   44,    42,    108,   74,    149,   1159,  
+    266,   268,   74,    76,    181,   333,   103,   333,   967,   198,   
+    85,    277,   108,   53,    428,   42,    236,   135,   44,    135,   
+    74,    44,    71,    1413,  2022,  421,   38,    1093,  1190,  1260,  
+    140,   4830,  261,   3166,  261,   265,   197,   201,   261,   265,   
+    261,   265,   197,   201,   261,   41,    41,    41,    94,    229,   
+    265,   453,   261,   264,   261,   264,   261,   264,   165,   69,    
+    137,   40,    56,    37,    120,   101,   69,    137,   40,    120,   
+    133,   69,    137,   120,   261,   169,   120,   101,   69,    137,   
+    40,    88,    381,   162,   209,   85,    52,    51,    54,    84,    
+    51,    54,    52,    277,   59,    60,    162,   61,    309,   52,    
+    51,    149,   80,    117,   57,    54,    50,    373,   57,    53,    
+    48,    341,   61,    162,   194,   47,    38,    207,   121,   54,    
+    50,    38,    335,   121,   54,    50,    422,   855,   428,   139,   
+    44,    107,   396,   90,    41,    154,   41,    90,    37,    105,   
+    69,    105,   37,    58,    41,    90,    57,    169,   218,   41,    
+    58,    41,    58,    41,    58,    137,   58,    37,    137,   37,    
+    135,   37,    90,    69,    73,    185,   94,    101,   58,    57,    
+    90,    37,    58,    527,   1134,  94,    142,   47,    185,   186,   
+    89,    154,   57,    90,    57,    90,    57,    250,   57,    1018,  
+    89,    90,    57,    58,    57,    1018,  8601,  282,   153,   666,   
+    89,    250,   54,    50,    2618,  57,    986,   825,   1306,  217,   
+    602,   1274,  378,   1935,  2522,  719,   5882,  57,    314,   57,    
+    1754,  281,   3578,  57,    4634,  3322,  54,    50,    54,    50,    
+    54,    50,    54,    50,    54,    50,    54,    50,    54,    50,    
+    975,   1434,  185,   54,    50,    1017,  54,    50,    54,    50,    
+    54,    50,    54,    50,    54,    50,    537,   8218,  4217,  54,    
+    50,    54,    50,    54,    50,    54,    50,    54,    50,    54,    
+    50,    54,    50,    54,    50,    54,    50,    54,    50,    54,    
+    50,    2041,  54,    50,    54,    50,    1049,  54,    50,    8281,  
+    1562,  697,   90,    217,   346,   1513,  1509,  126,   73,    69,    
+    254,   105,   37,    94,    37,    94,    165,   70,    105,   37,    
+    3166,  37,    218,   158,   108,   94,    149,   47,    85,    1221,  
+    37,    37,    1799,  38,    53,    44,    743,   231,   231,   231,   
+    231,   231,   231,   231,   231,   1036,  85,    52,    51,    52,    
+    51,    117,   52,    51,    53,    52,    51,    309,   49,    85,    
+    49,    53,    52,    51,    85,    52,    51,    54,    50,    54,    
+    50,    54,    50,    54,    50,    181,   38,    341,   81,    858,   
+    2874,  6874,  410,   61,    117,   58,    38,    39,    46,    54,    
+    50,    54,    50,    54,    50,    54,    50,    54,    50,    90,    
+    54,    50,    54,    50,    54,    50,    54,    50,    49,    54,    
+    82,    58,    302,   140,   74,    49,    166,   90,    110,   38,    
+    39,    53,    90,    2759,  76,    88,    70,    39,    49,    2887,  
+    53,    102,   39,    1319,  3015,  90,    143,   346,   871,   1178,  
+    519,   1018,  335,   986,   271,   58,    495,   1050,  335,   1274,  
+    495,   2042,  8218,  39,    39,    2074,  39,    39,    679,   38,    
+    36583, 1786,  1287,  198,   85,    8583,  38,    117,   519,   333,   
+    71,    1502,  39,    44,    107,   53,    332,   53,    38,    798,   
+    44,    2247,  334,   76,    213,   760,   294,   88,    478,   69,    
+    2014,  38,    261,   190,   350,   38,    88,    158,   158,   382,   
+    70,    37,    231,   44,    103,   44,    135,   44,    743,   74,    
+    76,    42,    154,   207,   90,    55,    58,    1671,  149,   74,    
+    1607,  522,   44,    85,    333,   588,   199,   117,   39,    333,   
+    903,   268,   85,    743,   364,   74,    53,    935,   108,   42,    
+    1511,  44,    74,    140,   74,    44,    138,   437,   38,    333,   
+    85,    1319,  204,   74,    76,    74,    76,    103,   44,    263,   
+    44,    42,    333,   149,   519,   38,    199,   122,   39,    42,    
+    1543,  44,    39,    108,   71,    76,    167,   76,    39,    44,    
+    39,    71,    38,    85,    359,   42,    76,    74,    85,    39,    
+    70,    42,    44,    199,   199,   199,   231,   231,   1127,  74,    
+    44,    74,    44,    74,    53,    42,    44,    333,   39,    39,    
+    743,   1575,  36,    68,    68,    36,    63,    63,    11719, 3399,  
+    229,   165,   39,    44,    327,   57,    423,   167,   39,    71,    
+    71,    3463,  536,   11623, 54,    50,    2055,  1735,  391,   55,    
+    58,    524,   245,   54,    50,    53,    236,   53,    81,    80,    
+    54,    50,    54,    50,    54,    50,    54,    50,    54,    50,    
+    54,    50,    54,    50,    54,    50,    85,    54,    50,    149,   
+    112,   117,   149,   49,    54,    50,    54,    50,    54,    50,    
+    117,   57,    49,    121,   53,    55,    85,    167,   4327,  34,    
+    117,   55,    117,   54,    50,    53,    57,    53,    49,    85,    
+    333,   85,    121,   85,    841,   54,    53,    50,    56,    48,    
+    56,    837,   54,    57,    50,    57,    54,    50,    53,    54,    
+    50,    85,    327,   38,    1447,  70,    999,   199,   199,   199,   
+    103,   87,    57,    56,    58,    87,    58,    153,   90,    98,    
+    90,    391,   839,   615,   71,    487,   455,   3943,  117,   1455,  
+    314,   1710,  143,   570,   47,    410,   1466,  44,    935,   1575,  
+    999,   143,   551,   46,    263,   46,    967,   53,    1159,  263,   
+    53,    174,   1289,  1285,  2503,  333,   199,   39,    1415,  71,    
+    39,    743,   53,    271,   711,   207,   53,    839,   53,    1799,  
+    71,    39,    108,   76,    140,   135,   103,   871,   108,   44,    
+    271,   309,   935,   79,    53,    1735,  245,   711,   271,   615,   
+    271,   2343,  1007,  42,    44,    42,    1703,  492,   245,   655,   
+    333,   76,    42,    1447,  106,   140,   74,    76,    85,    34,    
+    149,   807,   333,   108,   1159,  172,   42,    268,   333,   149,   
+    76,    42,    1543,  106,   300,   74,    135,   149,   333,   1383,  
+    44,    42,    44,    74,    204,   42,    44,    333,   28135, 3182,  
+    149,   34279, 18215, 2215,  39,    1482,  140,   422,   71,    7898,  
+    1274,  1946,  74,    108,   122,   202,   258,   268,   90,    236,   
+    986,   140,   1562,  2138,  108,   58,    2810,  591,   841,   837,   
+    841,   229,   581,   841,   837,   41,    73,    41,    73,    137,   
+    265,   133,   37,    229,   357,   841,   837,   73,    137,   265,   
+    233,   837,   73,    137,   169,   41,    233,   837,   841,   837,   
+    841,   837,   841,   837,   841,   837,   841,   837,   841,   901,   
+    809,   57,    805,   57,    197,   809,   57,    805,   57,    197,   
+    809,   57,    805,   57,    197,   809,   57,    805,   57,    197,   
+    809,   57,    805,   57,    197,   94,    1613,  135,   871,   71,    
+    39,    39,    327,   135,   39,    39,    39,    39,    39,    39,    
+    103,   71,    39,    39,    39,    39,    39,    39,    71,    39,    
+    135,   231,   135,   135,   39,    327,   551,   103,   167,   551,   
+    89,    1434,  3226,  506,   474,   506,   506,   367,   1018,  1946,  
+    1402,  954,   1402,  314,   90,    1082,  218,   2266,  666,   1210,  
+    186,   570,   2042,  58,    5850,  154,   2010,  154,   794,   2266,  
+    378,   2266,  3738,  39,    39,    39,    39,    39,    39,    17351, 
+    34,    3074,  7692,  63,    63,    
+  };
+
+int sqlite3Fts5UnicodeCategory(int iCode) { 
+  int iRes = -1;
+  int iHi;
+  int iLo;
+  int ret;
+  u16 iKey;
+
+  if( iCode>=(1<<20) ){
+    return 0;
+  }
+  iLo = aFts5UnicodeBlock[(iCode>>16)];
+  iHi = aFts5UnicodeBlock[1+(iCode>>16)];
+  iKey = (iCode & 0xFFFF);
+  while( iHi>iLo ){
+    int iTest = (iHi + iLo) / 2;
+    assert( iTest>=iLo && iTest<iHi );
+    if( iKey>=aFts5UnicodeMap[iTest] ){
+      iRes = iTest;
+      iLo = iTest+1;
+    }else{
+      iHi = iTest;
+    }
+  }
+
+  if( iRes<0 ) return 0;
+  if( iKey>=(aFts5UnicodeMap[iRes]+(aFts5UnicodeData[iRes]>>5)) ) return 0;
+  ret = aFts5UnicodeData[iRes] & 0x1F;
+  if( ret!=30 ) return ret;
+  return ((iKey - aFts5UnicodeMap[iRes]) & 0x01) ? 5 : 9;
+}
+
+void sqlite3Fts5UnicodeAscii(u8 *aArray, u8 *aAscii){
+  int i = 0;
+  int iTbl = 0;
+  while( i<128 ){
+    int bToken = aArray[ aFts5UnicodeData[iTbl] & 0x1F ];
+    int n = (aFts5UnicodeData[iTbl] >> 5) + i;
+    for(; i<128 && i<n; i++){
+      aAscii[i] = bToken;
+    }
+    iTbl++;
+  }
+}
+

diff --git a/ext/fts5/test/fts5cat.test b/ext/fts5/test/fts5cat.test
new file mode 100644 (file)
index 0000000..483f64b
--- /dev/null
+++ b/ext/fts5/test/fts5cat.test
@@ -0,0 +1,59 @@
+# 2016 Jan 15
+#
+# The author disclaims copyright to this source code.  In place of
+# a legal notice, here is a blessing:
+#
+#    May you do good and not evil.
+#    May you find forgiveness for yourself and forgive others.
+#    May you share freely, never taking more than you give.
+#
+#*************************************************************************
+#
+
+source [file join [file dirname [info script]] fts5_common.tcl]
+ifcapable !fts5 { finish_test ; return }
+set ::testprefix fts5cat
+
+
+do_execsql_test 1.0 {
+  CREATE VIRTUAL TABLE t1 USING fts5(x, tokenize="unicode61 categories 'L*'");
+  INSERT INTO t1 VALUES ('Unlike1option2values3and4column5names');
+}
+
+do_execsql_test 1.1 {
+  SELECT rowid FROM t1('option');
+} {1}
+
+do_execsql_test 1.2 {
+  CREATE VIRTUAL TABLE t2 USING fts5(x);
+  CREATE VIRTUAL TABLE t2t USING fts5vocab(t2, row);
+
+  CREATE VIRTUAL TABLE t3 USING fts5(
+    x, tokenize="unicode61 categories 'L* N* Co Mn'"
+  );
+  CREATE VIRTUAL TABLE t3t USING fts5vocab(t3, row);
+
+  CREATE VIRTUAL TABLE t4 USING fts5(
+    x, tokenize="unicode61 categories 'L* N* Co M*'"
+  );
+  CREATE VIRTUAL TABLE t4t USING fts5vocab(t4, row);
+
+  INSERT INTO t2 VALUES ('สนามกีฬา');
+  INSERT INTO t3 VALUES ('สนามกีฬา');
+  INSERT INTO t4 VALUES ('สนามกีฬา');
+}
+
+do_execsql_test 1.3 {
+  SELECT * FROM t2t
+} {สนามก 1 1 ฬา 1 1}
+
+do_execsql_test 1.4 {
+  SELECT * FROM t3t
+} {สนามกีฬา 1 1}
+
+do_execsql_test 1.5 {
+  SELECT * FROM t4t
+} {สนามกีฬา 1 1}
+
+
+finish_test

diff --git a/manifest b/manifest
index 0519b9126b5cb77952e8d85132f7d0821e0cdfe9..c56b0d6dcc6c0e585d1d5794738c66d5af283d6d 100644 (file)
--- a/manifest
+++ b/manifest
@@ -1,5 +1,5 @@
-C Add\sa\stest-control\sto\sdisable\sthe\sskip-scan\soptimization.
-D 2018-07-12T19:14:39.361
+C Add\sthe\s"categories"\soption\sto\sthe\sunicode61\stokenizer\sin\sfts5.
+D 2018-07-13T19:52:43.227
 F .fossil-settings/empty-dirs dbb81e8fc0401ac46a1491ab34a7f2c7c0452f2f06b54ebb845d024ca8283ef1
 F .fossil-settings/ignore-glob 35175cdfcf539b2318cb04a9901442804be81cd677d8b889fcc9149c21f239ea
 F Makefile.in 0a3a6c81e6fcb969ff9106e882f0a08547014ba463cb6beca4c4efaecc924ee6
@@ -104,15 +104,15 @@ F ext/fts3/tool/fts3cov.sh c331d006359456cf6f8f953e37f2b9c7d568f3863f00bb5f7eb87
 F ext/fts3/tool/fts3view.c 202801a2056995b763864d60c2dee744d46f1677
 F ext/fts3/unicode/CaseFolding.txt 8c678ca52ecc95e16bc7afc2dbf6fc9ffa05db8c
 F ext/fts3/unicode/UnicodeData.txt cd07314edb62d49fde34debdaf92fa2aa69011e7
-F ext/fts3/unicode/mkunicode.tcl ab0543a3b2399092ea2dd75df1bef333405b0d7f6b8c4951a0fbb60e780cb69f
-F ext/fts3/unicode/parseunicode.tcl da577d1384810fb4e2b209bf3313074353193e95
+F ext/fts3/unicode/mkunicode.tcl 0069320b64db6ee269c5e95f1f150d070fbf0a863fc7b3549d7e52bd068fb118
+F ext/fts3/unicode/parseunicode.tcl 024ae0bdd96309d7b8fc479148191e9b3001dc74017a3f65f9a27de3b3ff968b
 F ext/fts5/extract_api_docs.tcl a36e54ec777172ddd3f9a88daf593b00848368e0
 F ext/fts5/fts5.h 62f3e33ceeb9a428db139f9c012186b371da1cc7
-F ext/fts5/fts5Int.h eda28e3a0a5d87c412e8355fe35da875b04cb389908c8eb0d867ad662adbc491
+F ext/fts5/fts5Int.h 6354c8f830487a48b7a0872c9eb33b4258a57f9e3c315df9da95c896d70a732b
 F ext/fts5/fts5_aux.c ca666a3bbe07c5a3bbe9fffaea19c935a1efaf337333e28bad7bdd1971ffd093
 F ext/fts5/fts5_buffer.c 1dd1ec0446b3acfc2d7d407eb894762a461613e2695273f48e449bfd13e973ff
 F ext/fts5/fts5_config.c 5af9c360e99669d29f06492c370892394aba0857
-F ext/fts5/fts5_expr.c 4ff82037692dcaf42ddf4f379a08815b4bd8784194a083994fec51eef31ef2dd
+F ext/fts5/fts5_expr.c 5aef080ba3c8947e22f38ce1ff9fe548e4a740e72b77241f35ed941ae128d2c7
 F ext/fts5/fts5_hash.c 32be400cf761868c9db33efe81a06eb19a17c5402ad477ee9efb51301546dd55
 F ext/fts5/fts5_index.c 22b71d0e9e4b3ddd123a39ae27174e0012da2806f91b64087a68584f13f189de
 F ext/fts5/fts5_main.c da46761a7e9b582083fcb9f5a3ee50086205fb91f4e68d984a9946e64218e297
@@ -120,8 +120,8 @@ F ext/fts5/fts5_storage.c 4bec8a1b3905978b22a67bca5f4a3cfdb94af234cf51efb36f4f2d
 F ext/fts5/fts5_tcl.c 39bcbae507f594aad778172fa914cad0f585bf92fd3b078c686e249282db0d95
 F ext/fts5/fts5_test_mi.c 65864ba1e5c34a61d409c4c587e0bbe0466eb4f8f478d85dc42a92caad1338e6
 F ext/fts5/fts5_test_tok.c ffd657dd67e7fcdb31bf63fb60b6d867299a581d0f46e97086abacd66c2a9b26
-F ext/fts5/fts5_tokenize.c 2ce7b44183538ec46b7907726262ee43ffdd39a8
-F ext/fts5/fts5_unicode2.c b450b209b157d598f7b9df9f837afb75a14c24bf
+F ext/fts5/fts5_tokenize.c ebd13d034f3dc7c841e1c32c364a4fca5cc2e05a0b91682a93fa1e6defcd4292
+F ext/fts5/fts5_unicode2.c bcf4afd86a00be1608deac305b914ca101161e43feed026818fcdde19e9b30e2
 F ext/fts5/fts5_varint.c a5aceacda04dafcbae725413d7a16818ecd65738
 F ext/fts5/fts5_vocab.c 1cd79854cb21543e66507b25b0578bc1b20aa6a1349b7feceb8e8fed0e7a77a6
 F ext/fts5/fts5parse.y eb526940f892ade5693f22ffd6c4f2702543a9059942772526eac1fde256bb05
@@ -145,6 +145,7 @@ F ext/fts5/test/fts5aux.test 8e7f9c96f8570f48402f2fd7b74e1ccfadd319fa56ef6773218
 F ext/fts5/test/fts5auxdata.test eacc97ff04892f1a5f3d4df5a73f8bcbc3955ea1d12c9f24137eb1fc079e7611
 F ext/fts5/test/fts5bigpl.test 6466c89b38439f0aba26ac09e232a6b963f29b1cbe1304f6a664fe1e7a8f5fd3
 F ext/fts5/test/fts5bigtok.test 541119e616c637caea925a8c028c37c2c29e94383e00aa2f9198d530724b6e36
+F ext/fts5/test/fts5cat.test daba0b80659460b0cb60bd1f40b402478a761fe7ea414c3c94c2be25568cc33a
 F ext/fts5/test/fts5colset.test a30473451321bbf0b6218af62e96b4ae5fa99931cfdb210b5ecc804623b30f75
 F ext/fts5/test/fts5columnsize.test 45459ce4dd9fd853b6044cdc9674921bff89e3d840f348ca8c1630f9edbf5482
 F ext/fts5/test/fts5config.test 60094712debc59286c59aef0e6cf511c37d866802776a825ce437d26afe0817f
@@ -1746,7 +1747,8 @@ F vsixtest/vsixtest.tcl 6a9a6ab600c25a91a7acc6293828957a386a8a93
 F vsixtest/vsixtest.vcxproj.data 2ed517e100c66dc455b492e1a33350c1b20fbcdc
 F vsixtest/vsixtest.vcxproj.filters 37e51ffedcdb064aad6ff33b6148725226cd608e
 F vsixtest/vsixtest_TemporaryKey.pfx e5b1b036facdb453873e7084e1cae9102ccc67a0
-P 6145f5b3bacdee8f377892ac8b885941336c4cc543052f6373e21c1f99e1eca8
-R cc992c4b8820025ab207851df0931eaf
+P 650a3fe03d61068e06f3097878a777ca8ed713a769444cdb3f8be7d1e19a0b83
+R dca8016ca1ceb389549f1bc1385b49e3
+T +closed 8e59dc7807d15d4fca6ee7b6a624447748eb24afb48f902fca27c2d550d84f1c
 U dan
-Z 0f26e8f4a97d1c9c90af6cbfe9ae4cb9
+Z 28a0d1fcc85b29bafa0cd6ceb5a53164

diff --git a/manifest.uuid b/manifest.uuid
index 35261a67bbb78a77fd8149e8970c6b8e86d3b9b0..97be00cc68e29a3d54ab8393913c77fa602c82a9 100644 (file)
--- a/manifest.uuid
+++ b/manifest.uuid
@@ -1 +1 @@
-650a3fe03d61068e06f3097878a777ca8ed713a769444cdb3f8be7d1e19a0b83
\ No newline at end of file
+80d2b9e635e3100f90cffdcffa5b5038da6fbbfccc9f5777c59a4ae760d4cb62
\ No newline at end of file