SQLite

    'f',       'g',       'h',       'h',       'i',       'i'|HIBIT, 
    'k',       'l',       'l'|HIBIT, 'l',       'm',       'n',       
    'o'|HIBIT, 'p',       'r',       'r'|HIBIT, 'r',       's',       
    's'|HIBIT, 't',       'u',       'u'|HIBIT, 'v',       'w',       
    'w',       'x',       'y',       'z',       'h',       't',       
    'w',       'y',       'a',       'a'|HIBIT, 'a'|HIBIT, 'a'|HIBIT, 
    'e',       'e'|HIBIT, 'e'|HIBIT, 'i',       'o',       'o'|HIBIT, 
    'o'|HIBIT, 'o'|HIBIT, 'u',       'u'|HIBIT, 'u'|HIBIT, 'y',       
  };

  unsigned int key = (((unsigned int)c)<<3) | 0x00000007;
  int iRes = 0;
  int iHi = sizeof(aDia)/sizeof(aDia[0]) - 1;
  int iLo = 0;
  while( iHi>=iLo ){

** is a diacritical modifier character.
*/
int sqlite3FtsUnicodeIsdiacritic(int c){
  unsigned int mask0 = 0x08029FDF;
  unsigned int mask1 = 0x000361F8;
  if( c<768 || c>817 ) return 0;
  return (c < 768+32) ?
      (mask0 & ((unsigned int)1 << (c-768))) :
      (mask1 & ((unsigned int)1 << (c-768-32)));
}


/*
** Interpret the argument as a unicode codepoint. If the codepoint
** is an upper case character that has a lower case equivalent,
** return the codepoint corresponding to the lower case version.

    incr i

    puts -nonewline [format "%5d" [expr ($iCode<<3) + $nRange-1]]
    puts -nonewline ", "
  }
  puts ""
  puts "  \};"
  puts "#define HIBIT ((char)0x80)"
  puts "  char aChar\[\] = \{"
  puts -nonewline "    '\\0',      "
  set i 1
  foreach c $aChar f $aFlag {
    if { $f } {
      set str "'$c'|HIBIT, "
    } else {
      set str "'$c',       "
    }
    if {$c == ""} { set str "'\\0',      " }

    if {($i % 6)==0} {puts "" ; puts -nonewline "    " }
    incr i
    puts -nonewline "$str"
  }

  puts "*/"
  puts "int ${zFunc}\(int c)\{"
  puts "  unsigned int mask0 = [format "0x%08X" $i1];"
  puts "  unsigned int mask1 = [format "0x%08X" $i2];"

  puts "  if( c<$iFirst || c>$iLast ) return 0;"
  puts "  return (c < $iFirst+32) ?"
  puts "      (mask0 & ((unsigned int)1 << (c-$iFirst))) :"
  puts "      (mask1 & ((unsigned int)1 << (c-$iFirst-32)));"
  puts "\}"
}


#-------------------------------------------------------------------------

proc an_load_separator_ranges {} {

  set aMapArray [intarray $aMap]
  set aDataArray [intarray $aData]
  puts [code {
    static u16 aFts5UnicodeBlock[] = {$aBlockArray};
    static u16 aFts5UnicodeMap[] = {$aMapArray};
    static u16 aFts5UnicodeData[] = {$aDataArray};

    int sqlite3Fts5UnicodeCategory(u32 iCode) { 
      int iRes = -1;
      int iHi;
      int iLo;
      int ret;
      u16 iKey;

      if( iCode>=(1<<20) ){

        if( aCP[iCP].iCode==i ){
          sqlite3Fts5UnicodeCatParse(aCP[iCP].zCat, aArray);
          iCP++;
        }else{
          aArray[0] = 1;
        }

        c = sqlite3Fts5UnicodeCategory((u32)i);
        if( aArray[c]==0 ){
          *piCode = i;
          return 1;
        }
      }

      return 0;

/**************************************************************************
** Interface to automatically generated code in fts5_unicode2.c. 
*/
int sqlite3Fts5UnicodeIsdiacritic(int c);
int sqlite3Fts5UnicodeFold(int c, int bRemoveDiacritic);

int sqlite3Fts5UnicodeCatParse(const char*, u8*);
int sqlite3Fts5UnicodeCategory(u32 iCode);
void sqlite3Fts5UnicodeAscii(u8*, u8*);
/*
** End of interface to code in fts5_unicode2.c.
**************************************************************************/

#endif

    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, aArr[sqlite3Fts5UnicodeCategory((u32)iCode)]);
}

static void fts5ExprFold(
  sqlite3_context *pCtx,          /* Function call context */
  int nArg,                       /* Number of args */
  sqlite3_value **apVal           /* Function arguments */
){

      int i;
      u32 mask;
      memset(aUsed, 0, sizeof(aUsed));
      for(iLvl=0; iLvl<pStruct->nLevel; iLvl++){
        for(iSeg=0; iSeg<pStruct->aLevel[iLvl].nSeg; iSeg++){
          int iId = pStruct->aLevel[iLvl].aSeg[iSeg].iSegid;
          if( iId<=FTS5_MAX_SEGMENT ){
            aUsed[(iId-1) / 32] |= (u32)1 << ((iId-1) % 32);
          }
        }
      }

      for(i=0; aUsed[i]==0xFFFFFFFF; i++);
      mask = aUsed[i];
      for(iSegid=0; mask & ((u32)1 << iSegid); iSegid++);
      iSegid += 1 + i*32;

#ifdef SQLITE_DEBUG
      for(iLvl=0; iLvl<pStruct->nLevel; iLvl++){
        for(iSeg=0; iSeg<pStruct->aLevel[iLvl].nSeg; iSeg++){
          assert_nc( iSegid!=pStruct->aLevel[iLvl].aSeg[iSeg].iSegid );
        }

  ** copy is followed by FTS5_DATA_ZERO_PADDING 0x00 bytes, which prevents
  ** buffer overreads even if the record is corrupt.  */
  n = sqlite3_value_bytes(apVal[1]);
  aBlob = sqlite3_value_blob(apVal[1]);
  nSpace = n + FTS5_DATA_ZERO_PADDING;
  a = (u8*)sqlite3Fts5MallocZero(&rc, nSpace);
  if( a==0 ) goto decode_out;
  if( n>0 ) memcpy(a, aBlob, n);


  fts5DecodeRowid(iRowid, &iSegid, &bDlidx, &iHeight, &iPgno);

  fts5DebugRowid(&rc, &s, iRowid);
  if( bDlidx ){
    Fts5Data dlidx;
    Fts5DlidxLvl lvl;

  if( idxNum==1 ){
    const char *zByte = (const char *)sqlite3_value_text(apVal[0]);
    int nByte = sqlite3_value_bytes(apVal[0]);
    pCsr->zInput = sqlite3_malloc(nByte+1);
    if( pCsr->zInput==0 ){
      rc = SQLITE_NOMEM;
    }else{
      if( nByte>0 ) memcpy(pCsr->zInput, zByte, nByte);
      pCsr->zInput[nByte] = 0;
      rc = pTab->tok.xTokenize(
          pTab->pTok, (void*)pCsr, 0, zByte, nByte, fts5tokCb
      );
    }
  }

      while( zCsr<zTerm ){
        int iCode;
        int bToken;
        READ_UTF8(zCsr, zTerm, iCode);
        if( iCode<128 ){
          p->aTokenChar[iCode] = (unsigned char)bTokenChars;
        }else{
          bToken = p->aCategory[sqlite3Fts5UnicodeCategory((u32)iCode)];
          assert( (bToken==0 || bToken==1) ); 
          assert( (bTokenChars==0 || bTokenChars==1) );
          if( bToken!=bTokenChars && sqlite3Fts5UnicodeIsdiacritic(iCode)==0 ){
            int i;
            for(i=0; i<nNew; i++){
              if( aNew[i]>iCode ) break;
            }

/*
** Return true if, for the purposes of tokenizing with the tokenizer
** passed as the first argument, codepoint iCode is considered a token 
** character (not a separator).
*/
static int fts5UnicodeIsAlnum(Unicode61Tokenizer *p, int iCode){
  return (
    p->aCategory[sqlite3Fts5UnicodeCategory((u32)iCode)]
    ^ fts5UnicodeIsException(p, iCode)
  );
}

static int fts5UnicodeTokenize(
  Fts5Tokenizer *pTokenizer,
  void *pCtx,

    'f',       'g',       'h',       'h',       'i',       'i'|HIBIT, 
    'k',       'l',       'l'|HIBIT, 'l',       'm',       'n',       
    'o'|HIBIT, 'p',       'r',       'r'|HIBIT, 'r',       's',       
    's'|HIBIT, 't',       'u',       'u'|HIBIT, 'v',       'w',       
    'w',       'x',       'y',       'z',       'h',       't',       
    'w',       'y',       'a',       'a'|HIBIT, 'a'|HIBIT, 'a'|HIBIT, 
    'e',       'e'|HIBIT, 'e'|HIBIT, 'i',       'o',       'o'|HIBIT, 
    'o'|HIBIT, 'o'|HIBIT, 'u',       'u'|HIBIT, 'u'|HIBIT, 'y',       
  };

  unsigned int key = (((unsigned int)c)<<3) | 0x00000007;
  int iRes = 0;
  int iHi = sizeof(aDia)/sizeof(aDia[0]) - 1;
  int iLo = 0;
  while( iHi>=iLo ){

** is a diacritical modifier character.
*/
int sqlite3Fts5UnicodeIsdiacritic(int c){
  unsigned int mask0 = 0x08029FDF;
  unsigned int mask1 = 0x000361F8;
  if( c<768 || c>817 ) return 0;
  return (c < 768+32) ?
      (mask0 & ((unsigned int)1 << (c-768))) :
      (mask1 & ((unsigned int)1 << (c-768-32)));
}


/*
** Interpret the argument as a unicode codepoint. If the codepoint
** is an upper case character that has a lower case equivalent,
** return the codepoint corresponding to the lower case version.

  
  else if( c>=66560 && c<66600 ){
    ret = c + 40;
  }

  return ret;
}


int sqlite3Fts5UnicodeCatParse(const char *zCat, u8 *aArray){ 
  aArray[0] = 1;
  switch( zCat[0] ){
    case 'C':
          switch( zCat[1] ){
            case 'c': aArray[1] = 1; break;

    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(u32 iCode) { 
  int iRes = -1;
  int iHi;
  int iLo;
  int ret;
  u16 iKey;

  if( iCode>=(1<<20) ){

    int n = (aFts5UnicodeData[iTbl] >> 5) + i;
    for(; i<128 && i<n; i++){
      aAscii[i] = bToken;
    }
    iTbl++;
  }
}

# If SQLITE_ENABLE_FTS5 is defined, omit this file.
ifcapable !fts5 {
  finish_test
  return
}

proc fts3_unicode_path {file} {
  file join .. [file dirname [info script]] .. .. fts3 unicode $file
}

source [fts3_unicode_path parseunicode.tcl]
set testprefix fts5unicode3

set CF [fts3_unicode_path CaseFolding.txt]
set UD [fts3_unicode_path UnicodeData.txt]