SQLite

*/
struct Fts5Global {
  fts5_api api;                   /* User visible part of object (see fts5.h) */
  sqlite3 *db;                    /* Associated database connection */ 
  i64 iNextId;                    /* Used to allocate unique cursor ids */
  Fts5Auxiliary *pAux;            /* First in list of all aux. functions */
  Fts5TokenizerModule *pTok;      /* First in list of all tokenizer modules */
  Fts5TokenizerModule *pDfltTok;  /* Default tokenizer module */
  Fts5Cursor *pCsr;               /* First in list of all open cursors */
};

/*
** Each auxiliary function registered with the FTS5 module is represented
** by an object of the following type. All such objects are stored as part
** of the Fts5Global.pAux list.

}


static int fts5FindRankFunction(Fts5Cursor *pCsr){
  Fts5Table *pTab = (Fts5Table*)(pCsr->base.pVtab);
  Fts5Config *pConfig = pTab->pConfig;
  int rc = SQLITE_OK;
  Fts5Auxiliary *pAux = 0;
  const char *zRank = pCsr->zRank;
  const char *zRankArgs = pCsr->zRankArgs;

  if( zRankArgs ){
    char *zSql = sqlite3_mprintf("SELECT %s", zRankArgs);
    if( zSql==0 ){
      rc = SQLITE_NOMEM;

      }
    }
  }
  return rc;
}

static void fts5SetVtabError(Fts5Table *p, const char *zFormat, ...){

  va_list ap;                     /* ... printf arguments */
  va_start(ap, zFormat);
  assert( p->base.zErrMsg==0 );
  p->base.zErrMsg = sqlite3_vmprintf(zFormat, ap);
  va_end(ap);
}

    pNew->zName = (char*)&pNew[1];
    strcpy(pNew->zName, zName);
    pNew->pUserData = pUserData;
    pNew->x = *pTokenizer;
    pNew->xDestroy = xDestroy;
    pNew->pNext = pGlobal->pTok;
    pGlobal->pTok = pNew;
    if( pNew->pNext==0 ){
      pGlobal->pDfltTok = pNew;
    }
  }else{
    rc = SQLITE_NOMEM;
  }

  return rc;
}

  void **ppUserData,
  fts5_tokenizer *pTokenizer      /* Populate this object */
){
  Fts5Global *pGlobal = (Fts5Global*)pApi;
  int rc = SQLITE_OK;
  Fts5TokenizerModule *pTok;

  if( zName==0 ){
    pTok = pGlobal->pDfltTok;
  }else{
    for(pTok=pGlobal->pTok; pTok; pTok=pTok->pNext){
      if( sqlite3_stricmp(zName, pTok->zName)==0 ) break;
    }
  }

  if( pTok ){
    *pTokenizer = pTok->x;
    *ppUserData = pTok->pUserData;
  }else{
    memset(pTokenizer, 0, sizeof(fts5_tokenizer));

  const char **azArg,
  int nArg,
  Fts5Tokenizer **ppTok,
  fts5_tokenizer **ppTokApi
){
  Fts5TokenizerModule *pMod = 0;
  int rc = SQLITE_OK;

  if( nArg==0 ){
    pMod = pGlobal->pDfltTok;
  }else{
    for(pMod=pGlobal->pTok; pMod; pMod=pMod->pNext){
      if( sqlite3_stricmp(azArg[0], pMod->zName)==0 ) break;
    }
  }

  if( pMod==0 ){

*/

#include "fts5.h"
#include <string.h>
#include <assert.h>

/**************************************************************************
** Start of ascii tokenizer implementation.
*/

/*
** For tokenizers with no "unicode" modifier, the set of token characters
** is the same as the set of ASCII range alphanumeric characters. 
*/
static unsigned char aAsciiTokenChar[128] = {
  0, 0, 0, 0, 0, 0, 0, 0,   0, 0, 0, 0, 0, 0, 0, 0,   /* 0x00..0x0F */
  0, 0, 0, 0, 0, 0, 0, 0,   0, 0, 0, 0, 0, 0, 0, 0,   /* 0x10..0x1F */
  0, 0, 0, 0, 0, 0, 0, 0,   0, 0, 0, 0, 0, 0, 0, 0,   /* 0x20..0x2F */
  1, 1, 1, 1, 1, 1, 1, 1,   1, 1, 0, 0, 0, 0, 0, 0,   /* 0x30..0x3F */
  0, 1, 1, 1, 1, 1, 1, 1,   1, 1, 1, 1, 1, 1, 1, 1,   /* 0x40..0x4F */
  1, 1, 1, 1, 1, 1, 1, 1,   1, 1, 1, 0, 0, 0, 0, 0,   /* 0x50..0x5F */
  0, 1, 1, 1, 1, 1, 1, 1,   1, 1, 1, 1, 1, 1, 1, 1,   /* 0x60..0x6F */
  1, 1, 1, 1, 1, 1, 1, 1,   1, 1, 1, 0, 0, 0, 0, 0,   /* 0x70..0x7F */
};

typedef struct AsciiTokenizer AsciiTokenizer;
struct AsciiTokenizer {
  unsigned char aTokenChar[128];
};

static void fts5AsciiAddExceptions(
  AsciiTokenizer *p, 
  const char *zArg, 
  int bTokenChars
){
  int i;
  for(i=0; zArg[i]; i++){
    if( (zArg[i] & 0x80)==0 ){
      p->aTokenChar[(int)zArg[i]] = (unsigned char)bTokenChars;
    }
  }
}

/*
** Create a "ascii" tokenizer.
*/
static int fts5AsciiCreate(
  void *pCtx, 
  const char **azArg, int nArg,
  Fts5Tokenizer **ppOut
){
  int rc = SQLITE_OK;
  AsciiTokenizer *p = 0;
  if( nArg%2 ){
    rc = SQLITE_ERROR;
  }else{
    p = sqlite3_malloc(sizeof(AsciiTokenizer));
    if( p==0 ){
      rc = SQLITE_NOMEM;
    }else{
      int i;
      memset(p, 0, sizeof(AsciiTokenizer));
      memcpy(p->aTokenChar, aAsciiTokenChar, sizeof(aAsciiTokenChar));
      for(i=0; rc==SQLITE_OK && i<nArg; i+=2){
        const char *zArg = azArg[i+1];
        if( 0==sqlite3_stricmp(azArg[i], "tokenchars") ){
          fts5AsciiAddExceptions(p, zArg, 1);
        }else
        if( 0==sqlite3_stricmp(azArg[i], "separators") ){
          fts5AsciiAddExceptions(p, zArg, 0);
        }else{
          rc = SQLITE_ERROR;
        }
      }
    }
  }

  *ppOut = (Fts5Tokenizer*)p;
  return rc;
}

/*
** Delete a "ascii" tokenizer.
*/
static void fts5AsciiDelete(Fts5Tokenizer *p){
  sqlite3_free(p);
}


static void asciiFold(char *aOut, const char *aIn, int nByte){
  int i;
  for(i=0; i<nByte; i++){
    char c = aIn[i];
    if( c>='A' && c<='Z' ) c += 32;
    aOut[i] = c;
  }
}

/*
** Tokenize some text using the ascii tokenizer.
*/
static int fts5AsciiTokenize(
  Fts5Tokenizer *pTokenizer,
  void *pCtx,
  const char *pText, int nText,
  int (*xToken)(void*, const char*, int nToken, int iStart, int iEnd)
){
  AsciiTokenizer *p = (AsciiTokenizer*)pTokenizer;
  int rc = SQLITE_OK;
  int ie;
  int is = 0;

  char aFold[64];
  int nFold = sizeof(aFold);
  char *pFold = aFold;
  unsigned char *a = p->aTokenChar;

  while( is<nText && rc==SQLITE_OK ){
    int nByte;

    /* Skip any leading divider characters. */
    while( is<nText && ((pText[is]&0x80)==0 && a[(int)pText[is]]==0) ){
      is++;
    }
    if( is==nText ) break;

    /* Count the token characters */
    ie = is+1;
    while( ie<nText && ((pText[ie]&0x80) || a[(int)pText[ie]] ) ){
      ie++;
    }

    /* Fold to lower case */
    nByte = ie-is;
    if( nByte>nFold ){
      if( pFold!=aFold ) sqlite3_free(pFold);
      pFold = sqlite3_malloc(nByte*2);
      if( pFold==0 ){
        rc = SQLITE_NOMEM;
        break;
      }
      nFold = nByte*2;
    }
    asciiFold(pFold, &pText[is], nByte);

    /* Invoke the token callback */
    rc = xToken(pCtx, pFold, nByte, is, ie);
    is = ie+1;
  }
  
  if( pFold!=aFold ) sqlite3_free(pFold);

    while( zIn!=zTerm && (*zIn & 0xc0)==0x80 ){            \
      c = (c<<6) + (0x3f & *(zIn++));                      \
    }                                                      \
    if( c<0x80                                             \
        || (c&0xFFFFF800)==0xD800                          \
        || (c&0xFFFFFFFE)==0xFFFE ){  c = 0xFFFD; }        \
  }


#define WRITE_UTF8(zOut, c) {                          \
  if( c<0x00080 ){                                     \
    *zOut++ = (unsigned char)(c&0xFF);                 \
  }                                                    \
  else if( c<0x00800 ){                                \
    *zOut++ = 0xC0 + (unsigned char)((c>>6)&0x1F);     \

  }                                                    \
}

#endif /* ifndef SQLITE_AMALGAMATION */

typedef struct Unicode61Tokenizer Unicode61Tokenizer;
struct Unicode61Tokenizer {
  unsigned char aTokenChar[128];  /* ASCII range token characters */
  char *aFold;                    /* Buffer to fold text into */
  int nFold;                      /* Size of aFold[] in bytes */
  int bRemoveDiacritic;           /* True if remove_diacritics=1 is set */
  int nException;
  int *aiException;
};

static int fts5UnicodeAddExceptions(
  Unicode61Tokenizer *p,          /* Tokenizer object */

      int nNew = p->nException;
      const unsigned char *zCsr = (const unsigned char*)z;
      const unsigned char *zTerm = (const unsigned char*)&z[n];
      while( zCsr<zTerm ){
        int iCode;
        int bToken;
        READ_UTF8(zCsr, zTerm, iCode);
        if( iCode<128 ){
          p->aTokenChar[iCode] = bTokenChars;
        }else{
          bToken = sqlite3Fts5UnicodeIsalnum(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;
            }
            memmove(&aNew[i+1], &aNew[i], (nNew-i)*sizeof(int));
            aNew[i] = iCode;
            nNew++;
          }
        }
      }
      p->aiException = aNew;
      p->nException = nNew;
    }else{
      rc = SQLITE_NOMEM;
    }

        iHi = iTest-1;
      }
    }
  }

  return 0;
}

/*
** Delete a "unicode61" tokenizer.
*/
static void fts5UnicodeDelete(Fts5Tokenizer *pTok){
  if( pTok ){
    Unicode61Tokenizer *p = (Unicode61Tokenizer*)pTok;
    sqlite3_free(p->aiException);
    sqlite3_free(p->aFold);
    sqlite3_free(p);
  }
  return;
}

/*
** Create a "unicode61" tokenizer.
*/
static int fts5UnicodeCreate(
  void *pCtx, 
  const char **azArg, int nArg,
  Fts5Tokenizer **ppOut
){
  int rc = SQLITE_OK;             /* Return code */
  Unicode61Tokenizer *p = 0;      /* New tokenizer object */ 

  if( nArg%2 ){
    rc = SQLITE_ERROR;
  }else{
    p = (Unicode61Tokenizer*)sqlite3_malloc(sizeof(Unicode61Tokenizer));
    if( p ){
      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;
      }
      for(i=0; rc==SQLITE_OK && i<nArg; i+=2){
        const char *zArg = azArg[i+1];
        if( 0==sqlite3_stricmp(azArg[i], "remove_diacritics") ){
          if( (zArg[0]!='0' && zArg[0]!='1') || zArg[1] ){
            rc = SQLITE_ERROR;
          }
          p->bRemoveDiacritic = (zArg[0]=='1');
        }else
        if( 0==sqlite3_stricmp(azArg[i], "tokenchars") ){
          rc = fts5UnicodeAddExceptions(p, zArg, 1);
        }else
        if( 0==sqlite3_stricmp(azArg[i], "separators") ){
          rc = fts5UnicodeAddExceptions(p, zArg, 0);
        }else{
          rc = SQLITE_ERROR;
        }
      }
    }else{
      rc = SQLITE_NOMEM;
    }
    if( rc!=SQLITE_OK ){
      fts5UnicodeDelete((Fts5Tokenizer*)p);
      p = 0;
    }
    *ppOut = (Fts5Tokenizer*)p;
  }
  return rc;
}











/*
** 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){
  assert( (sqlite3Fts5UnicodeIsalnum(iCode) & 0xFFFFFFFE)==0 );
  return sqlite3Fts5UnicodeIsalnum(iCode) ^ fts5UnicodeIsException(p, iCode);
}




static int fts5UnicodeTokenize(
  Fts5Tokenizer *pTokenizer,
  void *pCtx,
  const char *pText, int nText,
  int (*xToken)(void*, const char*, int nToken, int iStart, int iEnd)
){
  Unicode61Tokenizer *p = (Unicode61Tokenizer*)pTokenizer;



  int rc = SQLITE_OK;
  unsigned char *a = p->aTokenChar;

  unsigned char *zTerm = (unsigned char*)&pText[nText];
  unsigned char *zCsr = (unsigned char *)pText;

  /* Output buffer */
  char *aFold = p->aFold;
  int nFold = p->nFold;

  /* Each iteration of this loop gobbles up a contiguous run of separators,
  ** then the next token.  */
  while( rc==SQLITE_OK ){
    int iCode;                    /* non-ASCII codepoint read from input */
    char *zOut = aFold;
    int is;
    int ie;

    /* Skip any separator characters. */
    while( 1 ){
      if( zCsr>=zTerm ) goto tokenize_done;
      if( *zCsr & 0x80 ) {
        /* A character outside of the ascii range. Skip past it if it is
        ** a separator character. Or break out of the loop if it is not. */
        is = zCsr - (unsigned char*)pText;



        READ_UTF8(zCsr, zTerm, iCode);
        if( fts5UnicodeIsAlnum(p, iCode) ){
          goto non_ascii_tokenchar;
        }
      }else{
        if( a[*zCsr] ){
          is = zCsr - (unsigned char*)pText;
          goto ascii_tokenchar;
        }
        zCsr++;
      }
    }


    /* Run through the tokenchars. Fold them into the output buffer along
    ** the way.  */
    while( zCsr<zTerm ){

      /* Grow the output buffer so that there is sufficient space to fit the
      ** largest possible utf-8 character.  */
      if( (zOut-aFold)+6>nFold ){


        aFold = sqlite3_malloc(nFold*2);
        if( aFold==0 ){
          rc = SQLITE_NOMEM;
          goto tokenize_done;



        }
        memcpy(aFold, p->aFold, nFold);
        sqlite3_free(p->aFold);
        p->aFold = aFold;
        p->nFold = nFold = nFold*2;
      }

      if( *zCsr & 0x80 ){
        /* An non-ascii-range character. Fold it into the output buffer if
        ** it is a token character, or break out of the loop if it is not. */
        READ_UTF8(zCsr, zTerm, iCode);
        if( fts5UnicodeIsAlnum(p,iCode)||sqlite3Fts5UnicodeIsdiacritic(iCode) ){
 non_ascii_tokenchar:
          iCode = sqlite3Fts5UnicodeFold(iCode, p->bRemoveDiacritic);
          if( iCode ) WRITE_UTF8(zOut, iCode);
        }else{
          break;
        }
      }else if( a[*zCsr]==0 ){
        /* An ascii-range separator character. End of token. */
        break; 
      }else{
 ascii_tokenchar:
        if( *zCsr>='A' && *zCsr<='Z' ){
          *zOut++ = *zCsr + 32;
        }else{
          *zOut++ = *zCsr;
        }



        zCsr++;
      }
      ie = zCsr - (unsigned char*)pText;
    }

    /* Invoke the token callback */
    rc = xToken(pCtx, aFold, zOut-aFold, is, ie);
  }
  
 tokenize_done:

  if( rc==SQLITE_DONE ) rc = SQLITE_OK;
  return rc;
}

/**************************************************************************
** Start of porter stemmer implementation.
*/

  int rc = SQLITE_OK;
  PorterTokenizer *pRet;
  void *pUserdata = 0;

  pRet = (PorterTokenizer*)sqlite3_malloc(sizeof(PorterTokenizer));
  if( pRet ){
    memset(pRet, 0, sizeof(PorterTokenizer));
    rc = pApi->xFindTokenizer(pApi, "ascii", &pUserdata, &pRet->tokenizer);
  }else{
    rc = SQLITE_NOMEM;
  }
  if( rc==SQLITE_OK ){
    rc = pRet->tokenizer.xCreate(pUserdata, 0, 0, &pRet->pTokenizer);
  }

** Register all built-in tokenizers with FTS5.
*/
int sqlite3Fts5TokenizerInit(fts5_api *pApi){
  struct BuiltinTokenizer {
    const char *zName;
    fts5_tokenizer x;
  } aBuiltin[] = {

    { "unicode61", {fts5UnicodeCreate, fts5UnicodeDelete, fts5UnicodeTokenize}},
    { "ascii",     {fts5AsciiCreate, fts5AsciiDelete, fts5AsciiTokenize }},
    { "porter",    {fts5PorterCreate, fts5PorterDelete, fts5PorterTokenize }},
  };
  
  int rc = SQLITE_OK;             /* Return code */
  int i;                          /* To iterate through builtin functions */

  for(i=0; rc==SQLITE_OK && i<sizeof(aBuiltin)/sizeof(aBuiltin[0]); i++){
    rc = pApi->xCreateTokenizer(pApi,

    DELETE FROM t1;
    INSERT INTO t1 VALUES('$doc');
    SELECT count(*) FROM t1 WHERE t1 MATCH '$near';
  " $res]
}

execsql { 
  CREATE VIRTUAL TABLE t1 USING fts5(x, tokenize = 'ascii tokenchars .') 
}

do_near_test 1.1 ". . a . . . b . ." { NEAR(a b, 5) } 1
do_near_test 1.2 ". . a . . . b . ." { NEAR(a b, 4) } 1
do_near_test 1.3 ". . a . . . b . ." { NEAR(a b, 3) } 1
do_near_test 1.4 ". . a . . . b . ." { NEAR(a b, 2) } 0

  DROP TABLE ft1;
}
do_execsql_test 1.3 {
  CREATE VIRTUAL TABLE ft1 USING fts5(x, tokenize = 'porter');
  DROP TABLE ft1;
}
do_execsql_test 1.4 {
  CREATE VIRTUAL TABLE ft1 USING fts5(x, tokenize = 'porter ascii');
  DROP TABLE ft1;
}

do_execsql_test 2.0 {
  CREATE VIRTUAL TABLE ft1 USING fts5(x, tokenize=porter);
  INSERT INTO ft1 VALUES('embedded databases');
}

do_catchsql_test 4.2 {
  CREATE VIRTUAL TABLE ft2 USING fts5(x y)
} {1 {parse error in "x y"}}

#-------------------------------------------------------------------------
# Test the "separators" and "tokenchars" options a bit.
#
foreach {tn tokenizer} {1 ascii 2 unicode61} {
  reset_db
  set T "$tokenizer tokenchars ',.:' separators 'xyz'"
  execsql "CREATE VIRTUAL TABLE t1 USING fts5(x, tokenize = \"$T\")"
  do_execsql_test 5.$tn.1 {
    INSERT INTO t1 VALUES('abcxdefyghizjkl.mno,pqr:stu/vwx+yz');
  }
  foreach {tn2 token res} {

    foreach {z s e} [sqlite3_fts5_tokenize db {$tokenizer} {$input}] {
      lappend ret [set z]
    }
    set ret
  }] [list {*}$output]]
}

foreach {tn t} {1 ascii 2 unicode61} {
  tokenize_test 1.$tn.0 $t {A B C D} {a b c d}
  tokenize_test 1.$tn.1 $t {May you share freely,} {may you share freely}
  tokenize_test 1.$tn.2 $t {..May...you.shAre.freely} {may you share freely}
  tokenize_test 1.$tn.3 $t {} {}
}

#-------------------------------------------------------------------------
# Check that "unicode61" really is the default tokenizer.
#

do_execsql_test 2.0 "
  CREATE VIRTUAL TABLE t1 USING fts5(x);
  CREATE VIRTUAL TABLE t2 USING fts5(x, tokenize = unicode61);
  CREATE VIRTUAL TABLE t3 USING fts5(x, tokenize = ascii);
  INSERT INTO t1 VALUES('\xC0\xC8\xCC');
  INSERT INTO t2 VALUES('\xC0\xC8\xCC');
  INSERT INTO t3 VALUES('\xC0\xC8\xCC');
"
breakpoint
do_execsql_test 2.1 "
  SELECT 't1' FROM t1 WHERE t1 MATCH '\xE0\xE8\xEC';
  SELECT 't2' FROM t2 WHERE t2 MATCH '\xE0\xE8\xEC';
  SELECT 't3' FROM t3 WHERE t3 MATCH '\xE0\xE8\xEC';
" {t1 t2}


finish_test