SQLite

Check-in [366a70b086]
Login

Many hyperlinks are disabled.
Use anonymous login to enable hyperlinks.

Overview
Comment:Allow virtual tables to contain multiple full-text-indexed columns. Added a magic column "_all" which can be used for querying all columns in a table at once.

For now, each posting list stores position/offset information for multiple columns. We may implement separate posting lists for separate columns at some future point. (CVS 3408)

Downloads: Tarball | ZIP archive
Timelines: family | ancestors | descendants | both | trunk
Files: files | file ages | folders
SHA1: 366a70b086c817bddecd83053472ec76ef20f309
User & Date: adamd 2006-09-13 02:18:20.000
Context
2006-09-13
12:36
Fix the FTS1 test cases and add new tests. Comments added to the FTS1 code. (CVS 3409) (check-in: 528036c828 user: drh tags: trunk)
02:18
Allow virtual tables to contain multiple full-text-indexed columns. Added a magic column "_all" which can be used for querying all columns in a table at once.

For now, each posting list stores position/offset information for multiple columns. We may implement separate posting lists for separate columns at some future point. (CVS 3408) (check-in: 366a70b086 user: adamd tags: trunk)

2006-09-12
23:36
Answer queries for a particular rowid in a full-text table by looking up that rowid directly rather than by performing a table scan. (CVS 3407) (check-in: 877d5558b1 user: adamd tags: trunk)
Changes
Unified Diff Ignore Whitespace Patch
Changes to ext/fts1/fts1.c.
35
36
37
38
39
40
41


















42
43
44
45
46
47
48
#if 0
# define TRACE(A)  printf A; fflush(stdout)
#else
# define TRACE(A)
#endif

/* utility functions */



















/* We encode variable-length integers in little-endian order using seven bits
 * per byte as follows:
**
** KEY:
**         A = 0xxxxxxx    7 bits of data and one flag bit
**         B = 1xxxxxxx    7 bits of data and one flag bit







>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>







35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
#if 0
# define TRACE(A)  printf A; fflush(stdout)
#else
# define TRACE(A)
#endif

/* utility functions */

typedef struct StringBuffer {
  int len;  /* length, not including null terminator */
  char *s;
} StringBuffer;

void initStringBuffer(StringBuffer *sb){
  sb->len = 0;
  sb->s = malloc(1);
  sb->s[0] = '\0';
}

void append(StringBuffer *sb, const char *zFrom){
  int nFrom = strlen(zFrom);
  sb->s = realloc(sb->s, sb->len + nFrom + 1);
  strcpy(sb->s + sb->len, zFrom);
  sb->len += nFrom;
}

/* We encode variable-length integers in little-endian order using seven bits
 * per byte as follows:
**
** KEY:
**         A = 0xxxxxxx    7 bits of data and one flag bit
**         B = 1xxxxxxx    7 bits of data and one flag bit
103
104
105
106
107
108
109
110
111
112
113
114
115
116





117
118
119
120
121
122
123
 * A document list holds a sorted list of varint-encoded document IDs.
 *
 * A doclist with type DL_POSITIONS_OFFSETS is stored like this:
 *
 * array {
 *   varint docid;
 *   array {
 *     varint position;     (delta from previous position plus 1, or 0 for end)
 *     varint startOffset;  (delta from previous startOffset)
 *     varint endOffset;    (delta from startOffset)
 *   }
 * }
 *
 * Here, array { X } means zero or more occurrences of X, adjacent in memory.





 *
 * A doclist with type DL_POSITIONS is like the above, but holds only docids
 * and positions without offset information.
 *
 * A doclist with type DL_DOCIDS is like the above, but holds only docids
 * without positions or offset information.
 *







|






>
>
>
>
>







121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
 * A document list holds a sorted list of varint-encoded document IDs.
 *
 * A doclist with type DL_POSITIONS_OFFSETS is stored like this:
 *
 * array {
 *   varint docid;
 *   array {
 *     varint position;     (delta from previous position plus POS_BASE)
 *     varint startOffset;  (delta from previous startOffset)
 *     varint endOffset;    (delta from startOffset)
 *   }
 * }
 *
 * Here, array { X } means zero or more occurrences of X, adjacent in memory.
 *
 * A position list may hold positions for text in multiple columns.  A position
 * POS_COLUMN is followed by a varint containing the index of the column for
 * following positions in the list.  Any positions appearing before any
 * occurrences of POS_COLUMN are for column 0.
 *
 * A doclist with type DL_POSITIONS is like the above, but holds only docids
 * and positions without offset information.
 *
 * A doclist with type DL_DOCIDS is like the above, but holds only docids
 * without positions or offset information.
 *
144
145
146
147
148
149
150

151
152
153






154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
  DL_POSITIONS_OFFSETS    /* docids + positions + offsets */
} DocListType;

typedef struct DocList {
  char *pData;
  int nData;
  DocListType iType;

  int iLastPos;       /* the last position written */
  int iLastOffset;    /* the last start offset written */
} DocList;







/* Initialize a new DocList to hold the given data. */
static void docListInit(DocList *d, DocListType iType,
                        const char *pData, int nData){
  d->nData = nData;
  if( nData>0 ){
    d->pData = malloc(nData);
    memcpy(d->pData, pData, nData);
  } else {
    d->pData = NULL;
  }
  d->iType = iType;
  d->iLastPos = 0;
  d->iLastOffset = 0;
}

/* Create a new dynamically-allocated DocList. */
static DocList *docListNew(DocListType iType){
  DocList *d = (DocList *) malloc(sizeof(DocList));
  docListInit(d, iType, 0, 0);
  return d;







>



>
>
>
>
>
>












|
|







167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
  DL_POSITIONS_OFFSETS    /* docids + positions + offsets */
} DocListType;

typedef struct DocList {
  char *pData;
  int nData;
  DocListType iType;
  int iLastColumn;    /* the last column written */
  int iLastPos;       /* the last position written */
  int iLastOffset;    /* the last start offset written */
} DocList;

enum {
  POS_END = 0,        /* end of this position list */
  POS_COLUMN,         /* followed by new column number */
  POS_BASE
};

/* Initialize a new DocList to hold the given data. */
static void docListInit(DocList *d, DocListType iType,
                        const char *pData, int nData){
  d->nData = nData;
  if( nData>0 ){
    d->pData = malloc(nData);
    memcpy(d->pData, pData, nData);
  } else {
    d->pData = NULL;
  }
  d->iType = iType;
  d->iLastColumn = 0;
  d->iLastPos = d->iLastOffset = 0;
}

/* Create a new dynamically-allocated DocList. */
static DocList *docListNew(DocListType iType){
  DocList *d = (DocList *) malloc(sizeof(DocList));
  docListInit(d, iType, 0, 0);
  return d;
198
199
200
201
202
203
204
205

206
207
208
209
210
211










212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230


231
232


233

234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253

254
255
256
257
258
259
260
261
262
263
264

265
266
267
268
269
270
271
  memcpy(d->pData + d->nData, c, n);
  d->nData += n;
}

static void docListAddDocid(DocList *d, sqlite_int64 iDocid){
  appendVarint(d, iDocid);
  if( d->iType>=DL_POSITIONS ){
    appendVarint(d, 0);  /* initially empty position list */

    d->iLastPos = 0;
  }
}

/* helper function for docListAddPos and docListAddPosOffset */
static void addPos(DocList *d, int iPos) {










  appendVarint(d, iPos-d->iLastPos+1);
  d->iLastPos = iPos;
}

/* Add a position to the last position list in a doclist. */
static void docListAddPos(DocList *d, int iPos){
  assert( d->iType==DL_POSITIONS );
  assert( d->nData>0 );
  --d->nData;  /* remove previous terminator */
  addPos(d, iPos);
  appendVarint(d, 0);  /* add new terminator */
}

static void docListAddPosOffset(DocList *d, int iPos,
                                int iStartOffset, int iEndOffset){
  assert( d->iType==DL_POSITIONS_OFFSETS );
  assert( d->nData>0 );
  --d->nData;  /* remove previous terminator */
  addPos(d, iPos);


  appendVarint(d, iStartOffset-d->iLastOffset);
  d->iLastOffset = iStartOffset;


  appendVarint(d, iEndOffset-iStartOffset);

  appendVarint(d, 0);  /* add new terminator */
}

/*
** A DocListReader object is a cursor into a doclist.  Initialize
** the cursor to the beginning of the doclist by calling readerInit().
** Then use routines
**
**      peekDocid()
**      readDocid()
**      readPosition()
**      skipPositionList()
**      and so forth...
**
** to read information out of the doclist.  When we reach the end
** of the doclist, atEnd() returns TRUE.
*/
typedef struct DocListReader {
  DocList *pDoclist;  /* The document list we are stepping through */
  char *p;            /* Pointer to next unread byte in the doclist */

  int iLastPos;  /* the last position read, or -1 when not in a position list */
} DocListReader;

/*
** Initialize the DocListReader r to point to the beginning of pDoclist.
*/
static void readerInit(DocListReader *r, DocList *pDoclist){
  r->pDoclist = pDoclist;
  if( pDoclist!=NULL ){
    r->p = pDoclist->pData;
  }

  r->iLastPos = -1;
}

/*
** Return TRUE if we have reached then end of pReader and there is
** nothing else left to read.
*/







|
>
|




|
>
>
>
>
>
>
>
>
>
>
|




|

<
<
|
|


|


<
<
|
>
>


>
>

>
|



















>











>







228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259


260
261
262
263
264
265
266


267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
  memcpy(d->pData + d->nData, c, n);
  d->nData += n;
}

static void docListAddDocid(DocList *d, sqlite_int64 iDocid){
  appendVarint(d, iDocid);
  if( d->iType>=DL_POSITIONS ){
    appendVarint(d, POS_END);  /* initially empty position list */
    d->iLastColumn = 0;
    d->iLastPos = d->iLastOffset = 0;
  }
}

/* helper function for docListAddPos and docListAddPosOffset */
static void addPos(DocList *d, int iColumn, int iPos){
  assert( d->nData>0 );
  --d->nData;  /* remove previous terminator */
  if( iColumn!=d->iLastColumn ){
    assert( iColumn>d->iLastColumn );
    appendVarint(d, POS_COLUMN);
    appendVarint(d, iColumn);
    d->iLastColumn = iColumn;
    d->iLastPos = d->iLastOffset = 0;
  }
  assert( iPos>=d->iLastPos );
  appendVarint(d, iPos-d->iLastPos+POS_BASE);
  d->iLastPos = iPos;
}

/* Add a position to the last position list in a doclist. */
static void docListAddPos(DocList *d, int iColumn, int iPos){
  assert( d->iType==DL_POSITIONS );


  addPos(d, iColumn, iPos);
  appendVarint(d, POS_END);  /* add new terminator */
}

static void docListAddPosOffset(DocList *d, int iColumn, int iPos,
                                int iStartOffset, int iEndOffset){
  assert( d->iType==DL_POSITIONS_OFFSETS );


  addPos(d, iColumn, iPos);

  assert( iStartOffset>=d->iLastOffset );
  appendVarint(d, iStartOffset-d->iLastOffset);
  d->iLastOffset = iStartOffset;

  assert( iEndOffset>=iStartOffset );
  appendVarint(d, iEndOffset-iStartOffset);

  appendVarint(d, POS_END);  /* add new terminator */
}

/*
** A DocListReader object is a cursor into a doclist.  Initialize
** the cursor to the beginning of the doclist by calling readerInit().
** Then use routines
**
**      peekDocid()
**      readDocid()
**      readPosition()
**      skipPositionList()
**      and so forth...
**
** to read information out of the doclist.  When we reach the end
** of the doclist, atEnd() returns TRUE.
*/
typedef struct DocListReader {
  DocList *pDoclist;  /* The document list we are stepping through */
  char *p;            /* Pointer to next unread byte in the doclist */
  int iLastColumn;
  int iLastPos;  /* the last position read, or -1 when not in a position list */
} DocListReader;

/*
** Initialize the DocListReader r to point to the beginning of pDoclist.
*/
static void readerInit(DocListReader *r, DocList *pDoclist){
  r->pDoclist = pDoclist;
  if( pDoclist!=NULL ){
    r->p = pDoclist->pData;
  }
  r->iLastColumn = -1;
  r->iLastPos = -1;
}

/*
** Return TRUE if we have reached then end of pReader and there is
** nothing else left to read.
*/
287
288
289
290
291
292
293

294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315

316
317






318
319
320
321
322
323
324

325
326
327
328
329
330
331

332
333
334
335
336
337
338
339
*/
static sqlite_int64 readDocid(DocListReader *pReader){
  sqlite_int64 ret;
  assert( !atEnd(pReader) );
  assert( pReader->iLastPos==-1 );
  pReader->p += getVarint(pReader->p, &ret);
  if( pReader->pDoclist->iType>=DL_POSITIONS ){

    pReader->iLastPos = 0;
  }
  return ret;
}

/* Read the next position from a position list.
 * Returns the position, or -1 at the end of the list. */
static int readPosition(DocListReader *pReader){
  int i;
  int iType = pReader->pDoclist->iType;

  if( pReader->iLastPos==-1 ){
    return -1;
  }
  assert( !atEnd(pReader) );

  if( iType<DL_POSITIONS ){
    return -1;
  }
  pReader->p += getVarint32(pReader->p, &i);
  if( i==0 ){
    pReader->iLastPos = -1;

    return -1;
  }






  pReader->iLastPos += ((int) i)-1;
  if( iType>=DL_POSITIONS_OFFSETS ){
    /* Skip over offsets, ignoring them for now. */
    int iStart, iEnd;
    pReader->p += getVarint32(pReader->p, &iStart);
    pReader->p += getVarint32(pReader->p, &iEnd);
  }

  return pReader->iLastPos;
}

/* Skip past the end of a position list. */
static void skipPositionList(DocListReader *pReader){
  DocList *p = pReader->pDoclist;
  if( p && p->iType>=DL_POSITIONS ){

    while( readPosition(pReader)!=-1 ){}
  }
}

/* Skip over a docid, including its position list if the doclist has
 * positions. */
static void skipDocument(DocListReader *pReader){
  readDocid(pReader);







>





|

|












|
|
>


>
>
>
>
>
>
|






>







>
|







331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
*/
static sqlite_int64 readDocid(DocListReader *pReader){
  sqlite_int64 ret;
  assert( !atEnd(pReader) );
  assert( pReader->iLastPos==-1 );
  pReader->p += getVarint(pReader->p, &ret);
  if( pReader->pDoclist->iType>=DL_POSITIONS ){
    pReader->iLastColumn = 0;
    pReader->iLastPos = 0;
  }
  return ret;
}

/* Read the next position and column index from a position list.
 * Returns the position, or -1 at the end of the list. */
static int readPosition(DocListReader *pReader, int *iColumn){
  int i;
  int iType = pReader->pDoclist->iType;

  if( pReader->iLastPos==-1 ){
    return -1;
  }
  assert( !atEnd(pReader) );

  if( iType<DL_POSITIONS ){
    return -1;
  }
  pReader->p += getVarint32(pReader->p, &i);
  if( i==POS_END ){
    pReader->iLastColumn = pReader->iLastPos = -1;
    *iColumn = -1;
    return -1;
  }
  if( i==POS_COLUMN ){
    pReader->p += getVarint32(pReader->p, &pReader->iLastColumn);
    pReader->iLastPos = 0;
    pReader->p += getVarint32(pReader->p, &i);
    assert( i>=POS_BASE );
  }
  pReader->iLastPos += ((int) i)-POS_BASE;
  if( iType>=DL_POSITIONS_OFFSETS ){
    /* Skip over offsets, ignoring them for now. */
    int iStart, iEnd;
    pReader->p += getVarint32(pReader->p, &iStart);
    pReader->p += getVarint32(pReader->p, &iEnd);
  }
  *iColumn = pReader->iLastColumn;
  return pReader->iLastPos;
}

/* Skip past the end of a position list. */
static void skipPositionList(DocListReader *pReader){
  DocList *p = pReader->pDoclist;
  if( p && p->iType>=DL_POSITIONS ){
    int iColumn;
    while( readPosition(pReader, &iColumn)!=-1 ){}
  }
}

/* Skip over a docid, including its position list if the doclist has
 * positions. */
static void skipDocument(DocListReader *pReader){
  readDocid(pReader);
388
389
390
391
392
393
394





























395
396
397
398
399
400
401
      printf(")");
    }
  }
  printf("\n");
  fflush(stdout);
}
#endif /* SQLITE_DEBUG */






























/* Helper function for docListUpdate() and docListAccumulate().
** Splices a doclist element into the doclist represented by r,
** leaving r pointing after the newly spliced element.
*/
static void docListSpliceElement(DocListReader *r, sqlite_int64 iDocid,
                                 const char *pSource, int nSource){







>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>







442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
      printf(")");
    }
  }
  printf("\n");
  fflush(stdout);
}
#endif /* SQLITE_DEBUG */

/* Trim the given doclist to contain only positions in column [iRestrictColumn],
 * discarding any docids without any remaining positions. */
static void docListRestrictColumn(DocList *in, int iRestrictColumn){
  DocListReader r;
  DocList out;

  assert( in->iType>=DL_POSITIONS );
  readerInit(&r, in);
  docListInit(&out, DL_POSITIONS, NULL, 0);

  while( !atEnd(&r) ){
    sqlite_int64 iDocid = readDocid(&r);
    int match = 0;
    int iPos, iColumn;
    while( (iPos = readPosition(&r, &iColumn)) != -1 ){
      if( iColumn==iRestrictColumn ){
        if( !match ){
          docListAddDocid(&out, iDocid);
          match = 1;
        }
        docListAddPos(&out, iColumn, iPos);
      }
    }
  }

  docListDestroy(in);
  *in = out;
}

/* Helper function for docListUpdate() and docListAccumulate().
** Splices a doclist element into the doclist represented by r,
** leaving r pointing after the newly spliced element.
*/
static void docListSpliceElement(DocListReader *r, sqlite_int64 iDocid,
                                 const char *pSource, int nSource){
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529

530
531
532
533
534
535
536
537
538
539
*/
static void mergePosList(
  DocListReader *pLeft,    /* Left position list */
  DocListReader *pRight,   /* Right position list */
  sqlite_int64 iDocid,     /* The docid from pLeft and pRight */
  DocList *pOut            /* Write the merged document record here */
){
  int iLeftPos = readPosition(pLeft);
  int iRightPos = readPosition(pRight);
  int match = 0;

  /* Loop until we've reached the end of both position lists. */
  while( iLeftPos!=-1 && iRightPos!=-1 ){
    if( iLeftPos+1==iRightPos ){
      if( !match ){
        docListAddDocid(pOut, iDocid);
        match = 1;
      }
      if( pOut->iType>=DL_POSITIONS ){
        docListAddPos(pOut, iRightPos);
      }
      iLeftPos = readPosition(pLeft);
      iRightPos = readPosition(pRight);
    }else if( iRightPos<iLeftPos+1 ){

      iRightPos = readPosition(pRight);
    }else{
      iLeftPos = readPosition(pLeft);
    }
  }
  if( iLeftPos>=0 ) skipPositionList(pLeft);
  if( iRightPos>=0 ) skipPositionList(pRight);
}

/* We have two doclists:  pLeft and pRight.







|
|




|





|

|
|
|
>
|

|







589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
*/
static void mergePosList(
  DocListReader *pLeft,    /* Left position list */
  DocListReader *pRight,   /* Right position list */
  sqlite_int64 iDocid,     /* The docid from pLeft and pRight */
  DocList *pOut            /* Write the merged document record here */
){
  int iLeftCol, iLeftPos = readPosition(pLeft, &iLeftCol);
  int iRightCol, iRightPos = readPosition(pRight, &iRightCol);
  int match = 0;

  /* Loop until we've reached the end of both position lists. */
  while( iLeftPos!=-1 && iRightPos!=-1 ){
    if( iLeftCol==iRightCol && iLeftPos+1==iRightPos ){
      if( !match ){
        docListAddDocid(pOut, iDocid);
        match = 1;
      }
      if( pOut->iType>=DL_POSITIONS ){
        docListAddPos(pOut, iRightCol, iRightPos);
      }
      iLeftPos = readPosition(pLeft, &iLeftCol);
      iRightPos = readPosition(pRight, &iRightCol);
    }else if( iRightCol<iLeftCol ||
              iRightCol==iLeftCol && iRightPos<iLeftPos+1 ){
      iRightPos = readPosition(pRight, &iRightCol);
    }else{
      iLeftPos = readPosition(pLeft, &iLeftCol);
    }
  }
  if( iLeftPos>=0 ) skipPositionList(pLeft);
  if( iRightPos>=0 ) skipPositionList(pRight);
}

/* We have two doclists:  pLeft and pRight.
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699







700
701
702
703
704
705
706
  }
  while( docidLeft>0 ){
    docListAddDocid(pOut, docidLeft);
    docidLeft = nextValidDocid(&left);
  }
}

/* Duplicate a string; the caller must free() the returned string.
 * (We don't use strdup() since it's not part of the standard C library and
 * may not be available everywhere.) */
static char *string_dup(const char *s){
  int n = strlen(s);
  char *str = malloc(n + 1);
  memcpy(str, s, n);
  str[n] = '\0';
  return str;
}








/* Format a string, replacing each occurrence of the % character with
 * zName.  This may be more convenient than sqlite_mprintf()
 * when one string is used repeatedly in a format string.
 * The caller must free() the returned string. */
static char *string_format(const char *zFormat, const char *zName){
  const char *p;







<
<
<
|
<





>
>
>
>
>
>
>







767
768
769
770
771
772
773



774

775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
  }
  while( docidLeft>0 ){
    docListAddDocid(pOut, docidLeft);
    docidLeft = nextValidDocid(&left);
  }
}




static char *string_dup_n(const char *s, int n){

  char *str = malloc(n + 1);
  memcpy(str, s, n);
  str[n] = '\0';
  return str;
}

/* Duplicate a string; the caller must free() the returned string.
 * (We don't use strdup() since it's not part of the standard C library and
 * may not be available everywhere.) */
static char *string_dup(const char *s){
  return string_dup_n(s, strlen(s));
}

/* Format a string, replacing each occurrence of the % character with
 * zName.  This may be more convenient than sqlite_mprintf()
 * when one string is used repeatedly in a format string.
 * The caller must free() the returned string. */
static char *string_format(const char *zFormat, const char *zName){
  const char *p;
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
  free(zCommand);
  return rc;
}

/* end utility functions */

typedef enum QueryType {
  QUERY_GENERIC,     /* table scan */
  QUERY_ROWID,       /* lookup by rowid */
  QUERY_FULLTEXT    /* full-text search */
} QueryType;

/* TODO(shess) CHUNK_MAX controls how much data we allow in segment 0
** before we start aggregating into larger segments.  Lower CHUNK_MAX
** means that for a given input we have more individual segments per
** term, which means more rows in the table and a bigger index (due to
** both more rows and bigger rowids).  But it also reduces the average







|
|
|







834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
  free(zCommand);
  return rc;
}

/* end utility functions */

typedef enum QueryType {
  QUERY_GENERIC,   /* table scan */
  QUERY_ROWID,     /* lookup by rowid */
  QUERY_FULLTEXT   /* QUERY_FULLTEXT + [i] is a full-text search for column i*/
} QueryType;

/* TODO(shess) CHUNK_MAX controls how much data we allow in segment 0
** before we start aggregating into larger segments.  Lower CHUNK_MAX
** means that for a given input we have more individual segments per
** term, which means more rows in the table and a bigger index (due to
** both more rows and bigger rowids).  But it also reduces the average
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810


811
812
813
814
815
816
817
818
819
820
/* These must exactly match the enum above. */
/* TODO(adam): Is there some risk that a statement (in particular,
** pTermSelectStmt) will be used in two cursors at once, e.g.  if a
** query joins a virtual table to itself?  If so perhaps we should
** move some of these to the cursor object.
*/
static const char *const fulltext_zStatement[MAX_STMT] = {
  /* CONTENT_INSERT */ "insert into %_content (rowid, content) values (?, ?)",
  /* CONTENT_SELECT */ "select content from %_content where rowid = ?",
  /* CONTENT_DELETE */ "delete from %_content where rowid = ?",

  /* TERM_SELECT */
  "select rowid, doclist from %_term where term = ? and segment = ?",
  /* TERM_SELECT_ALL */
  "select doclist from %_term where term = ? order by segment",
  /* TERM_INSERT */
  "insert into %_term (rowid, term, segment, doclist) values (?, ?, ?, ?)",
  /* TERM_UPDATE */ "update %_term set doclist = ? where rowid = ?",
  /* TERM_DELETE */ "delete from %_term where rowid = ?",
};

typedef struct fulltext_vtab {
  sqlite3_vtab base;
  sqlite3 *db;
  const char *zName;               /* virtual table name */


  sqlite3_tokenizer *pTokenizer;   /* tokenizer for inserts and queries */
  int nColumn;                     /* Number of columns */
  char **azColumn;                 /* Names of all columns */

  /* Precompiled statements which we keep as long as the table is
  ** open.
  */
  sqlite3_stmt *pFulltextStatements[MAX_STMT];
} fulltext_vtab;








|
|
















>
>

<
<







873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900


901
902
903
904
905
906
907
/* These must exactly match the enum above. */
/* TODO(adam): Is there some risk that a statement (in particular,
** pTermSelectStmt) will be used in two cursors at once, e.g.  if a
** query joins a virtual table to itself?  If so perhaps we should
** move some of these to the cursor object.
*/
static const char *const fulltext_zStatement[MAX_STMT] = {
  /* CONTENT_INSERT */ NULL,  /* generated in contentInsertStatement() */
  /* CONTENT_SELECT */ "select * from %_content where rowid = ?",
  /* CONTENT_DELETE */ "delete from %_content where rowid = ?",

  /* TERM_SELECT */
  "select rowid, doclist from %_term where term = ? and segment = ?",
  /* TERM_SELECT_ALL */
  "select doclist from %_term where term = ? order by segment",
  /* TERM_INSERT */
  "insert into %_term (rowid, term, segment, doclist) values (?, ?, ?, ?)",
  /* TERM_UPDATE */ "update %_term set doclist = ? where rowid = ?",
  /* TERM_DELETE */ "delete from %_term where rowid = ?",
};

typedef struct fulltext_vtab {
  sqlite3_vtab base;
  sqlite3 *db;
  const char *zName;               /* virtual table name */
  int nColumns;                    /* number of columns in virtual table */
  const char *zColumnNames;        /* all column names, separated by commas */
  sqlite3_tokenizer *pTokenizer;   /* tokenizer for inserts and queries */



  /* Precompiled statements which we keep as long as the table is
  ** open.
  */
  sqlite3_stmt *pFulltextStatements[MAX_STMT];
} fulltext_vtab;

829
830
831
832
833
834
835

















836
837
838
839
840
841
842
843
844


845
846

847
848
849
850
851
852
853
} fulltext_cursor;

static struct fulltext_vtab *cursor_vtab(fulltext_cursor *c){
  return (fulltext_vtab *) c->base.pVtab;
}

static const sqlite3_module fulltextModule;   /* forward declaration */


















/* Puts a freshly-prepared statement determined by iStmt in *ppStmt.
** If the indicated statement has never been prepared, it is prepared
** and cached, otherwise the cached version is reset.
*/
static int sql_get_statement(fulltext_vtab *v, fulltext_statement iStmt,
                             sqlite3_stmt **ppStmt){
  assert( iStmt<MAX_STMT );
  if( v->pFulltextStatements[iStmt]==NULL ){


    int rc = sql_prepare(v->db, v->zName, &v->pFulltextStatements[iStmt],
                         fulltext_zStatement[iStmt]);

    if( rc!=SQLITE_OK ) return rc;
  } else {
    int rc = sqlite3_reset(v->pFulltextStatements[iStmt]);
    if( rc!=SQLITE_OK ) return rc;
  }

  *ppStmt = v->pFulltextStatements[iStmt];







>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>









>
>

|
>







916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
} fulltext_cursor;

static struct fulltext_vtab *cursor_vtab(fulltext_cursor *c){
  return (fulltext_vtab *) c->base.pVtab;
}

static const sqlite3_module fulltextModule;   /* forward declaration */

/* Return a dynamically generated statement of the form
 *   insert into %_content (rowid, ...) values (?, ...)
 */
static const char *contentInsertStatement(fulltext_vtab *v){
  StringBuffer sb;
  int i;

  initStringBuffer(&sb);
  append(&sb, "insert into %_content (rowid, ");
  append(&sb, v->zColumnNames);
  append(&sb, ") values (?");
  for(i=0; i<v->nColumns; ++i)
    append(&sb, ", ?");
  append(&sb, ")");
  return sb.s;
}

/* Puts a freshly-prepared statement determined by iStmt in *ppStmt.
** If the indicated statement has never been prepared, it is prepared
** and cached, otherwise the cached version is reset.
*/
static int sql_get_statement(fulltext_vtab *v, fulltext_statement iStmt,
                             sqlite3_stmt **ppStmt){
  assert( iStmt<MAX_STMT );
  if( v->pFulltextStatements[iStmt]==NULL ){
    const char *zStmt = iStmt==CONTENT_INSERT_STMT ? contentInsertStatement(v) : 
                                                     fulltext_zStatement[iStmt];
    int rc = sql_prepare(v->db, v->zName, &v->pFulltextStatements[iStmt],
                         zStmt);
    if( iStmt==CONTENT_INSERT_STMT ) free((void *) zStmt);
    if( rc!=SQLITE_OK ) return rc;
  } else {
    int rc = sqlite3_reset(v->pFulltextStatements[iStmt]);
    if( rc!=SQLITE_OK ) return rc;
  }

  *ppStmt = v->pFulltextStatements[iStmt];
901
902
903
904
905
906
907
908
909
910
911

912
913
914
915
916
917
918
919
920

921
922
923
924









925
926



927
928
929






930
931
932
933
934
935
936
937
938
939
940
941

942
943
944
945
946


947
948

949
950
951
952
953
954
955
static int sql_single_step_statement(fulltext_vtab *v,
                                     fulltext_statement iStmt,
                                     sqlite3_stmt **ppStmt){
  int rc = sql_step_statement(v, iStmt, ppStmt);
  return (rc==SQLITE_DONE) ? SQLITE_OK : rc;
}

/* insert into %_content (rowid, content) values ([rowid], [zContent]) */
static int content_insert(fulltext_vtab *v, sqlite3_value *rowid,
                          const char *pContent, int nContent){
  sqlite3_stmt *s;

  int rc = sql_get_statement(v, CONTENT_INSERT_STMT, &s);
  if( rc!=SQLITE_OK ) return rc;

  rc = sqlite3_bind_value(s, 1, rowid);
  if( rc!=SQLITE_OK ) return rc;

  assert( nContent>=0 );
  rc = sqlite3_bind_text(s, 2, pContent, nContent, SQLITE_STATIC);
  if( rc!=SQLITE_OK ) return rc;


  return sql_single_step_statement(v, CONTENT_INSERT_STMT, &s);
}










/* select content from %_content where rowid = [iRow]
 * The caller must delete the returned string. */



static int content_select(fulltext_vtab *v, sqlite_int64 iRow,
                          char **ppContent, int *pnContent){
  sqlite3_stmt *s;






  int rc = sql_get_statement(v, CONTENT_SELECT_STMT, &s);
  if( rc!=SQLITE_OK ) return rc;

  rc = sqlite3_bind_int64(s, 1, iRow);
  if( rc!=SQLITE_OK ) return rc;

  rc = sql_step_statement(v, CONTENT_SELECT_STMT, &s);
  if( rc!=SQLITE_ROW ) return rc;

  *pnContent = sqlite3_column_bytes(s, 0);
  *ppContent = malloc(*pnContent);
  memcpy(*ppContent, sqlite3_column_blob(s, 0), *pnContent);


  /* We expect only one row.  We must execute another sqlite3_step()
   * to complete the iteration; otherwise the table will remain locked. */
  rc = sqlite3_step(s);
  if( rc==SQLITE_DONE ) return SQLITE_OK;



  free(*ppContent);

  return rc;
}

/* delete from %_content where rowid = [iRow ] */
static int content_delete(fulltext_vtab *v, sqlite_int64 iRow){
  sqlite3_stmt *s;
  int rc = sql_get_statement(v, CONTENT_DELETE_STMT, &s);







|

|

>






|
|
|
>




>
>
>
>
>
>
>
>
>
|
|
>
>
>

|

>
>
>
>
>
>
|








|
|
|
>




|
>
>
|
|
>







1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
static int sql_single_step_statement(fulltext_vtab *v,
                                     fulltext_statement iStmt,
                                     sqlite3_stmt **ppStmt){
  int rc = sql_step_statement(v, iStmt, ppStmt);
  return (rc==SQLITE_DONE) ? SQLITE_OK : rc;
}

/* insert into %_content (rowid, ...) values ([rowid], [pValues]) */
static int content_insert(fulltext_vtab *v, sqlite3_value *rowid,
                          sqlite3_value **pValues){
  sqlite3_stmt *s;
  int i;
  int rc = sql_get_statement(v, CONTENT_INSERT_STMT, &s);
  if( rc!=SQLITE_OK ) return rc;

  rc = sqlite3_bind_value(s, 1, rowid);
  if( rc!=SQLITE_OK ) return rc;

  for(i=0; i<v->nColumns; ++i){
    rc = sqlite3_bind_value(s, 2+i, pValues[i]);
    if( rc!=SQLITE_OK ) return rc;
  }

  return sql_single_step_statement(v, CONTENT_INSERT_STMT, &s);
}

void freeStringArray(int nString, const char **pString){
  int i;

  for (i=0 ; i < nString ; ++i) {
    free((void *) pString[i]);
  }
  free((void *) pString);
}

/* select * from %_content where rowid = [iRow]
 * The caller must delete the returned array and all strings in it.
 *
 * TODO: Perhaps we should return pointer/length strings here for consistency
 * with other code which uses pointer/length. */
static int content_select(fulltext_vtab *v, sqlite_int64 iRow,
                          const char ***pValues){
  sqlite3_stmt *s;
  const char **values;
  int i;
  int rc;

  *pValues = NULL;

  rc = sql_get_statement(v, CONTENT_SELECT_STMT, &s);
  if( rc!=SQLITE_OK ) return rc;

  rc = sqlite3_bind_int64(s, 1, iRow);
  if( rc!=SQLITE_OK ) return rc;

  rc = sql_step_statement(v, CONTENT_SELECT_STMT, &s);
  if( rc!=SQLITE_ROW ) return rc;

  values = (const char **) malloc(v->nColumns * sizeof(const char *));
  for(i=0; i<v->nColumns; ++i){
    values[i] = string_dup(sqlite3_column_text(s, i));
  }

  /* We expect only one row.  We must execute another sqlite3_step()
   * to complete the iteration; otherwise the table will remain locked. */
  rc = sqlite3_step(s);
  if( rc==SQLITE_DONE ){
    *pValues = values;
    return SQLITE_OK;
  }

  freeStringArray(v->nColumns, values);
  return rc;
}

/* delete from %_content where rowid = [iRow ] */
static int content_delete(fulltext_vtab *v, sqlite_int64 iRow){
  sqlite3_stmt *s;
  int rc = sql_get_statement(v, CONTENT_DELETE_STMT, &s);
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
}

/* Load the segment doclists for term pTerm and merge them in
** appropriate order into out.  Returns SQLITE_OK if successful.  If
** there are no segments for pTerm, successfully returns an empty
** doclist in out.
*/
static int term_select_all(fulltext_vtab *v, const char *pTerm, int nTerm,
                           DocList *out){
  DocList doclist;
  sqlite3_stmt *s;
  int rc = sql_get_statement(v, TERM_SELECT_ALL_STMT, &s);
  if( rc!=SQLITE_OK ) return rc;

  rc = sqlite3_bind_text(s, 1, pTerm, nTerm, SQLITE_STATIC);
  if( rc!=SQLITE_OK ) return rc;







|
|







1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
}

/* Load the segment doclists for term pTerm and merge them in
** appropriate order into out.  Returns SQLITE_OK if successful.  If
** there are no segments for pTerm, successfully returns an empty
** doclist in out.
*/
static int term_select_all(fulltext_vtab *v, int iColumn,
                           const char *pTerm, int nTerm, DocList *out){
  DocList doclist;
  sqlite3_stmt *s;
  int rc = sql_get_statement(v, TERM_SELECT_ALL_STMT, &s);
  if( rc!=SQLITE_OK ) return rc;

  rc = sqlite3_bind_text(s, 1, pTerm, nTerm, SQLITE_STATIC);
  if( rc!=SQLITE_OK ) return rc;
1016
1017
1018
1019
1020
1021
1022




1023
1024
1025
1026
1027
1028
1029
    ** could skip the malloc() involved with the following call.  For
    ** now, I'd rather keep this logic similar to index_insert_term().
    ** We could additionally drop elements when we see deletes, but
    ** that would require a distinct version of docListAccumulate().
    */
    docListInit(&old, doclist.iType,
                sqlite3_column_blob(s, 0), sqlite3_column_bytes(s, 0));





    /* doclist contains the newer data, so write it over old.  Then
    ** steal accumulated result for doclist.
    */
    docListAccumulate(&old, &doclist);
    docListDestroy(&doclist);
    doclist = old;







>
>
>
>







1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
    ** could skip the malloc() involved with the following call.  For
    ** now, I'd rather keep this logic similar to index_insert_term().
    ** We could additionally drop elements when we see deletes, but
    ** that would require a distinct version of docListAccumulate().
    */
    docListInit(&old, doclist.iType,
                sqlite3_column_blob(s, 0), sqlite3_column_bytes(s, 0));

    if( iColumn<v->nColumns ){   /* querying a single column */
      docListRestrictColumn(&old, iColumn);
    }

    /* doclist contains the newer data, so write it over old.  Then
    ** steal accumulated result for doclist.
    */
    docListAccumulate(&old, &doclist);
    docListDestroy(&doclist);
    doclist = old;
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206

1207



1208

1209

1210
1211
1212





1213

1214

1215
1216
1217





1218
1219
1220

1221



1222


1223
1224
1225
1226
1227

1228
1229

1230
1231

1232
1233
1234
1235

1236


1237


1238
1239

1240


1241
1242

1243
1244

1245



1246



1247





1248
1249

1250
1251



1252

1253


1254




1255







1256

1257


1258

1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270

1271
1272



1273




1274
1275
1276

1277
1278
1279








1280
1281
1282




1283

1284




1285

1286
1287
1288

1289

1290
1291
1292


1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
  }

  if( v->pTokenizer!=NULL ){
    v->pTokenizer->pModule->xDestroy(v->pTokenizer);
    v->pTokenizer = NULL;
  }
  
  free(v->azColumn);
  free((void *) v->zName);
  free(v);
}

/*
** Token types for parsing the arguments to xConnect or xCreate.
*/
#define TOKEN_EOF         0    /* End of file */
#define TOKEN_SPACE       1    /* Any kind of whitespace */
#define TOKEN_ID          2    /* An identifier */
#define TOKEN_STRING      3    /* A string literal */
#define TOKEN_PUNCT       4    /* A single punctuation character */

/*
** If X is a character that can be used in an identifier then
** IdChar(X) will be true.  Otherwise it is false.
**
** For ASCII, any character with the high-order bit set is
** allowed in an identifier.  For 7-bit characters, 
** sqlite3IsIdChar[X] must be 1.
**
** Ticket #1066.  the SQL standard does not allow '$' in the
** middle of identfiers.  But many SQL implementations do. 
** SQLite will allow '$' in identifiers for compatibility.
** But the feature is undocumented.
*/
static const char isIdChar[] = {
/* x0 x1 x2 x3 x4 x5 x6 x7 x8 x9 xA xB xC xD xE xF */
    0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,  /* 2x */
    1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0,  /* 3x */
    0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,  /* 4x */
    1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1,  /* 5x */
    0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,  /* 6x */
    1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0,  /* 7x */
};
#define IdChar(C)  (((c=C)&0x80)!=0 || (c>0x1f && isIdChar[c-0x20]))


/*
** Return the length of the token that begins at z[0]. 
** Store the token type in *tokenType before returning.
*/
static int getToken(const char *z, int *tokenType){
  int i, c;
  switch( *z ){
    case 0: {
      *tokenType = TOKEN_EOF;
      return 0;
    }
    case ' ': case '\t': case '\n': case '\f': case '\r': {
      for(i=1; isspace(z[i]); i++){}
      *tokenType = TOKEN_SPACE;
      return i;
    }
    case '\'':
    case '"': {
      int delim = z[0];
      for(i=1; (c=z[i])!=0; i++){
        if( c==delim ){
          if( z[i+1]==delim ){
            i++;
          }else{
            break;
          }
        }
      }
      *tokenType = TOKEN_STRING;
      return i + (c!=0);
    }
    case '[': {
      for(i=1, c=z[0]; c!=']' && (c=z[i])!=0; i++){}
      *tokenType = TOKEN_ID;
      return i;
    }
    default: {
      if( !IdChar(*z) ){
        break;
      }
      for(i=1; IdChar(z[i]); i++){}
      *tokenType = TOKEN_ID;
      return i;
    }
  }
  *tokenType = TOKEN_PUNCT;
  return 1;
}


/*



** A token extracted from a string is an instance of the following

** structure.

*/
typedef struct Token {
  const char *z;       /* Pointer to token text.  Not '\000' terminated */





  short int n;         /* Length of the token text in bytes. */

} Token;


/*
** Given a input string (which is really one of the argv[] parameters





** passed into xConnect or xCreate) split the string up into tokens.
** Return an array of pointers to '\000' terminated strings, one string
** for each non-whitespace token.

**



** The returned array is terminated by a single NULL pointer.


**
** Space to hold the returned array is obtained from a single
** malloc and should be freed by passing the return value to free().
** The individual strings within the token list are all a part of
** the single memory allocation and will all be freed at once.

*/
static char **tokenizeString(const char *z, int *pnToken){

  int nToken = 0;
  Token *aToken = malloc( strlen(z) * sizeof(aToken[0]) );

  int n = 1;
  int e, i;
  int totalSize = 0;
  char **azToken;

  char *zCopy;


  while( n>0 ){


    n = getToken(z, &e);
    if( e!=TOKEN_SPACE ){

      aToken[nToken].z = z;


      aToken[nToken].n = n;
      nToken++;

      totalSize += n+1;
    }

    z += n;



  }



  azToken = (char**)malloc( nToken*sizeof(char*) + totalSize );





  zCopy = (char*)&azToken[nToken];
  nToken--;

  for(i=0; i<nToken; i++){
    azToken[i] = zCopy;



    n = aToken[i].n;

    memcpy(zCopy, aToken[i].z, n);


    zCopy[n] = 0;




    zCopy += n+1;







  }

  azToken[nToken] = 0;


  free(aToken);

  *pnToken = nToken;
  return azToken;
}

/*
** Remove the first nSkip tokens from a token list as well
** as all "(", ",", and ")" tokens from a token list.
**
** The memory for a token list comes from a single malloc().
** This routine just rearranges the pointers within that allocation.
** The token list is still freed by a single free().
*/

static void removeDelimiterTokens(char **azTokens, int nSkip, int *pnToken){
  int i, j, c;



  for(i=nSkip, j=0; azTokens[i]; i++){




    c = azTokens[i][0];
    if( c=='(' || c==',' || c==')' ) continue;
    azTokens[j++] = azTokens[i];

  }
  azTokens[j] = 0;
  *pnToken = j;








}








/* Current interface:




** argv[0]    - module name

** argv[1]    - database name
** argv[2...] - arguments.
**

** Arguments:

**
**   tokenizer  NAME(ARG1,ARG2,...)
**   columns(C1,C2,C3,...)



** argv[3] - tokenizer name (optional, a sensible default is provided)
** argv[4..] - passed to tokenizer (optional based on tokenizer)
**/
static int fulltextConnect(
  sqlite3 *db,
  void *pAux,
  int argc, const char *const*argv,
  sqlite3_vtab **ppVTab,
  char **pzErr
){
  int rc, i;
  fulltext_vtab *v = 0;
  const sqlite3_tokenizer_module *m = NULL;
  char **azToken = 0;
  int seen_tokenizer = 0;
  int seen_columns = 0;

  assert( argc>=3 );
  v = (fulltext_vtab *) malloc(sizeof(fulltext_vtab));
  if( v==0 ) goto connect_failed;
  memset(v, 0, sizeof(*v));
  v->db = db;
  v->zName = string_dup(argv[2]);
  v->pTokenizer = NULL;

  /* Process arguments to the module */
  for(i=3; i<argc; i++){
    int nToken;
    azToken = tokenizeString(argv[i], &nToken);
    if( azToken==0 ) goto connect_failed;
    removeDelimiterTokens(azToken, 0, &nToken);
    if( nToken>=2 && strcmp(azToken[0],"tokenizer")==0 ){
      if( seen_tokenizer ){
        *pzErr = sqlite3_mprintf("multiple tokenizer definitions");
        goto connect_failed;
      }
      seen_tokenizer = 1;
      if( !strcmp(azToken[1], "simple") ){
        sqlite3Fts1SimpleTokenizerModule(&m);
      } else {
        *pzErr = sqlite3_mprintf("unknown tokenizer: %s", azToken[1]);
        goto connect_failed;
      }
      rc = m->xCreate(nToken-2, (const char *const*)&azToken[2],&v->pTokenizer);
      v->pTokenizer->pModule = m;
      m = 0;
      if( rc ){
        goto connect_failed;
      }
    }else if( nToken>=2 && strcmp(azToken[0], "columns")==0 ){
      if( seen_columns ){
        *pzErr = sqlite3_mprintf("multiple column definitions");
        goto connect_failed;
      }
      removeDelimiterTokens(azToken, 1, &nToken);
      v->nColumn = nToken;
      v->azColumn = azToken;
      azToken = 0;
      seen_columns = 1;
    }else{
      *pzErr = sqlite3_mprintf("bad argument: %s", argv[i]);
      goto connect_failed;
    }
    free(azToken);
    azToken = 0;
  }

  /* Put in default values for the column names and the tokenizer if
  ** none is specified in the arguments.
  */
  if( !seen_tokenizer ){
    sqlite3Fts1SimpleTokenizerModule(&m);      
    rc = m->xCreate(0, 0, &v->pTokenizer);
    v->pTokenizer->pModule = m;
    if( rc!=SQLITE_OK ){
      goto connect_failed;
    }
    m = 0;
  }
  if( !seen_columns ){
    v->nColumn = 1;
    v->azColumn = malloc( sizeof(char*)*2 );
    if( v->azColumn==0 ) goto connect_failed;
    v->azColumn[0] = "content";
    v->azColumn[1] = 0;
  }

  /* TODO: verify the existence of backing tables foo_content, foo_term */

  rc = sqlite3_declare_vtab(db, "create table x(content text)");
  if( rc!=SQLITE_OK ) return rc;

  memset(v->pFulltextStatements, 0, sizeof(v->pFulltextStatements));

  *ppVTab = &v->base;
  TRACE(("FTS1 Connect %p\n", v));
  return SQLITE_OK;

connect_failed:
  if( v ){
    fulltext_vtab_destroy(v);
  }
  free(azToken);
  return SQLITE_ERROR;
}

static int fulltextCreate(sqlite3 *db, void *pAux,
                          int argc, const char *const*argv,
                          sqlite3_vtab **ppVTab, char **pzErr){
  int rc;
  assert( argc>=3 );
  TRACE(("FTS1 Create\n"));

  /* The %_content table holds the text of each full-text item, with
  ** the rowid used as the docid.
  **
  ** The %_term table maps each term to a document list blob
  ** containing elements sorted by ascending docid, each element
  ** encoded as:







|




<
<
<
<
<
<
<
<
|
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
|
|
<
<
<
<
<
<
<
<
<
|
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<



>
|
>
>
>
|
>
|
>
|
|
|
>
>
>
>
>
|
>
|
>
|
|
<
>
>
>
>
>
|
<
<
>
|
>
>
>
|
>
>
|
<
<
<
<
>
|
<
>
|
<
>
|
|
|
|
>
|
>
>
|
>
>
|
<
>
|
>
>
|
|
>
|

>
|
>
>
>
|
>
>
>
|
>
>
>
>
>
|
<
>
|
<
>
>
>
|
>
|
>
>
|
>
>
>
>
|
>
>
>
>
>
>
>
|
>
|
>
>
|
>
|
|


<
<
<
<
<
<
<
<
>
|
|
>
>
>
|
>
>
>
>
|
|
|
>
|
|
|
>
>
>
>
>
>
>
>
|
|

>
>
>
>

>
|
>
>
>
>
|
>
|
<
|
>
|
>
|
|
<
>
>
|
|
<
<







<
<
<
<
<
<
|
<
<
<
<
<
<
<
|
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<


<
|
<
<
<
|
<
<
<
<
<
|

<
<
<
<
<
<
<







1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258








1259























1260
1261









1262




































1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288

1289
1290
1291
1292
1293
1294


1295
1296
1297
1298
1299
1300
1301
1302
1303




1304
1305

1306
1307

1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320

1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345

1346
1347

1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379








1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422

1423
1424
1425
1426
1427
1428

1429
1430
1431
1432


1433
1434
1435
1436
1437
1438
1439






1440







1441

































































1442
1443

1444



1445





1446
1447







1448
1449
1450
1451
1452
1453
1454
  }

  if( v->pTokenizer!=NULL ){
    v->pTokenizer->pModule->xDestroy(v->pTokenizer);
    v->pTokenizer = NULL;
  }
  
  free((void *) v->zColumnNames);
  free((void *) v->zName);
  free(v);
}









/* Return true if the string s begins with the string t, ignoring case. */























static int startsWithIgnoreCase(const char *s, const char *t){
  while( *t )









    if( tolower(*s++)!=tolower(*t++) ) return 0;




































  return 1;
}

const char *kTokenize = "tokenize";

static int isTokenize(const char *arg){
  return startsWithIgnoreCase(arg, kTokenize);
}

static const char *tokenizerSpec(const char *arg){
  return arg + strlen(kTokenize);
}

typedef struct TableSpec {
  const char *zName;
  int nColumns;
  const char * const *zColumnNames;
  char *zTokenizer;
  char *zTokenizerArg;
} TableSpec;

void destroyTableSpec(TableSpec *p) {
  free(p->zTokenizer);
  free(p->zTokenizerArg);
}


/* Parse a CREATE VIRTUAL TABLE statement, which looks like this:
 *
 * CREATE VIRTUAL TABLE email
 *        USING fts1(subject, body, tokenize mytokenizer(myarg))
 *
 * We return parsed information in a TableSpec structure.


 * 
 */
int parseSpec(TableSpec *pSpec, int argc, const char * const *argv){
  int i;
  assert( argc>=3 );
  /* Current interface:
  ** argv[0] - module name
  ** argv[1] - database name
  ** argv[2] - table name




  ** argv[3..] - columns, optionally followed by tokenizer specification
  */

  pSpec->zName = argv[2];
  for (i=3; i<argc && !isTokenize(argv[i]); ++i)

    ;
  pSpec->nColumns = i-3;
  if( pSpec->nColumns<1) return SQLITE_ERROR;
  pSpec->zColumnNames = &argv[3];
  pSpec->zTokenizer = pSpec->zTokenizerArg = NULL;
  if( i<argc ){  /* we have a tokenizer */
    const char *start, *end;
    assert( isTokenize(argv[i]) );
    start = tokenizerSpec(argv[i]);
    while( isspace(*start) ){
      ++start;
    }
    end = start;

    while( isalnum(*end) ){
      ++end;
    }
    pSpec->zTokenizer = string_dup_n(start, end-start);

    start = end;
    while( isspace(*start) ){
      ++start;
    }
    if( *start=='(' ){  /* tokenizer has an argument */
      ++start;
      end = strchr(start, ')');
      if( !end ) return SQLITE_ERROR;
      pSpec->zTokenizerArg = string_dup_n(start, end-start);
    }
  }
  return SQLITE_OK;
}

/* Concatenate an array of strings into a single string, separating with commas.
 * The caller must free the returned string. */
static char *commaConcatenate(int nColumns, const char * const *zColumns){
  StringBuffer buf;
  int i;


  initStringBuffer(&buf);
  for(i=0; i<nColumns ; ++i){

    if( i>0 ){
      append(&buf, ", ");
    }
    append(&buf, zColumns[i]);
  }

  return buf.s;
}

static char *fulltextSchema(char *name, int nColumns,
                            const char * const *zColumns, int magic){
  StringBuffer buf;
  int i;

  initStringBuffer(&buf);
  append(&buf, "create table ");
  append(&buf, name);
  append(&buf, "(");
  for(i=0; i<nColumns; ++i){
    if( i>0 ){
      append(&buf, ", ");
    }
    append(&buf, zColumns[i]);
    append(&buf, " text");
  }
  if( magic ){
    append(&buf, ", _all text");
  }
  append(&buf, ")");
  return buf.s;
}









static int connect(sqlite3 *db, TableSpec *spec,
                   sqlite3_vtab **ppVTab, char **pzErr){
  int rc;
  fulltext_vtab *v = 0;
  const sqlite3_tokenizer_module *m = NULL;
  char *schema;

  v = (fulltext_vtab *) malloc(sizeof(fulltext_vtab));
  if( v==0 ) return SQLITE_ERROR;
  memset(v, 0, sizeof(*v));
  /* sqlite will initialize v->base */
  v->db = db;
  v->zName = string_dup(spec->zName);
  v->nColumns = spec->nColumns;
  v->zColumnNames = commaConcatenate(spec->nColumns, spec->zColumnNames);

  if( spec->zTokenizer == NULL ){
    sqlite3Fts1SimpleTokenizerModule(&m);
  } else {
    /* TODO(shess) For now, add new tokenizers as else if clauses. */
    if( !strcmp(spec->zTokenizer, "simple") ){
      sqlite3Fts1SimpleTokenizerModule(&m);
    } else {
      *pzErr = sqlite3_mprintf("unknown tokenizer: %s", spec->zTokenizer);
      rc = SQLITE_ERROR;
      goto err;
    }
  }

  /* TODO: Support multiple arguments to tokenizers. */
  rc = m->xCreate(1, &spec->zTokenizerArg, &v->pTokenizer);
  if( rc!=SQLITE_OK ) goto err;
  v->pTokenizer->pModule = m;

  /* TODO: verify the existence of backing tables foo_content, foo_term */

  schema = fulltextSchema("x", spec->nColumns, spec->zColumnNames, 1);
  rc = sqlite3_declare_vtab(db, schema);
  free(schema);
  if( rc!=SQLITE_OK ) goto err;

  memset(v->pFulltextStatements, 0, sizeof(v->pFulltextStatements));


  *ppVTab = &v->base;
  TRACE(("FTS1 Connect %p\n", v));

  return rc;

err:

  fulltext_vtab_destroy(v);
  return rc;
}



static int fulltextConnect(
  sqlite3 *db,
  void *pAux,
  int argc, const char *const*argv,
  sqlite3_vtab **ppVTab,
  char **pzErr
){






  TableSpec spec;







  int rc = parseSpec(&spec, argc, argv);

































































  if( rc!=SQLITE_OK ) return rc;


  rc = connect(db, &spec, ppVTab, pzErr);



  destroyTableSpec(&spec);





  return rc;
}








  /* The %_content table holds the text of each full-text item, with
  ** the rowid used as the docid.
  **
  ** The %_term table maps each term to a document list blob
  ** containing elements sorted by ascending docid, each element
  ** encoded as:
1435
1436
1437
1438
1439
1440
1441












1442


1443

1444
1445
1446
1447

1448



1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465

1466
1467
1468
1469
1470
1471
1472
1473
  ** information in preference to higher-segment information.
  */
  /* TODO(shess) Provide a VACUUM type operation which both removes
  ** deleted elements which are no longer necessary, and duplicated
  ** elements.  I suspect this will probably not be necessary in
  ** practice, though.
  */












  rc = sql_exec(db, argv[2],


    "create table %_content(content text);"

    "create table %_term(term text, segment integer, doclist blob, "
                        "primary key(term, segment));");
  if( rc!=SQLITE_OK ) return rc;


  return fulltextConnect(db, pAux, argc, argv, ppVTab, pzErr);



}

/* Decide how to handle an SQL query.
 * At the moment, MATCH queries can include implicit boolean ANDs; we
 * haven't implemented phrase searches or OR yet. */
static int fulltextBestIndex(sqlite3_vtab *pVTab, sqlite3_index_info *pInfo){
  int i;

  for(i=0; i<pInfo->nConstraint; ++i){
    const struct sqlite3_index_constraint *pConstraint;
    pConstraint = &pInfo->aConstraint[i];
    if( pConstraint->usable ) {
      if( pConstraint->iColumn==-1 &&
          pConstraint->op==SQLITE_INDEX_CONSTRAINT_EQ ){
        pInfo->idxNum = QUERY_ROWID;      /* lookup by rowid */
      } else if( pConstraint->iColumn==0 &&
                 pConstraint->op==SQLITE_INDEX_CONSTRAINT_MATCH ){

        pInfo->idxNum = QUERY_FULLTEXT;   /* full-text search */
      } else continue;

      pInfo->aConstraintUsage[i].argvIndex = 1;
      pInfo->aConstraintUsage[i].omit = 1;

      /* An arbitrary value for now.
       * TODO: Perhaps rowid matches should be considered cheaper than







>
>
>
>
>
>
>
>
>
>
>
>
|
>
>
|
>


|

>
|
>
>
>


|
<
<










|

>
|







1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512


1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
  ** information in preference to higher-segment information.
  */
  /* TODO(shess) Provide a VACUUM type operation which both removes
  ** deleted elements which are no longer necessary, and duplicated
  ** elements.  I suspect this will probably not be necessary in
  ** practice, though.
  */
static int fulltextCreate(sqlite3 *db, void *pAux,
                          int argc, const char * const *argv,
                          sqlite3_vtab **ppVTab, char **pzErr){
  int rc;
  TableSpec spec;
  char *schema;
  TRACE(("FTS1 Create\n"));

  rc = parseSpec(&spec, argc, argv);
  if( rc!=SQLITE_OK ) return rc;

  schema = fulltextSchema("%_content", spec.nColumns, spec.zColumnNames, 0);
  rc = sql_exec(db, spec.zName, schema);
  free(schema);
  if( rc!=SQLITE_OK ) goto out;

  rc = sql_exec(db, spec.zName,
    "create table %_term(term text, segment integer, doclist blob, "
                        "primary key(term, segment));");
  if( rc!=SQLITE_OK ) goto out;

  rc = connect(db, &spec, ppVTab, pzErr);

out:
  destroyTableSpec(&spec);
  return rc;
}

/* Decide how to handle an SQL query. */


static int fulltextBestIndex(sqlite3_vtab *pVTab, sqlite3_index_info *pInfo){
  int i;

  for(i=0; i<pInfo->nConstraint; ++i){
    const struct sqlite3_index_constraint *pConstraint;
    pConstraint = &pInfo->aConstraint[i];
    if( pConstraint->usable ) {
      if( pConstraint->iColumn==-1 &&
          pConstraint->op==SQLITE_INDEX_CONSTRAINT_EQ ){
        pInfo->idxNum = QUERY_ROWID;      /* lookup by rowid */
      } else if( pConstraint->iColumn>=0 &&
                 pConstraint->op==SQLITE_INDEX_CONSTRAINT_MATCH ){
        /* full-text search */
        pInfo->idxNum = QUERY_FULLTEXT + pConstraint->iColumn;
      } else continue;

      pInfo->aConstraintUsage[i].argvIndex = 1;
      pInfo->aConstraintUsage[i].omit = 1;

      /* An arbitrary value for now.
       * TODO: Perhaps rowid matches should be considered cheaper than
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539

static int fulltextNext(sqlite3_vtab_cursor *pCursor){
  fulltext_cursor *c = (fulltext_cursor *) pCursor;
  sqlite_int64 iDocid;
  int rc;

  TRACE(("FTS1 Next %p\n", pCursor));
  if( c->iCursorType != QUERY_FULLTEXT ){
    /* TODO(shess) Handle SQLITE_SCHEMA AND SQLITE_BUSY. */
    rc = sqlite3_step(c->pStmt);
    switch( rc ){
      case SQLITE_ROW:
        c->eof = 0;
        return SQLITE_OK;
      case SQLITE_DONE:







|







1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599

static int fulltextNext(sqlite3_vtab_cursor *pCursor){
  fulltext_cursor *c = (fulltext_cursor *) pCursor;
  sqlite_int64 iDocid;
  int rc;

  TRACE(("FTS1 Next %p\n", pCursor));
  if( c->iCursorType < QUERY_FULLTEXT ){
    /* TODO(shess) Handle SQLITE_SCHEMA AND SQLITE_BUSY. */
    rc = sqlite3_step(c->pStmt);
    switch( rc ){
      case SQLITE_ROW:
        c->eof = 0;
        return SQLITE_OK;
      case SQLITE_DONE:
1581
1582
1583
1584
1585
1586
1587

1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
** is the first term of a phrase query, go ahead and evaluate the phrase
** query and return the doclist for the entire phrase query.
**
** The result is stored in pTerm->doclist.
*/
static int docListOfTerm(
  fulltext_vtab *v,     /* The full text index */

  QueryTerm *pQTerm,    /* Term we are looking for, or 1st term of a phrase */
  DocList **ppResult    /* Write the result here */
){
  DocList *pLeft, *pRight, *pNew;
  int i, rc;

  pLeft = docListNew(DL_POSITIONS);
  rc = term_select_all(v, pQTerm->pTerm, pQTerm->nTerm, pLeft);
  if( rc ) return rc;
  for(i=1; i<=pQTerm->nPhrase; i++){
    pRight = docListNew(DL_POSITIONS);
    rc = term_select_all(v, pQTerm[i].pTerm, pQTerm[i].nTerm, pRight);
    if( rc ){
      docListDelete(pLeft);
      return rc;
    }
    pNew = docListNew(i<pQTerm->nPhrase ? DL_POSITIONS : DL_DOCIDS);
    docListPhraseMerge(pLeft, pRight, pNew);
    docListDelete(pLeft);
    docListDelete(pRight);
    pLeft = pNew;
  }
  *ppResult = pLeft;
  return SQLITE_OK;
}



/* Parse a query string into a Query structure.
 *
 * We could, in theory, allow query strings to be complicated
 * nested expressions with precedence determined by parentheses.
 * But none of the major search engines do this.  (Perhaps the
 * feeling is that an parenthesized expression is two complex of







>







|



|













<
<







1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673


1674
1675
1676
1677
1678
1679
1680
** is the first term of a phrase query, go ahead and evaluate the phrase
** query and return the doclist for the entire phrase query.
**
** The result is stored in pTerm->doclist.
*/
static int docListOfTerm(
  fulltext_vtab *v,     /* The full text index */
  int iColumn,           /* column to restrict to */
  QueryTerm *pQTerm,    /* Term we are looking for, or 1st term of a phrase */
  DocList **ppResult    /* Write the result here */
){
  DocList *pLeft, *pRight, *pNew;
  int i, rc;

  pLeft = docListNew(DL_POSITIONS);
  rc = term_select_all(v, iColumn, pQTerm->pTerm, pQTerm->nTerm, pLeft);
  if( rc ) return rc;
  for(i=1; i<=pQTerm->nPhrase; i++){
    pRight = docListNew(DL_POSITIONS);
    rc = term_select_all(v, iColumn, pQTerm[i].pTerm, pQTerm[i].nTerm, pRight);
    if( rc ){
      docListDelete(pLeft);
      return rc;
    }
    pNew = docListNew(i<pQTerm->nPhrase ? DL_POSITIONS : DL_DOCIDS);
    docListPhraseMerge(pLeft, pRight, pNew);
    docListDelete(pLeft);
    docListDelete(pRight);
    pLeft = pNew;
  }
  *ppResult = pLeft;
  return SQLITE_OK;
}



/* Parse a query string into a Query structure.
 *
 * We could, in theory, allow query strings to be complicated
 * nested expressions with precedence determined by parentheses.
 * But none of the major search engines do this.  (Perhaps the
 * feeling is that an parenthesized expression is two complex of
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
    iInput = i;
    if( i<nInput ){
      assert( pInput[i]=='"' );
      inPhrase = !inPhrase;
    }
  }

  if(inPhrase) {  /* unmatched quote */
    queryDestroy(pQuery);
    return SQLITE_ERROR;
  }
  return SQLITE_OK;
}

/* Perform a full-text query using the search expression in
** pInput[0..nInput-1].  Return a list of matching documents
** in pResult.
*/
static int fulltextQuery(fulltext_vtab *v, const char *pInput, int nInput,
                          DocList **pResult){
  Query q;
  int i, rc;
  DocList *pLeft = NULL;
  DocList *pRight, *pNew;
  int nNot = 0;

  rc = parseQuery(v, pInput, nInput, &q);
  if( rc!=SQLITE_OK ) return rc;

  /* Merge AND terms. */
  for(i = 0 ; i < q.nTerms; i += q.pTerms[i].nPhrase + 1){

    if( q.pTerms[i].isNot ){
      /* Handle all NOT terms in a separate pass */
      nNot++;
      continue;
    }

    rc = docListOfTerm(v, &q.pTerms[i], &pRight);
    if( rc ){
      queryDestroy(&q);
      return rc;
    }
    if( pLeft==0 ){
      pLeft = pRight;
    }else{







|










|
|


















|







1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
    iInput = i;
    if( i<nInput ){
      assert( pInput[i]=='"' );
      inPhrase = !inPhrase;
    }
  }

  if( inPhrase ){  /* unmatched quote */
    queryDestroy(pQuery);
    return SQLITE_ERROR;
  }
  return SQLITE_OK;
}

/* Perform a full-text query using the search expression in
** pInput[0..nInput-1].  Return a list of matching documents
** in pResult.
*/
static int fulltextQuery(fulltext_vtab *v, int iColumn,
                         const char *pInput, int nInput, DocList **pResult){
  Query q;
  int i, rc;
  DocList *pLeft = NULL;
  DocList *pRight, *pNew;
  int nNot = 0;

  rc = parseQuery(v, pInput, nInput, &q);
  if( rc!=SQLITE_OK ) return rc;

  /* Merge AND terms. */
  for(i = 0 ; i < q.nTerms; i += q.pTerms[i].nPhrase + 1){

    if( q.pTerms[i].isNot ){
      /* Handle all NOT terms in a separate pass */
      nNot++;
      continue;
    }

    rc = docListOfTerm(v, iColumn, &q.pTerms[i], &pRight);
    if( rc ){
      queryDestroy(&q);
      return rc;
    }
    if( pLeft==0 ){
      pLeft = pRight;
    }else{
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
  }

  if( nNot && pLeft==0 ){
    /* We do not yet know how to handle a query of only NOT terms */
    return SQLITE_ERROR;
  }


  /* Do the EXCEPT terms */
  for(i=0; i<q.nTerms;  i += q.pTerms[i].nPhrase + 1){
    if( !q.pTerms[i].isNot ) continue;
    rc = docListOfTerm(v, &q.pTerms[i], &pRight);
    if( rc ){
      queryDestroy(&q);
      docListDelete(pLeft);
      return rc;
    }
    pNew = docListNew(DL_DOCIDS);
    docListExceptMerge(pLeft, pRight, pNew);







<



|







1864
1865
1866
1867
1868
1869
1870

1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
  }

  if( nNot && pLeft==0 ){
    /* We do not yet know how to handle a query of only NOT terms */
    return SQLITE_ERROR;
  }


  /* Do the EXCEPT terms */
  for(i=0; i<q.nTerms;  i += q.pTerms[i].nPhrase + 1){
    if( !q.pTerms[i].isNot ) continue;
    rc = docListOfTerm(v, iColumn, &q.pTerms[i], &pRight);
    if( rc ){
      queryDestroy(&q);
      docListDelete(pLeft);
      return rc;
    }
    pNew = docListNew(DL_DOCIDS);
    docListExceptMerge(pLeft, pRight, pNew);
1833
1834
1835
1836
1837
1838
1839

1840
1841











1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860

1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875


1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886

1887
1888



1889
1890
1891

1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905

1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926

1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947

static int fulltextFilter(sqlite3_vtab_cursor *pCursor,
                          int idxNum, const char *idxStr,
                          int argc, sqlite3_value **argv){
  fulltext_cursor *c = (fulltext_cursor *) pCursor;
  fulltext_vtab *v = cursor_vtab(c);
  int rc;


  TRACE(("FTS1 Filter %p\n",pCursor));











  c->iCursorType = idxNum;
  switch( idxNum ){
    case QUERY_GENERIC:
      rc = sql_prepare(v->db, v->zName, &c->pStmt,
                       "select rowid, content from %_content");
      break;

    case QUERY_ROWID:
      rc = sql_prepare(v->db, v->zName, &c->pStmt,
                       "select rowid, content from %_content where rowid = ?");
      if( rc!=SQLITE_OK ) return rc;

      rc = sqlite3_bind_int64(c->pStmt, 1, sqlite3_value_int64(argv[0]));
      break;

    case QUERY_FULLTEXT:   /* full-text search */
    {
      const char *zQuery = (const char *)sqlite3_value_text(argv[0]);
      DocList *pResult;

      assert( argc==1 );
      rc = fulltextQuery(v, zQuery, -1, &pResult);
      if( rc!=SQLITE_OK ) return rc;
      readerInit(&c->result, pResult);
      rc = sql_prepare(v->db, v->zName, &c->pStmt,
                       "select rowid, content from %_content where rowid = ?");
      break;
    }

    default:
      assert( 0 );
  }

  if( rc!=SQLITE_OK ) return rc;



  return fulltextNext(pCursor);
}

static int fulltextEof(sqlite3_vtab_cursor *pCursor){
  fulltext_cursor *c = (fulltext_cursor *) pCursor;
  return c->eof;
}

static int fulltextColumn(sqlite3_vtab_cursor *pCursor,
                          sqlite3_context *pContext, int idxCol){
  fulltext_cursor *c = (fulltext_cursor *) pCursor;

  const char *s;




  assert( idxCol==0 );
  s = (const char *) sqlite3_column_text(c->pStmt, 1);
  sqlite3_result_text(pContext, s, -1, SQLITE_TRANSIENT);


  return SQLITE_OK;
}

static int fulltextRowid(sqlite3_vtab_cursor *pCursor, sqlite_int64 *pRowid){
  fulltext_cursor *c = (fulltext_cursor *) pCursor;

  *pRowid = sqlite3_column_int64(c->pStmt, 0);
  return SQLITE_OK;
}

/* Build a hash table containing all terms in pText. */
static int buildTerms(fts1Hash *terms, sqlite3_tokenizer *pTokenizer,
                       const char *pText, int nText, sqlite_int64 iDocid){

  sqlite3_tokenizer_cursor *pCursor;
  const char *pToken;
  int nTokenBytes;
  int iStartOffset, iEndOffset, iPosition;
  int rc;

  assert( nText>=0 );

  rc = pTokenizer->pModule->xOpen(pTokenizer, pText, nText, &pCursor);
  if( rc!=SQLITE_OK ) return rc;

  pCursor->pTokenizer = pTokenizer;
  fts1HashInit(terms, FTS1_HASH_STRING, 1);
  while( SQLITE_OK==pTokenizer->pModule->xNext(pCursor,
                                               &pToken, &nTokenBytes,
                                               &iStartOffset, &iEndOffset,
                                               &iPosition) ){
    DocList *p;

    /* Positions can't be negative; we use -1 as a terminator internally. */
    if( iPosition<0 ) {

      rc = SQLITE_ERROR;  
      goto err;
    }

    p = fts1HashFind(terms, pToken, nTokenBytes);
    if( p==NULL ){
      p = docListNew(DL_POSITIONS_OFFSETS);
      docListAddDocid(p, iDocid);
      fts1HashInsert(terms, pToken, nTokenBytes, p);
    }
    docListAddPosOffset(p, iPosition, iStartOffset, iEndOffset);
  }

err:
  /* TODO(shess) Check return?  Should this be able to cause errors at
  ** this point?  Actually, same question about sqlite3_finalize(),
  ** though one could argue that failure there means that the data is
  ** not durable.  *ponder*
  */
  pTokenizer->pModule->xClose(pCursor);
  return rc;







>


>
>
>
>
>
>
>
>
>
>
>



<
<



|
<
|
<
<


|



>

|
|

<
<


|
<
<
|
|
<

>
>
|










>


>
>
>
|
|
|
>











|
|
|
>






<
<
|



<







|
>
|
<








|


<







1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914


1915
1916
1917
1918

1919


1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930


1931
1932
1933


1934
1935

1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980


1981
1982
1983
1984

1985
1986
1987
1988
1989
1990
1991
1992
1993
1994

1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005

2006
2007
2008
2009
2010
2011
2012

static int fulltextFilter(sqlite3_vtab_cursor *pCursor,
                          int idxNum, const char *idxStr,
                          int argc, sqlite3_value **argv){
  fulltext_cursor *c = (fulltext_cursor *) pCursor;
  fulltext_vtab *v = cursor_vtab(c);
  int rc;
  StringBuffer sb;

  TRACE(("FTS1 Filter %p\n",pCursor));

  initStringBuffer(&sb);
  append(&sb, "select rowid, ");
  append(&sb, v->zColumnNames);
  append(&sb, " from %_content");
  if( idxNum != QUERY_GENERIC) {
    append(&sb, " where rowid = ?"); 
  }
  rc = sql_prepare(v->db, v->zName, &c->pStmt, sb.s);
  if( rc!=SQLITE_OK ) goto out;

  c->iCursorType = idxNum;
  switch( idxNum ){
    case QUERY_GENERIC:


      break;

    case QUERY_ROWID:
      rc = sqlite3_bind_int64(c->pStmt, 1, sqlite3_value_int64(argv[0]));

      if( rc!=SQLITE_OK ) goto out;


      break;

    default:   /* full-text search */
    {
      const char *zQuery = (const char *)sqlite3_value_text(argv[0]);
      DocList *pResult;
      assert( idxNum<=QUERY_FULLTEXT+v->nColumns);
      assert( argc==1 );
      rc = fulltextQuery(v, idxNum-QUERY_FULLTEXT, zQuery, -1, &pResult);
      if( rc!=SQLITE_OK ) goto out;
      readerInit(&c->result, pResult);


      break;
    }
  }



  rc = fulltextNext(pCursor);


out:
  free(sb.s);
  return rc;
}

static int fulltextEof(sqlite3_vtab_cursor *pCursor){
  fulltext_cursor *c = (fulltext_cursor *) pCursor;
  return c->eof;
}

static int fulltextColumn(sqlite3_vtab_cursor *pCursor,
                          sqlite3_context *pContext, int idxCol){
  fulltext_cursor *c = (fulltext_cursor *) pCursor;
  fulltext_vtab *v = cursor_vtab(c);
  const char *s;

  if( idxCol==v->nColumns ){  /* a request for _all */
    sqlite3_result_null(pContext);
  } else {
    assert( idxCol<v->nColumns );
    s = (const char *) sqlite3_column_text(c->pStmt, idxCol+1);
    sqlite3_result_text(pContext, s, -1, SQLITE_TRANSIENT);
  }

  return SQLITE_OK;
}

static int fulltextRowid(sqlite3_vtab_cursor *pCursor, sqlite_int64 *pRowid){
  fulltext_cursor *c = (fulltext_cursor *) pCursor;

  *pRowid = sqlite3_column_int64(c->pStmt, 0);
  return SQLITE_OK;
}

/* Add all terms/positions in [zText] to the given hash table. */
static int buildTerms(fulltext_vtab *v, fts1Hash *terms, int iColumn,
                      const char *zText, int nText, sqlite_int64 iDocid){
  sqlite3_tokenizer *pTokenizer = v->pTokenizer;
  sqlite3_tokenizer_cursor *pCursor;
  const char *pToken;
  int nTokenBytes;
  int iStartOffset, iEndOffset, iPosition;
  int rc;



  rc = pTokenizer->pModule->xOpen(pTokenizer, zText, nText, &pCursor);
  if( rc!=SQLITE_OK ) return rc;

  pCursor->pTokenizer = pTokenizer;

  while( SQLITE_OK==pTokenizer->pModule->xNext(pCursor,
                                               &pToken, &nTokenBytes,
                                               &iStartOffset, &iEndOffset,
                                               &iPosition) ){
    DocList *p;

    /* Positions can't be negative; we use -1 as a terminator internally. */
    if( iPosition<0 ){
      pTokenizer->pModule->xClose(pCursor);
      return SQLITE_ERROR;

    }

    p = fts1HashFind(terms, pToken, nTokenBytes);
    if( p==NULL ){
      p = docListNew(DL_POSITIONS_OFFSETS);
      docListAddDocid(p, iDocid);
      fts1HashInsert(terms, pToken, nTokenBytes, p);
    }
    docListAddPosOffset(p, iColumn, iPosition, iStartOffset, iEndOffset);
  }


  /* TODO(shess) Check return?  Should this be able to cause errors at
  ** this point?  Actually, same question about sqlite3_finalize(),
  ** though one could argue that failure there means that the data is
  ** not durable.  *ponder*
  */
  pTokenizer->pModule->xClose(pCursor);
  return rc;
2013
2014
2015
2016
2017
2018
2019
2020
2021

2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035

2036
2037
2038
2039
2040
2041
2042

2043
2044
2045
2046
2047
2048
2049
2050
2051
2052
2053
2054
2055

2056
2057
2058
2059
2060
2061


2062
2063
2064

2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076



2077
2078
2079
2080
2081
2082
2083
  docListDestroy(&doclist);
  return rc;
}

/* Insert a row into the full-text index; set *piRowid to be the ID of the
 * new row. */
static int index_insert(fulltext_vtab *v, sqlite3_value *pRequestRowid,
                        const char *pText, int nText,
                        sqlite_int64 *piRowid){

  fts1Hash terms;  /* maps term string -> PosList */
  fts1HashElem *e;
  int rc;

  assert( nText>=0 );

  rc = content_insert(v, pRequestRowid, pText, nText);
  if( rc!=SQLITE_OK ) return rc;
  *piRowid = sqlite3_last_insert_rowid(v->db);

  if( !pText || !nText ) return SQLITE_OK;   /* nothing to index */

  rc = buildTerms(&terms, v->pTokenizer, pText, nText, *piRowid);
  if( rc!=SQLITE_OK ) return rc;


  for(e=fts1HashFirst(&terms); e; e=fts1HashNext(e)){
    DocList *p = fts1HashData(e);
    rc = index_insert_term(v, fts1HashKey(e), fts1HashKeysize(e), p);
    if( rc!=SQLITE_OK ) break;
  }


  for(e=fts1HashFirst(&terms); e; e=fts1HashNext(e)){
    DocList *p = fts1HashData(e);
    docListDelete(p);
  }
  fts1HashClear(&terms);
  return rc;
}

/* Delete a row from the full-text index. */
static int index_delete(fulltext_vtab *v, sqlite_int64 iRow){
  char *pText = 0;
  int nText = 0;
  fts1Hash terms;

  fts1HashElem *e;
  DocList doclist;

  int rc = content_select(v, iRow, &pText, &nText);
  if( rc!=SQLITE_OK ) return rc;



  rc = buildTerms(&terms, v->pTokenizer, pText, nText, iRow);
  free(pText);
  if( rc!=SQLITE_OK ) return rc;


  /* Delete by inserting a doclist with no positions.  This will
  ** overwrite existing data as it is merged forward by
  ** index_insert_term().
  */
  docListInit(&doclist, DL_POSITIONS_OFFSETS, 0, 0);
  docListAddDocid(&doclist, iRow);

  for(e=fts1HashFirst(&terms); e; e=fts1HashNext(e)){
    rc = index_insert_term(v, fts1HashKey(e), fts1HashKeysize(e), &doclist);
    if( rc!=SQLITE_OK ) break;
  }



  for(e=fts1HashFirst(&terms); e; e=fts1HashNext(e)){
    DocList *p = fts1HashData(e);
    docListDelete(p);
  }
  fts1HashClear(&terms);
  docListDestroy(&doclist);








|

>




<
<
|



|
|
|
|
>







>










|
<

>



|


>
>
|
<
|
>












>
>
>







2078
2079
2080
2081
2082
2083
2084
2085
2086
2087
2088
2089
2090
2091


2092
2093
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
2105
2106
2107
2108
2109
2110
2111
2112
2113
2114
2115
2116
2117
2118
2119

2120
2121
2122
2123
2124
2125
2126
2127
2128
2129
2130

2131
2132
2133
2134
2135
2136
2137
2138
2139
2140
2141
2142
2143
2144
2145
2146
2147
2148
2149
2150
2151
2152
2153
2154
  docListDestroy(&doclist);
  return rc;
}

/* Insert a row into the full-text index; set *piRowid to be the ID of the
 * new row. */
static int index_insert(fulltext_vtab *v, sqlite3_value *pRequestRowid,
                        sqlite3_value **pValues,
                        sqlite_int64 *piRowid){
  int i;
  fts1Hash terms;  /* maps term string -> PosList */
  fts1HashElem *e;
  int rc;



  rc = content_insert(v, pRequestRowid, pValues);
  if( rc!=SQLITE_OK ) return rc;
  *piRowid = sqlite3_last_insert_rowid(v->db);

  fts1HashInit(&terms, FTS1_HASH_STRING, 1);
  for(i = 0; i < v->nColumns ; ++i){
    rc = buildTerms(v, &terms, i, sqlite3_value_text(pValues[i]), -1, *piRowid);
    if( rc!=SQLITE_OK ) goto out;
  }

  for(e=fts1HashFirst(&terms); e; e=fts1HashNext(e)){
    DocList *p = fts1HashData(e);
    rc = index_insert_term(v, fts1HashKey(e), fts1HashKeysize(e), p);
    if( rc!=SQLITE_OK ) break;
  }

out:
  for(e=fts1HashFirst(&terms); e; e=fts1HashNext(e)){
    DocList *p = fts1HashData(e);
    docListDelete(p);
  }
  fts1HashClear(&terms);
  return rc;
}

/* Delete a row from the full-text index. */
static int index_delete(fulltext_vtab *v, sqlite_int64 iRow){
  const char **pValues;

  fts1Hash terms;
  int i;
  fts1HashElem *e;
  DocList doclist;

  int rc = content_select(v, iRow, &pValues);
  if( rc!=SQLITE_OK ) return rc;

  fts1HashInit(&terms, FTS1_HASH_STRING, 1);
  for(i = 0 ; i < v->nColumns; ++i) {
    rc = buildTerms(v, &terms, i, pValues[i], -1, iRow);

    if( rc!=SQLITE_OK ) goto out;
  }

  /* Delete by inserting a doclist with no positions.  This will
  ** overwrite existing data as it is merged forward by
  ** index_insert_term().
  */
  docListInit(&doclist, DL_POSITIONS_OFFSETS, 0, 0);
  docListAddDocid(&doclist, iRow);

  for(e=fts1HashFirst(&terms); e; e=fts1HashNext(e)){
    rc = index_insert_term(v, fts1HashKey(e), fts1HashKeysize(e), &doclist);
    if( rc!=SQLITE_OK ) break;
  }

out:
  freeStringArray(v->nColumns, pValues);
  for(e=fts1HashFirst(&terms); e; e=fts1HashNext(e)){
    DocList *p = fts1HashData(e);
    docListDelete(p);
  }
  fts1HashClear(&terms);
  docListDestroy(&doclist);

2094
2095
2096
2097
2098
2099
2100
2101
2102
2103

2104
2105
2106
2107
2108
2109
2110
2111
2112
    return index_delete(v, sqlite3_value_int64(ppArg[0]));
  }

  if( sqlite3_value_type(ppArg[0]) != SQLITE_NULL ){
    return SQLITE_ERROR;   /* an update; not yet supported */
  }

  assert( nArg==3 );    /* ppArg[1] = rowid, ppArg[2] = content */
  return index_insert(v, ppArg[1],
                      sqlite3_value_blob(ppArg[2]),

                      sqlite3_value_bytes(ppArg[2]),
                      pRowid);
}

static const sqlite3_module fulltextModule = {
  0,
  fulltextCreate,
  fulltextConnect,
  fulltextBestIndex,







|
|
|
>
|
|







2165
2166
2167
2168
2169
2170
2171
2172
2173
2174
2175
2176
2177
2178
2179
2180
2181
2182
2183
2184
    return index_delete(v, sqlite3_value_int64(ppArg[0]));
  }

  if( sqlite3_value_type(ppArg[0]) != SQLITE_NULL ){
    return SQLITE_ERROR;   /* an update; not yet supported */
  }

  /* ppArg[1] = rowid
   * ppArg[2..2+v->nColumns-1] = values
   * ppArg[2+v->nColumns] = value for _all (we ignore this) */
  assert( nArg==2+v->nColumns+1);    

  return index_insert(v, ppArg[1], &ppArg[2], pRowid);
}

static const sqlite3_module fulltextModule = {
  0,
  fulltextCreate,
  fulltextConnect,
  fulltextBestIndex,
Changes to ext/fts1/fts1_tokenizer1.c.
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
static const sqlite3_tokenizer_module simpleTokenizerModule;

static int isDelim(simple_tokenizer *t, unsigned char c){
  return c<0x80 && t->delim[c];
}

static int simpleCreate(
  int argc, const char **argv,
  sqlite3_tokenizer **ppTokenizer
){
  simple_tokenizer *t;

  t = (simple_tokenizer *) calloc(sizeof(simple_tokenizer), 1);
  /* TODO(shess) Delimiters need to remain the same from run to run,
  ** else we need to reindex.  One solution would be a meta-table to







|







49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
static const sqlite3_tokenizer_module simpleTokenizerModule;

static int isDelim(simple_tokenizer *t, unsigned char c){
  return c<0x80 && t->delim[c];
}

static int simpleCreate(
  int argc, const char * const *argv,
  sqlite3_tokenizer **ppTokenizer
){
  simple_tokenizer *t;

  t = (simple_tokenizer *) calloc(sizeof(simple_tokenizer), 1);
  /* TODO(shess) Delimiters need to remain the same from run to run,
  ** else we need to reindex.  One solution would be a meta-table to