/ Check-in [0fc0ea20]
Login

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

Overview
Comment:Add syntax to fts5 used to specify that a phrase or NEAR group should match a subset of columns. For example "[col1 col2 ...] : <phrase>".
Downloads: Tarball | ZIP archive | SQL archive
Timelines: family | ancestors | descendants | both | fts5
Files: files | file ages | folders
SHA1: 0fc0ea20920615f3e48ea2dbe2b7dcd979b0993e
User & Date: dan 2015-05-29 15:55:30
Context
2015-05-29
19:00
Add extra tests and fixes for multi-column matches. check-in: ae6794ff user: dan tags: fts5
15:55
Add syntax to fts5 used to specify that a phrase or NEAR group should match a subset of columns. For example "[col1 col2 ...] : <phrase>". check-in: 0fc0ea20 user: dan tags: fts5
2015-05-28
19:57
Optimizations for fts5 queries that match against a specific column. check-in: b29ac50a user: dan tags: fts5
Changes
Hide Diffs Unified Diffs Ignore Whitespace Patch

Changes to ext/fts5/fts5Int.h.

507
508
509
510
511
512
513

514
515
516
517
518
519
520
...
573
574
575
576
577
578
579






580
581
582
583
584
585
586
587
588
589
590
591
592
593
*/
typedef struct Fts5Expr Fts5Expr;
typedef struct Fts5ExprNode Fts5ExprNode;
typedef struct Fts5Parse Fts5Parse;
typedef struct Fts5Token Fts5Token;
typedef struct Fts5ExprPhrase Fts5ExprPhrase;
typedef struct Fts5ExprNearset Fts5ExprNearset;


struct Fts5Token {
  const char *p;                  /* Token text (not NULL terminated) */
  int n;                          /* Size of buffer p in bytes */
};

/* Parse a MATCH expression. */
................................................................................
);

Fts5ExprNearset *sqlite3Fts5ParseNearset(
  Fts5Parse*, 
  Fts5ExprNearset*,
  Fts5ExprPhrase* 
);







void sqlite3Fts5ParsePhraseFree(Fts5ExprPhrase*);
void sqlite3Fts5ParseNearsetFree(Fts5ExprNearset*);
void sqlite3Fts5ParseNodeFree(Fts5ExprNode*);

void sqlite3Fts5ParseSetDistance(Fts5Parse*, Fts5ExprNearset*, Fts5Token*);
void sqlite3Fts5ParseSetColumn(Fts5Parse*, Fts5ExprNearset*, Fts5Token*);
void sqlite3Fts5ParseFinished(Fts5Parse *pParse, Fts5ExprNode *p);
void sqlite3Fts5ParseNear(Fts5Parse *pParse, Fts5Token*);

/*
** End of interface to code in fts5_expr.c.
**************************************************************************/








>







 







>
>
>
>
>
>






|







507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
...
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
*/
typedef struct Fts5Expr Fts5Expr;
typedef struct Fts5ExprNode Fts5ExprNode;
typedef struct Fts5Parse Fts5Parse;
typedef struct Fts5Token Fts5Token;
typedef struct Fts5ExprPhrase Fts5ExprPhrase;
typedef struct Fts5ExprNearset Fts5ExprNearset;
typedef struct Fts5ExprColset Fts5ExprColset;

struct Fts5Token {
  const char *p;                  /* Token text (not NULL terminated) */
  int n;                          /* Size of buffer p in bytes */
};

/* Parse a MATCH expression. */
................................................................................
);

Fts5ExprNearset *sqlite3Fts5ParseNearset(
  Fts5Parse*, 
  Fts5ExprNearset*,
  Fts5ExprPhrase* 
);

Fts5ExprColset *sqlite3Fts5ParseColset(
  Fts5Parse*, 
  Fts5ExprColset*, 
  Fts5Token *
);

void sqlite3Fts5ParsePhraseFree(Fts5ExprPhrase*);
void sqlite3Fts5ParseNearsetFree(Fts5ExprNearset*);
void sqlite3Fts5ParseNodeFree(Fts5ExprNode*);

void sqlite3Fts5ParseSetDistance(Fts5Parse*, Fts5ExprNearset*, Fts5Token*);
void sqlite3Fts5ParseSetColset(Fts5Parse*, Fts5ExprNearset*, Fts5ExprColset*);
void sqlite3Fts5ParseFinished(Fts5Parse *pParse, Fts5ExprNode *p);
void sqlite3Fts5ParseNear(Fts5Parse *pParse, Fts5Token*);

/*
** End of interface to code in fts5_expr.c.
**************************************************************************/

Changes to ext/fts5/fts5_expr.c.

75
76
77
78
79
80
81










82
83
84
85
86
87
88
89
90
91
92
93
94
95
...
132
133
134
135
136
137
138


139
140
141
142
143
144
145
...
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
...
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346

347
348
349
350
351
352
353
...
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
...
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732

733
734
735
736
737
738
739
740
741
742
743
744

745
746













747
748



749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
....
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
....
1231
1232
1233
1234
1235
1236
1237

1238
1239
1240
1241
1242
1243
1244
....
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
....
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
....
1459
1460
1461
1462
1463
1464
1465
1466



1467







1468
1469
1470
1471
1472
1473
1474
....
1526
1527
1528
1529
1530
1531
1532
1533

1534
1535
1536
1537
1538
1539
1540
1541
struct Fts5ExprPhrase {
  Fts5ExprNode *pNode;            /* FTS5_STRING node this phrase is part of */
  Fts5Buffer poslist;             /* Current position list */
  int nTerm;                      /* Number of entries in aTerm[] */
  Fts5ExprTerm aTerm[0];          /* Terms that make up this phrase */
};











/*
** One or more phrases that must appear within a certain token distance of
** each other within each matching document.
*/
struct Fts5ExprNearset {
  int nNear;                      /* NEAR parameter */
  int iCol;                       /* Column to search (-1 -> all columns) */
  int nPhrase;                    /* Number of entries in aPhrase[] array */
  Fts5ExprPhrase *apPhrase[0];    /* Array of phrase pointers */
};


/*
** Parse context.
................................................................................
  while( fts5ExprIsspace(*z) ) z++;

  pToken->p = z;
  pToken->n = 1;
  switch( *z ){
    case '(':  tok = FTS5_LP;    break;
    case ')':  tok = FTS5_RP;    break;


    case ':':  tok = FTS5_COLON; break;
    case ',':  tok = FTS5_COMMA; break;
    case '+':  tok = FTS5_PLUS;  break;
    case '*':  tok = FTS5_STAR;  break;
    case '\0': tok = FTS5_EOF;   break;

    case '"': {
................................................................................
      pNew->nPhrase = 1;
      pNew->apExprPhrase = apPhrase;
      pNew->apExprPhrase[0] = pCopy;

      pNode->eType = FTS5_STRING;
      pNode->pNear = pNear;

      pNear->iCol = -1;
      pNear->nPhrase = 1;
      pNear->apPhrase[0] = pCopy;

      pCopy->nTerm = pOrig->nTerm;
      pCopy->pNode = pNode;
    }else{
      /* At least one allocation failed. Free them all. */
................................................................................
**
** SQLITE_OK is returned if an error occurs, or an SQLite error code 
** otherwise. It is not considered an error code if the current rowid is 
** not a match.
*/
static int fts5ExprPhraseIsMatch(
  Fts5Expr *pExpr,                /* Expression pPhrase belongs to */
  int iCol,                       /* If >=0, search for matches in iCol only */
  Fts5ExprPhrase *pPhrase,        /* Phrase object to initialize */
  int *pbMatch                    /* OUT: Set to true if really a match */
){
  Fts5PoslistWriter writer = {0};
  Fts5PoslistReader aStatic[4];
  Fts5PoslistReader *aIter = aStatic;
  int i;
  int rc = SQLITE_OK;


  fts5BufferZero(&pPhrase->poslist);

  /* If the aStatic[] array is not large enough, allocate a large array
  ** using sqlite3_malloc(). This approach could be improved upon. */
  if( pPhrase->nTerm>(sizeof(aStatic) / sizeof(aStatic[0])) ){
    int nByte = sizeof(Fts5PoslistReader) * pPhrase->nTerm;
................................................................................
** contain any entries for column iCol, return 0.
*/
static int fts5ExprExtractCol(
  const u8 **pa,                  /* IN/OUT: Pointer to poslist */
  int n,                          /* IN: Size of poslist in bytes */
  int iCol                        /* Column to extract from poslist */
){
  int ii;
  int iCurrent = 0;
  const u8 *p = *pa;
  const u8 *pEnd = &p[n];         /* One byte past end of position list */
  u8 prev = 0;

  while( iCol!=iCurrent ){
    /* Advance pointer p until it points to pEnd or an 0x01 byte that is
................................................................................
  Fts5Expr *pExpr,                /* Expression that pNear is a part of */
  Fts5ExprNode *pNode             /* The "NEAR" node (FTS5_STRING) */
){
  Fts5ExprNearset *pNear = pNode->pNear;
  int rc = SQLITE_OK;

  while( 1 ){
    int i;

    if( pNear->nPhrase==1 && pNear->apPhrase[0]->nTerm==1 ){
      /* If this "NEAR" object is actually a single phrase that consists 
      ** of a single term only, then grab pointers into the poslist
      ** managed by the fts5_index.c iterator object. This is much faster 
      ** than synthesizing a new poslist the way we have to for more
      ** complicated phrase or NEAR expressions.  */
      Fts5ExprPhrase *pPhrase = pNear->apPhrase[0];
      Fts5IndexIter *pIter = pPhrase->aTerm[0].pIter;
      assert( pPhrase->poslist.nSpace==0 );
      rc = sqlite3Fts5IterPoslist(pIter, 
          (const u8**)&pPhrase->poslist.p, &pPhrase->poslist.n, &pNode->iRowid
      );


      /* If the term may match any column, then this must be a match. 
      ** Return immediately in this case. Otherwise, try to find the
      ** part of the poslist that corresponds to the required column.
      ** If it can be found, return. If it cannot, the next iteration
      ** of the loop will test the next rowid in the database for this
      ** term.  */
      if( pNear->iCol<0 ) return rc;

      pPhrase->poslist.n = fts5ExprExtractCol(
          (const u8**)&pPhrase->poslist.p,
          pPhrase->poslist.n,

          pNear->iCol
      );













      if( pPhrase->poslist.n ) return rc;
    }else{




      /* Advance the iterators until they all point to the same rowid */
      rc = fts5ExprNearNextRowidMatch(pExpr, pNode);
      if( rc!=SQLITE_OK || pNode->bEof ) break;

      /* Check that each phrase in the nearset matches the current row.
      ** Populate the pPhrase->poslist buffers at the same time. If any
      ** phrase is not a match, break out of the loop early.  */
      for(i=0; rc==SQLITE_OK && i<pNear->nPhrase; i++){
        Fts5ExprPhrase *pPhrase = pNear->apPhrase[i];
        if( pPhrase->nTerm>1 || pNear->iCol>=0 ){
          int bMatch = 0;
          rc = fts5ExprPhraseIsMatch(pExpr, pNear->iCol, pPhrase, &bMatch);
          if( bMatch==0 ) break;
        }else{
          rc = sqlite3Fts5IterPoslistBuffer(
              pPhrase->aTerm[0].pIter, &pPhrase->poslist
              );
        }
      }

      if( rc==SQLITE_OK && i==pNear->nPhrase ){
        int bMatch = 1;
        if( pNear->nPhrase>1 ){
          rc = fts5ExprNearIsMatch(pNear, &bMatch);
................................................................................
    if( pNear==0 ){
      int nByte = sizeof(Fts5ExprNearset) + SZALLOC * sizeof(Fts5ExprPhrase*);
      pRet = sqlite3_malloc(nByte);
      if( pRet==0 ){
        pParse->rc = SQLITE_NOMEM;
      }else{
        memset(pRet, 0, nByte);
        pRet->iCol = -1;
      }
    }else if( (pNear->nPhrase % SZALLOC)==0 ){
      int nNew = pNear->nPhrase + SZALLOC;
      int nByte = sizeof(Fts5ExprNearset) + nNew * sizeof(Fts5ExprPhrase*);

      pRet = (Fts5ExprNearset*)sqlite3_realloc(pNear, nByte);
      if( pRet==0 ){
................................................................................
*/
void sqlite3Fts5ParseNearsetFree(Fts5ExprNearset *pNear){
  if( pNear ){
    int i;
    for(i=0; i<pNear->nPhrase; i++){
      fts5ExprPhraseFree(pNear->apPhrase[i]);
    }

    sqlite3_free(pNear);
  }
}

void sqlite3Fts5ParseFinished(Fts5Parse *pParse, Fts5ExprNode *p){
  assert( pParse->pExpr==0 );
  pParse->pExpr = p;
................................................................................
        pParse, "fts5: syntax error near \"%.*s\"", pTok->n, pTok->p
    );
  }
}

void sqlite3Fts5ParseSetDistance(
  Fts5Parse *pParse, 
  Fts5ExprNearset *pNear, 
  Fts5Token *p
){
  int nNear = 0;
  int i;
  if( p->n ){
    for(i=0; i<p->n; i++){
      char c = (char)p->p[i];
................................................................................
    }
  }else{
    nNear = FTS5_DEFAULT_NEARDIST;
  }
  pNear->nNear = nNear;
}














































void sqlite3Fts5ParseSetColumn(
  Fts5Parse *pParse, 
  Fts5ExprNearset *pNear, 

  Fts5Token *p
){


  if( pParse->rc==SQLITE_OK ){

    char *z = 0;
    int rc = fts5ParseStringFromToken(p, &z);
    if( rc==SQLITE_OK ){
      Fts5Config *pConfig = pParse->pConfig;
      int i;
      for(i=0; i<pConfig->nCol; i++){
        if( 0==sqlite3_stricmp(pConfig->azCol[i], z) ){
          pNear->iCol = i;
          break;
        }
      }
      if( i==pConfig->nCol ){
        sqlite3Fts5ParseError(pParse, "no such column: %s", z);
      }
      sqlite3_free(z);
    }else{
      pParse->rc = rc;
    }























  }
}

/*
** Allocate and return a new expression object. If anything goes wrong (i.e.
** OOM error), leave an error code in pParse and return NULL.
*/
................................................................................
  if( pExpr->eType==FTS5_STRING ){
    Fts5ExprNearset *pNear = pExpr->pNear;
    int i; 
    int iTerm;

    zRet = fts5PrintfAppend(zRet, "[%s ", zNearsetCmd);
    if( zRet==0 ) return 0;
    if( pNear->iCol>=0 ){



      zRet = fts5PrintfAppend(zRet, "-col %d ", pNear->iCol);







      if( zRet==0 ) return 0;
    }

    if( pNear->nPhrase>1 ){
      zRet = fts5PrintfAppend(zRet, "-near %d ", pNear->nNear);
      if( zRet==0 ) return 0;
    }
................................................................................
static char *fts5ExprPrint(Fts5Config *pConfig, Fts5ExprNode *pExpr){
  char *zRet = 0;
  if( pExpr->eType==FTS5_STRING ){
    Fts5ExprNearset *pNear = pExpr->pNear;
    int i; 
    int iTerm;

    if( pNear->iCol>=0 ){

      zRet = fts5PrintfAppend(zRet, "%s : ", pConfig->azCol[pNear->iCol]);
      if( zRet==0 ) return 0;
    }

    if( pNear->nPhrase>1 ){
      zRet = fts5PrintfAppend(zRet, "NEAR(");
      if( zRet==0 ) return 0;
    }







>
>
>
>
>
>
>
>
>
>






|







 







>
>







 







<







 







|








>







 







<







 







<









|
|
|
|
>







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


>
>
>










|

|




|







 







<







 







>







 







|







 







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


>
>

>




|
|
|
<



|






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







 







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







 







|
>
|







75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
...
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
...
283
284
285
286
287
288
289

290
291
292
293
294
295
296
...
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
...
672
673
674
675
676
677
678

679
680
681
682
683
684
685
...
723
724
725
726
727
728
729

730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751

752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
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
794
795
796
797
798
799
800
....
1175
1176
1177
1178
1179
1180
1181

1182
1183
1184
1185
1186
1187
1188
....
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
....
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
....
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
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
....
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
....
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
struct Fts5ExprPhrase {
  Fts5ExprNode *pNode;            /* FTS5_STRING node this phrase is part of */
  Fts5Buffer poslist;             /* Current position list */
  int nTerm;                      /* Number of entries in aTerm[] */
  Fts5ExprTerm aTerm[0];          /* Terms that make up this phrase */
};

/*
** If a NEAR() clump may only match a specific set of columns, then
** Fts5ExprNearset.pColset points to an object of the following type.
** Each entry in the aiCol[] array
*/
struct Fts5ExprColset {
  int nCol;
  int aiCol[1];
};

/*
** One or more phrases that must appear within a certain token distance of
** each other within each matching document.
*/
struct Fts5ExprNearset {
  int nNear;                      /* NEAR parameter */
  Fts5ExprColset *pColset;        /* Columns to search (NULL -> all columns) */
  int nPhrase;                    /* Number of entries in aPhrase[] array */
  Fts5ExprPhrase *apPhrase[0];    /* Array of phrase pointers */
};


/*
** Parse context.
................................................................................
  while( fts5ExprIsspace(*z) ) z++;

  pToken->p = z;
  pToken->n = 1;
  switch( *z ){
    case '(':  tok = FTS5_LP;    break;
    case ')':  tok = FTS5_RP;    break;
    case '[':  tok = FTS5_LSP;   break;
    case ']':  tok = FTS5_RSP;   break;
    case ':':  tok = FTS5_COLON; break;
    case ',':  tok = FTS5_COMMA; break;
    case '+':  tok = FTS5_PLUS;  break;
    case '*':  tok = FTS5_STAR;  break;
    case '\0': tok = FTS5_EOF;   break;

    case '"': {
................................................................................
      pNew->nPhrase = 1;
      pNew->apExprPhrase = apPhrase;
      pNew->apExprPhrase[0] = pCopy;

      pNode->eType = FTS5_STRING;
      pNode->pNear = pNear;


      pNear->nPhrase = 1;
      pNear->apPhrase[0] = pCopy;

      pCopy->nTerm = pOrig->nTerm;
      pCopy->pNode = pNode;
    }else{
      /* At least one allocation failed. Free them all. */
................................................................................
**
** SQLITE_OK is returned if an error occurs, or an SQLite error code 
** otherwise. It is not considered an error code if the current rowid is 
** not a match.
*/
static int fts5ExprPhraseIsMatch(
  Fts5Expr *pExpr,                /* Expression pPhrase belongs to */
  Fts5ExprColset *pColset,        /* Restrict matches to these columns */
  Fts5ExprPhrase *pPhrase,        /* Phrase object to initialize */
  int *pbMatch                    /* OUT: Set to true if really a match */
){
  Fts5PoslistWriter writer = {0};
  Fts5PoslistReader aStatic[4];
  Fts5PoslistReader *aIter = aStatic;
  int i;
  int rc = SQLITE_OK;
  int iCol = pColset ? pColset->aiCol[0] : -1;

  fts5BufferZero(&pPhrase->poslist);

  /* If the aStatic[] array is not large enough, allocate a large array
  ** using sqlite3_malloc(). This approach could be improved upon. */
  if( pPhrase->nTerm>(sizeof(aStatic) / sizeof(aStatic[0])) ){
    int nByte = sizeof(Fts5PoslistReader) * pPhrase->nTerm;
................................................................................
** contain any entries for column iCol, return 0.
*/
static int fts5ExprExtractCol(
  const u8 **pa,                  /* IN/OUT: Pointer to poslist */
  int n,                          /* IN: Size of poslist in bytes */
  int iCol                        /* Column to extract from poslist */
){

  int iCurrent = 0;
  const u8 *p = *pa;
  const u8 *pEnd = &p[n];         /* One byte past end of position list */
  u8 prev = 0;

  while( iCol!=iCurrent ){
    /* Advance pointer p until it points to pEnd or an 0x01 byte that is
................................................................................
  Fts5Expr *pExpr,                /* Expression that pNear is a part of */
  Fts5ExprNode *pNode             /* The "NEAR" node (FTS5_STRING) */
){
  Fts5ExprNearset *pNear = pNode->pNear;
  int rc = SQLITE_OK;

  while( 1 ){


    if( pNear->nPhrase==1 && pNear->apPhrase[0]->nTerm==1 ){
      /* If this "NEAR" object is actually a single phrase that consists 
      ** of a single term only, then grab pointers into the poslist
      ** managed by the fts5_index.c iterator object. This is much faster 
      ** than synthesizing a new poslist the way we have to for more
      ** complicated phrase or NEAR expressions.  */
      Fts5ExprPhrase *pPhrase = pNear->apPhrase[0];
      Fts5IndexIter *pIter = pPhrase->aTerm[0].pIter;
      Fts5ExprColset *pColset = pNear->pColset;
      const u8 *pPos;
      int nPos;

      rc = sqlite3Fts5IterPoslist(pIter, &pPos, &nPos, &pNode->iRowid);

      /* If the term may match any column, then this must be a match. 
      ** Return immediately in this case. Otherwise, try to find the
      ** part of the poslist that corresponds to the required column.
      ** If it can be found, return. If it cannot, the next iteration
      ** of the loop will test the next rowid in the database for this
      ** term.  */
      if( pColset==0 ){

        assert( pPhrase->poslist.nSpace==0 );
        pPhrase->poslist.p = (u8*)pPos;
        pPhrase->poslist.n = nPos;
      }else if( pColset->nCol==1 ){
        assert( pPhrase->poslist.nSpace==0 );
        pPhrase->poslist.n = fts5ExprExtractCol(&pPos, nPos, pColset->aiCol[0]);
        pPhrase->poslist.p = (u8*)pPos;
      }else{
        int i;
        fts5BufferZero(&pPhrase->poslist);
        for(i=0; i<pColset->nCol; i++){
          const u8 *pSub = pPos;
          int nSub = fts5ExprExtractCol(&pSub, nPos, pColset->aiCol[i]);
          if( nSub ){
            fts5BufferAppendBlob(&rc, &pPhrase->poslist, nSub, pSub);
          }
        }
      }

      if( pPhrase->poslist.n ) return rc;
    }else{
      int i;

      assert( pNear->pColset==0 || pNear->pColset->nCol==1 );

      /* Advance the iterators until they all point to the same rowid */
      rc = fts5ExprNearNextRowidMatch(pExpr, pNode);
      if( rc!=SQLITE_OK || pNode->bEof ) break;

      /* Check that each phrase in the nearset matches the current row.
      ** Populate the pPhrase->poslist buffers at the same time. If any
      ** phrase is not a match, break out of the loop early.  */
      for(i=0; rc==SQLITE_OK && i<pNear->nPhrase; i++){
        Fts5ExprPhrase *pPhrase = pNear->apPhrase[i];
        if( pPhrase->nTerm>1 || pNear->pColset ){
          int bMatch = 0;
          rc = fts5ExprPhraseIsMatch(pExpr, pNear->pColset, pPhrase, &bMatch);
          if( bMatch==0 ) break;
        }else{
          rc = sqlite3Fts5IterPoslistBuffer(
              pPhrase->aTerm[0].pIter, &pPhrase->poslist
          );
        }
      }

      if( rc==SQLITE_OK && i==pNear->nPhrase ){
        int bMatch = 1;
        if( pNear->nPhrase>1 ){
          rc = fts5ExprNearIsMatch(pNear, &bMatch);
................................................................................
    if( pNear==0 ){
      int nByte = sizeof(Fts5ExprNearset) + SZALLOC * sizeof(Fts5ExprPhrase*);
      pRet = sqlite3_malloc(nByte);
      if( pRet==0 ){
        pParse->rc = SQLITE_NOMEM;
      }else{
        memset(pRet, 0, nByte);

      }
    }else if( (pNear->nPhrase % SZALLOC)==0 ){
      int nNew = pNear->nPhrase + SZALLOC;
      int nByte = sizeof(Fts5ExprNearset) + nNew * sizeof(Fts5ExprPhrase*);

      pRet = (Fts5ExprNearset*)sqlite3_realloc(pNear, nByte);
      if( pRet==0 ){
................................................................................
*/
void sqlite3Fts5ParseNearsetFree(Fts5ExprNearset *pNear){
  if( pNear ){
    int i;
    for(i=0; i<pNear->nPhrase; i++){
      fts5ExprPhraseFree(pNear->apPhrase[i]);
    }
    sqlite3_free(pNear->pColset);
    sqlite3_free(pNear);
  }
}

void sqlite3Fts5ParseFinished(Fts5Parse *pParse, Fts5ExprNode *p){
  assert( pParse->pExpr==0 );
  pParse->pExpr = p;
................................................................................
        pParse, "fts5: syntax error near \"%.*s\"", pTok->n, pTok->p
    );
  }
}

void sqlite3Fts5ParseSetDistance(
  Fts5Parse *pParse, 
  Fts5ExprNearset *pNear,
  Fts5Token *p
){
  int nNear = 0;
  int i;
  if( p->n ){
    for(i=0; i<p->n; i++){
      char c = (char)p->p[i];
................................................................................
    }
  }else{
    nNear = FTS5_DEFAULT_NEARDIST;
  }
  pNear->nNear = nNear;
}

/*
** The second argument passed to this function may be NULL, or it may be
** an existing Fts5ExprColset object. This function returns a pointer to
** a new colset object containing the contents of (p) with new value column
** number iCol appended. 
**
** If an OOM error occurs, store an error code in pParse and return NULL.
** The old colset object (if any) is not freed in this case.
*/
static Fts5ExprColset *fts5ParseColset(
  Fts5Parse *pParse,              /* Store SQLITE_NOMEM here if required */
  Fts5ExprColset *p,              /* Existing colset object */
  int iCol                        /* New column to add to colset object */
){
  int nCol = p ? p->nCol : 0;     /* Num. columns already in colset object */
  Fts5ExprColset *pNew;           /* New colset object to return */

  assert( pParse->rc==SQLITE_OK );
  assert( iCol>=0 && iCol<pParse->pConfig->nCol );

  pNew = sqlite3_realloc(p, sizeof(Fts5ExprColset) + sizeof(int)*nCol);
  if( pNew==0 ){
    pParse->rc = SQLITE_NOMEM;
  }else{
    int *aiCol = pNew->aiCol;
    int i, j;
    for(i=0; i<nCol; i++){
      if( aiCol[i]==iCol ) return pNew;
      if( aiCol[i]>iCol ) break;
    }
    for(j=nCol; j>i; j--){
      aiCol[j] = aiCol[j-1];
    }
    aiCol[i] = iCol;
    pNew->nCol = nCol+1;

#ifndef NDEBUG
    /* Check that the array is in order and contains no duplicate entries. */
    for(i=1; i<pNew->nCol; i++) assert( pNew->aiCol[i]>pNew->aiCol[i-1] );
#endif
  }

  return pNew;
}

Fts5ExprColset *sqlite3Fts5ParseColset(
  Fts5Parse *pParse,              /* Store SQLITE_NOMEM here if required */

  Fts5ExprColset *pColset,        /* Existing colset object */
  Fts5Token *p
){
  Fts5ExprColset *pRet = 0;

  if( pParse->rc==SQLITE_OK ){
    int iCol;
    char *z = 0;
    int rc = fts5ParseStringFromToken(p, &z);
    if( rc==SQLITE_OK ){
      Fts5Config *pConfig = pParse->pConfig;
      sqlite3Fts5Dequote(z);
      for(iCol=0; iCol<pConfig->nCol; iCol++){
        if( 0==sqlite3_stricmp(pConfig->azCol[iCol], z) ){

          break;
        }
      }
      if( iCol==pConfig->nCol ){
        sqlite3Fts5ParseError(pParse, "no such column: %s", z);
      }
      sqlite3_free(z);
    }else{
      pParse->rc = rc;
    }

    if( pParse->rc==SQLITE_OK ){
      pRet = fts5ParseColset(pParse, pColset, iCol);
    }
  }

  if( pParse->rc!=SQLITE_OK ){
    assert( pRet==0 );
    sqlite3_free(pColset);
  }

  return pRet;
}

void sqlite3Fts5ParseSetColset(
  Fts5Parse *pParse, 
  Fts5ExprNearset *pNear, 
  Fts5ExprColset *pColset 
){
  if( pNear ){
    pNear->pColset = pColset;
  }else{
    sqlite3_free(pColset);
  }
}

/*
** Allocate and return a new expression object. If anything goes wrong (i.e.
** OOM error), leave an error code in pParse and return NULL.
*/
................................................................................
  if( pExpr->eType==FTS5_STRING ){
    Fts5ExprNearset *pNear = pExpr->pNear;
    int i; 
    int iTerm;

    zRet = fts5PrintfAppend(zRet, "[%s ", zNearsetCmd);
    if( zRet==0 ) return 0;
    if( pNear->pColset ){
      int *aiCol = pNear->pColset->aiCol;
      int nCol = pNear->pColset->nCol;
      if( nCol==1 ){
        zRet = fts5PrintfAppend(zRet, "-col %d ", aiCol[0]);
      }else{
        zRet = fts5PrintfAppend(zRet, "-col {%d", aiCol[0]);
        for(i=1; i<pNear->pColset->nCol; i++){
          zRet = fts5PrintfAppend(zRet, " %d", aiCol[i]);
        }
        zRet = fts5PrintfAppend(zRet, "} ");
      }
      if( zRet==0 ) return 0;
    }

    if( pNear->nPhrase>1 ){
      zRet = fts5PrintfAppend(zRet, "-near %d ", pNear->nNear);
      if( zRet==0 ) return 0;
    }
................................................................................
static char *fts5ExprPrint(Fts5Config *pConfig, Fts5ExprNode *pExpr){
  char *zRet = 0;
  if( pExpr->eType==FTS5_STRING ){
    Fts5ExprNearset *pNear = pExpr->pNear;
    int i; 
    int iTerm;

    if( pNear->pColset ){
      int iCol = pNear->pColset->aiCol[0];
      zRet = fts5PrintfAppend(zRet, "%s : ", pConfig->azCol[iCol]);
      if( zRet==0 ) return 0;
    }

    if( pNear->nPhrase>1 ){
      zRet = fts5PrintfAppend(zRet, "NEAR(");
      if( zRet==0 ) return 0;
    }

Changes to ext/fts5/fts5_index.c.

4475
4476
4477
4478
4479
4480
4481
4482
4483
4484
4485
4486
4487
4488
4489
....
4490
4491
4492
4493
4494
4495
4496
4497
4498
4499
4500
4501
4502
4503
4504
}

/*
** Move to the next matching rowid. 
*/
int sqlite3Fts5IterNext(Fts5IndexIter *pIter){
  assert( pIter->pIndex->rc==SQLITE_OK );
  fts5BufferZero(&pIter->poslist);
  fts5MultiIterNext(pIter->pIndex, pIter->pMulti, 0, 0);
  return fts5IndexReturn(pIter->pIndex);
}

/*
** Move to the next matching term/rowid. Used by the fts5vocab module.
*/
................................................................................
int sqlite3Fts5IterNextScan(Fts5IndexIter *pIter){
  Fts5Index *p = pIter->pIndex;
  Fts5MultiSegIter *pMulti = pIter->pMulti;

  assert( pIter->pIndex->rc==SQLITE_OK );
  assert( pMulti );

  fts5BufferZero(&pIter->poslist);
  fts5MultiIterNext(p, pMulti, 0, 0);
  if( p->rc==SQLITE_OK ){
    Fts5SegIter *pSeg = &pMulti->aSeg[ pMulti->aFirst[1].iFirst ];
    if( pSeg->pLeaf && pSeg->term.p[0]!=FTS5_MAIN_PREFIX ){
      fts5DataRelease(pSeg->pLeaf);
      pSeg->pLeaf = 0;
    }







<







 







<







4475
4476
4477
4478
4479
4480
4481

4482
4483
4484
4485
4486
4487
4488
....
4489
4490
4491
4492
4493
4494
4495

4496
4497
4498
4499
4500
4501
4502
}

/*
** Move to the next matching rowid. 
*/
int sqlite3Fts5IterNext(Fts5IndexIter *pIter){
  assert( pIter->pIndex->rc==SQLITE_OK );

  fts5MultiIterNext(pIter->pIndex, pIter->pMulti, 0, 0);
  return fts5IndexReturn(pIter->pIndex);
}

/*
** Move to the next matching term/rowid. Used by the fts5vocab module.
*/
................................................................................
int sqlite3Fts5IterNextScan(Fts5IndexIter *pIter){
  Fts5Index *p = pIter->pIndex;
  Fts5MultiSegIter *pMulti = pIter->pMulti;

  assert( pIter->pIndex->rc==SQLITE_OK );
  assert( pMulti );


  fts5MultiIterNext(p, pMulti, 0, 0);
  if( p->rc==SQLITE_OK ){
    Fts5SegIter *pSeg = &pMulti->aSeg[ pMulti->aFirst[1].iFirst ];
    if( pSeg->pLeaf && pSeg->term.p[0]!=FTS5_MAIN_PREFIX ){
      fts5DataRelease(pSeg->pLeaf);
      pSeg->pLeaf = 0;
    }

Changes to ext/fts5/fts5parse.y.

91
92
93
94
95
96
97
98
99
100
101

















102
103
104
105
106
107
108
exprlist(A) ::= exprlist(X) cnearset(Y). {
  A = sqlite3Fts5ParseNode(pParse, FTS5_AND, X, Y, 0);
}

cnearset(A) ::= nearset(X). { 
  A = sqlite3Fts5ParseNode(pParse, FTS5_STRING, 0, 0, X); 
}
cnearset(A) ::= STRING(X) COLON nearset(Y). { 
  sqlite3Fts5ParseSetColumn(pParse, Y, &X);
  A = sqlite3Fts5ParseNode(pParse, FTS5_STRING, 0, 0, Y); 
}


















%type nearset     {Fts5ExprNearset*}
%type nearphrases {Fts5ExprNearset*}
%destructor nearset { sqlite3Fts5ParseNearsetFree($$); }
%destructor nearphrases { sqlite3Fts5ParseNearsetFree($$); }

nearset(A) ::= phrase(X). { A = sqlite3Fts5ParseNearset(pParse, 0, X); }







|
|


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







91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
exprlist(A) ::= exprlist(X) cnearset(Y). {
  A = sqlite3Fts5ParseNode(pParse, FTS5_AND, X, Y, 0);
}

cnearset(A) ::= nearset(X). { 
  A = sqlite3Fts5ParseNode(pParse, FTS5_STRING, 0, 0, X); 
}
cnearset(A) ::= colset(X) COLON nearset(Y). { 
  sqlite3Fts5ParseSetColset(pParse, Y, X);
  A = sqlite3Fts5ParseNode(pParse, FTS5_STRING, 0, 0, Y); 
}

%type colset {Fts5ExprColset*}
%destructor colset { sqlite3_free($$); }
%type colsetlist {Fts5ExprColset*}
%destructor colsetlist { sqlite3_free($$); }

colset(A) ::= LSP colsetlist(X) RSP. { A = X; }
colset(A) ::= STRING(X). {
  A = sqlite3Fts5ParseColset(pParse, 0, &X);
}

colsetlist(A) ::= colsetlist(Y) STRING(X). { 
  A = sqlite3Fts5ParseColset(pParse, Y, &X); }
colsetlist(A) ::= STRING(X). { 
  A = sqlite3Fts5ParseColset(pParse, 0, &X); 
}


%type nearset     {Fts5ExprNearset*}
%type nearphrases {Fts5ExprNearset*}
%destructor nearset { sqlite3Fts5ParseNearsetFree($$); }
%destructor nearphrases { sqlite3Fts5ParseNearsetFree($$); }

nearset(A) ::= phrase(X). { A = sqlite3Fts5ParseNearset(pParse, 0, X); }

Changes to ext/fts5/test/fts5ac.test.

121
122
123
124
125
126
127

128
129
130
131


















132
133
134
135
136
137
138
139
140
141
...
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
...
357
358
359
360
361
362
363
364
365
366
367














368
369
370
371
372
373
374
    95  {h b n j t k i h o q u}             {w n g i t o k c a m y p f l x c p}
    96  {f c x p y r b m o l m o a}         {p c a q s u n n x d c f a o}
    97  {u h h k m n k}                     {u b v n u a o c}
    98  {s p e t c z d f n w f}             {l s f j b l c e s h}
    99  {r c v w i v h a t a c v c r e}     {h h u m g o f b a e o}
}


# Usage:
#
#   poslist aCol ?-pc VARNAME? ?-near N? ?-col C? -- phrase1 phrase2...
#


















proc poslist {aCol args} {
  set O(-near) 10
  set O(-col)  -1
  set O(-pc)   ""

  set nOpt [lsearch -exact $args --]
  if {$nOpt<0} { error "no -- option" }

  foreach {k v} [lrange $args 0 [expr $nOpt-1]] {
    if {[info exists O($k)]==0} { error "unrecognized option $k" }
................................................................................
      set A($j,$i) [list]
    }
  }

  set iCol -1
  foreach col $aCol {
    incr iCol
    if {$O(-col)>=0 && $O(-col)!=$iCol} continue

    set nToken [llength $col]

    set iFL [expr $O(-near) >= $nToken ? $nToken - 1 : $O(-near)]
    for { } {$iFL < $nToken} {incr iFL} {
      for {set iPhrase 0} {$iPhrase<$nPhrase} {incr iPhrase} {
        set B($iPhrase) [list]
      }
................................................................................
    } $res
  }

  #-------------------------------------------------------------------------
  # Queries on a specific column.
  #
  foreach {tn expr} {
    1 "x:a"
    2 "y:a"
    3 "x:b"
    4 "y:b"














  } {
    set res [matchdata 1 $expr]
    do_execsql_test $tn2.3.$tn.[llength $res] { 
      SELECT rowid, fts5_test_poslist(xx) FROM xx WHERE xx match $expr
    } $res
  }








>




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


|







 







|
<







 







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







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
147
148
149
150
151
152
153
154
155
156
157
158
159
160
...
176
177
178
179
180
181
182
183

184
185
186
187
188
189
190
...
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
    95  {h b n j t k i h o q u}             {w n g i t o k c a m y p f l x c p}
    96  {f c x p y r b m o l m o a}         {p c a q s u n n x d c f a o}
    97  {u h h k m n k}                     {u b v n u a o c}
    98  {s p e t c z d f n w f}             {l s f j b l c e s h}
    99  {r c v w i v h a t a c v c r e}     {h h u m g o f b a e o}
}

#-------------------------------------------------------------------------
# Usage:
#
#   poslist aCol ?-pc VARNAME? ?-near N? ?-col C? -- phrase1 phrase2...
#
# This command is used to test if a document (set of column values) matches
# the logical equivalent of a single FTS5 NEAR() clump and, if so, return
# the equivalent of an FTS5 position list.
#
# Parameter $aCol is passed a list of the column values for the document
# to test. Parameters $phrase1 and so on are the phrases.
#
# The result is a list of phrase hits. Each phrase hit is formatted as
# three integers separated by "." characters, in the following format:
#
#   <phrase number> . <column number> . <token offset>
#
# Options:
#
#   -near N        (NEAR distance. Default 10)
#   -col  C        (List of column indexes to match against)
#   -pc   VARNAME  (variable in caller frame to use for phrase numbering)
#
proc poslist {aCol args} {
  set O(-near) 10
  set O(-col)  {}
  set O(-pc)   ""

  set nOpt [lsearch -exact $args --]
  if {$nOpt<0} { error "no -- option" }

  foreach {k v} [lrange $args 0 [expr $nOpt-1]] {
    if {[info exists O($k)]==0} { error "unrecognized option $k" }
................................................................................
      set A($j,$i) [list]
    }
  }

  set iCol -1
  foreach col $aCol {
    incr iCol
    if {$O(-col)!="" && [lsearch $O(-col) $iCol]<0} continue

    set nToken [llength $col]

    set iFL [expr $O(-near) >= $nToken ? $nToken - 1 : $O(-near)]
    for { } {$iFL < $nToken} {incr iFL} {
      for {set iPhrase 0} {$iPhrase<$nPhrase} {incr iPhrase} {
        set B($iPhrase) [list]
      }
................................................................................
    } $res
  }

  #-------------------------------------------------------------------------
  # Queries on a specific column.
  #
  foreach {tn expr} {
    1.1 "x:a"
    1.2 "y:a"
    1.3 "x:b"
    1.4 "y:b"
    2.1 "[x]:a"
    2.2 "[y]:a"
    2.3 "[x]:b"
    2.4 "[y]:b"

    3.1 "[x y]:a"
    3.2 "[y x]:a"
    3.3 "[x x]:b"
    3.4 "[y y]:b"

    4.1 {["x" "y"]:a}
    4.2 {["y" x]:a}
    4.3 {[x "x"]:b}
    4.4 {["y" y]:b}
  } {
    set res [matchdata 1 $expr]
    do_execsql_test $tn2.3.$tn.[llength $res] { 
      SELECT rowid, fts5_test_poslist(xx) FROM xx WHERE xx match $expr
    } $res
  }