SQLite

Check-in [70fc69eed9]
Login

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

Overview
Comment:Explicitly limit the size of fts5 tokens to 32768 bytes.
Downloads: Tarball | ZIP archive
Timelines: family | ancestors | descendants | both | trunk
Files: files | file ages | folders
SHA1: 70fc69eed9b09159899d7cbd1416a59d04210a63
User & Date: dan 2016-03-23 15:04:00.239
Context
2016-03-23
15:53
Remove an unused local variable from FTS5. (check-in: 0ed693c29f user: drh tags: trunk)
15:04
Explicitly limit the size of fts5 tokens to 32768 bytes. (check-in: 70fc69eed9 user: dan tags: trunk)
13:46
Update a requirement mark. No changes to code. (check-in: 412984642a user: drh tags: trunk)
Changes
Unified Diff Ignore Whitespace Patch
Changes to ext/fts5/fts5Int.h.
43
44
45
46
47
48
49




50
51
52
53
54
55
56
** Constants for the largest and smallest possible 64-bit signed integers.
*/
# define LARGEST_INT64  (0xffffffff|(((i64)0x7fffffff)<<32))
# define SMALLEST_INT64 (((i64)-1) - LARGEST_INT64)

#endif






/*
** Maximum number of prefix indexes on single FTS5 table. This must be
** less than 32. If it is set to anything large than that, an #error
** directive in fts5_index.c will cause the build to fail.
*/
#define FTS5_MAX_PREFIX_INDEXES 31







>
>
>
>







43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
** Constants for the largest and smallest possible 64-bit signed integers.
*/
# define LARGEST_INT64  (0xffffffff|(((i64)0x7fffffff)<<32))
# define SMALLEST_INT64 (((i64)-1) - LARGEST_INT64)

#endif

/* Truncate very long tokens to this many bytes. Hard limit is 
** (65536-1-1-4-9)==65521 bytes. The limiting factor is the 16-bit offset
** field that occurs at the start of each leaf page (see fts5_index.c). */
#define FTS5_MAX_TOKEN_SIZE 32768

/*
** Maximum number of prefix indexes on single FTS5 table. This must be
** less than 32. If it is set to anything large than that, an #error
** directive in fts5_index.c will cause the build to fail.
*/
#define FTS5_MAX_PREFIX_INDEXES 31
Changes to ext/fts5/fts5_expr.c.
1489
1490
1491
1492
1493
1494
1495

1496
1497
1498
1499
1500
1501
1502
  TokenCtx *pCtx = (TokenCtx*)pContext;
  Fts5ExprPhrase *pPhrase = pCtx->pPhrase;

  UNUSED_PARAM2(iUnused1, iUnused2);

  /* If an error has already occurred, this is a no-op */
  if( pCtx->rc!=SQLITE_OK ) return pCtx->rc;


  if( pPhrase && pPhrase->nTerm>0 && (tflags & FTS5_TOKEN_COLOCATED) ){
    Fts5ExprTerm *pSyn;
    int nByte = sizeof(Fts5ExprTerm) + sizeof(Fts5Buffer) + nToken+1;
    pSyn = (Fts5ExprTerm*)sqlite3_malloc(nByte);
    if( pSyn==0 ){
      rc = SQLITE_NOMEM;







>







1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
  TokenCtx *pCtx = (TokenCtx*)pContext;
  Fts5ExprPhrase *pPhrase = pCtx->pPhrase;

  UNUSED_PARAM2(iUnused1, iUnused2);

  /* If an error has already occurred, this is a no-op */
  if( pCtx->rc!=SQLITE_OK ) return pCtx->rc;
  if( nToken>FTS5_MAX_TOKEN_SIZE ) nToken = FTS5_MAX_TOKEN_SIZE;

  if( pPhrase && pPhrase->nTerm>0 && (tflags & FTS5_TOKEN_COLOCATED) ){
    Fts5ExprTerm *pSyn;
    int nByte = sizeof(Fts5ExprTerm) + sizeof(Fts5Buffer) + nToken+1;
    pSyn = (Fts5ExprTerm*)sqlite3_malloc(nByte);
    if( pSyn==0 ){
      rc = SQLITE_NOMEM;
2491
2492
2493
2494
2495
2496
2497

2498
2499
2500
2501
2502
2503
2504
){
  Fts5ExprCtx *p = (Fts5ExprCtx*)pCtx;
  Fts5Expr *pExpr = p->pExpr;
  int i;

  UNUSED_PARAM2(iUnused1, iUnused2);


  if( (tflags & FTS5_TOKEN_COLOCATED)==0 ) p->iOff++;
  for(i=0; i<pExpr->nPhrase; i++){
    Fts5ExprTerm *pTerm;
    if( p->aPopulator[i].bOk==0 ) continue;
    for(pTerm=&pExpr->apExprPhrase[i]->aTerm[0]; pTerm; pTerm=pTerm->pSynonym){
      int nTerm = (int)strlen(pTerm->zTerm);
      if( (nTerm==nToken || (nTerm<nToken && pTerm->bPrefix))







>







2492
2493
2494
2495
2496
2497
2498
2499
2500
2501
2502
2503
2504
2505
2506
){
  Fts5ExprCtx *p = (Fts5ExprCtx*)pCtx;
  Fts5Expr *pExpr = p->pExpr;
  int i;

  UNUSED_PARAM2(iUnused1, iUnused2);

  if( nToken>FTS5_MAX_TOKEN_SIZE ) nToken = FTS5_MAX_TOKEN_SIZE;
  if( (tflags & FTS5_TOKEN_COLOCATED)==0 ) p->iOff++;
  for(i=0; i<pExpr->nPhrase; i++){
    Fts5ExprTerm *pTerm;
    if( p->aPopulator[i].bOk==0 ) continue;
    for(pTerm=&pExpr->apExprPhrase[i]->aTerm[0]; pTerm; pTerm=pTerm->pSynonym){
      int nTerm = (int)strlen(pTerm->zTerm);
      if( (nTerm==nToken || (nTerm<nToken && pTerm->bPrefix))
Changes to ext/fts5/fts5_index.c.
2316
2317
2318
2319
2320
2321
2322












2323
2324
2325
2326
2327
2328
2329
2330
2331
2332
2333
2334
2335
2336
2337
2338
2339
2340

2341
2342
2343
2344
2345
2346
2347
2348
2349
2350
2351
2352
2353
2354
2355
2356
2357
2358
2359
2360
2361
2362
2363
2364
2365
2366
2367
2368
2369
2370
2371
2372
    pIter->iEndofDoclist = iTermOff + nExtra;
  }
  pIter->iPgidxOff = iPgidx;

  fts5SegIterLoadRowid(p, pIter);
  fts5SegIterLoadNPos(p, pIter);
}













/*
** Initialize the object pIter to point to term pTerm/nTerm within segment
** pSeg. If there is no such term in the index, the iterator is set to EOF.
**
** If an error occurs, Fts5Index.rc is set to an appropriate error code. If 
** an error has already occurred when this function is called, it is a no-op.
*/
static void fts5SegIterSeekInit(
  Fts5Index *p,                   /* FTS5 backend */
  const u8 *pTerm, int nTerm,     /* Term to seek to */
  int flags,                      /* Mask of FTS5INDEX_XXX flags */
  Fts5StructureSegment *pSeg,     /* Description of segment */
  Fts5SegIter *pIter              /* Object to populate */
){
  int iPg = 1;
  int bGe = (flags & FTS5INDEX_QUERY_SCAN);
  int bDlidx = 0;                 /* True if there is a doclist-index */


  assert( bGe==0 || (flags & FTS5INDEX_QUERY_DESC)==0 );
  assert( pTerm && nTerm );
  memset(pIter, 0, sizeof(*pIter));
  pIter->pSeg = pSeg;

  /* This block sets stack variable iPg to the leaf page number that may
  ** contain term (pTerm/nTerm), if it is present in the segment. */
  if( p->pIdxSelect==0 ){
    Fts5Config *pConfig = p->pConfig;
    fts5IndexPrepareStmt(p, &p->pIdxSelect, sqlite3_mprintf(
          "SELECT pgno FROM '%q'.'%q_idx' WHERE "
          "segid=? AND term<=? ORDER BY term DESC LIMIT 1",
          pConfig->zDb, pConfig->zName
    ));
  }
  if( p->rc ) return;
  sqlite3_bind_int(p->pIdxSelect, 1, pSeg->iSegid);
  sqlite3_bind_blob(p->pIdxSelect, 2, pTerm, nTerm, SQLITE_STATIC);
  if( SQLITE_ROW==sqlite3_step(p->pIdxSelect) ){
    i64 val = sqlite3_column_int(p->pIdxSelect, 0);
    iPg = (int)(val>>1);
    bDlidx = (val & 0x0001);
  }
  p->rc = sqlite3_reset(p->pIdxSelect);

  if( iPg<pSeg->pgnoFirst ){
    iPg = pSeg->pgnoFirst;
    bDlidx = 0;
  }

  pIter->iLeafPgno = iPg - 1;







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


















>








|
<
<
<
<
<
<
<

|
|
|
|



|







2316
2317
2318
2319
2320
2321
2322
2323
2324
2325
2326
2327
2328
2329
2330
2331
2332
2333
2334
2335
2336
2337
2338
2339
2340
2341
2342
2343
2344
2345
2346
2347
2348
2349
2350
2351
2352
2353
2354
2355
2356
2357
2358
2359
2360
2361
2362







2363
2364
2365
2366
2367
2368
2369
2370
2371
2372
2373
2374
2375
2376
2377
2378
    pIter->iEndofDoclist = iTermOff + nExtra;
  }
  pIter->iPgidxOff = iPgidx;

  fts5SegIterLoadRowid(p, pIter);
  fts5SegIterLoadNPos(p, pIter);
}

static sqlite3_stmt *fts5IdxSelectStmt(Fts5Index *p){
  if( p->pIdxSelect==0 ){
    Fts5Config *pConfig = p->pConfig;
    fts5IndexPrepareStmt(p, &p->pIdxSelect, sqlite3_mprintf(
          "SELECT pgno FROM '%q'.'%q_idx' WHERE "
          "segid=? AND term<=? ORDER BY term DESC LIMIT 1",
          pConfig->zDb, pConfig->zName
    ));
  }
  return p->pIdxSelect;
}

/*
** Initialize the object pIter to point to term pTerm/nTerm within segment
** pSeg. If there is no such term in the index, the iterator is set to EOF.
**
** If an error occurs, Fts5Index.rc is set to an appropriate error code. If 
** an error has already occurred when this function is called, it is a no-op.
*/
static void fts5SegIterSeekInit(
  Fts5Index *p,                   /* FTS5 backend */
  const u8 *pTerm, int nTerm,     /* Term to seek to */
  int flags,                      /* Mask of FTS5INDEX_XXX flags */
  Fts5StructureSegment *pSeg,     /* Description of segment */
  Fts5SegIter *pIter              /* Object to populate */
){
  int iPg = 1;
  int bGe = (flags & FTS5INDEX_QUERY_SCAN);
  int bDlidx = 0;                 /* True if there is a doclist-index */
  sqlite3_stmt *pIdxSelect = 0;

  assert( bGe==0 || (flags & FTS5INDEX_QUERY_DESC)==0 );
  assert( pTerm && nTerm );
  memset(pIter, 0, sizeof(*pIter));
  pIter->pSeg = pSeg;

  /* This block sets stack variable iPg to the leaf page number that may
  ** contain term (pTerm/nTerm), if it is present in the segment. */
  pIdxSelect = fts5IdxSelectStmt(p);







  if( p->rc ) return;
  sqlite3_bind_int(pIdxSelect, 1, pSeg->iSegid);
  sqlite3_bind_blob(pIdxSelect, 2, pTerm, nTerm, SQLITE_STATIC);
  if( SQLITE_ROW==sqlite3_step(pIdxSelect) ){
    i64 val = sqlite3_column_int(pIdxSelect, 0);
    iPg = (int)(val>>1);
    bDlidx = (val & 0x0001);
  }
  p->rc = sqlite3_reset(pIdxSelect);

  if( iPg<pSeg->pgnoFirst ){
    iPg = pSeg->pgnoFirst;
    bDlidx = 0;
  }

  pIter->iLeafPgno = iPg - 1;
3548
3549
3550
3551
3552
3553
3554












3555
3556
3557
3558
3559
3560
3561
#ifdef SQLITE_DEBUG
      for(iLvl=0; iLvl<pStruct->nLevel; iLvl++){
        for(iSeg=0; iSeg<pStruct->aLevel[iLvl].nSeg; iSeg++){
          assert( iSegid!=pStruct->aLevel[iLvl].aSeg[iSeg].iSegid );
        }
      }
      assert( iSegid>0 && iSegid<=FTS5_MAX_SEGMENT );












#endif
    }
  }

  return iSegid;
}








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







3554
3555
3556
3557
3558
3559
3560
3561
3562
3563
3564
3565
3566
3567
3568
3569
3570
3571
3572
3573
3574
3575
3576
3577
3578
3579
#ifdef SQLITE_DEBUG
      for(iLvl=0; iLvl<pStruct->nLevel; iLvl++){
        for(iSeg=0; iSeg<pStruct->aLevel[iLvl].nSeg; iSeg++){
          assert( iSegid!=pStruct->aLevel[iLvl].aSeg[iSeg].iSegid );
        }
      }
      assert( iSegid>0 && iSegid<=FTS5_MAX_SEGMENT );

      {
        sqlite3_stmt *pIdxSelect = fts5IdxSelectStmt(p);
        if( p->rc==SQLITE_OK ){
          int rc;
          u8 aBlob[2] = {0xff, 0xff};
          sqlite3_bind_int(pIdxSelect, 1, iSegid);
          sqlite3_bind_blob(pIdxSelect, 2, aBlob, 2, SQLITE_STATIC);
          assert( sqlite3_step(pIdxSelect)!=SQLITE_ROW );
          p->rc = sqlite3_reset(pIdxSelect);
        }
      }
#endif
    }
  }

  return iSegid;
}

3793
3794
3795
3796
3797
3798
3799



3800
3801
3802
3803
3804
3805
3806
  }
}

static void fts5WriteFlushLeaf(Fts5Index *p, Fts5SegWriter *pWriter){
  static const u8 zero[] = { 0x00, 0x00, 0x00, 0x00 };
  Fts5PageWriter *pPage = &pWriter->writer;
  i64 iRowid;




  assert( (pPage->pgidx.n==0)==(pWriter->bFirstTermInPage) );

  /* Set the szLeaf header field. */
  assert( 0==fts5GetU16(&pPage->buf.p[2]) );
  fts5PutU16(&pPage->buf.p[2], (u16)pPage->buf.n);








>
>
>







3811
3812
3813
3814
3815
3816
3817
3818
3819
3820
3821
3822
3823
3824
3825
3826
3827
  }
}

static void fts5WriteFlushLeaf(Fts5Index *p, Fts5SegWriter *pWriter){
  static const u8 zero[] = { 0x00, 0x00, 0x00, 0x00 };
  Fts5PageWriter *pPage = &pWriter->writer;
  i64 iRowid;

static int nCall = 0;
nCall++;

  assert( (pPage->pgidx.n==0)==(pWriter->bFirstTermInPage) );

  /* Set the szLeaf header field. */
  assert( 0==fts5GetU16(&pPage->buf.p[2]) );
  fts5PutU16(&pPage->buf.p[2], (u16)pPage->buf.n);

Changes to ext/fts5/fts5_storage.c.
365
366
367
368
369
370
371

372
373
374
375
376
377
378
  int nToken,                     /* Size of token in bytes */
  int iUnused1,                   /* Start offset of token */
  int iUnused2                    /* End offset of token */
){
  Fts5InsertCtx *pCtx = (Fts5InsertCtx*)pContext;
  Fts5Index *pIdx = pCtx->pStorage->pIndex;
  UNUSED_PARAM2(iUnused1, iUnused2);

  if( (tflags & FTS5_TOKEN_COLOCATED)==0 || pCtx->szCol==0 ){
    pCtx->szCol++;
  }
  return sqlite3Fts5IndexWrite(pIdx, pCtx->iCol, pCtx->szCol-1, pToken, nToken);
}

/*







>







365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
  int nToken,                     /* Size of token in bytes */
  int iUnused1,                   /* Start offset of token */
  int iUnused2                    /* End offset of token */
){
  Fts5InsertCtx *pCtx = (Fts5InsertCtx*)pContext;
  Fts5Index *pIdx = pCtx->pStorage->pIndex;
  UNUSED_PARAM2(iUnused1, iUnused2);
  if( nToken>FTS5_MAX_TOKEN_SIZE ) nToken = FTS5_MAX_TOKEN_SIZE;
  if( (tflags & FTS5_TOKEN_COLOCATED)==0 || pCtx->szCol==0 ){
    pCtx->szCol++;
  }
  return sqlite3Fts5IndexWrite(pIdx, pCtx->iCol, pCtx->szCol-1, pToken, nToken);
}

/*
811
812
813
814
815
816
817

818
819
820
821
822
823
824
  int bPresent;
  int ii;
  int rc = SQLITE_OK;
  int iPos;
  int iCol;

  UNUSED_PARAM2(iUnused1, iUnused2);


  if( (tflags & FTS5_TOKEN_COLOCATED)==0 || pCtx->szCol==0 ){
    pCtx->szCol++;
  }

  switch( pCtx->pConfig->eDetail ){
    case FTS5_DETAIL_FULL:







>







812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
  int bPresent;
  int ii;
  int rc = SQLITE_OK;
  int iPos;
  int iCol;

  UNUSED_PARAM2(iUnused1, iUnused2);
  if( nToken>FTS5_MAX_TOKEN_SIZE ) nToken = FTS5_MAX_TOKEN_SIZE;

  if( (tflags & FTS5_TOKEN_COLOCATED)==0 || pCtx->szCol==0 ){
    pCtx->szCol++;
  }

  switch( pCtx->pConfig->eDetail ){
    case FTS5_DETAIL_FULL:
Changes to ext/fts5/test/fts5simple.test.
443
444
445
446
447
448
449




















450
451
  execsql { INSERT INTO x1(x1) VALUES('optimize'); }
  execsql { DELETE FROM x1 WHERE rowid = 4; }
} {}
do_execsql_test 20.2 {
  INSERT INTO x1(x1) VALUES('optimize');
  INSERT INTO x1(x1) VALUES('integrity-check');
} {}





















finish_test







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


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
  execsql { INSERT INTO x1(x1) VALUES('optimize'); }
  execsql { DELETE FROM x1 WHERE rowid = 4; }
} {}
do_execsql_test 20.2 {
  INSERT INTO x1(x1) VALUES('optimize');
  INSERT INTO x1(x1) VALUES('integrity-check');
} {}

#-------------------------------------------------------------------------
reset_db
set doc "a b [string repeat x 100000]"
do_execsql_test 21.0 {
  CREATE VIRTUAL TABLE x1 USING fts5(x);
  INSERT INTO x1(rowid, x) VALUES(11111, $doc);
  INSERT INTO x1(rowid, x) VALUES(11112, $doc);
}
do_execsql_test 21.1 {
  INSERT INTO x1(x1) VALUES('integrity-check');
}
do_execsql_test 21.2 {
  SELECT rowid FROM x1($doc);
} {11111 11112}
do_execsql_test 21.3 {
  DELETE FROM x1 WHERE rowid=11111;
  INSERT INTO x1(x1) VALUES('integrity-check');
  SELECT rowid FROM x1($doc);
} {11112}

finish_test