SQLite

Check-in [cc83991caa]
Login

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

Overview
Comment:If a prefix index of size N is not present, use a prefix index of size N+1 along with the terms index for queries for prefixes of length N.
Downloads: Tarball | ZIP archive
Timelines: family | ancestors | descendants | both | fts3-prefix-search
Files: files | file ages | folders
SHA1: cc83991caae7c7d647432d5711b6cd80228c3002
User & Date: dan 2011-05-25 19:17:32.713
Context
2011-05-28
15:57
Minor changes made while planning a larger change. (check-in: 84097a4c75 user: dan tags: fts3-prefix-search)
2011-05-25
19:17
If a prefix index of size N is not present, use a prefix index of size N+1 along with the terms index for queries for prefixes of length N. (check-in: cc83991caa user: dan tags: fts3-prefix-search)
18:47
Merge trunk changes into experimental fts3-prefix-search branch. (check-in: f0f0a03db2 user: dan tags: fts3-prefix-search)
Changes
Unified Diff Ignore Whitespace Patch
Changes to ext/fts3/fts3.c.
2233
2234
2235
2236
2237
2238
2239
2240
2241
2242
2243
2244
2245
2246
2247
2248
2249
2250
2251
2252
2253
2254
2255
2256
2257
2258
2259
2260
2261
2262
2263
2264
2265
2266
2267
2268
2269
2270
2271
2272
2273
2274
2275
2276
2277
2278
2279
2280
2281
    }
    pCsr->apSegment = apNew;
  }
  pCsr->apSegment[pCsr->nSegment++] = pNew;
  return SQLITE_OK;
}

/*
** Set up a cursor object for iterating through a full-text index or a 
** single level therein.
*/
int sqlite3Fts3SegReaderCursor(
  Fts3Table *p,                   /* FTS3 table handle */
  int iIndex,                     /* Index to search (from 0 to p->nIndex-1) */
  int iLevel,                     /* Level of segments to scan */
  const char *zTerm,              /* Term to query for */
  int nTerm,                      /* Size of zTerm in bytes */
  int isPrefix,                   /* True for a prefix search */
  int isScan,                     /* True to scan from zTerm to EOF */
  Fts3SegReaderCursor *pCsr       /* Cursor object to populate */
){
  int rc = SQLITE_OK;
  int rc2;
  int iAge = 0;
  sqlite3_stmt *pStmt = 0;

  assert( iIndex>=0 && iIndex<p->nIndex );
  assert( iLevel==FTS3_SEGCURSOR_ALL
      ||  iLevel==FTS3_SEGCURSOR_PENDING 
      ||  iLevel>=0
  );
  assert( iLevel<FTS3_SEGDIR_MAXLEVEL );
  assert( FTS3_SEGCURSOR_ALL<0 && FTS3_SEGCURSOR_PENDING<0 );
  assert( iLevel==FTS3_SEGCURSOR_ALL || (zTerm==0 && isPrefix==1) );
  assert( isPrefix==0 || isScan==0 );

  /* "isScan" is only set to true by the ft4aux module, an ordinary
  ** full-text tables. */
  assert( isScan==0 || p->aIndex==0 );

  memset(pCsr, 0, sizeof(Fts3SegReaderCursor));

  /* If iLevel is less than 0 and this is not a scan, include a seg-reader 
  ** for the pending-terms. If this is a scan, then this call must be being
  ** made by an fts4aux module, not an FTS table. In this case calling
  ** Fts3SegReaderPending might segfault, as the data structures used by 
  ** fts4aux are not completely populated. So it's easiest to filter these
  ** calls out here.  */
  if( iLevel<0 && p->aIndex ){







<
<
<
<
|











<


<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<







2233
2234
2235
2236
2237
2238
2239




2240
2241
2242
2243
2244
2245
2246
2247
2248
2249
2250
2251

2252
2253
















2254
2255
2256
2257
2258
2259
2260
    }
    pCsr->apSegment = apNew;
  }
  pCsr->apSegment[pCsr->nSegment++] = pNew;
  return SQLITE_OK;
}





static int fts3SegReaderCursor(
  Fts3Table *p,                   /* FTS3 table handle */
  int iIndex,                     /* Index to search (from 0 to p->nIndex-1) */
  int iLevel,                     /* Level of segments to scan */
  const char *zTerm,              /* Term to query for */
  int nTerm,                      /* Size of zTerm in bytes */
  int isPrefix,                   /* True for a prefix search */
  int isScan,                     /* True to scan from zTerm to EOF */
  Fts3SegReaderCursor *pCsr       /* Cursor object to populate */
){
  int rc = SQLITE_OK;
  int rc2;

  sqlite3_stmt *pStmt = 0;

















  /* If iLevel is less than 0 and this is not a scan, include a seg-reader 
  ** for the pending-terms. If this is a scan, then this call must be being
  ** made by an fts4aux module, not an FTS table. In this case calling
  ** Fts3SegReaderPending might segfault, as the data structures used by 
  ** fts4aux are not completely populated. So it's easiest to filter these
  ** calls out here.  */
  if( iLevel<0 && p->aIndex ){
2306
2307
2308
2309
2310
2311
2312
2313
2314
2315
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
      if( iStartBlock && zTerm ){
        sqlite3_int64 *pi = (isPrefix ? &iLeavesEndBlock : 0);
        rc = fts3SelectLeaf(p, zTerm, nTerm, zRoot, nRoot, &iStartBlock, pi);
        if( rc!=SQLITE_OK ) goto finished;
        if( isPrefix==0 && isScan==0 ) iLeavesEndBlock = iStartBlock;
      }
 
      rc = sqlite3Fts3SegReaderNew(iAge, iStartBlock, iLeavesEndBlock,
          iEndBlock, zRoot, nRoot, &pSeg
      );
      if( rc!=SQLITE_OK ) goto finished;
      rc = fts3SegReaderCursorAppend(pCsr, pSeg);
      iAge++;
    }
  }

 finished:
  rc2 = sqlite3_reset(pStmt);
  if( rc==SQLITE_DONE ) rc = rc2;

  return rc;
}













































static int fts3TermSegReaderCursor(
  Fts3Cursor *pCsr,               /* Virtual table cursor handle */
  const char *zTerm,              /* Term to query for */
  int nTerm,                      /* Size of zTerm in bytes */
  int isPrefix,                   /* True for a prefix search */
  Fts3SegReaderCursor **ppSegcsr  /* OUT: Allocated seg-reader cursor */
){
  Fts3SegReaderCursor *pSegcsr;   /* Object to allocate and return */
  int rc = SQLITE_NOMEM;          /* Return code */

  pSegcsr = sqlite3_malloc(sizeof(Fts3SegReaderCursor));
  if( pSegcsr ){
    int i;
    int nCost = 0;
    int bFound = 0;               /* True once an index has been found */
    Fts3Table *p = (Fts3Table *)pCsr->base.pVtab;

    if( isPrefix ){
      for(i=1; i<p->nIndex; i++){
        if( p->aIndex[i].nPrefix==nTerm ){
          bFound = 1;
          rc = sqlite3Fts3SegReaderCursor(
              p, i, FTS3_SEGCURSOR_ALL, zTerm, nTerm, 0, 0, pSegcsr);


          break;









        }
      }
    }

    if( bFound==0 ){
      rc = sqlite3Fts3SegReaderCursor(
          p, 0, FTS3_SEGCURSOR_ALL, zTerm, nTerm, isPrefix, 0, pSegcsr







|
|



<









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




















|




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







2285
2286
2287
2288
2289
2290
2291
2292
2293
2294
2295
2296

2297
2298
2299
2300
2301
2302
2303
2304
2305
2306
2307
2308
2309
2310
2311
2312
2313
2314
2315
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
2379
2380
2381
2382
2383
2384
2385
2386
2387
2388
2389
2390
2391
2392
      if( iStartBlock && zTerm ){
        sqlite3_int64 *pi = (isPrefix ? &iLeavesEndBlock : 0);
        rc = fts3SelectLeaf(p, zTerm, nTerm, zRoot, nRoot, &iStartBlock, pi);
        if( rc!=SQLITE_OK ) goto finished;
        if( isPrefix==0 && isScan==0 ) iLeavesEndBlock = iStartBlock;
      }
 
      rc = sqlite3Fts3SegReaderNew(pCsr->nSegment+1, 
          iStartBlock, iLeavesEndBlock, iEndBlock, zRoot, nRoot, &pSeg
      );
      if( rc!=SQLITE_OK ) goto finished;
      rc = fts3SegReaderCursorAppend(pCsr, pSeg);

    }
  }

 finished:
  rc2 = sqlite3_reset(pStmt);
  if( rc==SQLITE_DONE ) rc = rc2;

  return rc;
}

/*
** Set up a cursor object for iterating through a full-text index or a 
** single level therein.
*/
int sqlite3Fts3SegReaderCursor(
  Fts3Table *p,                   /* FTS3 table handle */
  int iIndex,                     /* Index to search (from 0 to p->nIndex-1) */
  int iLevel,                     /* Level of segments to scan */
  const char *zTerm,              /* Term to query for */
  int nTerm,                      /* Size of zTerm in bytes */
  int isPrefix,                   /* True for a prefix search */
  int isScan,                     /* True to scan from zTerm to EOF */
  Fts3SegReaderCursor *pCsr       /* Cursor object to populate */
){
  assert( iIndex>=0 && iIndex<p->nIndex );
  assert( iLevel==FTS3_SEGCURSOR_ALL
      ||  iLevel==FTS3_SEGCURSOR_PENDING 
      ||  iLevel>=0
  );
  assert( iLevel<FTS3_SEGDIR_MAXLEVEL );
  assert( FTS3_SEGCURSOR_ALL<0 && FTS3_SEGCURSOR_PENDING<0 );
  assert( isPrefix==0 || isScan==0 );

  /* "isScan" is only set to true by the ft4aux module, an ordinary
  ** full-text tables. */
  assert( isScan==0 || p->aIndex==0 );

  memset(pCsr, 0, sizeof(Fts3SegReaderCursor));

  return fts3SegReaderCursor(
      p, iIndex, iLevel, zTerm, nTerm, isPrefix, isScan, pCsr
  );
}

static int fts3SegReaderCursorAddZero(
  Fts3Table *p,
  const char *zTerm,
  int nTerm,
  Fts3SegReaderCursor *pCsr
){
  return fts3SegReaderCursor(p, 0, FTS3_SEGCURSOR_ALL, zTerm, nTerm, 0, 0,pCsr);
}


static int fts3TermSegReaderCursor(
  Fts3Cursor *pCsr,               /* Virtual table cursor handle */
  const char *zTerm,              /* Term to query for */
  int nTerm,                      /* Size of zTerm in bytes */
  int isPrefix,                   /* True for a prefix search */
  Fts3SegReaderCursor **ppSegcsr  /* OUT: Allocated seg-reader cursor */
){
  Fts3SegReaderCursor *pSegcsr;   /* Object to allocate and return */
  int rc = SQLITE_NOMEM;          /* Return code */

  pSegcsr = sqlite3_malloc(sizeof(Fts3SegReaderCursor));
  if( pSegcsr ){
    int i;
    int nCost = 0;
    int bFound = 0;               /* True once an index has been found */
    Fts3Table *p = (Fts3Table *)pCsr->base.pVtab;

    if( isPrefix ){
      for(i=1; bFound==0 && i<p->nIndex; i++){
        if( p->aIndex[i].nPrefix==nTerm ){
          bFound = 1;
          rc = sqlite3Fts3SegReaderCursor(
              p, i, FTS3_SEGCURSOR_ALL, zTerm, nTerm, 0, 0, pSegcsr);
        }
      }

      for(i=1; bFound==0 && i<p->nIndex; i++){
        if( p->aIndex[i].nPrefix==nTerm+1 ){
          bFound = 1;
          rc = sqlite3Fts3SegReaderCursor(
              p, i, FTS3_SEGCURSOR_ALL, zTerm, nTerm, 1, 0, pSegcsr
          );
          if( rc==SQLITE_OK ){
            rc = fts3SegReaderCursorAddZero(p, zTerm, nTerm, pSegcsr);
          }
        }
      }
    }

    if( bFound==0 ){
      rc = sqlite3Fts3SegReaderCursor(
          p, 0, FTS3_SEGCURSOR_ALL, zTerm, nTerm, isPrefix, 0, pSegcsr
Changes to ext/fts3/fts3_write.c.
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
  int rc;
  sqlite3_stmt *pStmt = 0;

  assert( iLevel==FTS3_SEGCURSOR_ALL || iLevel>=0 );
  assert( iLevel<FTS3_SEGDIR_MAXLEVEL );
  assert( iIndex>=0 && iIndex<p->nIndex );

  if( iLevel==FTS3_SEGCURSOR_ALL ){
    /* "SELECT * FROM %_segdir WHERE level BETWEEN ? AND ? ORDER BY ..." */
    rc = fts3SqlStmt(p, SQL_SELECT_LEVEL_RANGE, &pStmt, 0);
    if( rc==SQLITE_OK ){ 
      sqlite3_bind_int(pStmt, 1, iIndex*FTS3_SEGDIR_MAXLEVEL);
      sqlite3_bind_int(pStmt, 2, (iIndex+1)*FTS3_SEGDIR_MAXLEVEL-1);
    }
  }else{







|







418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
  int rc;
  sqlite3_stmt *pStmt = 0;

  assert( iLevel==FTS3_SEGCURSOR_ALL || iLevel>=0 );
  assert( iLevel<FTS3_SEGDIR_MAXLEVEL );
  assert( iIndex>=0 && iIndex<p->nIndex );

  if( iLevel<0 ){
    /* "SELECT * FROM %_segdir WHERE level BETWEEN ? AND ? ORDER BY ..." */
    rc = fts3SqlStmt(p, SQL_SELECT_LEVEL_RANGE, &pStmt, 0);
    if( rc==SQLITE_OK ){ 
      sqlite3_bind_int(pStmt, 1, iIndex*FTS3_SEGDIR_MAXLEVEL);
      sqlite3_bind_int(pStmt, 2, (iIndex+1)*FTS3_SEGDIR_MAXLEVEL-1);
    }
  }else{
Changes to test/fts3prefix.test.
189
190
191
192
193
194
195






196
197
} {{four five six}}
do_execsql_test 4.3 {
  SELECT * FROM t3 WHERE t3 MATCH 'four*'
} {{four five six}}
do_execsql_test 4.4 {
  SELECT * FROM t3 WHERE t3 MATCH 's*'
} {{four five six} {seven eight nine}}







finish_test







>
>
>
>
>
>


189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
} {{four five six}}
do_execsql_test 4.3 {
  SELECT * FROM t3 WHERE t3 MATCH 'four*'
} {{four five six}}
do_execsql_test 4.4 {
  SELECT * FROM t3 WHERE t3 MATCH 's*'
} {{four five six} {seven eight nine}}
do_execsql_test 4.5 {
  SELECT * FROM t3 WHERE t3 MATCH 'sev*'
} {{seven eight nine}}
do_execsql_test 4.6 {
  SELECT * FROM t3 WHERE t3 MATCH 'one*'
} {{one two three}}

finish_test