/ Check-in [7958cbba]
Login

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

Overview
Comment:Move to an O(NlogN) algorithm for the priority queue. An insertion sort was way too slow.
Downloads: Tarball | ZIP archive | SQL archive
Timelines: family | ancestors | descendants | both | word-fuzzer
Files: files | file ages | folders
SHA1: 7958cbba736a599c1293b06602eec43dfe4fd7d1
User & Date: drh 2011-03-30 01:43:00
Context
2011-04-01
20:28
Add additional test data and documentation to the fuzzer virtual table. Closed-Leaf check-in: a6a81d4f user: drh tags: word-fuzzer
2011-03-30
01:43
Move to an O(NlogN) algorithm for the priority queue. An insertion sort was way too slow. check-in: 7958cbba user: drh tags: word-fuzzer
2011-03-29
23:41
Add support for rowid. check-in: 2cf4158f user: drh tags: word-fuzzer
Changes
Hide Diffs Unified Diffs Ignore Whitespace Patch

Changes to src/test_fuzzer.c.

62
63
64
65
66
67
68

69
70
71
72
73
74
75
..
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
...
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
...
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
...
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
...
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
394
395
...
404
405
406
407
408
409
410
411
412
413
414

415
416
417
418
419
420
421
...
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
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
...
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
*/
struct fuzzer_stem {
  char *zBasis;              /* Word being fuzzed */
  int nBasis;                /* Length of the zBasis string */
  const fuzzer_rule *pRule;  /* Current rule to apply */
  int n;                     /* Apply pRule at this character offset */
  fuzzer_cost rBaseCost;     /* Base cost of getting to zBasis */

  fuzzer_stem *pNext;        /* Next stem in rCost order */
  fuzzer_stem *pHash;        /* Next stem with same hash on zBasis */
};

/* 
** A fuzzer virtual-table object 
*/
................................................................................
  char *zClassName;          /* Name of this class.  Default: "fuzzer" */
  fuzzer_rule *pRule;        /* All active rules in this fuzzer */
  fuzzer_rule *pNewRule;     /* New rules to add when last cursor expires */
  int nCursor;               /* Number of active cursors */
};

#define FUZZER_HASH  4001    /* Hash table size */


/* A fuzzer cursor object */
struct fuzzer_cursor {
  sqlite3_vtab_cursor base;  /* Base class - must be first */
  sqlite3_int64 iRowid;      /* The rowid of the current word */
  fuzzer_vtab *pVtab;        /* The virtual table this cursor belongs to */
  fuzzer_cost rLimit;        /* Maximum cost of any term */
  fuzzer_stem *pStem;        /* Sorted list of stems for generating new terms */
  fuzzer_stem *pDone;        /* Stems already processed to completion */


  char *zBuf;                /* Temporary use buffer */
  int nBuf;                  /* Bytes allocated for zBuf */

  fuzzer_rule nullRule;      /* Null rule used first */
  fuzzer_stem *apHash[FUZZER_HASH]; /* Hash of previously generated terms */
};

/* Methods for the fuzzer module */
static int fuzzerConnect(
  sqlite3 *db,
................................................................................
      pX = fuzzerMergeRules(a[i], pX);
    }
    p->pRule = fuzzerMergeRules(p->pRule, pX);
  }
  p->nCursor++;
  return SQLITE_OK;
}












/*
** Free up all the memory allocated by a cursor.  Set it rLimit to 0
** to indicate that it is at EOF.
*/
static void fuzzerClearCursor(fuzzer_cursor *pCur, int clearHash){
  if( pCur->pStem==0 && pCur->pDone==0 ) clearHash = 0;
  do{
    while( pCur->pStem ){

      fuzzer_stem *pStem = pCur->pStem;
      pCur->pStem = pStem->pNext;
      sqlite3_free(pStem);
    }

    pCur->pStem = pCur->pDone;
    pCur->pDone = 0;
  }while( pCur->pStem );
  pCur->rLimit = (fuzzer_cost)0;
  if( clearHash ) memset(pCur->apHash, 0, sizeof(pCur->apHash));


}

/*
** Close a fuzzer cursor.
*/
static int fuzzerClose(sqlite3_vtab_cursor *cur){
  fuzzer_cursor *pCur = (fuzzer_cursor *)cur;
................................................................................
  return h % FUZZER_HASH;
}

/*
** Current cost of a stem
*/
static fuzzer_cost fuzzerCost(fuzzer_stem *pStem){
  return pStem->rBaseCost + pStem->pRule->rCost;
}

#if 0
/*
** Print a description of a fuzzer_stem on stderr.
*/
static void fuzzerStemPrint(
................................................................................
    );
  }else{
    char *zBuf = 0;
    int nBuf = 0;
    if( fuzzerRender(pStem, &zBuf, &nBuf)!=SQLITE_OK ) return;
    fprintf(stderr, "%s[%s](%d)-->{%s}(%d)%s",
      zPrefix,
      pStem->zBasis, pStem->rBaseCost, zBuf, fuzzerCost(pStem),
      zSuffix
    );
    sqlite3_free(zBuf);
  }
}
#endif

................................................................................
      if( pRule->nFrom==0
       || memcmp(&pStem->zBasis[pStem->n], pRule->zFrom, pRule->nFrom)==0
      ){
        /* Found a rewrite case.  Make sure it is not a duplicate */
        int rc = fuzzerSeen(pCur, pStem);
        if( rc<0 ) return -1;
        if( rc==0 ){

          return 1;
        }
      }
    }
    pStem->n = -1;
    pStem->pRule = pRule->pNext;
    if( pStem->pRule && fuzzerCost(pStem)>pCur->rLimit ) pStem->pRule = 0;
  }
  return 0;
}

/*


** Insert pNew into the list at pList.  Return a pointer to the new

























































** list.  The insert is done such the pNew is in the correct order
** according to fuzzer_stem.zBaseCost+fuzzer_stem.pRule->rCost.
*/
static fuzzer_stem *fuzzerInsert(fuzzer_stem *pList, fuzzer_stem *pNew){
  fuzzer_cost c1;


  if( pList==0 ){




    pNew->pNext = 0;

    return pNew;
  }
  c1 = fuzzerCost(pNew);
  if( c1 <= fuzzerCost(pList) ){


    pNew->pNext = pList;
    return pNew;




  }else{
    fuzzer_stem *pPrev;
    pPrev = pList;
    while( pPrev->pNext && fuzzerCost(pPrev->pNext)<c1 ){
      pPrev = pPrev->pNext;


    }
    pNew->pNext = pPrev->pNext;
    pPrev->pNext = pNew;
    return pList;
  }












}

/*
** Allocate a new fuzzer_stem.  Add it to the hash table but do not
** link it into either the pCur->pStem or pCur->pDone lists.
*/
static fuzzer_stem *fuzzerNewStem(
................................................................................
  if( pNew==0 ) return 0;
  memset(pNew, 0, sizeof(*pNew));
  pNew->zBasis = (char*)&pNew[1];
  pNew->nBasis = strlen(zWord);
  memcpy(pNew->zBasis, zWord, pNew->nBasis+1);
  pNew->pRule = pCur->pVtab->pRule;
  pNew->n = -1;
  pNew->rBaseCost = rBaseCost;
  h = fuzzerHash(pNew->zBasis);
  pNew->pHash = pCur->apHash[h];
  pCur->apHash[h] = pNew;

  return pNew;
}


/*
** Advance a cursor to its next row of output
*/
................................................................................

  pCur->iRowid++;

  /* Use the element the cursor is currently point to to create
  ** a new stem and insert the new stem into the priority queue.
  */
  pStem = pCur->pStem;
  if( fuzzerCost(pStem)>0 ){
    rc = fuzzerRender(pStem, &pCur->zBuf, &pCur->nBuf);
    if( rc==SQLITE_NOMEM ) return SQLITE_NOMEM;
    pNew = fuzzerNewStem(pCur, pCur->zBuf, fuzzerCost(pStem));
    if( pNew ){
      if( fuzzerAdvance(pCur, pNew)==0 ){
        pNew->pNext = pCur->pDone;
        pCur->pDone = pNew;
      }else{
        pCur->pStem = fuzzerInsert(pStem, pNew);
        if( pCur->pStem==pNew ){
          return SQLITE_OK;
        }
      }
    }else{
      return SQLITE_NOMEM;
    }
  }

  /* Adjust the priority queue so that the first element of the
  ** stem list is the next lowest cost word.
  */
  while( (pStem = pCur->pStem)!=0 ){
    if( fuzzerAdvance(pCur, pStem) ){

      pCur->pStem = pStem = fuzzerInsert(pStem->pNext, pStem);
      if( (rc = fuzzerSeen(pCur, pStem))!=0 ){
        if( rc<0 ) return SQLITE_NOMEM;
        continue;
      }
      return SQLITE_OK;  /* New word found */
    }
    pCur->pStem = pStem->pNext;
    pStem->pNext = pCur->pDone;
    pCur->pDone = pStem;
    if( pCur->pStem ){
      rc = fuzzerSeen(pCur, pCur->pStem);
      if( rc<0 ) return SQLITE_NOMEM;
      if( rc==0 ){
        return SQLITE_OK;
      }
    }
  }
................................................................................
    /* the "word" column */
    if( fuzzerRender(pCur->pStem, &pCur->zBuf, &pCur->nBuf)==SQLITE_NOMEM ){
      return SQLITE_NOMEM;
    }
    sqlite3_result_text(ctx, pCur->zBuf, -1, SQLITE_TRANSIENT);
  }else if( i==1 ){
    /* the "distance" column */
    sqlite3_result_int(ctx, fuzzerCost(pCur->pStem));
  }else{
    /* All other columns are NULL */
    sqlite3_result_null(ctx);
  }
  return SQLITE_OK;
}








>







 







>







|

>
>


>







 







>
>
>
>
>
>
>
>
>
>
>






|
<
|
>
|
|
|
<
>
|

|
<
|
>
>







 







|







 







|







 







>












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



|
|
>

<
>
>
>
>

>
|

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

<
<
<

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







 







|



>







 







|


|





|
<













>
|






|


|







 







|







62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
..
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
106
107
...
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
...
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
...
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
...
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
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
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
485
486
487
488
...
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
...
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536

537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
...
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
*/
struct fuzzer_stem {
  char *zBasis;              /* Word being fuzzed */
  int nBasis;                /* Length of the zBasis string */
  const fuzzer_rule *pRule;  /* Current rule to apply */
  int n;                     /* Apply pRule at this character offset */
  fuzzer_cost rBaseCost;     /* Base cost of getting to zBasis */
  fuzzer_cost rCostX;        /* Precomputed rBaseCost + pRule->rCost */
  fuzzer_stem *pNext;        /* Next stem in rCost order */
  fuzzer_stem *pHash;        /* Next stem with same hash on zBasis */
};

/* 
** A fuzzer virtual-table object 
*/
................................................................................
  char *zClassName;          /* Name of this class.  Default: "fuzzer" */
  fuzzer_rule *pRule;        /* All active rules in this fuzzer */
  fuzzer_rule *pNewRule;     /* New rules to add when last cursor expires */
  int nCursor;               /* Number of active cursors */
};

#define FUZZER_HASH  4001    /* Hash table size */
#define FUZZER_NQUEUE  20    /* Number of slots on the stem queue */

/* A fuzzer cursor object */
struct fuzzer_cursor {
  sqlite3_vtab_cursor base;  /* Base class - must be first */
  sqlite3_int64 iRowid;      /* The rowid of the current word */
  fuzzer_vtab *pVtab;        /* The virtual table this cursor belongs to */
  fuzzer_cost rLimit;        /* Maximum cost of any term */
  fuzzer_stem *pStem;        /* Stem with smallest rCostX */
  fuzzer_stem *pDone;        /* Stems already processed to completion */
  fuzzer_stem *aQueue[FUZZER_NQUEUE];  /* Queue of stems with higher rCostX */
  int mxQueue;               /* Largest used index in aQueue[] */
  char *zBuf;                /* Temporary use buffer */
  int nBuf;                  /* Bytes allocated for zBuf */
  int nStem;                 /* Number of stems allocated */
  fuzzer_rule nullRule;      /* Null rule used first */
  fuzzer_stem *apHash[FUZZER_HASH]; /* Hash of previously generated terms */
};

/* Methods for the fuzzer module */
static int fuzzerConnect(
  sqlite3 *db,
................................................................................
      pX = fuzzerMergeRules(a[i], pX);
    }
    p->pRule = fuzzerMergeRules(p->pRule, pX);
  }
  p->nCursor++;
  return SQLITE_OK;
}

/*
** Free all stems in a list.
*/
static void fuzzerClearStemList(fuzzer_stem *pStem){
  while( pStem ){
    fuzzer_stem *pNext = pStem->pNext;
    sqlite3_free(pStem);
    pStem = pNext;
  }
}

/*
** Free up all the memory allocated by a cursor.  Set it rLimit to 0
** to indicate that it is at EOF.
*/
static void fuzzerClearCursor(fuzzer_cursor *pCur, int clearHash){
  int i;

  fuzzerClearStemList(pCur->pStem);
  fuzzerClearStemList(pCur->pDone);
  for(i=0; i<FUZZER_NQUEUE; i++) fuzzerClearStemList(pCur->aQueue[i]);
  pCur->rLimit = (fuzzer_cost)0;
  if( clearHash && pCur->nStem ){

    pCur->mxQueue = 0;
    pCur->pStem = 0;
    pCur->pDone = 0;
    memset(pCur->aQueue, 0, sizeof(pCur->aQueue));

    memset(pCur->apHash, 0, sizeof(pCur->apHash));
  }
  pCur->nStem = 0;
}

/*
** Close a fuzzer cursor.
*/
static int fuzzerClose(sqlite3_vtab_cursor *cur){
  fuzzer_cursor *pCur = (fuzzer_cursor *)cur;
................................................................................
  return h % FUZZER_HASH;
}

/*
** Current cost of a stem
*/
static fuzzer_cost fuzzerCost(fuzzer_stem *pStem){
  return pStem->rCostX = pStem->rBaseCost + pStem->pRule->rCost;
}

#if 0
/*
** Print a description of a fuzzer_stem on stderr.
*/
static void fuzzerStemPrint(
................................................................................
    );
  }else{
    char *zBuf = 0;
    int nBuf = 0;
    if( fuzzerRender(pStem, &zBuf, &nBuf)!=SQLITE_OK ) return;
    fprintf(stderr, "%s[%s](%d)-->{%s}(%d)%s",
      zPrefix,
      pStem->zBasis, pStem->rBaseCost, zBuf, pStem->,
      zSuffix
    );
    sqlite3_free(zBuf);
  }
}
#endif

................................................................................
      if( pRule->nFrom==0
       || memcmp(&pStem->zBasis[pStem->n], pRule->zFrom, pRule->nFrom)==0
      ){
        /* Found a rewrite case.  Make sure it is not a duplicate */
        int rc = fuzzerSeen(pCur, pStem);
        if( rc<0 ) return -1;
        if( rc==0 ){
          fuzzerCost(pStem);
          return 1;
        }
      }
    }
    pStem->n = -1;
    pStem->pRule = pRule->pNext;
    if( pStem->pRule && fuzzerCost(pStem)>pCur->rLimit ) pStem->pRule = 0;
  }
  return 0;
}

/*
** The two input stem lists are both sorted in order of increasing
** rCostX.  Merge them together into a single list, sorted by rCostX, and
** return a pointer to the head of that new list.
*/
static fuzzer_stem *fuzzerMergeStems(fuzzer_stem *pA, fuzzer_stem *pB){
  fuzzer_stem head;
  fuzzer_stem *pTail;

  pTail =  &head;
  while( pA && pB ){
    if( pA->rCostX<=pB->rCostX ){
      pTail->pNext = pA;
      pTail = pA;
      pA = pA->pNext;
    }else{
      pTail->pNext = pB;
      pTail = pB;
      pB = pB->pNext;
    }
  }
  if( pA==0 ){
    pTail->pNext = pB;
  }else{
    pTail->pNext = pA;
  }
  return head.pNext;
}

/*
** Load pCur->pStem with the lowest-cost stem.  Return a pointer
** to the lowest-cost stem.
*/
static fuzzer_stem *fuzzerLowestCostStem(fuzzer_cursor *pCur){
  fuzzer_stem *pBest, *pX;
  int iBest;
  int i;

  if( pCur->pStem==0 ){
    iBest = -1;
    pBest = 0;
    for(i=0; i<=pCur->mxQueue; i++){
      pX = pCur->aQueue[i];
      if( pX==0 ) continue;
      if( pBest==0 || pBest->rCostX>pX->rCostX ){
        pBest = pX;
        iBest = i;
      }
    } 
    if( pBest ){
      pCur->aQueue[iBest] = pBest->pNext;
      pBest->pNext = 0;
      pCur->pStem = pBest;
    }
  }
  return pCur->pStem;
}

/*
** Insert pNew into queue of pending stems.  Then find the stem
** with the lowest rCostX and move it into pCur->pStem.
** list.  The insert is done such the pNew is in the correct order
** according to fuzzer_stem.zBaseCost+fuzzer_stem.pRule->rCost.
*/
static fuzzer_stem *fuzzerInsert(fuzzer_cursor *pCur, fuzzer_stem *pNew){
  fuzzer_stem *pX;
  int i;


  /* If pCur->pStem exists and is greater than pNew, then make pNew
  ** the new pCur->pStem and insert the old pCur->pStem instead.
  */
  if( (pX = pCur->pStem)!=0 && pX->rCostX>pNew->rCostX ){
    pNew->pNext = 0;
    pCur->pStem = pNew;
    pNew = pX;
  }



  /* Insert the new value */
  pNew->pNext = 0;
  pX = pNew;
  for(i=0; i<=pCur->mxQueue; i++){
    if( pCur->aQueue[i] ){
      pX = fuzzerMergeStems(pX, pCur->aQueue[i]);
      pCur->aQueue[i] = 0;
    }else{




      pCur->aQueue[i] = pX;
      break;
    }



  }
  if( i>pCur->mxQueue ){
    if( i<FUZZER_NQUEUE ){
      pCur->mxQueue = i;
      pCur->aQueue[i] = pX;
    }else{
      assert( pCur->mxQueue==FUZZER_NQUEUE-1 );
      pX = fuzzerMergeStems(pX, pCur->aQueue[FUZZER_NQUEUE-1]);
      pCur->aQueue[FUZZER_NQUEUE-1] = pX;
    }
  }

  return fuzzerLowestCostStem(pCur);
}

/*
** Allocate a new fuzzer_stem.  Add it to the hash table but do not
** link it into either the pCur->pStem or pCur->pDone lists.
*/
static fuzzer_stem *fuzzerNewStem(
................................................................................
  if( pNew==0 ) return 0;
  memset(pNew, 0, sizeof(*pNew));
  pNew->zBasis = (char*)&pNew[1];
  pNew->nBasis = strlen(zWord);
  memcpy(pNew->zBasis, zWord, pNew->nBasis+1);
  pNew->pRule = pCur->pVtab->pRule;
  pNew->n = -1;
  pNew->rBaseCost = pNew->rCostX = rBaseCost;
  h = fuzzerHash(pNew->zBasis);
  pNew->pHash = pCur->apHash[h];
  pCur->apHash[h] = pNew;
  pCur->nStem++;
  return pNew;
}


/*
** Advance a cursor to its next row of output
*/
................................................................................

  pCur->iRowid++;

  /* Use the element the cursor is currently point to to create
  ** a new stem and insert the new stem into the priority queue.
  */
  pStem = pCur->pStem;
  if( pStem->rCostX>0 ){
    rc = fuzzerRender(pStem, &pCur->zBuf, &pCur->nBuf);
    if( rc==SQLITE_NOMEM ) return SQLITE_NOMEM;
    pNew = fuzzerNewStem(pCur, pCur->zBuf, pStem->rCostX);
    if( pNew ){
      if( fuzzerAdvance(pCur, pNew)==0 ){
        pNew->pNext = pCur->pDone;
        pCur->pDone = pNew;
      }else{
        if( fuzzerInsert(pCur, pNew)==pNew ){

          return SQLITE_OK;
        }
      }
    }else{
      return SQLITE_NOMEM;
    }
  }

  /* Adjust the priority queue so that the first element of the
  ** stem list is the next lowest cost word.
  */
  while( (pStem = pCur->pStem)!=0 ){
    if( fuzzerAdvance(pCur, pStem) ){
      pCur->pStem = 0;
      pStem = fuzzerInsert(pCur, pStem);
      if( (rc = fuzzerSeen(pCur, pStem))!=0 ){
        if( rc<0 ) return SQLITE_NOMEM;
        continue;
      }
      return SQLITE_OK;  /* New word found */
    }
    pCur->pStem = 0;
    pStem->pNext = pCur->pDone;
    pCur->pDone = pStem;
    if( fuzzerLowestCostStem(pCur) ){
      rc = fuzzerSeen(pCur, pCur->pStem);
      if( rc<0 ) return SQLITE_NOMEM;
      if( rc==0 ){
        return SQLITE_OK;
      }
    }
  }
................................................................................
    /* the "word" column */
    if( fuzzerRender(pCur->pStem, &pCur->zBuf, &pCur->nBuf)==SQLITE_NOMEM ){
      return SQLITE_NOMEM;
    }
    sqlite3_result_text(ctx, pCur->zBuf, -1, SQLITE_TRANSIENT);
  }else if( i==1 ){
    /* the "distance" column */
    sqlite3_result_int(ctx, pCur->pStem->rCostX);
  }else{
    /* All other columns are NULL */
    sqlite3_result_null(ctx);
  }
  return SQLITE_OK;
}