SQLite

Check-in [0888838371]
Login

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

Overview
Comment:Update the spellfix virtual table extension so that an explicit "top = ?" constraint works even if there is also a "distance < ?" or "distance <= ?" constraint.
Downloads: Tarball | ZIP archive
Timelines: family | ancestors | descendants | both | trunk
Files: files | file ages | folders
SHA1: 0888838371608558f31d5bcb5fed1c8861aa52c1
User & Date: dan 2015-08-05 15:29:32.743
Context
2015-08-05
19:35
Add tests for the matchinfo-like test function. Fix problems found in test and fts5 code by doing so. (check-in: 9e3aafe44a user: dan tags: trunk)
15:29
Update the spellfix virtual table extension so that an explicit "top = ?" constraint works even if there is also a "distance < ?" or "distance <= ?" constraint. (check-in: 0888838371 user: dan tags: trunk)
08:01
Fix some problems in test code detected by cppcheck. (check-in: c9ff4ca4a0 user: dan tags: trunk)
Changes
Unified Diff Ignore Whitespace Patch
Changes to ext/misc/spellfix.c.
1766
1767
1768
1769
1770
1771
1772

1773
1774
1775
1776
1777
1778
1779
};

/* Fuzzy-search cursor object */
struct spellfix1_cursor {
  sqlite3_vtab_cursor base;    /* Base class - must be first */
  spellfix1_vtab *pVTab;       /* The table to which this cursor belongs */
  char *zPattern;              /* rhs of MATCH clause */

  int nRow;                    /* Number of rows of content */
  int nAlloc;                  /* Number of allocated rows */
  int iRow;                    /* Current row of content */
  int iLang;                   /* Value of the langid= constraint */
  int iTop;                    /* Value of the top= constraint */
  int iScope;                  /* Value of the scope= constraint */
  int nSearch;                 /* Number of vocabulary items checked */







>







1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
};

/* Fuzzy-search cursor object */
struct spellfix1_cursor {
  sqlite3_vtab_cursor base;    /* Base class - must be first */
  spellfix1_vtab *pVTab;       /* The table to which this cursor belongs */
  char *zPattern;              /* rhs of MATCH clause */
  int idxNum;                  /* idxNum value passed to xFilter() */
  int nRow;                    /* Number of rows of content */
  int nAlloc;                  /* Number of allocated rows */
  int iRow;                    /* Current row of content */
  int iLang;                   /* Value of the langid= constraint */
  int iTop;                    /* Value of the top= constraint */
  int iScope;                  /* Value of the scope= constraint */
  int nSearch;                 /* Number of vocabulary items checked */
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
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
2155
2156
2157
2158
2159
2160
2161
2162
2163
2164
2165
2166
2167
2168
2169
2170
  spellfix1ResetCursor(pCur);
  spellfix1ResizeCursor(pCur, 0);
  sqlite3_free(pCur->zPattern);
  sqlite3_free(pCur);
  return SQLITE_OK;
}

/*
** Search for terms of these forms:
**
**   (A)    word MATCH $str
**   (B)    langid == $langid
**   (C)    top = $top
**   (D)    scope = $scope
**   (E)    distance < $distance
**   (F)    distance <= $distance


**   (G)    rowid = $rowid

**
** The plan number is a bit mask formed with these bits:
**
**   0x01   (A) is found
**   0x02   (B) is found
**   0x04   (C) is found
**   0x08   (D) is found
**   0x10   (E) is found
**   0x20   (F) is found
**   0x40   (G) is found
**
** filter.argv[*] values contains $str, $langid, $top, $scope and $rowid
** if specified and in that order.
*/
static int spellfix1BestIndex(sqlite3_vtab *tab, sqlite3_index_info *pIdxInfo){
  int iPlan = 0;
  int iLangTerm = -1;
  int iTopTerm = -1;
  int iScopeTerm = -1;
  int iDistTerm = -1;
  int iRowidTerm = -1;
  int i;
  const struct sqlite3_index_constraint *pConstraint;
  pConstraint = pIdxInfo->aConstraint;
  for(i=0; i<pIdxInfo->nConstraint; i++, pConstraint++){
    if( pConstraint->usable==0 ) continue;

    /* Terms of the form:  word MATCH $str */
    if( (iPlan & 1)==0 
     && pConstraint->iColumn==SPELLFIX_COL_WORD
     && pConstraint->op==SQLITE_INDEX_CONSTRAINT_MATCH
    ){
      iPlan |= 1;
      pIdxInfo->aConstraintUsage[i].argvIndex = 1;
      pIdxInfo->aConstraintUsage[i].omit = 1;
    }

    /* Terms of the form:  langid = $langid  */
    if( (iPlan & 2)==0
     && pConstraint->iColumn==SPELLFIX_COL_LANGID
     && pConstraint->op==SQLITE_INDEX_CONSTRAINT_EQ
    ){
      iPlan |= 2;
      iLangTerm = i;
    }

    /* Terms of the form:  top = $top */
    if( (iPlan & 4)==0
     && pConstraint->iColumn==SPELLFIX_COL_TOP
     && pConstraint->op==SQLITE_INDEX_CONSTRAINT_EQ
    ){
      iPlan |= 4;
      iTopTerm = i;
    }

    /* Terms of the form:  scope = $scope */
    if( (iPlan & 8)==0
     && pConstraint->iColumn==SPELLFIX_COL_SCOPE
     && pConstraint->op==SQLITE_INDEX_CONSTRAINT_EQ
    ){
      iPlan |= 8;
      iScopeTerm = i;
    }

    /* Terms of the form:  distance < $dist or distance <= $dist */
    if( (iPlan & (16|32))==0
     && pConstraint->iColumn==SPELLFIX_COL_DISTANCE
     && (pConstraint->op==SQLITE_INDEX_CONSTRAINT_LT
          || pConstraint->op==SQLITE_INDEX_CONSTRAINT_LE)
    ){
      iPlan |= pConstraint->op==SQLITE_INDEX_CONSTRAINT_LT ? 16 : 32;




      iDistTerm = i;
    }

    /* Terms of the form:  distance < $dist or distance <= $dist */
    if( (iPlan & 64)==0
     && pConstraint->iColumn<0
     && pConstraint->op==SQLITE_INDEX_CONSTRAINT_EQ
    ){
      iPlan |= 64;
      iRowidTerm = i;
    }
  }
  if( iPlan&1 ){
    int idx = 2;
    pIdxInfo->idxNum = iPlan;
    if( pIdxInfo->nOrderBy==1
     && pIdxInfo->aOrderBy[0].iColumn==SPELLFIX_COL_SCORE
     && pIdxInfo->aOrderBy[0].desc==0
    ){
      pIdxInfo->orderByConsumed = 1;  /* Default order by iScore */
    }
    if( iPlan&2 ){
      pIdxInfo->aConstraintUsage[iLangTerm].argvIndex = idx++;
      pIdxInfo->aConstraintUsage[iLangTerm].omit = 1;
    }
    if( iPlan&4 ){
      pIdxInfo->aConstraintUsage[iTopTerm].argvIndex = idx++;
      pIdxInfo->aConstraintUsage[iTopTerm].omit = 1;
    }
    if( iPlan&8 ){
      pIdxInfo->aConstraintUsage[iScopeTerm].argvIndex = idx++;
      pIdxInfo->aConstraintUsage[iScopeTerm].omit = 1;
    }
    if( iPlan&(16|32) ){
      pIdxInfo->aConstraintUsage[iDistTerm].argvIndex = idx++;
      pIdxInfo->aConstraintUsage[iDistTerm].omit = 1;
    }
    pIdxInfo->estimatedCost = 1e5;
  }else if( (iPlan & 64) ){
    pIdxInfo->idxNum = 64;
    pIdxInfo->aConstraintUsage[iRowidTerm].argvIndex = 1;
    pIdxInfo->aConstraintUsage[iRowidTerm].omit = 1;
    pIdxInfo->estimatedCost = 5;
  }else{
    pIdxInfo->idxNum = 0;
    pIdxInfo->estimatedCost = 1e50;
  }







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

|
|
<
<
<
<
<
<
<


















|



|





|



|




|



|




|



|




|




|
>
>
>
>




|



|



|








|



|



|



|




|
|







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
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
2155
2156
2157
2158
2159
2160
2161
2162
2163
2164
2165
2166
2167
2168
  spellfix1ResetCursor(pCur);
  spellfix1ResizeCursor(pCur, 0);
  sqlite3_free(pCur->zPattern);
  sqlite3_free(pCur);
  return SQLITE_OK;
}




#define SPELLFIX_IDXNUM_MATCH  0x01         /* word MATCH $str */
#define SPELLFIX_IDXNUM_LANGID 0x02         /* langid == $langid */
#define SPELLFIX_IDXNUM_TOP    0x04         /* top = $top */
#define SPELLFIX_IDXNUM_SCOPE  0x08         /* scope = $scope */
#define SPELLFIX_IDXNUM_DISTLT 0x10         /* distance < $distance */
#define SPELLFIX_IDXNUM_DISTLE 0x20         /* distance <= $distance */
#define SPELLFIX_IDXNUM_ROWID  0x40         /* rowid = $rowid */
#define SPELLFIX_IDXNUM_DIST   (0x10|0x20)  /* DISTLT and DISTLE */

/*
**
** The plan number is a bitmask of the SPELLFIX_IDXNUM_* values defined
** above.







**
** filter.argv[*] values contains $str, $langid, $top, $scope and $rowid
** if specified and in that order.
*/
static int spellfix1BestIndex(sqlite3_vtab *tab, sqlite3_index_info *pIdxInfo){
  int iPlan = 0;
  int iLangTerm = -1;
  int iTopTerm = -1;
  int iScopeTerm = -1;
  int iDistTerm = -1;
  int iRowidTerm = -1;
  int i;
  const struct sqlite3_index_constraint *pConstraint;
  pConstraint = pIdxInfo->aConstraint;
  for(i=0; i<pIdxInfo->nConstraint; i++, pConstraint++){
    if( pConstraint->usable==0 ) continue;

    /* Terms of the form:  word MATCH $str */
    if( (iPlan & SPELLFIX_IDXNUM_MATCH)==0 
     && pConstraint->iColumn==SPELLFIX_COL_WORD
     && pConstraint->op==SQLITE_INDEX_CONSTRAINT_MATCH
    ){
      iPlan |= SPELLFIX_IDXNUM_MATCH;
      pIdxInfo->aConstraintUsage[i].argvIndex = 1;
      pIdxInfo->aConstraintUsage[i].omit = 1;
    }

    /* Terms of the form:  langid = $langid  */
    if( (iPlan & SPELLFIX_IDXNUM_LANGID)==0
     && pConstraint->iColumn==SPELLFIX_COL_LANGID
     && pConstraint->op==SQLITE_INDEX_CONSTRAINT_EQ
    ){
      iPlan |= SPELLFIX_IDXNUM_LANGID;
      iLangTerm = i;
    }

    /* Terms of the form:  top = $top */
    if( (iPlan & SPELLFIX_IDXNUM_TOP)==0
     && pConstraint->iColumn==SPELLFIX_COL_TOP
     && pConstraint->op==SQLITE_INDEX_CONSTRAINT_EQ
    ){
      iPlan |= SPELLFIX_IDXNUM_TOP;
      iTopTerm = i;
    }

    /* Terms of the form:  scope = $scope */
    if( (iPlan & SPELLFIX_IDXNUM_SCOPE)==0
     && pConstraint->iColumn==SPELLFIX_COL_SCOPE
     && pConstraint->op==SQLITE_INDEX_CONSTRAINT_EQ
    ){
      iPlan |= SPELLFIX_IDXNUM_SCOPE;
      iScopeTerm = i;
    }

    /* Terms of the form:  distance < $dist or distance <= $dist */
    if( (iPlan & SPELLFIX_IDXNUM_DIST)==0
     && pConstraint->iColumn==SPELLFIX_COL_DISTANCE
     && (pConstraint->op==SQLITE_INDEX_CONSTRAINT_LT
          || pConstraint->op==SQLITE_INDEX_CONSTRAINT_LE)
    ){
      if( pConstraint->op==SQLITE_INDEX_CONSTRAINT_LT ){
        iPlan |= SPELLFIX_IDXNUM_DISTLT;
      }else{
        iPlan |= SPELLFIX_IDXNUM_DISTLE;
      }
      iDistTerm = i;
    }

    /* Terms of the form:  distance < $dist or distance <= $dist */
    if( (iPlan & SPELLFIX_IDXNUM_ROWID)==0
     && pConstraint->iColumn<0
     && pConstraint->op==SQLITE_INDEX_CONSTRAINT_EQ
    ){
      iPlan |= SPELLFIX_IDXNUM_ROWID;
      iRowidTerm = i;
    }
  }
  if( iPlan&SPELLFIX_IDXNUM_MATCH ){
    int idx = 2;
    pIdxInfo->idxNum = iPlan;
    if( pIdxInfo->nOrderBy==1
     && pIdxInfo->aOrderBy[0].iColumn==SPELLFIX_COL_SCORE
     && pIdxInfo->aOrderBy[0].desc==0
    ){
      pIdxInfo->orderByConsumed = 1;  /* Default order by iScore */
    }
    if( iPlan&SPELLFIX_IDXNUM_LANGID ){
      pIdxInfo->aConstraintUsage[iLangTerm].argvIndex = idx++;
      pIdxInfo->aConstraintUsage[iLangTerm].omit = 1;
    }
    if( iPlan&SPELLFIX_IDXNUM_TOP ){
      pIdxInfo->aConstraintUsage[iTopTerm].argvIndex = idx++;
      pIdxInfo->aConstraintUsage[iTopTerm].omit = 1;
    }
    if( iPlan&SPELLFIX_IDXNUM_SCOPE ){
      pIdxInfo->aConstraintUsage[iScopeTerm].argvIndex = idx++;
      pIdxInfo->aConstraintUsage[iScopeTerm].omit = 1;
    }
    if( iPlan&SPELLFIX_IDXNUM_DIST ){
      pIdxInfo->aConstraintUsage[iDistTerm].argvIndex = idx++;
      pIdxInfo->aConstraintUsage[iDistTerm].omit = 1;
    }
    pIdxInfo->estimatedCost = 1e5;
  }else if( (iPlan & SPELLFIX_IDXNUM_ROWID) ){
    pIdxInfo->idxNum = SPELLFIX_IDXNUM_ROWID;
    pIdxInfo->aConstraintUsage[iRowidTerm].argvIndex = 1;
    pIdxInfo->aConstraintUsage[iRowidTerm].omit = 1;
    pIdxInfo->estimatedCost = 5;
  }else{
    pIdxInfo->idxNum = 0;
    pIdxInfo->estimatedCost = 1e50;
  }
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
      iDist = editdist1(p->zPattern, zK1, 0);
    }
    if( iDist<0 ){
      p->rc = SQLITE_NOMEM;
      break;
    }
    pCur->nSearch++;
    iScore = spellfix1Score(iDist,iRank);







    if( p->iMaxDist>=0 ){
      if( iDist>p->iMaxDist ) continue;
      if( pCur->nRow>=pCur->nAlloc-1 ){
        spellfix1ResizeCursor(pCur, pCur->nAlloc*2 + 10);
        if( pCur->a==0 ) break;
      }

      idx = pCur->nRow;

    }else if( pCur->nRow<pCur->nAlloc ){
      idx = pCur->nRow;
    }else if( iScore<iWorst ){
      idx = idxWorst;
      sqlite3_free(pCur->a[idx].zWord);
    }else{
      continue;
    }

    pCur->a[idx].zWord = sqlite3_mprintf("%s", sqlite3_column_text(pStmt, 1));
    if( pCur->a[idx].zWord==0 ){
      p->rc = SQLITE_NOMEM;
      break;
    }
    pCur->a[idx].iRowid = sqlite3_column_int64(pStmt, 0);
    pCur->a[idx].iRank = iRank;







|
>
>
>
>
>
>
>


|



>
|
>
|







>







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
      iDist = editdist1(p->zPattern, zK1, 0);
    }
    if( iDist<0 ){
      p->rc = SQLITE_NOMEM;
      break;
    }
    pCur->nSearch++;
    
    /* If there is a "distance < $dist" or "distance <= $dist" constraint,
    ** check if this row meets it. If not, jump back up to the top of the
    ** loop to process the next row. Otherwise, if the row does match the
    ** distance constraint, check if the pCur->a[] array is already full.
    ** If it is and no explicit "top = ?" constraint was present in the
    ** query, grow the array to ensure there is room for the new entry. */
    assert( (p->iMaxDist>=0)==((pCur->idxNum & SPELLFIX_IDXNUM_DIST) ? 1 : 0) );
    if( p->iMaxDist>=0 ){
      if( iDist>p->iMaxDist ) continue;
      if( pCur->nRow>=pCur->nAlloc && (pCur->idxNum & SPELLFIX_IDXNUM_TOP)==0 ){
        spellfix1ResizeCursor(pCur, pCur->nAlloc*2 + 10);
        if( pCur->a==0 ) break;
      }
    }

    iScore = spellfix1Score(iDist,iRank);
    if( pCur->nRow<pCur->nAlloc ){
      idx = pCur->nRow;
    }else if( iScore<iWorst ){
      idx = idxWorst;
      sqlite3_free(pCur->a[idx].zWord);
    }else{
      continue;
    }

    pCur->a[idx].zWord = sqlite3_mprintf("%s", sqlite3_column_text(pStmt, 1));
    if( pCur->a[idx].zWord==0 ){
      p->rc = SQLITE_NOMEM;
      break;
    }
    pCur->a[idx].iRowid = sqlite3_column_int64(pStmt, 0);
    pCur->a[idx].iRank = iRank;
2357
2358
2359
2360
2361
2362
2363
2364
2365
2366
2367

2368
2369
2370
2371
2372
2373
2374

/*
** This version of the xFilter method work if the MATCH term is present
** and we are doing a scan.
*/
static int spellfix1FilterForMatch(
  spellfix1_cursor *pCur,
  int idxNum,
  int argc,
  sqlite3_value **argv
){

  const unsigned char *zMatchThis;   /* RHS of the MATCH operator */
  EditDist3FromString *pMatchStr3 = 0; /* zMatchThis as an editdist string */
  char *zPattern;                    /* Transliteration of zMatchThis */
  int nPattern;                      /* Length of zPattern */
  int iLimit = 20;                   /* Max number of rows of output */
  int iScope = 3;                    /* Use this many characters of zClass */
  int iLang = 0;                     /* Language code */







<



>







2365
2366
2367
2368
2369
2370
2371

2372
2373
2374
2375
2376
2377
2378
2379
2380
2381
2382

/*
** This version of the xFilter method work if the MATCH term is present
** and we are doing a scan.
*/
static int spellfix1FilterForMatch(
  spellfix1_cursor *pCur,

  int argc,
  sqlite3_value **argv
){
  int idxNum = pCur->idxNum;
  const unsigned char *zMatchThis;   /* RHS of the MATCH operator */
  EditDist3FromString *pMatchStr3 = 0; /* zMatchThis as an editdist string */
  char *zPattern;                    /* Transliteration of zMatchThis */
  int nPattern;                      /* Length of zPattern */
  int iLimit = 20;                   /* Max number of rows of output */
  int iScope = 3;                    /* Use this many characters of zClass */
  int iLang = 0;                     /* Language code */
2472
2473
2474
2475
2476
2477
2478
2479
2480
2481
2482
2483

2484
2485
2486
2487
2488
2489
2490
}

/*
** This version of xFilter handles a full-table scan case
*/
static int spellfix1FilterForFullScan(
  spellfix1_cursor *pCur,
  int idxNum,
  int argc,
  sqlite3_value **argv
){
  int rc = SQLITE_OK;

  char *zSql;
  spellfix1_vtab *pVTab = pCur->pVTab;
  spellfix1ResetCursor(pCur);
  assert( idxNum==0 || idxNum==64 );
  zSql = sqlite3_mprintf(
     "SELECT word, rank, NULL, langid, id FROM \"%w\".\"%w_vocab\"%s",
     pVTab->zDbName, pVTab->zTableName,







<




>







2480
2481
2482
2483
2484
2485
2486

2487
2488
2489
2490
2491
2492
2493
2494
2495
2496
2497
2498
}

/*
** This version of xFilter handles a full-table scan case
*/
static int spellfix1FilterForFullScan(
  spellfix1_cursor *pCur,

  int argc,
  sqlite3_value **argv
){
  int rc = SQLITE_OK;
  int idxNum = pCur->idxNum;
  char *zSql;
  spellfix1_vtab *pVTab = pCur->pVTab;
  spellfix1ResetCursor(pCur);
  assert( idxNum==0 || idxNum==64 );
  zSql = sqlite3_mprintf(
     "SELECT word, rank, NULL, langid, id FROM \"%w\".\"%w_vocab\"%s",
     pVTab->zDbName, pVTab->zTableName,
2517
2518
2519
2520
2521
2522
2523

2524
2525
2526
2527
2528
2529
2530
2531
2532
2533
2534
static int spellfix1Filter(
  sqlite3_vtab_cursor *cur, 
  int idxNum, const char *idxStr,
  int argc, sqlite3_value **argv
){
  spellfix1_cursor *pCur = (spellfix1_cursor *)cur;
  int rc;

  if( idxNum & 1 ){
    rc = spellfix1FilterForMatch(pCur, idxNum, argc, argv);
  }else{
    rc = spellfix1FilterForFullScan(pCur, idxNum, argc, argv);
  }
  return rc;
}


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







>

|

|







2525
2526
2527
2528
2529
2530
2531
2532
2533
2534
2535
2536
2537
2538
2539
2540
2541
2542
2543
static int spellfix1Filter(
  sqlite3_vtab_cursor *cur, 
  int idxNum, const char *idxStr,
  int argc, sqlite3_value **argv
){
  spellfix1_cursor *pCur = (spellfix1_cursor *)cur;
  int rc;
  pCur->idxNum = idxNum;
  if( idxNum & 1 ){
    rc = spellfix1FilterForMatch(pCur, argc, argv);
  }else{
    rc = spellfix1FilterForFullScan(pCur, argc, argv);
  }
  return rc;
}


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