SQLite

Check-in [37a1de02d1]
Login

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

Overview
Comment:Add the experimental FTS3 matchinfo() function. Provides details of the match that may be used for result ranking and other purposes.
Downloads: Tarball | ZIP archive
Timelines: family | ancestors | descendants | both | trunk
Files: files | file ages | folders
SHA1: 37a1de02d1d8a34604f1bee896eaf579d4ba149a
User & Date: dan 2009-12-22 18:56:19.000
Context
2009-12-22
23:52
Move the query flattener turn-off from a pragma to an sqlite3_test_control() call. Make provisions (not yet implemented) to turn off other optimizers using the same call. (check-in: 4a97c623f4 user: drh tags: trunk)
18:56
Add the experimental FTS3 matchinfo() function. Provides details of the match that may be used for result ranking and other purposes. (check-in: 37a1de02d1 user: dan tags: trunk)
00:29
Add a pragma to disable the query flattener - for use during testing. (check-in: 1d8550e5c8 user: drh tags: trunk)
Changes
Unified Diff Ignore Whitespace Patch
Changes to ext/fts3/fts3.c.
953
954
955
956
957
958
959




960
961
962











963
964

965
966
967
968
969
970
971
  sqlite3_int64 iVal              /* Write this value to the list */
){
  assert( iVal-*piPrev > 0 || (*piPrev==0 && iVal==0) );
  *pp += sqlite3Fts3PutVarint(*pp, iVal-*piPrev);
  *piPrev = iVal;
}





static void fts3PoslistCopy(char **pp, char **ppPoslist){
  char *pEnd = *ppPoslist;
  char c = 0;











  while( *pEnd | c ) c = *pEnd++ & 0x80;
  pEnd++;

  if( pp ){
    int n = (int)(pEnd - *ppPoslist);
    char *p = *pp;
    memcpy(p, *ppPoslist, n);
    p += n;
    *pp = p;
  }







>
>
>
>



>
>
>
>
>
>
>
>
>
>
>


>







953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
  sqlite3_int64 iVal              /* Write this value to the list */
){
  assert( iVal-*piPrev > 0 || (*piPrev==0 && iVal==0) );
  *pp += sqlite3Fts3PutVarint(*pp, iVal-*piPrev);
  *piPrev = iVal;
}

/*
** When this function is called, *ppPoslist is assumed to point to the 
** start of a position-list.
*/
static void fts3PoslistCopy(char **pp, char **ppPoslist){
  char *pEnd = *ppPoslist;
  char c = 0;

  /* The end of a position list is marked by a zero encoded as an FTS3 
  ** varint. A single 0x00 byte. Except, if the 0x00 byte is preceded by
  ** a byte with the 0x80 bit set, then it is not a varint 0, but the tail
  ** of some other, multi-byte, value.
  **
  ** The following block moves pEnd to point to the first byte that is not 
  ** immediately preceded by a byte with the 0x80 bit set. Then increments
  ** pEnd once more so that it points to the byte immediately following the
  ** last byte in the position-list.
  */
  while( *pEnd | c ) c = *pEnd++ & 0x80;
  pEnd++;

  if( pp ){
    int n = (int)(pEnd - *ppPoslist);
    char *p = *pp;
    memcpy(p, *ppPoslist, n);
    p += n;
    *pp = p;
  }
983
984
985
986
987
988
989



















990
991
992
993
994
995
996
    char *p = *pp;
    memcpy(p, *ppPoslist, n);
    p += n;
    *pp = p;
  }
  *ppPoslist = pEnd;
}




















/*
** Value used to signify the end of an offset-list. This is safe because
** it is not possible to have a document with 2^31 terms.
*/
#define OFFSET_LIST_END 0x7fffffff








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







999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
    char *p = *pp;
    memcpy(p, *ppPoslist, n);
    p += n;
    *pp = p;
  }
  *ppPoslist = pEnd;
}

/*
** This function is used to count the entries in a column-list (delta-encoded
** list of term offsets within a single column of a single row).
*/
static int fts3ColumnlistCount(char **ppCollist){
  char *pEnd = *ppCollist;
  char c = 0;
  int nEntry = 0;

  /* A column-list is terminated by either a 0x01 or 0x00. */
  while( 0xFE & (*pEnd | c) ){
    c = *pEnd++ & 0x80;
    if( !c ) nEntry++;
  }

  *ppCollist = pEnd;
  return nEntry;
}

/*
** Value used to signify the end of an offset-list. This is safe because
** it is not possible to have a document with 2^31 terms.
*/
#define OFFSET_LIST_END 0x7fffffff

1695
1696
1697
1698
1699
1700
1701
1702

1703
1704
1705
1706
1707
1708
1709
1710
1711
1712



1713



1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
** Evaluate the full-text expression pExpr against fts3 table pTab. Store
** the resulting doclist in *paOut and *pnOut.
*/
static int evalFts3Expr(
  Fts3Table *p,                   /* Virtual table handle */
  Fts3Expr *pExpr,                /* Parsed fts3 expression */
  char **paOut,                   /* OUT: Pointer to malloc'd result buffer */
  int *pnOut                      /* OUT: Size of buffer at *paOut */

){
  int rc = SQLITE_OK;             /* Return code */

  /* Zero the output parameters. */
  *paOut = 0;
  *pnOut = 0;

  if( pExpr ){
    if( pExpr->eType==FTSQUERY_PHRASE ){
      int isReqPos = (pExpr->pParent && pExpr->pParent->eType==FTSQUERY_NEAR);



      rc = fts3PhraseSelect(p, pExpr->pPhrase, isReqPos, paOut, pnOut);



    }else{
      char *aLeft;
      char *aRight;
      int nLeft;
      int nRight;

      if( SQLITE_OK==(rc = evalFts3Expr(p, pExpr->pRight, &aRight, &nRight))
       && SQLITE_OK==(rc = evalFts3Expr(p, pExpr->pLeft, &aLeft, &nLeft))
      ){
        assert( pExpr->eType==FTSQUERY_NEAR || pExpr->eType==FTSQUERY_OR     
            || pExpr->eType==FTSQUERY_AND  || pExpr->eType==FTSQUERY_NOT
        );
        switch( pExpr->eType ){
          case FTSQUERY_NEAR: {
            Fts3Expr *pLeft;
            Fts3Expr *pRight;
            int mergetype = MERGE_NEAR;
            int nParam1;
            int nParam2;
            char *aBuffer;
           
            if( pExpr->pParent && pExpr->pParent->eType==FTSQUERY_NEAR ){
              mergetype = MERGE_POS_NEAR;
            }







|
>








|
|
>
>
>
|
>
>
>






|
|








|







1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
** Evaluate the full-text expression pExpr against fts3 table pTab. Store
** the resulting doclist in *paOut and *pnOut.
*/
static int evalFts3Expr(
  Fts3Table *p,                   /* Virtual table handle */
  Fts3Expr *pExpr,                /* Parsed fts3 expression */
  char **paOut,                   /* OUT: Pointer to malloc'd result buffer */
  int *pnOut,                     /* OUT: Size of buffer at *paOut */
  int isReqPos                    /* Require positions in output buffer */
){
  int rc = SQLITE_OK;             /* Return code */

  /* Zero the output parameters. */
  *paOut = 0;
  *pnOut = 0;

  if( pExpr ){
    assert( pExpr->eType==FTSQUERY_PHRASE 
         || pExpr->eType==FTSQUERY_NEAR 
         || isReqPos==0
    );
    if( pExpr->eType==FTSQUERY_PHRASE ){
      rc = fts3PhraseSelect(p, pExpr->pPhrase, 
          isReqPos || (pExpr->pParent && pExpr->pParent->eType==FTSQUERY_NEAR),
          paOut, pnOut
      );
    }else{
      char *aLeft;
      char *aRight;
      int nLeft;
      int nRight;

      if( 0==(rc = evalFts3Expr(p, pExpr->pRight, &aRight, &nRight, isReqPos))
       && 0==(rc = evalFts3Expr(p, pExpr->pLeft, &aLeft, &nLeft, isReqPos))
      ){
        assert( pExpr->eType==FTSQUERY_NEAR || pExpr->eType==FTSQUERY_OR     
            || pExpr->eType==FTSQUERY_AND  || pExpr->eType==FTSQUERY_NOT
        );
        switch( pExpr->eType ){
          case FTSQUERY_NEAR: {
            Fts3Expr *pLeft;
            Fts3Expr *pRight;
            int mergetype = isReqPos ? MERGE_POS_NEAR : MERGE_NEAR;
            int nParam1;
            int nParam2;
            char *aBuffer;
           
            if( pExpr->pParent && pExpr->pParent->eType==FTSQUERY_NEAR ){
              mergetype = MERGE_POS_NEAR;
            }
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
    }

    rc = sqlite3Fts3ExprParse(p->pTokenizer, p->azColumn, p->nColumn, 
        iCol, zQuery, -1, &pCsr->pExpr
    );
    if( rc!=SQLITE_OK ) return rc;

    rc = evalFts3Expr(p, pCsr->pExpr, &pCsr->aDoclist, &pCsr->nDoclist);
    pCsr->pNextId = pCsr->aDoclist;
    pCsr->iPrevId = 0;
  }

  if( rc!=SQLITE_OK ) return rc;
  return fts3NextMethod(pCursor);
}







|







1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
    }

    rc = sqlite3Fts3ExprParse(p->pTokenizer, p->azColumn, p->nColumn, 
        iCol, zQuery, -1, &pCsr->pExpr
    );
    if( rc!=SQLITE_OK ) return rc;

    rc = evalFts3Expr(p, pCsr->pExpr, &pCsr->aDoclist, &pCsr->nDoclist, 0);
    pCsr->pNextId = pCsr->aDoclist;
    pCsr->iPrevId = 0;
  }

  if( rc!=SQLITE_OK ) return rc;
  return fts3NextMethod(pCursor);
}
1987
1988
1989
1990
1991
1992
1993



























































































































































































1994
1995
1996
1997
1998
1999
2000
** Implementation of xRollback(). Discard the contents of the pending-terms
** hash-table. Any changes made to the database are reverted by SQLite.
*/
static int fts3RollbackMethod(sqlite3_vtab *pVtab){
  sqlite3Fts3PendingTermsClear((Fts3Table *)pVtab);
  return SQLITE_OK;
}




























































































































































































/*
** Helper function used by the implementation of the overloaded snippet(),
** offsets() and optimize() SQL functions.
**
** If the value passed as the third argument is a blob of size
** sizeof(Fts3Cursor*), then the blob contents are copied to the 







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







2029
2030
2031
2032
2033
2034
2035
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
2171
2172
2173
2174
2175
2176
2177
2178
2179
2180
2181
2182
2183
2184
2185
2186
2187
2188
2189
2190
2191
2192
2193
2194
2195
2196
2197
2198
2199
2200
2201
2202
2203
2204
2205
2206
2207
2208
2209
2210
2211
2212
2213
2214
2215
2216
2217
2218
2219
2220
2221
2222
2223
2224
2225
2226
2227
2228
2229
** Implementation of xRollback(). Discard the contents of the pending-terms
** hash-table. Any changes made to the database are reverted by SQLite.
*/
static int fts3RollbackMethod(sqlite3_vtab *pVtab){
  sqlite3Fts3PendingTermsClear((Fts3Table *)pVtab);
  return SQLITE_OK;
}

/*
** The following flags affect the format of the blob of unsigned int values
** returned by the matchinfo() function. The format is defined as follows:
**
**   Integer 0: Number of 'simple queries' that make up the FTS3 query.
**   Integer 1: Number of columns in queried table.
**
** followed by the data for (query 0, column 0), (query 0, column 1) ...
** (query 1, column 0) and so on.
**
** The first integer in each data is the number of hits that the simple
** query has in the current column.
**
** If the GLOBALCOUNT flag is set, then this is followed by the total
** number of hits the simple query has in the current column of *all*
** selected rows.
**
** If the POSITIONLIST flag is set, then this is followed by <local-count>
** integers - the positions of each of the hits for the current column/query.
*/
#define FTS3_MATCHINFO_GLOBALCOUNT  0x00000001
#define FTS3_MATCHINFO_POSITIONLIST 0x00000002

typedef struct MatchInfo MatchInfo;
struct MatchInfo {
  int rc;                         /* Return code. SQLITE_OK if no error */
  sqlite3_int64 iDocid;           /* Docid of entry to return data for */
  Fts3Table *pTab;                /* FTS3 Virtual table */
  int flags;                      /* Output flags (see above) */
  int nQuery;                     /* Number of simple queries */

  /* Malloced output buffer */
  unsigned int *aOut;
  int nOut;
  int nAlloc;
};

static void fts3MatchInfoAppend(MatchInfo *pInfo, unsigned int iVal){
  if( pInfo->rc!=SQLITE_OK ) return;

  if( pInfo->nOut==pInfo->nAlloc ){
    int nNew = pInfo->nAlloc*2+100;
    unsigned int *aNew = (unsigned int *)sqlite3_realloc(
        pInfo->aOut, nNew * sizeof(unsigned int)
    );
    if( !aNew ){
      pInfo->rc = SQLITE_NOMEM;
      return;
    }
    pInfo->aOut = aNew;
    pInfo->nAlloc = nNew;
  }

  pInfo->aOut[pInfo->nOut++] = iVal;
}

/*
** Iterate through each simple query that makes up the query expression 
** implemented by the cursor passed as the first argument.
*/
static void fts3ExprMatchInfo(
  sqlite3_context *pCtx,
  Fts3Expr *pExpr,
  MatchInfo *pInfo
){
  int eType = pExpr->eType;
  if( eType==FTSQUERY_NOT || pInfo->rc ){
    return;
  }else if( eType!=FTSQUERY_PHRASE && eType!=FTSQUERY_NEAR ){
    assert( pExpr->pLeft && pExpr->pRight );
    fts3ExprMatchInfo(pCtx, pExpr->pLeft, pInfo);
    if( pInfo->rc==SQLITE_OK ){
      fts3ExprMatchInfo(pCtx, pExpr->pRight, pInfo);
    }
  }else{
    Fts3Table *pTab = pInfo->pTab;

    /* If it is not loaded already, load the doclist for this simple query
    ** from the FTS3 full-text index. 
    */
    if( pExpr->isLoaded==0 ){
      pInfo->rc = evalFts3Expr(pTab,pExpr,&pExpr->aDoclist,&pExpr->nDoclist,1);
      if( pInfo->rc ) return;
      pExpr->isLoaded = 1;
    }

    /* If aDoclist is not NULL, search for the doclist entry in pExpr->aDoclist
    ** associated with the docid pInfo->iDocid.
    */
    if( pExpr->aDoclist ){
      char *pEnd = &pExpr->aDoclist[pExpr->nDoclist];
      sqlite3_int64 iSearch = pInfo->iDocid;

      if( pExpr->pCurrent==0 ){
        assert( pExpr->iDocid==0 );
        pExpr->pCurrent = pExpr->aDoclist;
        fts3GetDeltaVarint(&pExpr->pCurrent, &pExpr->iDocid);
      }

      while( pExpr->iDocid<iSearch && pExpr->pCurrent<pEnd ){
        /* Skip pCurrent to the start of the next doclist entry */
        fts3PoslistCopy(0, &pExpr->pCurrent);
        if( pExpr->pCurrent<pEnd ){
          fts3GetDeltaVarint(&pExpr->pCurrent, &pExpr->iDocid);
        }
      }

      if( pExpr->iDocid==iSearch ){
        int i;
        for(i=0; i<pTab->nColumn; i++){
          unsigned int iLocalOff;

          /* Add space for the "local-count" field. */
          iLocalOff = pInfo->nOut;
          fts3MatchInfoAppend(pInfo, 0);
          if( pInfo->rc ) return;

          /* If the GLOBALCOUNT field is required, write the global-count
          ** value for this query/column to the output buffer.
          */
          if( pInfo->flags&FTS3_MATCHINFO_GLOBALCOUNT ){
            if( !pExpr->aHist ){
              char *pCsr = pExpr->aDoclist;

              /* Allocate a zeroed buffer to store the global-counts 
              ** corresponding to this simple query for each table column. 
              */
              int nByte = sizeof(unsigned int)*pTab->nColumn;
              pExpr->aHist = (unsigned int *)sqlite3_malloc(nByte);
              if( !pExpr->aHist ){
                pInfo->rc = SQLITE_NOMEM;
                return;
              }
              memset(pExpr->aHist, 0, nByte);

              /* Scan the entire doclist to populate Fts3Expr.aHist[]. */ 
              while( pCsr<pEnd ){
                while( *pCsr++ & 0x80 );
                while( *pCsr ){
                  sqlite3_int64 iCol = 0;
                  if( *pCsr==0x01 ) pCsr += sqlite3Fts3GetVarint(pCsr, &iCol);
                  pExpr->aHist[iCol] += fts3ColumnlistCount(&pCsr);
                }
                pCsr++;
              }
            }

            fts3MatchInfoAppend(pInfo, pExpr->aHist[i]);
          }

          if( i==0 ){
            if( *pExpr->pCurrent==0x01 ) continue;
          }else{
            sqlite3_int64 iCol;
            char *pList = pExpr->pCurrent;
            if( *pList==0x00 ) continue;
            pList++;
            pList += sqlite3Fts3GetVarint(pList, &iCol);
            if( iCol!=i ) continue;
            pExpr->pCurrent = pList;
          }

          if( pInfo->flags&FTS3_MATCHINFO_POSITIONLIST ){
            int nLocal = 0;
            sqlite3_int64 iOffset = 0;
            char *pList = pExpr->pCurrent;
            while( *pList&0xFE ){
              fts3GetDeltaVarint(&pList, &iOffset); iOffset -= 2;
              fts3MatchInfoAppend(pInfo, iOffset);
              nLocal++;
            }
            pExpr->pCurrent = pList;
            pInfo->aOut[iLocalOff] = nLocal;
          }else{
            pInfo->aOut[iLocalOff] = fts3ColumnlistCount(&pExpr->pCurrent);
          }
        }
        pExpr->pCurrent++;
        if( pExpr->pCurrent<pEnd ){
          fts3GetDeltaVarint(&pExpr->pCurrent, &pExpr->iDocid);
        }
      }
    }
    pInfo->nQuery++;
  }
}

/*
** Helper function used by the implementation of the overloaded snippet(),
** offsets() and optimize() SQL functions.
**
** If the value passed as the third argument is a blob of size
** sizeof(Fts3Cursor*), then the blob contents are copied to the 
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
      sqlite3_result_text(pContext, "Index already optimal", -1, SQLITE_STATIC);
      break;
    default:
      sqlite3_result_error_code(pContext, rc);
      break;
  }
}





























































/*
** This routine implements the xFindFunction method for the FTS3
** virtual table.
*/
static int fts3FindFunctionMethod(
  sqlite3_vtab *pVtab,            /* Virtual table handle */
  int nArg,                       /* Number of SQL function arguments */
  const char *zName,              /* Name of SQL function */
  void (**pxFunc)(sqlite3_context*,int,sqlite3_value**), /* OUT: Result */
  void **ppArg                    /* Unused */
){
  struct Overloaded {
    const char *zName;
    void (*xFunc)(sqlite3_context*,int,sqlite3_value**);
  } aOverload[] = {
    { "snippet", fts3SnippetFunc },
    { "offsets", fts3OffsetsFunc },
    { "optimize", fts3OptimizeFunc },

  };
  int i;                          /* Iterator variable */

  UNUSED_PARAMETER(pVtab);
  UNUSED_PARAMETER(nArg);
  UNUSED_PARAMETER(ppArg);








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



















>







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
2393
2394
2395
2396
2397
2398
2399
2400
2401
2402
2403
2404
2405
2406
2407
2408
2409
2410
2411
2412
2413
2414
2415
2416
2417
2418
2419
2420
2421
2422
2423
2424
2425
2426
2427
2428
2429
2430
2431
2432
2433
2434
2435
2436
      sqlite3_result_text(pContext, "Index already optimal", -1, SQLITE_STATIC);
      break;
    default:
      sqlite3_result_error_code(pContext, rc);
      break;
  }
}

/*
** Implementation of the matchinfo() function for FTS3
*/
static void fts3MatchinfoFunc(
  sqlite3_context *pContext,      /* SQLite function call context */
  int nVal,                       /* Size of argument array */
  sqlite3_value **apVal           /* Array of arguments */
){
  Fts3Cursor *pCsr;               /* Cursor handle passed through apVal[0] */
  int flags = 0;


  if( nVal==2 ){
    int i;
    const unsigned char *zFlags = sqlite3_value_text(apVal[1]);
    for(i=0; zFlags[i]; i++){
      switch( zFlags[i] ){
        case 'g': flags |= FTS3_MATCHINFO_GLOBALCOUNT; break;
        case 'p': flags |= FTS3_MATCHINFO_POSITIONLIST; break;
        default: {
          char zErr[18];
          memcpy(zErr, "Unknown flag: \"%c\"", 18);
          zErr[16] = (char)zFlags[i];
          sqlite3_result_error(pContext, zErr, -1);
          return;
        }
      }
    }
  }else if( nVal!=1 ){
    sqlite3_result_error(pContext,
        "wrong number of arguments to function matchinfo()", -1);
    return;
  }

  if( SQLITE_OK==fts3FunctionArg(pContext, "matchinfo", apVal[0], &pCsr) ){
    MatchInfo ctx;
    memset(&ctx, 0, sizeof(ctx));
    ctx.iDocid = pCsr->iPrevId;
    ctx.pTab = (Fts3Table *)pCsr->base.pVtab;
    ctx.flags = flags;

    fts3MatchInfoAppend(&ctx, 0);
    fts3MatchInfoAppend(&ctx, ctx.pTab->nColumn);

    /* Iterate through each of the 'simple' queries that make up the query
    ** expression. A 'simple' query is a phrase (including token and token 
    ** prefix) or NEAR query. 
    */
    fts3ExprMatchInfo(pContext, pCsr->pExpr, &ctx);
    if( ctx.rc ){
      sqlite3_free(ctx.aOut);
      sqlite3_result_error_code(pContext, ctx.rc);
    }else{
      int nByte = ctx.nOut*sizeof(unsigned int);
      ctx.aOut[0] = ctx.nQuery;
      sqlite3_result_blob(pContext, ctx.aOut, nByte, sqlite3_free);
    }
  }
}

/*
** This routine implements the xFindFunction method for the FTS3
** virtual table.
*/
static int fts3FindFunctionMethod(
  sqlite3_vtab *pVtab,            /* Virtual table handle */
  int nArg,                       /* Number of SQL function arguments */
  const char *zName,              /* Name of SQL function */
  void (**pxFunc)(sqlite3_context*,int,sqlite3_value**), /* OUT: Result */
  void **ppArg                    /* Unused */
){
  struct Overloaded {
    const char *zName;
    void (*xFunc)(sqlite3_context*,int,sqlite3_value**);
  } aOverload[] = {
    { "snippet", fts3SnippetFunc },
    { "offsets", fts3OffsetsFunc },
    { "optimize", fts3OptimizeFunc },
    { "matchinfo", fts3MatchinfoFunc },
  };
  int i;                          /* Iterator variable */

  UNUSED_PARAMETER(pVtab);
  UNUSED_PARAMETER(nArg);
  UNUSED_PARAMETER(ppArg);

2280
2281
2282
2283
2284
2285
2286

2287
2288
2289
2290
2291
2292
2293
  ** the two scalar functions. If this is successful, register the
  ** module with sqlite.
  */
  if( SQLITE_OK==rc 
   && SQLITE_OK==(rc = sqlite3Fts3InitHashTable(db, pHash, "fts3_tokenizer"))
   && SQLITE_OK==(rc = sqlite3_overload_function(db, "snippet", -1))
   && SQLITE_OK==(rc = sqlite3_overload_function(db, "offsets", 1))

   && SQLITE_OK==(rc = sqlite3_overload_function(db, "optimize", 1))
  ){
    return sqlite3_create_module_v2(
        db, "fts3", &fts3Module, (void *)pHash, hashDestroy
    );
  }








>







2570
2571
2572
2573
2574
2575
2576
2577
2578
2579
2580
2581
2582
2583
2584
  ** the two scalar functions. If this is successful, register the
  ** module with sqlite.
  */
  if( SQLITE_OK==rc 
   && SQLITE_OK==(rc = sqlite3Fts3InitHashTable(db, pHash, "fts3_tokenizer"))
   && SQLITE_OK==(rc = sqlite3_overload_function(db, "snippet", -1))
   && SQLITE_OK==(rc = sqlite3_overload_function(db, "offsets", 1))
   && SQLITE_OK==(rc = sqlite3_overload_function(db, "matchinfo", -1))
   && SQLITE_OK==(rc = sqlite3_overload_function(db, "optimize", 1))
  ){
    return sqlite3_create_module_v2(
        db, "fts3", &fts3Module, (void *)pHash, hashDestroy
    );
  }

Changes to ext/fts3/fts3Int.h.
182
183
184
185
186
187
188










189
190
191
192
193
194
195
196







197
198
199
200
201
202
203
    int n;                   /* Number of bytes in buffer pointed to by z */
    int isPrefix;            /* True if token ends in with a "*" character */
  } aToken[1];               /* One entry for each token in the phrase */
};

/*
** A tree of these objects forms the RHS of a MATCH operator.










*/
struct Fts3Expr {
  int eType;                 /* One of the FTSQUERY_XXX values defined below */
  int nNear;                 /* Valid if eType==FTSQUERY_NEAR */
  Fts3Expr *pParent;         /* pParent->pLeft==this or pParent->pRight==this */
  Fts3Expr *pLeft;           /* Left operand */
  Fts3Expr *pRight;          /* Right operand */
  Fts3Phrase *pPhrase;       /* Valid if eType==FTSQUERY_PHRASE */







};

/*
** Candidate values for Fts3Query.eType. Note that the order of the first
** four values is in order of precedence when parsing expressions. For 
** example, the following:
**







>
>
>
>
>
>
>
>
>
>








>
>
>
>
>
>
>







182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
    int n;                   /* Number of bytes in buffer pointed to by z */
    int isPrefix;            /* True if token ends in with a "*" character */
  } aToken[1];               /* One entry for each token in the phrase */
};

/*
** A tree of these objects forms the RHS of a MATCH operator.
**
** If Fts3Expr.eType is either FTSQUERY_NEAR or FTSQUERY_PHRASE and isLoaded
** is true, then aDoclist points to a malloced buffer, size nDoclist bytes, 
** containing the results of the NEAR or phrase query in FTS3 doclist
** format. As usual, the initial "Length" field found in doclists stored
** on disk is omitted from this buffer.
**
** Variable pCurrent always points to the start of a docid field within
** aDoclist. Since the doclist is usually scanned in docid order, this can
** be used to accelerate seeking to the required docid within the doclist.
*/
struct Fts3Expr {
  int eType;                 /* One of the FTSQUERY_XXX values defined below */
  int nNear;                 /* Valid if eType==FTSQUERY_NEAR */
  Fts3Expr *pParent;         /* pParent->pLeft==this or pParent->pRight==this */
  Fts3Expr *pLeft;           /* Left operand */
  Fts3Expr *pRight;          /* Right operand */
  Fts3Phrase *pPhrase;       /* Valid if eType==FTSQUERY_PHRASE */

  int isLoaded;
  sqlite3_int64 iDocid;
  char *aDoclist;
  int nDoclist;
  char *pCurrent;
  unsigned int *aHist;
};

/*
** Candidate values for Fts3Query.eType. Note that the order of the first
** four values is in order of precedence when parsing expressions. For 
** example, the following:
**
Changes to ext/fts3/fts3_expr.c.
731
732
733
734
735
736
737


738
739
740
741
742
743
744
/*
** Free a parsed fts3 query expression allocated by sqlite3Fts3ExprParse().
*/
void sqlite3Fts3ExprFree(Fts3Expr *p){
  if( p ){
    sqlite3Fts3ExprFree(p->pLeft);
    sqlite3Fts3ExprFree(p->pRight);


    sqlite3_free(p);
  }
}

/****************************************************************************
*****************************************************************************
** Everything after this point is just test code.







>
>







731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
/*
** Free a parsed fts3 query expression allocated by sqlite3Fts3ExprParse().
*/
void sqlite3Fts3ExprFree(Fts3Expr *p){
  if( p ){
    sqlite3Fts3ExprFree(p->pLeft);
    sqlite3Fts3ExprFree(p->pRight);
    sqlite3_free(p->aDoclist);
    sqlite3_free(p->aHist);
    sqlite3_free(p);
  }
}

/****************************************************************************
*****************************************************************************
** Everything after this point is just test code.
Changes to ext/fts3/fts3_tokenizer.h.
140
141
142
143
144
145
146




147
148
  /* Tokenizer implementations will typically add additional fields */
};

struct sqlite3_tokenizer_cursor {
  sqlite3_tokenizer *pTokenizer;       /* Tokenizer for this cursor. */
  /* Tokenizer implementations will typically add additional fields */
};





#endif /* _FTS3_TOKENIZER_H_ */







>
>
>
>


140
141
142
143
144
145
146
147
148
149
150
151
152
  /* Tokenizer implementations will typically add additional fields */
};

struct sqlite3_tokenizer_cursor {
  sqlite3_tokenizer *pTokenizer;       /* Tokenizer for this cursor. */
  /* Tokenizer implementations will typically add additional fields */
};

int fts3_global_term_cnt(int iTerm, int iCol);
int fts3_term_cnt(int iTerm, int iCol);


#endif /* _FTS3_TOKENIZER_H_ */
Changes to test/fts3query.test.
85
86
87
88
89
90
91



























92
93
  }
} {}

do_test fts3query-3.2 {
  execsql { SELECT docid FROM foobar WHERE description MATCH '"high sp d"' }
} {}




























finish_test








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


85
86
87
88
89
90
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
  }
} {}

do_test fts3query-3.2 {
  execsql { SELECT docid FROM foobar WHERE description MATCH '"high sp d"' }
} {}

proc mit {blob} {
  set scan(littleEndian) i*
  set scan(bigEndian) I*
  binary scan $blob $scan($::tcl_platform(byteOrder)) r
  return $r
}
db func mit mit

do_test fts3query-3.3 {
  execsql { SELECT mit(matchinfo(foobar)) FROM foobar WHERE foobar MATCH 'the' }
} {{1 1 3}}
do_test fts3query-3.4 {
  execsql { 
    SELECT mit(matchinfo(foobar, 'g')) FROM foobar WHERE foobar MATCH 'the' 
  }
} {{1 1 3 3}}
do_test fts3query-3.5 {
  execsql { 
    SELECT mit(matchinfo(foobar, 'p')) FROM foobar WHERE foobar MATCH 'the' 
  }
} {{1 1 3 27 74 79}}
do_test fts3query-3.5 {
  execsql { 
    SELECT mit(matchinfo(foobar, 'pg')) FROM foobar WHERE foobar MATCH 'the' 
  }
} {{1 1 3 3 27 74 79}}

finish_test

Changes to test/fts3rnd.test.
143
144
145
146
147
148
149

150

151
152
153
154
155
156

157















158
159
160
161
162
163
164

proc simple_phrase {zPrefix} {
  set ret [list]

  set reg [string map {* {[^ ]*}} $zPrefix]
  set reg " $reg "


  foreach {key value} [array get ::t1] {

    foreach col $value {
      if {[regexp $reg " $col "]} {lappend ret $key}
    }
  }

  lsort -uniq -integer $ret

}
















proc simple_near {termlist nNear} {
  set ret [list]

  foreach {key value} [array get ::t1] {
    foreach v $value {








>
|
>

|



|
>

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







143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182

proc simple_phrase {zPrefix} {
  set ret [list]

  set reg [string map {* {[^ ]*}} $zPrefix]
  set reg " $reg "

  foreach key [lsort -integer [array names ::t1]] {
    set value $::t1($key)
    set cnt [list]
    foreach col $value {
      if {[regexp $reg " $col "]} { lappend ret $key ; break }
    }
  }

  #lsort -uniq -integer $ret
  set ret
}

proc simple_token_matchinfo {zToken} {
  foreach key [lsort -integer [array names ::t1]] {
    set value $::t1($key)
    set cnt [list]
    foreach col $value {
      lappend cnt [llength [lsearch -all $col $zToken]]
    }
    if {[lindex [lsort $cnt] end]} {
      lappend ret $key [concat 1 3 $cnt]
    }
  }
  
  set ret
} 

proc simple_near {termlist nNear} {
  set ret [list]

  foreach {key value} [array get ::t1] {
    foreach v $value {

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
}
proc setop_and {A B} {
  foreach b $B { set n($b) {} }
  set ret [list]
  foreach a $A { if {[info exists n($a)]} {lappend ret $a} }
  return $ret
}









set sqlite_fts3_enable_parentheses 1

foreach nodesize {50 500 1000 2000} {
  catch { array unset ::t1 }

  # Create the FTS3 table. Populate it (and the Tcl array) with 100 rows.
  #
  db transaction {
    catchsql { DROP TABLE t1 }
    execsql "CREATE VIRTUAL TABLE t1 USING fts3(a, b, c)"
    execsql "INSERT INTO t1(t1) VALUES('nodesize=$nodesize')"
    for {set i 0} {$i < 100} {incr i} { insert_row $i }
  }
  
  for {set iTest 0} {$iTest <= 100} {incr iTest} {
    catchsql COMMIT

    set DO_MALLOC_TEST 0
    set nRep 10
    if {$iTest==100 && $nodesize==50} { 
      set DO_MALLOC_TEST 1 
      set nRep 2







>
>
>
>
>
>
>
>















|







227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
}
proc setop_and {A B} {
  foreach b $B { set n($b) {} }
  set ret [list]
  foreach a $A { if {[info exists n($a)]} {lappend ret $a} }
  return $ret
}

proc mit {blob} {
  set scan(littleEndian) i*
  set scan(bigEndian) I*
  binary scan $blob $scan($::tcl_platform(byteOrder)) r
  return $r
}
db func mit mit

set sqlite_fts3_enable_parentheses 1

foreach nodesize {50 500 1000 2000} {
  catch { array unset ::t1 }

  # Create the FTS3 table. Populate it (and the Tcl array) with 100 rows.
  #
  db transaction {
    catchsql { DROP TABLE t1 }
    execsql "CREATE VIRTUAL TABLE t1 USING fts3(a, b, c)"
    execsql "INSERT INTO t1(t1) VALUES('nodesize=$nodesize')"
    for {set i 0} {$i < 100} {incr i} { insert_row $i }
  }
  
  for {set iTest 1} {$iTest <= 100} {incr iTest} {
    catchsql COMMIT

    set DO_MALLOC_TEST 0
    set nRep 10
    if {$iTest==100 && $nodesize==50} { 
      set DO_MALLOC_TEST 1 
      set nRep 2
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
    # the database for the set of documents containing each of these terms
    # is the same as the result obtained by scanning the contents of the Tcl 
    # array for each term.
    #
    for {set i 0} {$i < 10} {incr i} {
      set term [random_term]
      do_select_test fts3rnd-1.$nodesize.$iTest.1.$i {
        SELECT docid FROM t1 WHERE t1 MATCH $term
      } [simple_phrase $term]
    }

    # This time, use the first two characters of each term as a term prefix
    # to query for. Test that querying the Tcl array produces the same results
    # as querying the FTS3 table for the prefix.
    #
    for {set i 0} {$i < $nRep} {incr i} {







|
|







287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
    # the database for the set of documents containing each of these terms
    # is the same as the result obtained by scanning the contents of the Tcl 
    # array for each term.
    #
    for {set i 0} {$i < 10} {incr i} {
      set term [random_term]
      do_select_test fts3rnd-1.$nodesize.$iTest.1.$i {
        SELECT docid, mit(matchinfo(t1)) FROM t1 WHERE t1 MATCH $term
      } [simple_token_matchinfo $term]
    }

    # This time, use the first two characters of each term as a term prefix
    # to query for. Test that querying the Tcl array produces the same results
    # as querying the FTS3 table for the prefix.
    #
    for {set i 0} {$i < $nRep} {incr i} {