/ Check-in [fca59281]
Login

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

Overview
Comment:Minor code cleanup in FTS1. (CVS 3412)
Downloads: Tarball | ZIP archive | SQL archive
Timelines: family | ancestors | descendants | both | trunk
Files: files | file ages | folders
SHA1: fca592816767de397fbaf22cccdf1028fc5dfc91
User & Date: drh 2006-09-13 17:17:48
Context
2006-09-13
18:40
Earlier refactoring changed name in fts1.c but not fts1.h. (CVS 3413) check-in: d4edb803 user: shess tags: trunk
17:17
Minor code cleanup in FTS1. (CVS 3412) check-in: fca59281 user: drh tags: trunk
16:02
Implementation of "column:" modifiers in FTS1 queries. (CVS 3411) check-in: 820634f7 user: drh tags: trunk
Changes
Hide Diffs Unified Diffs Ignore Whitespace Patch

Changes to ext/fts1/fts1.c.

2029
2030
2031
2032
2033
2034
2035
2036

2037
2038
2039
2040
2041
2042
2043
....
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
2070
....
2120
2121
2122
2123
2124
2125
2126
2127
2128
2129
2130
2131
2132
2133
2134
....
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
....
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
 *
 */
typedef struct Query {
  fulltext_vtab *pFts;  /* The full text index */
  int nTerms;           /* Number of terms in the query */
  QueryTerm *pTerms;    /* Array of terms.  Space obtained from malloc() */
  int nextIsOr;         /* Set the isOr flag on the next inserted term */
  int iColumn;          /* Text word parsed must be in this column */

} Query;

/* Add a new term pTerm[0..nTerm-1] to the query *q.
*/
static void queryAdd(Query *q, const char *pTerm, int nTerm){
  QueryTerm *t;
  ++q->nTerms;
................................................................................
  memset(t, 0, sizeof(*t));
  t->pTerm = malloc(nTerm+1);
  memcpy(t->pTerm, pTerm, nTerm);
  t->pTerm[nTerm] = 0;
  t->nTerm = nTerm;
  t->isOr = q->nextIsOr;
  q->nextIsOr = 0;
  t->iColumn = q->iColumn;
  q->iColumn = -1;
}

/* Free all of the memory that was malloced in order to build *q.
*/
static void queryDestroy(Query *q){
  int i;
  for(i = 0; i < q->nTerms; ++i){
    free(q->pTerms[i].pTerm);
  }
  free(q->pTerms);
}

................................................................................
    rc = pModule->xNext(pCursor,
                        &pToken, &nToken,
                        &iBegin, &iEnd, &iPos);
    if( rc!=SQLITE_OK ) break;
    if( !inPhrase &&
        pSegment[iEnd]==':' &&
         (iCol = checkColumnSpecifier(pQuery->pFts, pToken, nToken))>=0 ){
      pQuery->iColumn = iCol;
      continue;
    }
    if( !inPhrase && pQuery->nTerms>0 && nToken==2
         && pSegment[iBegin]=='O' && pSegment[iBegin+1]=='R' ){
      pQuery->nextIsOr = 1;
      continue;
    }
................................................................................
  if( inPhrase && pQuery->nTerms>firstIndex ){
    pQuery->pTerms[firstIndex].nPhrase = pQuery->nTerms - firstIndex - 1;
  }

  return pModule->xClose(pCursor);
}

/* Parse a query string, yielding a Query object [pQuery], which the caller
 * must free. */



static int parseQuery(fulltext_vtab *v, const char *pInput, int nInput,
                      Query *pQuery){






  int iInput, inPhrase = 0;

  if( nInput<0 ) nInput = strlen(pInput);
  pQuery->nTerms = 0;
  pQuery->pTerms = NULL;
  pQuery->nextIsOr = 0;
  pQuery->iColumn = -1;

  pQuery->pFts = v;

  for(iInput=0; iInput<nInput; ++iInput){
    int i;
    for(i=iInput; i<nInput && pInput[i]!='"'; ++i)
      ;
    if( i>iInput ){
      tokenizeSegment(v->pTokenizer, pInput+iInput, i-iInput, inPhrase,
                       pQuery);
    }
    iInput = i;
    if( i<nInput ){
      assert( pInput[i]=='"' );
      inPhrase = !inPhrase;
    }
  }


  if( inPhrase ){  /* unmatched quote */
    queryDestroy(pQuery);
    return SQLITE_ERROR;
  }
  return SQLITE_OK;
}

/* Perform a full-text query using the search expression in
** pInput[0..nInput-1].  Return a list of matching documents
** in pResult.
**
** Queries must match column iColumn.  Or if iColumn>=nColumn
** they are allowed to match against any column.
*/
static int fulltextQuery(fulltext_vtab *v, int iColumn,
                         const char *pInput, int nInput, DocList **pResult){





  Query q;
  int i, rc;
  DocList *pLeft = NULL;
  DocList *pRight, *pNew;
  int nNot = 0;
  int iCol;

  rc = parseQuery(v, pInput, nInput, &q);
  if( rc!=SQLITE_OK ) return rc;

  /* Merge AND terms. */
  for(i = 0 ; i < q.nTerms; i += q.pTerms[i].nPhrase + 1){

    if( q.pTerms[i].isNot ){
      /* Handle all NOT terms in a separate pass */
      nNot++;
      continue;
    }

    iCol = q.pTerms[i].iColumn;
    if( iCol<0 ) iCol = iColumn;
    rc = docListOfTerm(v, iCol, &q.pTerms[i], &pRight);
    if( rc ){
      queryDestroy(&q);

      return rc;
    }
    if( pLeft==0 ){
      pLeft = pRight;
    }else{
      pNew = docListNew(DL_DOCIDS);
      if( q.pTerms[i].isOr ){
................................................................................
    /* We do not yet know how to handle a query of only NOT terms */
    return SQLITE_ERROR;
  }

  /* Do the EXCEPT terms */
  for(i=0; i<q.nTerms;  i += q.pTerms[i].nPhrase + 1){
    if( !q.pTerms[i].isNot ) continue;
    iCol = q.pTerms[i].iColumn;
    if( iCol<0 ) iCol = iColumn;
    rc = docListOfTerm(v, iCol, &q.pTerms[i], &pRight);
    if( rc ){
      queryDestroy(&q);

      docListDelete(pLeft);
      return rc;
    }
    pNew = docListNew(DL_DOCIDS);
    docListExceptMerge(pLeft, pRight, pNew);
    docListDelete(pRight);
    docListDelete(pLeft);
    pLeft = pNew;
  }

  queryDestroy(&q);
  *pResult = pLeft;
  return rc;
}

/*
** This is the xFilter interface for the virtual table.  See
** the virtual table xFilter method documentation for additional







|
>







 







|
|


|

|







 







|







 







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


|



|
>




|
<

|




|




>
|
|






|





|
|
>
>
>
>
>





<

|











|
<
<

<
>







 







|
<
<

<
>










|







2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
....
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
2070
2071
....
2121
2122
2123
2124
2125
2126
2127
2128
2129
2130
2131
2132
2133
2134
2135
....
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
2230
2231
2232
2233
2234
....
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
 *
 */
typedef struct Query {
  fulltext_vtab *pFts;  /* The full text index */
  int nTerms;           /* Number of terms in the query */
  QueryTerm *pTerms;    /* Array of terms.  Space obtained from malloc() */
  int nextIsOr;         /* Set the isOr flag on the next inserted term */
  int nextColumn;       /* Next word parsed must be in this column */
  int dfltColumn;       /* The default column */
} Query;

/* Add a new term pTerm[0..nTerm-1] to the query *q.
*/
static void queryAdd(Query *q, const char *pTerm, int nTerm){
  QueryTerm *t;
  ++q->nTerms;
................................................................................
  memset(t, 0, sizeof(*t));
  t->pTerm = malloc(nTerm+1);
  memcpy(t->pTerm, pTerm, nTerm);
  t->pTerm[nTerm] = 0;
  t->nTerm = nTerm;
  t->isOr = q->nextIsOr;
  q->nextIsOr = 0;
  t->iColumn = q->nextColumn;
  q->nextColumn = q->dfltColumn;
}

/* Free all of the dynamically allocated memory held by *q
*/
static void queryClear(Query *q){
  int i;
  for(i = 0; i < q->nTerms; ++i){
    free(q->pTerms[i].pTerm);
  }
  free(q->pTerms);
}

................................................................................
    rc = pModule->xNext(pCursor,
                        &pToken, &nToken,
                        &iBegin, &iEnd, &iPos);
    if( rc!=SQLITE_OK ) break;
    if( !inPhrase &&
        pSegment[iEnd]==':' &&
         (iCol = checkColumnSpecifier(pQuery->pFts, pToken, nToken))>=0 ){
      pQuery->nextColumn = iCol;
      continue;
    }
    if( !inPhrase && pQuery->nTerms>0 && nToken==2
         && pSegment[iBegin]=='O' && pSegment[iBegin+1]=='R' ){
      pQuery->nextIsOr = 1;
      continue;
    }
................................................................................
  if( inPhrase && pQuery->nTerms>firstIndex ){
    pQuery->pTerms[firstIndex].nPhrase = pQuery->nTerms - firstIndex - 1;
  }

  return pModule->xClose(pCursor);
}

/* Parse a query string, yielding a Query object pQuery.
**
** The calling function will need to queryClear() to clean up
** the dynamically allocated memory held by pQuery.
*/
static int parseQuery(

  fulltext_vtab *v,        /* The fulltext index */
  const char *zInput,      /* Input text of the query string */
  int nInput,              /* Size of the input text */
  int dfltColumn,          /* Default column of the index to match against */
  Query *pQuery            /* Write the parse results here. */
){
  int iInput, inPhrase = 0;

  if( nInput<0 ) nInput = strlen(zInput);
  pQuery->nTerms = 0;
  pQuery->pTerms = NULL;
  pQuery->nextIsOr = 0;
  pQuery->nextColumn = dfltColumn;
  pQuery->dfltColumn = dfltColumn;
  pQuery->pFts = v;

  for(iInput=0; iInput<nInput; ++iInput){
    int i;
    for(i=iInput; i<nInput && zInput[i]!='"'; ++i){}

    if( i>iInput ){
      tokenizeSegment(v->pTokenizer, zInput+iInput, i-iInput, inPhrase,
                       pQuery);
    }
    iInput = i;
    if( i<nInput ){
      assert( zInput[i]=='"' );
      inPhrase = !inPhrase;
    }
  }

  if( inPhrase ){
    /* unmatched quote */
    queryClear(pQuery);
    return SQLITE_ERROR;
  }
  return SQLITE_OK;
}

/* Perform a full-text query using the search expression in
** zInput[0..nInput-1].  Return a list of matching documents
** in pResult.
**
** Queries must match column iColumn.  Or if iColumn>=nColumn
** they are allowed to match against any column.
*/
static int fulltextQuery(
  fulltext_vtab *v,      /* The full text index */
  int iColumn,           /* Match against this column by default */
  const char *zInput,    /* The query string */
  int nInput,            /* Number of bytes in zInput[] */
  DocList **pResult      /* Write the result doclist here */
){
  Query q;
  int i, rc;
  DocList *pLeft = NULL;
  DocList *pRight, *pNew;
  int nNot = 0;


  rc = parseQuery(v, zInput, nInput, iColumn, &q);
  if( rc!=SQLITE_OK ) return rc;

  /* Merge AND terms. */
  for(i = 0 ; i < q.nTerms; i += q.pTerms[i].nPhrase + 1){

    if( q.pTerms[i].isNot ){
      /* Handle all NOT terms in a separate pass */
      nNot++;
      continue;
    }

    rc = docListOfTerm(v, q.pTerms[i].iColumn, &q.pTerms[i], &pRight);


    if( rc ){

      queryClear(&q);
      return rc;
    }
    if( pLeft==0 ){
      pLeft = pRight;
    }else{
      pNew = docListNew(DL_DOCIDS);
      if( q.pTerms[i].isOr ){
................................................................................
    /* We do not yet know how to handle a query of only NOT terms */
    return SQLITE_ERROR;
  }

  /* Do the EXCEPT terms */
  for(i=0; i<q.nTerms;  i += q.pTerms[i].nPhrase + 1){
    if( !q.pTerms[i].isNot ) continue;
    rc = docListOfTerm(v, q.pTerms[i].iColumn, &q.pTerms[i], &pRight);


    if( rc ){

      queryClear(&q);
      docListDelete(pLeft);
      return rc;
    }
    pNew = docListNew(DL_DOCIDS);
    docListExceptMerge(pLeft, pRight, pNew);
    docListDelete(pRight);
    docListDelete(pLeft);
    pLeft = pNew;
  }

  queryClear(&q);
  *pResult = pLeft;
  return rc;
}

/*
** This is the xFilter interface for the virtual table.  See
** the virtual table xFilter method documentation for additional