/* Build a hash table containing all terms in zText. */
static int build_terms(Hash *terms, sqlite3_tokenizer *pTokenizer,
const char *zText, sqlite_int64 iDocid){
sqlite3_tokenizer_cursor *pCursor;
const char *pToken;
int nTokenBytes;
int iStartOffset, iEndOffset, iPosition;
int rc = pTokenizer->pModule->xOpen(pTokenizer, zText, -1, &pCursor);
if( rc!=SQLITE_OK ) return rc;
pCursor->pTokenizer = pTokenizer;
HashInit(terms, HASH_STRING, 1);
while( SQLITE_OK==pTokenizer->pModule->xNext(pCursor,
&pToken, &nTokenBytes,
&iStartOffset, &iEndOffset,
&iPosition) ){
DocList *p;
/* Positions can't be negative; we use -1 as a terminator internally. */
if( iPosition<0 ) {
rc = SQLITE_ERROR;
goto err;
}
p = HashFind(terms, pToken, nTokenBytes);
if( p==NULL ){
p = docListNew(DL_POSITIONS_OFFSETS);
docListAddDocid(p, iDocid);
HashInsert(terms, pToken, nTokenBytes, p);
}
docListAddPosOffset(p, iPosition, iStartOffset, iEndOffset);
}
err:
/* TODO(shess) Check return? Should this be able to cause errors at
** this point? Actually, same question about sqlite3_finalize(),
** though one could argue that failure there means that the data is
** not durable. *ponder*
*/
pTokenizer->pModule->xClose(pCursor);
return rc;
}
static int build_terms(Hash *terms, sqlite3_tokenizer *pTokenizer,
const char *zText, sqlite_int64 iDocid){
sqlite3_tokenizer_cursor *pCursor;
const char *pToken;
int nTokenBytes;
int iStartOffset, iEndOffset, iPosition;
int rc = pTokenizer->pModule->xOpen(pTokenizer, zText, -1, &pCursor);
if( rc!=SQLITE_OK ) return rc;
pCursor->pTokenizer = pTokenizer;
HashInit(terms, HASH_STRING, 1);
while( SQLITE_OK==pTokenizer->pModule->xNext(pCursor,
&pToken, &nTokenBytes,
&iStartOffset, &iEndOffset,
&iPosition) ){
DocList *p;
if( iPosition<0 ) {
rc = SQLITE_ERROR;
goto err;
}
p = HashFind(terms, pToken, nTokenBytes);
if( p==NULL ){
p = docListNew(DL_POSITIONS_OFFSETS);
docListAddDocid(p, iDocid);
HashInsert(terms, pToken, nTokenBytes, p);
}
docListAddPosOffset(p, iPosition, iStartOffset, iEndOffset);
}
err:
pTokenizer->pModule->xClose(pCursor);
return rc;
}
