Thanks to visit codestin.com
Credit goes to github.com

Skip to content

Commit 5c920b1

Browse files
author
Alexander Korotkov
committed
Fix multiple occurences of the single lexeme.
1 parent f3c37f5 commit 5c920b1

File tree

1 file changed

+64
-28
lines changed

1 file changed

+64
-28
lines changed

‎rumtsquery.c

+64-28
Original file line numberDiff line numberDiff line change
@@ -256,18 +256,47 @@ extract_wraps(QueryItemWrap *wrap, ExtractContext *context, int level)
256256
{
257257
if (wrap->type == QI_VAL)
258258
{
259-
bytea *addinfo = (bytea *) palloc(VARHDRSZ + Max(level, 1) * MAX_ENCODED_LEN);
260-
unsigned char *ptr = (unsigned char *)VARDATA(addinfo);
259+
bytea *addinfo;
260+
unsigned char *ptr;
261261
int index = context->index;
262262

263-
context->entries[index] = PointerGetDatum(cstring_to_text_with_len(context->operand + wrap->distance, wrap->length));
264-
elog(NOTICE, "%s", text_to_cstring(DatumGetTextP(context->entries[index])));
263+
264+
for (index = 0; index < context->index; index++)
265+
{
266+
text *entry;
267+
entry = DatumGetByteaP(context->entries[index]);
268+
if (VARSIZE_ANY_EXHDR(entry) == wrap->length &&
269+
!memcmp(context->operand + wrap->distance, VARDATA_ANY(entry), wrap->length))
270+
break;
271+
}
272+
273+
if (index >= context->index)
274+
{
275+
index = context->index;
276+
addinfo = (bytea *) palloc(VARHDRSZ + 2 * Max(level, 1) * MAX_ENCODED_LEN);
277+
ptr = (unsigned char *) VARDATA(addinfo);
278+
context->entries[index] = PointerGetDatum(cstring_to_text_with_len(context->operand + wrap->distance, wrap->length));
279+
context->addInfo[index] = PointerGetDatum(addinfo);
280+
context->addInfoIsNull[index] = false;
281+
context->index++;
282+
/*ptrEnd = (unsigned char *) VARDATA(addinfo) + VARHDRSZ + 2 * Max(level, 1) * MAX_ENCODED_LEN;*/
283+
}
284+
else
285+
{
286+
addinfo = DatumGetByteaP(context->addInfo[index]);
287+
addinfo = (bytea *) repalloc(addinfo,
288+
VARSIZE(addinfo) + 2 * Max(level, 1) * MAX_ENCODED_LEN);
289+
context->addInfo[index] = PointerGetDatum(addinfo);
290+
ptr = (unsigned char *) VARDATA(addinfo) + VARSIZE_ANY_EXHDR(addinfo);
291+
/*ptrEnd = (unsigned char *) VARDATA(addinfo) + VARSIZE_ANY_EXHDR(addinfo) + 2 * Max(level, 1) * MAX_ENCODED_LEN;*/
292+
}
293+
/*elog(NOTICE, "%s", text_to_cstring(DatumGetTextP(context->entries[index])));*/
265294

266295
while (wrap->parent)
267296
{
268297
QueryItemWrap *parent = wrap->parent;
269298
uint32 sum;
270-
elog(NOTICE, "%d %d %d", parent->num, parent->sum, wrap->not);
299+
/*elog(NOTICE, "%d %d %d", parent->num, parent->sum, wrap->not);*/
271300
encode_varbyte((uint32) parent->num, &ptr);
272301
sum = (uint32)abs(parent->sum);
273302
sum <<= 2;
@@ -283,11 +312,9 @@ extract_wraps(QueryItemWrap *wrap, ExtractContext *context, int level)
283312
encode_varbyte(1, &ptr);
284313
encode_varbyte(4 | 1, &ptr);
285314
}
315+
/*Assert(ptr <= ptrEnd);*/
286316
SET_VARSIZE(addinfo, ptr - (unsigned char *)addinfo);
287-
288-
context->addInfo[index] = PointerGetDatum(addinfo);
289-
context->addInfoIsNull[index] = false;
290-
context->index++;
317+
/*elog(NOTICE, "%s", DatumGetPointer(DirectFunctionCall1(byteaout, PointerGetDatum(addinfo))));*/
291318
}
292319
else if (wrap->type == QI_OPR)
293320
{
@@ -325,27 +352,19 @@ ruminv_extract_tsquery(PG_FUNCTION_ARGS)
325352
QueryItem *item = GETQUERY(query);
326353
QueryItemWrap *wrap;
327354
ExtractContext context;
328-
int num = 1;
355+
int num = 1,
356+
count;
329357
bool extractNull;
330358

331359
wrap = make_query_item_wrap(item, NULL, false);
332-
*nentries = calc_wraps(wrap, &num);
360+
count = calc_wraps(wrap, &num);
333361
extractNull = check_allnegative(wrap);
334362
if (extractNull)
335-
(*nentries)++;
363+
count++;
336364

337-
entries = (Datum *) palloc(sizeof(Datum) * (*nentries));
338-
*addInfo = (Datum *) palloc(sizeof(Datum) * (*nentries));
339-
*addInfoIsNull = (bool *) palloc(sizeof(bool) * (*nentries));
340-
if (extractNull)
341-
{
342-
int i;
343-
*nullFlags = (bool *) palloc(sizeof(bool) * (*nentries));
344-
for (i = 0; i < *nentries - 1; i++)
345-
(*nullFlags)[i] = false;
346-
(*nullFlags)[*nentries - 1] = true;
347-
(*addInfoIsNull)[*nentries - 1] = true;
348-
}
365+
entries = (Datum *) palloc(sizeof(Datum) * count);
366+
*addInfo = (Datum *) palloc(sizeof(Datum) * count);
367+
*addInfoIsNull = (bool *) palloc(sizeof(bool) * count);
349368

350369
context.addInfo = *addInfo;
351370
context.addInfoIsNull = *addInfoIsNull;
@@ -355,6 +374,20 @@ ruminv_extract_tsquery(PG_FUNCTION_ARGS)
355374

356375
extract_wraps(wrap, &context, 0);
357376

377+
count = context.index;
378+
if (extractNull)
379+
{
380+
int i;
381+
382+
count++;
383+
*nullFlags = (bool *) palloc(sizeof(bool) * count);
384+
for (i = 0; i < count - 1; i++)
385+
(*nullFlags)[i] = false;
386+
(*nullFlags)[count - 1] = true;
387+
(*addInfoIsNull)[count - 1] = true;
388+
}
389+
*nentries = count;
390+
358391
/* elog(NOTICE, "%d", *nentries);
359392
for (i = 0; i < *nentries; i++)
360393
{
@@ -451,7 +484,7 @@ ruminv_tsvector_consistent(PG_FUNCTION_ARGS)
451484
ptr = (unsigned char *)VARDATA_ANY(DatumGetPointer(addInfo[i]));
452485
size = VARSIZE_ANY_EXHDR(DatumGetPointer(addInfo[i]));
453486

454-
/* elog(NOTICE, "%d %s", i, DatumGetPointer(DirectFunctionCall1(byteaout, addInfo[i])));*/
487+
/*elog(NOTICE, "%d %s", i, DatumGetPointer(DirectFunctionCall1(byteaout, addInfo[i])));*/
455488

456489
if (size == 0)
457490
{
@@ -473,7 +506,7 @@ ruminv_tsvector_consistent(PG_FUNCTION_ARGS)
473506

474507
index = num - 1;
475508

476-
/* elog(NOTICE, "a %d %d %d %d", i, index, sum, not);*/
509+
/*elog(NOTICE, "a %d %d %d %d", i, index, sum, not);*/
477510

478511
if (child)
479512
{
@@ -501,7 +534,10 @@ ruminv_tsvector_consistent(PG_FUNCTION_ARGS)
501534
nodes[index].sum++;
502535
}
503536

504-
child = &nodes[index];
537+
if (index == 0)
538+
child = NULL;
539+
else
540+
child = &nodes[index];
505541
}
506542
}
507543

@@ -511,7 +547,7 @@ ruminv_tsvector_consistent(PG_FUNCTION_ARGS)
511547
}
512548
else
513549
{
514-
/* for (i = 0; i < lastIndex; i++)
550+
/*for (i = 0; i < lastIndex; i++)
515551
{
516552
elog(NOTICE, "s %d %d %d %d", i, nodes[i].sum, nodes[i].parent, nodes[i].not);
517553
}*/

0 commit comments

Comments
 (0)