Movatterモバイル変換

[0]ホーム
URL:

Skip to content

Navigation Menu

Sign in
Appearance settings

Search code, repositories, users, issues, pull requests...

Provide feedback

We read every piece of feedback, and take your input very seriously.

Saved searches

Use saved searches to filter your results more quickly

 Sign up
Appearance settings

Commite81e574

Browse files
committed
Fix full text search to handle NOT above a phrase search correctly.
Queries such as '!(foo<->bar)' failed to find matching rows whenimplemented as a GiST or GIN index search. That's because offailing to handle phrase searches as tri-valued when consideringa query without any position information for the target tsvector.We can only say that the phrase operator might match, not that itdoes match; and therefore its NOT also might match. The previouscoding incorrectly inverted the approximate phrase result todecide that there was certainly no match.To fix, we need to make TS_phrase_execute return a real ternary result,and then bubble that up accurately in TS_execute. As long as we haveto do that anyway, we can simplify the baroque things TS_phrase_executewas doing internally to manage tri-valued searching with only a boolas explicit result.For now, I left the externally-visible result of TS_execute as a plainbool. There do not appear to be any outside callers that need todistinguish a three-way result, given that they passed in a flagsaying what to do in the absence of position data. This might needto change someday, but we wouldn't want to back-patch such a change.Although tsginidx.c has its own TS_execute_ternary implementation foruse at upper index levels, that sadly managed to get this case wrongas well :-(. Fixing it is a lot easier fortunately.Per bug #16388 from Charles Offenbacher. Back-patch to 9.6 wherephrase search was introduced.Discussion:https://postgr.es/m/16388-98cffba38d0b7e6e@postgresql.org
1 parent5ac2475 commite81e574

File tree

7 files changed

+734
-115
lines changed

7 files changed

+734
-115
lines changed

‎src/backend/utils/adt/tsginidx.c

Lines changed: 20 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -210,6 +210,11 @@ checkcondition_gin(void *checkval, QueryOperand *val, ExecPhraseData *data)
210210

211211
/*
212212
* Evaluate tsquery boolean expression using ternary logic.
213+
*
214+
* Note: the reason we can't use TS_execute() for this is that its API
215+
* for the checkcondition callback doesn't allow a MAYBE result to be
216+
* returned, but we might have MAYBEs in the gcv->check array.
217+
* Perhaps we should change that API.
213218
*/
214219
staticGinTernaryValue
215220
TS_execute_ternary(GinChkVal*gcv,QueryItem*curitem,boolin_phrase)
@@ -230,9 +235,19 @@ TS_execute_ternary(GinChkVal *gcv, QueryItem *curitem, bool in_phrase)
230235
switch (curitem->qoperator.oper)
231236
{
232237
caseOP_NOT:
233-
/* In phrase search, always return MAYBE since we lack positions */
238+
239+
/*
240+
* Below a phrase search, force NOT's result to MAYBE. We cannot
241+
* invert a TRUE result from the subexpression to FALSE, since
242+
* TRUE only says that the subexpression matches somewhere, not
243+
* that it matches everywhere, so there might be positions where
244+
* the NOT will match. We could invert FALSE to TRUE, but there's
245+
* little point in distinguishing TRUE from MAYBE, since a recheck
246+
* will have been forced already.
247+
*/
234248
if (in_phrase)
235249
returnGIN_MAYBE;
250+
236251
result=TS_execute_ternary(gcv,curitem+1,in_phrase);
237252
if (result==GIN_MAYBE)
238253
returnresult;
@@ -242,7 +257,8 @@ TS_execute_ternary(GinChkVal *gcv, QueryItem *curitem, bool in_phrase)
242257

243258
/*
244259
* GIN doesn't contain any information about positions, so treat
245-
* OP_PHRASE as OP_AND with recheck requirement
260+
* OP_PHRASE as OP_AND with recheck requirement, and always
261+
* reporting MAYBE not TRUE.
246262
*/
247263
*(gcv->need_recheck)= true;
248264
/* Pass down in_phrase == true in case there's a NOT below */
@@ -258,7 +274,8 @@ TS_execute_ternary(GinChkVal *gcv, QueryItem *curitem, bool in_phrase)
258274
val2=TS_execute_ternary(gcv,curitem+1,in_phrase);
259275
if (val2==GIN_FALSE)
260276
returnGIN_FALSE;
261-
if (val1==GIN_TRUE&&val2==GIN_TRUE)
277+
if (val1==GIN_TRUE&&val2==GIN_TRUE&&
278+
curitem->qoperator.oper!=OP_PHRASE)
262279
returnGIN_TRUE;
263280
else
264281
returnGIN_MAYBE;

0 commit comments

Comments
 (0)
[8]ページ先頭
©2009-2025 Movatter.jp