Patch for #2391: "Similar to" pattern matching does not operate as documented
От | Dhanaraj M |
---|---|
Тема | Patch for #2391: "Similar to" pattern matching does not operate as documented |
Дата | |
Msg-id | 4445E5C6.50609@sun.com обсуждение исходный текст |
Ответы |
Re: Patch for #2391: "Similar to" pattern matching does not operate as documented
|
Список | pgsql-patches |
Hi I attach the patch for this bug. I have run the regression test and passed. please review this and waiting for your reply. As explained in the mailing list, the parenthesis is appened when '|' operator is used without parenthesis. Thanks Dhanaraj =========================================================================== The following bug has been logged online: Bug reference: 2391 Logged by: Eric Noriega Email address: noriega ( at ) gwu ( dot ) edu PostgreSQL version: 7.0.5 Operating system: Linux Fedora core 4 Description: "Similar to" pattern matching does not operate as documented Details: As far as I can tell, this may be a bug in how the pattern matches. db=# select 'tab' similar to '(a|b)'; ?column? ---------- f db=# select 'tab' similar to 'a|b'; ?column? ---------- t The doc says: Like LIKE, the SIMILAR TO operator succeeds only if its pattern matches the entire string; this is unlike common regular expression practice, wherein the pattern may match any part of the string. If the second case is invalid as an expression (not clear in the docs:Parentheses may be used to group items into a single logical item), then the statement should fail, or return false, not return true. ================================================================================= * From: Tom Lane <tgl ( at ) sss ( dot ) pgh ( dot ) pa ( dot ) us> * To: "Eric Noriega" <noriega ( at ) gwu ( dot ) edu> * Subject: Re: BUG #2391: "Similar to" pattern matching does not operate as documented * Date: Thu, 13 Apr 2006 12:55:41 -0400 "Eric Noriega" <noriega ( at ) gwu ( dot ) edu> writes: > db=# select 'tab' similar to 'a|b'; > ?column? > ---------- > t Yeah, this is a bug ... the cause can be seen by looking at the underlying similar_escape() function, which converts a SIMILAR TO pattern into a POSIX regex pattern: regression=# select similar_escape('(a|b)', null); similar_escape ---------------- ^(a|b)$ (1 row) regression=# select similar_escape('a|b', null); similar_escape ---------------- ^a|b$ (1 row) regression=# I believe that in the second case, ^ and $ bind more tightly than | per POSIX rules. So we need to put parens around the pattern to prevent that. Thanks for the report! regards, tom lane ========================================================================= *** ./src/backend/utils/adt/regexp.c.orig Sun Mar 5 21:28:43 2006 --- ./src/backend/utils/adt/regexp.c Wed Apr 19 12:23:30 2006 *************** *** 522,527 **** --- 522,529 ---- elen; bool afterescape = false; int nquotes = 0; + int index; + bool paranFlag = false; /* This function is not strict, so must test explicitly */ if (PG_ARGISNULL(0)) *************** *** 549,559 **** errhint("Escape string must be empty or one character."))); } /* We need room for ^, $, and up to 2 output bytes per input byte */ ! result = (text *) palloc(VARHDRSZ + 2 + 2 * plen); r = VARDATA(result); *r++ = '^'; while (plen > 0) { --- 551,583 ---- errhint("Escape string must be empty or one character."))); } + /* Add parenthesis */ + for(index = 0; index < plen; index++) + { + char pchar = *(p+index); + + if (pchar == '"' || pchar == '%' || pchar == '_' || pchar == '\\' || + pchar == '.' || pchar == '?' || pchar == '{' || pchar == '(' || pchar == ')') + { + paranFlag = false; + break; + } + else if (pchar == '|') + paranFlag = true; + + } + /* We need room for ^, $, and up to 2 output bytes per input byte */ ! if (paranFlag == true) ! result = (text *) palloc(VARHDRSZ + 2 + 2 + 2 * plen); ! else ! result = (text *) palloc(VARHDRSZ + 2 + 2 * plen); ! r = VARDATA(result); *r++ = '^'; + if (paranFlag == true) + *r++ = '('; while (plen > 0) { *************** *** 593,598 **** --- 617,625 ---- p++, plen--; } + if (paranFlag == true) + *r++ = ')'; + *r++ = '$'; VARATT_SIZEP(result) = r - ((char *) result);
В списке pgsql-patches по дате отправления: