Bug #67355 [Com]: Problem with named subpattern in a condition on recursion.
| From: | lingtalfi at gmail dot com | Date: | Thu, 29 May 2014 06:34:24 +0000 |
| Subject: | Bug #67355 [Com]: Problem with named subpattern in a condition on recursion. | ||
| References: | 1 | Groups: | php.bugs |
| Request: | Send a blank email to php-bugs+get-185963@lists.php.net to get a copy of this message | ||
Edit report at https://bugs.php.net/bug.php?id=67355&edit=1
ID: 67355
Comment by: lingtalfi at gmail dot com
Reported by: lingtalfi at gmail dot com
Summary: Problem with named subpattern in a condition on
recursion.
Status: Feedback
Type: Bug
Package: PCRE related
Operating System: MacOSX 10.5.8
PHP Version: Irrelevant
Block user comment: N
Private report: N
New Comment:
Submitted to pcre
http://bugs.exim.org/show_bug.cgi?id=1484
Previous Comments:
------------------------------------------------------------------------
[2014-05-29 05:57:03] rasmus@php.net
Please test this using the command line pcretest tool. Most issues filed against PHP and PCRE
aren't PHP-related but rather PCRE-implementation-related. If pcretest and php differ, then it
is a PHP issue. If pcretest shows the same behaviour as PHP and you believe it is a bug, file it
against PCRE.
------------------------------------------------------------------------
[2014-05-28 20:10:54] lingtalfi at gmail dot com
Description:
------------
Using PHP Version 5.5.8 on macOSX 10.8.5
PCRE Library Version 8.33 2013-05-28
I may be wrong,
but I think there is a problem with the PCRE implementation in php.
The php test is here: http://ideone.com/NuORid
The perl equivalent test is here: http://ideone.com/RGHaLK
Now, I know perl and pcre are two different flavours, but in this case,
I was expecting the same behaviour.
About the php test:
I tried to use a condition on a recursion (which is supposed to always fail at the top level).
The condition has two alternatives: the yes condition and the no condition.
In the subject, there is no recursion at all, so the second alternative should be used,
and the first alternative ignored.
It seems that the pattern match only if the first alternative doesn't use a duplicate named
subpattern,
or, alternatively, if a backreference to the named subpattern uses the numbered version of the
subpattern (and not the name).
Test script:
---------------
http://ideone.com/NuORid
Expected result:
----------------
The script http://ideone.com/NuORid should output
"ok"
Actual result:
--------------
The script http://ideone.com/NuORid outputs "oops"
------------------------------------------------------------------------
--
Edit this bug report at https://bugs.php.net/bug.php?id=67355&edit=1