------- You are receiving this mail because: -------
You are on the CC list for the bug.
http://bugs.exim.org/show_bug.cgi?id=617
--- Comment #2 from Felipe Pena <felipensp@???> 2007-10-17 13:44:47 ---
Ok, thanks.
felipe@felipe:~$ gdb -q pcretest
Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1".
(gdb) r re-evil.txt
Starting program: /usr/local/bin/pcretest re-evil.txt
PCRE version 7.4 2007-09-21
/(?(DEFINE)(?<a>(?&b))(?<b>(?&a)))(?&a)/
abcdefghijklmnopqrstuvxz
Program received signal SIGSEGV, Segmentation fault.
0xb7ed6d29 in match () from /usr/local/lib/libpcre.so.0
This problem with recursion is expected, right? It's complicated to build a
prevention, i think. There various way for occur it...
--------------------------------------------
re> /(?+-a)/
Failed: (?R or (?[+-]digits must be followed by ) at offset 3
re> /(?-+a)/
Failed: unrecognized character after (? at offset 3
Shoudn't show the same message?
--------------------------------------------
re> /(?Ca)/
Failed: closing ) for (?C expected at offset 3
Wrong message?
--------------------------------------------
re> /(?(-1))/
Failed: reference to non-existent subpattern at offset 6
But accept:
re> /(?(+10))/
--------------------------------------------
re> /(?<>\d+)/
Failed: unrecognized character after (?< at offset 3
re> /\k''/
Failed: reference to non-existent subpattern at offset 3
Shouldn't the second show message "Failed: syntax error in subpattern name
(missing terminator)"? This occur with \k'`', for example.
--------------------------------------------
re> /(?(DEFINE)(?<b>a+b))(?<a>(a+(*THEN)b|(?&b)))/
data> aaab
0: aaab
1: <unset>
2: aaab
3: aaab
re> /(?<a>(a+(*THEN)b|(?&b)))(?(DEFINE)(?<b>a+b))/
data> aaab
0: aaab
1: aaab
2: aaab
In the last only 2 groups? It's correct?
Regards,
Felipe.
--
Configure bugmail:
http://bugs.exim.org/userprefs.cgi?tab=email
