https://bugs.exim.org/show_bug.cgi?id=1626
Bug ID: 1626
Summary: coredump at least since 8.35 with expression "[\]\)"
Product: PCRE
Version: 8.37
Hardware: x86-64
OS: Linux
Status: NEW
Severity: bug
Priority: medium
Component: Code
Assignee: ph10@???
Reporter: markuman@???
CC: pcre-dev@???
Created attachment 807
-->
https://bugs.exim.org/attachment.cgi?id=807&action=edit
example file where the crash is reproduceable with
GNU Octave is using libpcre for it's regexp & regexprep function.
This will coredump GNU Octave
file = "xlsread.m";
fid = fopen (file);
str = fread (fid, "char=>char").';
fclose (fid);
mainregex =
'((^|\n)(([\]\)}\w.]''+|[^''%])+|''[^''\n]*(''''[^''\n]*)*'')*)[^\n]*';
str = regexprep (str, mainregex, "$1");
By skipping the tokken with ?:[\]\) it works
file = "xlsread.m";
fid = fopen (file);
str = fread (fid, "char=>char").';
fclose (fid);
mainregex =
'((^|\n)((?:[\]\)}\w.]''+|[^''%])+|''[^''\n]*(''''[^''\n]*)*'')*)[^\n]*';
str = regexprep (str, mainregex, "$1");
I can reproduce it with:
Arch Linux
GNU Octave 3.8 & libpcre 8.36-2
GNU Octave 3.8 & libpcre 8.37-1
GNU Octave 4.0.rc3 & libpcre 8.36-2
Debian
GNU Octave 4.0.rc4 & libpcre 8.35-3
It works fine without coredump on Ubuntu
GNU Octave 3.8 & libpcre 8.31-2
--
You are receiving this mail because:
You are on the CC list for the bug.