Bug #13635: preg_match_all returns some matched subexpressions under the wrong array index

From: delme@jumeaux.bc.ca Operating system: linux / apache (module) PHP version: 4.0.6 PHP Bug Type: PCRE related Bug description: preg_match_all returns some matched subexpressions under the wrong array index In a moderately complex PCRE regex, some parenthised subexpressions appear in the wrong place in preg_match_all's returned array of matches. This script reproduces the problem (it could probably be pared down further but is already a lot more simple than the case that originally showed me the error): <? preg_match_all("/(.*?)(?:(\\w+)( *@ *(\\w+)?(?=\\W|\$))?|\$)/s", "foo[bar @ baz]", $matches); print "<pre>"; for($c=0; $c<count($matches[0])-1; $c++) { print "MATCH $c:\n"; print " 0: {$matches[0][$c]}\n"; print " 1: {$matches[1][$c]}\n"; print " 2: {$matches[2][$c]}\n"; print " 3: {$matches[3][$c]}\n"; print " 4: {$matches[4][$c]}\n\n"; } print "</pre>\n"; ?> output under 4.0.6 is: MATCH 0: 0: foo 1: 2: foo 3: @ baz 4: baz MATCH 1: 0: [bar @ baz 1: [ 2: bar 3: 4: MATCH 2: 0: ] 1: ] 2: 3: 4: the following equivalent perl 5.6.0 one-liner produces the expected result: perl -e '$_="foo[bar @ baz]"; while(/(.*?)(?:(\w+)( *@ *(\w+)?(?=\W|$))?|$)/g) { print "MATCH:\n &: $&\n 1: $1\n 2: $2\n 3: $3\n 4: $4\n\n"; }' my copy of php is compiled as an apache module, as follows: ./configure --with-apache=../apache_$APACHE_VERSION --with-gd=/usr/local --with-jpeg-dir=/usr/local --with-png-dir=/usr/local --with-xpm-dir=/usr/X11R6 --with-imap=/usr/local --with-mysql=/usr/local/mysql --with-zlib=/usr/local --enable-sysvshm=yes --enable-sysvsem=yes --with-config-file-path=/usr/local/lib --enable-track-vars --disable-magic-quotes --with-pcre-regex --enable-apc -- Edit bug report at: http://bugs.php.net/?id=13635&edit=1