id,summary,reporter,owner,description,type,status,priority,milestone,component,version,severity,resolution,keywords,cc
18007,is_serialized trouble for multibytes encoding,challet,,"Possible bug in the ''is_serialized'' function, in a multibytes environment :

== Context ==

* ''utf-8'' encoded database fields
* auto overlaoding singlebyte functions from conf (see http://www.php.net/manual/en/mbstring.overload.php)

Retrieving an option from the database, the ''get_option'' function try to deserialize it.

== Bug scenario ==
 * the ''strlen'' call is actually a ''mb_strlen'' one
 * the ''$length'' var is the number of characters (not bytes)
 * using {{{ $data[$length-1] }}} don't retrieve the last char but one before, it's actual position is undefined, depending on the other content of the string
 * this ''$lastc'' is not '';'' or ''}'' 
 * the string is understood has not serialized

== Patch proposal ==

here is a fix i made to make it work with multibytes string :
Not tested it for single bytes, but there is no reason for it to not work.
{{{
--- a/wp-includes/functions.php
+++ b/wp-includes/functions.php
@@ -256,7 +256,7 @@ function is_serialized( $data ) {
                return false;
        if ( ':' !== $data[1] )
                return false;
-       $lastc = $data[$length-1];
+       $lastc = substr($data, -1);
        if ( ';' !== $lastc && '}' !== $lastc )
                return false;
        $token = $data[0];
}}}

",defect (bug),new,normal,Future Release,Charset,3.2,normal,,has-patch 3.6-early,kurtpayne