Bug #52592 [Opn->Bgs]: mb_ereg_replace and the Greek capital Pi

Edit report at http://bugs.php.net/bug.php?id=52592&edit=1 ID: 52592 Updated by: aharvey@php.net Reported by: pj at ezgr dot net Summary: mb_ereg_replace and the Greek capital Pi -Status: Open +Status: Bogus Type: Bug Package: mbstring related Operating System: Centos 5.5 x64 PHP Version: 5.2.14 Block user comment: N New Comment: You need to also call mb_regex_encoding('UTF-8'); before using a UTF-8 regular expression. Previous Comments: ------------------------------------------------------------------------ [2010-08-12 14:36:15] pj at ezgr dot net Description: ------------ PHP: 5.2.14, Apache 2.2.15, mod_php While \s is supposed to match all whitespace, the greek unicode letter Pi (Π) whose code is 0xCEA0 is matched too and if replaced with something, it's stripped of its second byte (0xA0). Test script: --------------- <?php mb_internal_encoding('UTF-8'); $testStr = 'Π Π Π!'; $newStr = mb_ereg_replace('\s+','_',$testStr); echo $testStr; echo $newStr; echo urlencode($testStr); echo urlencode($newStr); ?> Expected result: ---------------- Π Π Π! Π__Π__Π! %CE%A0++%CE%A0++%CE%A0%21 %CE%A0__%CE%A0__%CE%A0%21 Actual result: -------------- Π Π Π! [non printable character]_[non printable character]_[non printable character]! %CE%A0++%CE%A0++%CE%A0%21 %CE_%CE_%CE_%21 ------------------------------------------------------------------------ -- Edit this bug report at http://bugs.php.net/bug.php?id=52592&edit=1