From: Ilmari Karonen <vyznev@users.mediawiki.org>
Date: Thu, 27 Nov 2008 20:11:38 +0000 (+0000)
Subject: (bug 6100) Strip Unicode BiDi embedding/override characters (U+202A - U+202E) from... 
X-Git-Tag: 1.31.0-rc.0~44187
X-Git-Url: http://git.cyclocoop.org/%7B%24www_url%7Dadmin/password.php?a=commitdiff_plain;h=ce16f36e954fd228ad1d605bc3650deb67e6183e;p=lhc%2Fweb%2Fwiklou.git

(bug 6100) Strip Unicode BiDi embedding/override characters (U+202A - U+202E) from titles.
NOTE: run maintenance/cleanupImages.php and cleanupTitles.php ASAP after deploying this!
---

diff --git a/RELEASE-NOTES b/RELEASE-NOTES
index f496598899..ef0d8bdc33 100644
--- a/RELEASE-NOTES
+++ b/RELEASE-NOTES
@@ -369,6 +369,10 @@ The following extensions are migrated into MediaWiki 1.14:
 * Honour unchecked "Leave a redirect behind" for moved subpages
 * (bug 16440) Broken 0-byte math renderings are now deleted and re-rendered
   when page is re-parsed.
+* (bug 6100) Unicode BiDi embedding/override characters (U+202A - U+202E) are
+  now automatically removed from titles; these characters can accidentally end
+  up in copy-and-pasted titles, and, by overriding normal bidirectional text
+  handling, can lead to annoying behavior such as text rendering backwards
 
 === API changes in 1.14 ===
 
diff --git a/includes/Title.php b/includes/Title.php
index aa9a739bdf..88118fdefa 100644
--- a/includes/Title.php
+++ b/includes/Title.php
@@ -2047,8 +2047,7 @@ class Title {
 		# Strip Unicode bidi override characters.
 		# Sometimes they slip into cut-n-pasted page titles, where the
 		# override chars get included in list displays.
-		$dbkey = str_replace( "\xE2\x80\x8E", '', $dbkey ); // 200E LEFT-TO-RIGHT MARK
-		$dbkey = str_replace( "\xE2\x80\x8F", '', $dbkey ); // 200F RIGHT-TO-LEFT MARK
+		$dbkey = preg_replace( '/\xE2\x80[\x8E\x8F\xAA-\xAE]/S', '', $dbkey );
 
 		# Clean up whitespace
 		#