From: Brion Vibber <brion@users.mediawiki.org>
Date: Thu, 18 Dec 2003 11:31:03 +0000 (+0000)
Subject: Hopefully slightly saner handling of character entities in links; now will
X-Git-Tag: 1.3.0beta1~1228
X-Git-Url: http://git.cyclocoop.org/%27.parametre_url%28%20%20%20generer_action_auteur%28%27charger_plugin%27%2C%20%27update_flux%27%29%2C%27update_flux%27%2C%20%27oui%27%29.%27?a=commitdiff_plain;h=aea531963521f8771529c1ea267e30e7b929439d;p=lhc%2Fweb%2Fwiklou.git

Hopefully slightly saner handling of character entities in links; now will
use UTF-8 characters on a UTF-8 wiki for the Latin-1 named entities; this
prevents corrupt in-wiki links for cases such as [[Wei&szlig; kreuz]] or
[[Br&ucirc;ker:Foo]]. There is a small chance this will harm interwiki
links to latin-1 wikis that don't accept UTF-8 incoming URLs, but on
balance this is a lesser harm. Raw bytes in such links can still be used
(%XX) if need be.

This shouldn't change behavior on Latin-1 wikis.
---

diff --git a/includes/Title.php b/includes/Title.php
index d0300ed4cc..2b3da1fae3 100644
--- a/includes/Title.php
+++ b/includes/Title.php
@@ -1,7 +1,6 @@
 <?
 # See title.doc
 
-/* private static */ $title_html_translation_table = array_flip( get_html_translation_table( HTML_ENTITIES ) );
 /* private static */ $title_interwiki_cache = array();
 
 class Title {
@@ -34,13 +33,19 @@ class Title {
 	
 	function newFromText( $text )
 	{	
-		global $title_html_translation_table;
+		static $trans;
 		$fname = "Title::newFromText";
 		wfProfileIn( $fname );
 		
 		# Note - mixing latin1 named entities and unicode numbered
 		# ones will result in a bad link.
-		$trans =& $title_html_translation_table;
+		if( !isset( $trans ) ) {
+			global $wgInputEncoding;
+			$trans = array_flip( get_html_translation_table( HTML_ENTITIES ) );
+			if( strcasecmp( "utf-8", $wgInputEncoding ) == 0 ) {
+			    $trans = array_map( "utf8_encode", $trans );
+			}
+		}
 
 		$text = strtr( $text, $trans );