Merge "Sanitizer: Allow attribute names to use any Unicode "Letter" or "Number""

[lhc/web/wiklou.git] / includes / Sanitizer.php

diff --git a/includes/Sanitizer.php b/includes/Sanitizer.php
index dd4a314..b08bc69 100644 (file)
--- a/includes/Sanitizer.php
+++ b/includes/Sanitizer.php
@@ -339,8 +339,8 @@ class Sanitizer {
         */
        static function getAttribsRegex() {
                if ( self::$attribsRegex === null ) {
         */
        static function getAttribsRegex() {
                if ( self::$attribsRegex === null ) {

-                       $attribFirst = '[:A-Z_a-z0-9]';
-                       $attrib = '[:A-Z_a-z-.0-9]';
+                       $attribFirst = "[:_\p{L}\p{N}]";
+                       $attrib = "[:_\.\-\p{L}\p{N}]";

                        $space = '[\x09\x0a\x0c\x0d\x20]';
                        self::$attribsRegex =
                                "/(?:^|$space)({$attribFirst}{$attrib}*)
                        $space = '[\x09\x0a\x0c\x0d\x20]';
                        self::$attribsRegex =
                                "/(?:^|$space)({$attribFirst}{$attrib}*)


@@ -351,7 +351,7 @@ class Sanitizer {
                                                | '([^']*)(?:'|\$)
                                                | (((?!$space|>).)*)
                                        )
                                                | '([^']*)(?:'|\$)
                                                | (((?!$space|>).)*)
                                        )

-                               )?(?=$space|\$)/sx";
+                               )?(?=$space|\$)/sxu";

                }
                return self::$attribsRegex;
        }
                }
                return self::$attribsRegex;
        }