Merge "resourceloader: Tidy up RL to simplify ResourceLoaderEditToolbarModule"

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

diff --git a/includes/Sanitizer.php b/includes/Sanitizer.php
index fbf3234..f88dd05 100644 (file)
--- a/includes/Sanitizer.php
+++ b/includes/Sanitizer.php
@@ -346,8 +346,8 @@ class Sanitizer {
                                  ($space*=$space*
                                        (?:
                                         # The attribute value: quoted or alone
-                                         \"([^<\"]*)\"
-                                        | '([^<']*)'
+                                         \"([^<\"]*)(?:\"|\$)
+                                        | '([^<']*)(?:'|\$)
                                         |  ([a-zA-Z0-9!#$%&()*,\\-.\\/:;<>?@[\\]^_`{|}~]+)
                                        )
                                )?(?=$space|\$)/sx";
@@ -454,15 +454,13 @@ class Sanitizer {
        public static function removeHTMLtags( $text, $processCallback = null,
                $args = array(), $extratags = array(), $removetags = array()
        ) {
-               global $wgUseTidy;
-
                extract( self::getRecognizedTagData( $extratags, $removetags ) );
 
                # Remove HTML comments
                $text = Sanitizer::removeHTMLcomments( $text );
                $bits = explode( '<', $text );
                $text = str_replace( '>', '&gt;', array_shift( $bits ) );
-               if ( !$wgUseTidy ) {
+               if ( !MWTidy::isEnabled() ) {
                        $tagstack = $tablestack = array();
                        foreach ( $bits as $x ) {
                                $regs = array();
@@ -538,7 +536,7 @@ class Sanitizer {
                                                        $badtag = true;
                                                } elseif ( in_array( $t, $tagstack ) && !isset( $htmlnest[$t] ) ) {
                                                        $badtag = true;
-                                               # Is it a self closed htmlpair ? (bug 5487)
+                                               #  Is it a self closed htmlpair ? (bug 5487)
                                                } elseif ( $brace == '/>' && isset( $htmlpairs[$t] ) ) {
                                                        $badtag = true;
                                                } elseif ( isset( $htmlsingleonly[$t] ) ) {
@@ -740,7 +738,7 @@ class Sanitizer {
 
                $out = array();
                foreach ( $attribs as $attribute => $value ) {
-                       #allow XML namespace declaration if RDFa is enabled
+                       # allow XML namespace declaration if RDFa is enabled
                        if ( $wgAllowRdfaAttributes && preg_match( self::XMLNS_ATTRIBUTE_PATTERN, $attribute ) ) {
                                if ( !preg_match( self::EVIL_URI_PATTERN, $value ) ) {
                                        $out[$attribute] = $value;
@@ -786,7 +784,7 @@ class Sanitizer {
                                || $attribute === 'itemref' || $attribute === 'itemscope'
                                || $attribute === 'itemtype'
                        ) {
-                               //Paranoia. Allow "simple" values but suppress javascript
+                               // Paranoia. Allow "simple" values but suppress javascript
                                if ( preg_match( self::EVIL_URI_PATTERN, $value ) ) {
                                        continue;
                                }
@@ -796,7 +794,7 @@ class Sanitizer {
                        #       validation code that can be used by tag hook handlers, etc
                        if ( $attribute === 'href' || $attribute === 'src' ) {
                                if ( !preg_match( $hrefExp, $value ) ) {
-                                       continue; //drop any href or src attributes not using an allowed protocol.
+                                       continue; // drop any href or src attributes not using an allowed protocol.
                                        // NOTE: this also drops all relative URLs
                                }
                        }
@@ -1699,7 +1697,7 @@ class Sanitizer {
                        # rbc
                        'rb'         => $common,
                        'rp'         => $common,
-                       'rt'         => $common, #array_merge( $common, array( 'rbspan' ) ),
+                       'rt'         => $common, # array_merge( $common, array( 'rbspan' ) ),
                        'rtc'         => $common,
 
                        # MathML root element, where used for extensions
@@ -1809,7 +1807,9 @@ class Sanitizer {
                        $host = preg_replace( $strip, '', $host );
 
                        // IPv6 host names are bracketed with [].  Url-decode these.
-                       if ( substr_compare( "//%5B", $host, 0, 5 ) === 0 && preg_match( '!^//%5B(.*?)%5D((:\d+)?)$!', $host, $matches ) ) {
+                       if ( substr_compare( "//%5B", $host, 0, 5 ) === 0 &&
+                               preg_match( '!^//%5B([0-9A-Fa-f:.]+)%5D((:\d+)?)$!', $host, $matches )
+                       ) {
                                $host = '//[' . $matches[1] . ']' . $matches[2];
                        }