Changeset 13181 for sites/trunk/trac.wordpress.org/templates/update-headers.php

Timestamp:

02/07/2024 05:58:19 AM (3 years ago)

Author:

dd32

Message:

Trac: Escape HTML within script tags, as DOMDocument doesn't like it. Remove CDATA entirely, I don't think it's needed.

File:

• sites/trunk/trac.wordpress.org/templates/update-headers.php (modified) (2 diffs)

sites/trunk/trac.wordpress.org/templates/update-headers.php

@@ -7 +7 @@
 function domdocument_from_url( $url ) {
         $html = file_get_contents( $url );
+
+        /*
+         * Escape HTML within Javascript strings.
+         * DomDocument doesn't handle HTML tags within Javascript strings.
+         * See https://stackoverflow.com/questions/40703313/php-domdocument-errors-while-parsing-unescaped-strings
+         */
+        $html = preg_replace_callback(
+                '!<script([^>]+)>(.*?)</script>!ism',
+                function( $m ) {
+                        $escaped = $m[2];
+                        $escaped = str_replace( array( '<', '>' ), array( '\x3C',  '\x3E' ), $escaped );
+                        return "<script{$m[1]}>{$escaped}</script>";
+                },
+                $html
+        );
 
         $doc = new DOMDocument();
@@ -42 +57 @@
         $html = preg_replace( '#<style([^>]*)><!\[CDATA\[(.+?)\]\]></style>#ism', "<style$1>$2</style>", $html );
 
-        // Escape CDATA tags in <script>
-        $html = preg_replace( '#<script([^>]*)><!\[CDATA\[(.+?)\]\]></script>#ism', "<script$1>//<![CDATA[\n$2\n//]]></script>", $html );
+        // Remove CDATA tags in <script>
+        $html = preg_replace( '#<script([^>]*)><!\[CDATA\[(.+?)\]\]></script>#ism', "<script$1>$2</script>", $html );
 
         // Remove trailing whitespace.