Ticket #15657: patch.txt

--- wp-includes/formatting.php  27 Nov 2010 03:12:10 -0000      1.1
+++ wp-includes/formatting.php  3 Dec 2010 05:54:41 -0000
@@ -1829,7 +1829,9 @@
 
                $text = apply_filters('the_content', $text);
                $text = str_replace(']]>', ']]>', $text);
-               $text = strip_tags($text);
+// jwz: do better tag stripping
+//             $text = strip_tags($text);
+               $text = wp_strip_all_tags($text);
                $excerpt_length = apply_filters('excerpt_length', 55);
                $excerpt_more = apply_filters('excerpt_more', ' ' . '[...]');
                $words = preg_split("/[\n\r\t ]+/", $text, $excerpt_length + 1, PREG_SPLIT_NO_EMPTY);
@@ -2770,6 +2772,16 @@
  */
 function wp_strip_all_tags($string, $remove_breaks = false) {
        $string = preg_replace( '@<(script|style)[^>]*?>.*?</\\1>@si', '', $string );
+
+       // jwz: convert P and P-like block tags to \n\n.
+       $string = preg_replace( '@</?\s*(P|UL|OL|DL|BLOCKQUOTE)\b[^>]*?>@si',
+                               "\n\n", $string );
+       // jwz: convert BR and other known block-display tags to \n.
+       $string = preg_replace( '@<(BR|DIV|LI|DT|DD|TR|TD|H\d)\b[^>]*?>@si',
+                               "\n", $string );
+       // jwz: compress newlines
+       $string = preg_replace( "@\n\n\n+@si", "\n\n", $string );
+
        $string = strip_tags($string);
 
        if ( $remove_breaks )