Index: wp-includes/formatting.php =================================================================== --- wp-includes/formatting.php (revision 18495) +++ wp-includes/formatting.php (working copy) @@ -981,11 +981,14 @@ * @since 2.0.4 * * @author Leonard Lin + * @author Kurt Payne * @license GPL * @copyright November 4, 2001 * @version 1.1 * @todo Make better - change loop condition to $text in 1.2 * @internal Modified by Scott Reilly (coffee2code) 02 Aug 2004 + * 1.2 Updated to fix bug 18312 - improper balancing of nested lists + * Added the ability to skip cleaning of certain tags * 1.1 Fixed handling of append/stack pop order of end text * Added Cleaning Hooks * 1.0 First Version @@ -994,103 +997,129 @@ * @return string Balanced text. */ function force_balance_tags( $text ) { - $tagstack = array(); - $stacksize = 0; - $tagqueue = ''; - $newtext = ''; $single_tags = array('br', 'hr', 'img', 'input'); // Known single-entity/self-closing tags - $nestable_tags = array('blockquote', 'div', 'span'); // Tags that can be immediately nested within themselves + $skip_tags = array('script', 'style'); // No fixing should be done inside these tags // WP bug fix for comments - in case you REALLY meant to type '< !--' $text = str_replace('< !--', '< !--', $text); // WP bug fix for LOVE <3 (and other situations with '<' before a number) $text = preg_replace('#<([0-9]{1})#', '<$1', $text); - while ( preg_match("/<(\/?[\w:]*)\s*([^>]*)>/", $text, $regex) ) { - $newtext .= $tagqueue; + // Copy the old text to a new field + $newtext = $text; - $i = strpos($text, $regex[0]); - $l = strlen($regex[0]); + // Keeping track of the tags we've seen + $stack = array(); - // clear the shifter - $tagqueue = ''; - // Pop or Push - if ( isset($regex[1][0]) && '/' == $regex[1][0] ) { // End Tag - $tag = strtolower(substr($regex[1],1)); - // if too many closing tags - if( $stacksize <= 0 ) { - $tag = ''; - // or close to be safe $tag = '/' . $tag; - } - // if stacktop value = tag close value then pop - else if ( $tagstack[$stacksize - 1] == $tag ) { // found closing tag - $tag = ''; // Close Tag - // Pop - array_pop( $tagstack ); - $stacksize--; - } else { // closing tag not at top, search for it - for ( $j = $stacksize-1; $j >= 0; $j-- ) { - if ( $tagstack[$j] == $tag ) { - // add tag to tagqueue - for ( $k = $stacksize-1; $k >= $j; $k--) { - $tagqueue .= ''; - $stacksize--; - } - break; - } - } - $tag = ''; - } - } else { // Begin Tag - $tag = strtolower($regex[1]); + // Start the position tracking for the new text at 0 + $newpos = 0; - // Tag Cleaning + // Start off with a skip level of 0. Skipping tags can be nested and we + // have to roll / unroll each one every time we open / close a skip tag + $skip_level = 0; - // If self-closing or '', don't do anything. - if ( substr($regex[2],-1) == '/' || $tag == '' ) { - // do nothing - } - // ElseIf it's a known single-entity tag but it doesn't close itself, do so - elseif ( in_array($tag, $single_tags) ) { - $regex[2] .= '/'; - } else { // Push the tag onto the stack - // If the top of the stack is the same as the tag we want to push, close previous tag - if ( $stacksize > 0 && !in_array($tag, $nestable_tags) && $tagstack[$stacksize - 1] == $tag ) { - $tagqueue = ''; - $stacksize--; - } - $stacksize = array_push ($tagstack, $tag); - } + // Look for matches + while ( preg_match('/<(\/?[\w:]*)\s*([^>]*)>/S', $text, $matches, PREG_OFFSET_CAPTURE, $pos) ) { - // Attributes - $attributes = $regex[2]; - if( !empty($attributes) ) - $attributes = ' '.$attributes; + /** + * Matches: + * [0][0] = whole match (e.g.
+ * [0][1] = numeric position in $text (e.g. 0) + * [1][0] = just the tag (e.g. div or /div) + * [1][1] = numeric position in $text (e.g. 1) + * [2][0] = any attributes (e.g. style="align:center;") + * [2][1] = numeric position in $text (e.g. 5) + */ - $tag = '<' . $tag . $attributes . '>'; - //If already queuing a close tag, then put this tag on, too - if ( !empty($tagqueue) ) { - $tagqueue .= $tag; - $tag = ''; - } + // Is it a closing tag or not? + $closing = ($matches[1][0][0] == '/'); + + // Just the tag, no brackets, slashes, or attributes + $tag = ( $closing ? strtolower(substr($matches[1][0], 1) ) : strtolower($matches[1][0])); + + // Find the position of this + $newpos = strpos($newtext, $matches[0][0], $newpos); + + // Skip any empty tags (most likely comments or malformed html tags ... + // we could make the doc worse by adding a closing tag to a bad html tag) + if ( empty($matches[1][0]) ) { + $pos = $matches[0][1] + 1; } - $newtext .= substr($text, 0, $i) . $tag; - $text = substr($text, $i + $l); - } + + // If it's a known single tag and it's a closing tag, discard it + elseif ( in_array($tag, $single_tags) && $closing ) { + if ( $skip_level <= 0 ) + $newtext = substr($newtext, 0, $newpos) . substr($newtext, $newpos + strlen($matches[0][0])); + + // Move to the next match + $pos = $matches[0][1] + 1; + } + + // If it's a known single tag and it's not properly closed, close it + elseif ( in_array($tag, $single_tags) && '/>' != substr($matches[0][0], -2) ) { + + // Add the properly closed tag to the new text + if ( $skip_level <= 0 ) + $newtext = substr($newtext, 0, $newpos) . substr($matches[0][0], 0, -1) . '/>' . substr($newtext, $newpos + strlen($matches[0][0])); + + // Move to the next match + $pos = $matches[0][1] + 1; + } + + // If it's an opening tag, add it to the stack + elseif ( !$closing ) { + + // If it's a self closing tag (regardless of whether we have it in our list) we don't + // want it in the stack, as it's legitimate XML, but it will mess up our balancing act + if ( '/>' != substr($matches[0][0], -2) && ($skip_level <= 0 || in_array($tag, $skip_tags)) ) + array_push($stack, $matches); + + // Are we in skip mode? + if ( in_array($tag, $skip_tags) ) + $skip_level++; + + // Move to the next match + $pos = $matches[0][1] + 1; + } + + // If we're here, it must be a closing tag + else { - // Clear Tag Queue - $newtext .= $tagqueue; + // If the stack is empty, the closing tag must be useless, skip it, move to the next match + if ( empty($stack) ) { + if ( $skip_level <= 0 ) + $newtext = substr($newtext, 0, $newpos) . substr($newtext, $newpos + strlen($matches[0][0])); + $pos = $matches[0][1] + 1; + } else { + $lasttag = array_pop($stack); - // Add Remaining text - $newtext .= $text; + // If the closing tag matches the last tag on the stack, just move to the next match + if ( strtolower($lasttag[1][0]) == $tag ) { + $pos = $matches[0][1] + 1; - // Empty Stack - while( $x = array_pop($tagstack) ) - $newtext .= ''; // Add remaining tags to close + // Can we stop skipping? + if ( $skip_level > 0 && in_array($tag, $skip_tags) ) + $skip_level--; + + // If it doesn't match, add in a closing tag, but don't remove it from the original text. This will force the tag to run through the + // loop again and continue to go up a level in the stack until it finds its match or until the stack is empty and the tag is discarded + } else { + if ( $skip_level <= 0 ) + $newtext = substr($newtext, 0, $newpos) . '' . substr($newtext, $newpos); + } + } + } + } + + // If there are still elements on the stack, add the closing tags at the bottom of the text + while ( !empty($stack) ) { + $tag = array_pop($stack); + $newtext .= ''; + } // WP fix for the bug with HTML comments - $newtext = str_replace("< !--","