closeOpenTags.php

<?php

/**
 * Close open HTML tags and get the ones left open
 * 
 * @author   Pedro Sanção <pedro at sancao do co>
 * @license  MIT Licence
 * 
 * @param string $html
 * @return array tags without closing pair
 */
function closeOpenTags(&$html) {
	$indexes = array();
	$matches = array();
	preg_match_all('#<(/?)([a-z0-9]+).*?(/?)>#m', $html, $matches);
	$tagNames = $matches[2];
	$limit = count($matches[0]);
	for ($k = 0; $k < $limit; $k++) {
		if (!empty($matches[3][$k])) { // self closing tags
		} elseif (empty($matches[1][$k])) { // tag open
			array_push($indexes, $k);
		} elseif ($tagNames[$k] === $tagNames[end($indexes)]) { // tag closed
			array_pop($indexes);
		} else { // malformed HTML
		}
	}
	if (!empty($indexes)) {
		$indexes = array_flip($indexes);
		$tags = array_intersect_key($tagNames, $indexes);
		foreach ($tags as &$tag) {
			$tag = sprintf('</%s>', $tag);
		}
		$html = trim($html) . join(array_reverse($tags));
		return array_intersect_key($matches[0], $indexes);
	}
	return array();
}