HTML Eingaben intelligent filtern

<?php
/**
 * class XHTMLFilter
 * needs PHP's built in DOM classes
 * 
 * Copyright 2010 Jan Oliver Oelerich
 * 
 * Filters HTML Code from any input source by defined rules.
 * One can specify the allowed tags and the context they're allowed in
 * 
 * Usage: 
 *      $filter = new XHTMLFilter();
 *      $filter->allowTag('em')
 *             ->allowTag('strong',array('em');
 *      echo $filter->filter($htmlSource);
 *     
 *      This Code allows em and strong Tags. em may only be
 *      used at the top layer and within no other tag, strong can be used
 *      within em.
 */
class XHTMLFilter
{
    /**
     * In this variable, the rules of which html elements are allowed where,
     * are stored
     * 
     * @var array
     */
    private $_allowed;
 
 
    /**
     * Defines a new rule for an element $name, which should be allowed in the
     * context $allowedIn
     * 
     * Example: $this->allowTag('div', array('strong'));
     * Allow 'div' ELements within 'strong' Elements (but not, e.g., within 'em')
     * 
     * @param string $name
     * @param array $allowedIn
     * @return XHTMLFilter
     */
    public function allowTag($name, $allowedIn = array()) {
        $this->_allowed[] = array('name' => $name, 'in' => $allowedIn);
        return $this;
    }
 
    /**
     * Accepts a string as paramter. Replaces every NOT allowed html Tag 
     * with htmlentities()
     * 
     * Wraps html in <div /> and starts the XML traversing
     * 
     * @param string $value
     * @return string 
     */
    public function filter($value)
    {
        $value = '<div>' . $value . '</div>';
 
        $doc = new DOMDocument();
        $doc->loadXML($value);
 
        $doc->replaceChild($this->traverse($doc->firstChild),$doc->firstChild);
 
        return $this->stripXML($doc->saveXML());
    }
 
    /**
     * removes <?xml ..?><div>xxx</div> and returns xxx
     * 
     * @param xml string $source
     * @return string 
     */
    private function stripXml($source) {
        $source = preg_replace('/<\?[^>]+\?>/','',$source);
        $source = str_replace(array('<div>','</div>'),array('',''),$source);
        return trim($source);
    }
 
    /**
     * traverses the xml tree and calls replaceNode for not-allowed tags
     * 
     * @param DOMNode $nodeObj
     * @return DOMNode 
     */
    private function traverse($nodeObj) {
        foreach($nodeObj->childNodes as $node) {
            if($node->hasChildNodes())
                $node = $this->traverse($node);
 
            if(!$this->is_allowed($node)) {
                $nodeObj->replaceChild($this->replaceNode($node),$node);
            }
        }
        return $nodeObj;
    }
 
    /**
     * check, if a Node is allowed in it's context defined by the Rules
     * 
     * @param DOMNode $node
     * @return boolean
     */
    private function is_allowed($node) {
        foreach($this->_allowed as $rule) {
            if(
                    ($rule['name']==$node->localName &&
                    in_array($node->parentNode->localName,$rule['in'])) ||
                    get_class($node) == 'DOMText' ||
                    $node->parentNode->localName == 'div'
              )
                return true;
        }
        return false;
    }
 
    /**
     * replaces a Node recursively with its htmlentities() html string
     * 
     * @param DOMNode $node
     * @return DOMText
     */
    private function replaceNode($node) {
        if($node->hasChildNodes()) {
            foreach($node->childNodes as $childnode) {
                $node->replaceChild($this->replaceNode($childnode),$childnode);
            }
        }
 
        $newNode = new DOMText($this->getHtmlString($node));
        return $newNode;
    }
 
    /**
     * Gets the HTML String of a Node (without children!!!)
     * 
     * @param DOMNode $node
     * @return string
     */
    private function getHtmlString($node) {
        if(is_null($node->localName))
            return $node->textContent;
 
        $str = '<' . $node->localName;
        if(!is_null($node->attributes))
            foreach($node->attributes as $attr)
                $str .= ' ' . $attr->name . '="' . $attr->value . '"';
 
        $str .= '>' . $node->textContent . '</';
        $str .= $node->localName . '>';
        return $str;
    }
}