document = new DOMDocument('1.0', 'UTF-8'); $this->document->scrictErrorChecking = false; $this->setContent($content); parent::__construct(); } /** * @return string */ public function getContent() { // strip the body tags from the output (which are automatically added by DOMDocument) return preg_replace ( array ( '/^\s*]*>/i', '/<\/body[^>]*>\s*$/i' ), null, $this->getDocument()->saveXML($this->getDocument()->documentElement->lastChild) ); } /** * @param string $content * @return bool */ public function setContent($content) { // Ensure that \r (carriage return) characters don't get replaced with " " entity by DOMDocument // apparently it's in the XML standard, but that messes it up because we're using HTML $content = str_replace(chr(13), '', $content); return @$this->getDocument()->loadHTML( '' . "$content" ); } /** * @return DOMDocument */ public function getDocument() { return $this->document; } /** * A simple convenience wrapper around DOMDocument::getElementsByTagName(). * * @param string $name * @return DOMNodeList */ public function getElementsByTagName($name) { return $this->getDocument()->getElementsByTagName($name); } /** * @see HTMLValue::getContent() */ public function forTemplate() { return $this->getContent(); } }