/**
* Remove script tags
*
* @access public
* @param string $data Input data
* @return string
*/
public function removeBlacklistedTags($data)
{
$dom = XmlParser::getDomDocument($data);
if ($dom === false) {
return '';
}
$xpath = new DOMXpath($dom);
$nodes = $xpath->query(implode(' | ', $this->tag_blacklist));
foreach ($nodes as $node) {
$node->parentNode->removeChild($node);
}
return $dom->saveXML();
}
/**
* Strip useless tags
*
* @access public
*/
public function stripGarbage()
{
$dom = XmlParser::getDomDocument($this->content);
if ($dom !== false) {
$xpath = new DOMXPath($dom);
foreach ($this->stripTags as $tag) {
$nodes = $xpath->query('//' . $tag);
if ($nodes !== false && $nodes->length > 0) {
Logger::setMessage(get_called_class() . ': Strip tag: "' . $tag . '"');
foreach ($nodes as $node) {
$node->parentNode->removeChild($node);
}
}
}
foreach ($this->stripAttributes as $attribute) {
$nodes = $xpath->query('//*[contains(@class, "' . $attribute . '") or contains(@id, "' . $attribute . '")]');
if ($nodes !== false && $nodes->length > 0) {
Logger::setMessage(get_called_class() . ': Strip attribute: "' . $attribute . '"');
foreach ($nodes as $node) {
if ($this->shouldRemove($dom, $node)) {
$node->parentNode->removeChild($node);
}
}
}
}
$this->content = $dom->saveXML($dom->documentElement);
}
}
/**
* Strip useless tags.
*
* @param string $content
*
* @return string
*/
public function stripGarbage($content)
{
$dom = XmlParser::getDomDocument($content);
if ($dom !== false) {
$xpath = new DOMXPath($dom);
$this->stripTags($xpath);
$this->stripAttributes($dom, $xpath);
$content = $dom->saveXML($dom->documentElement);
}
return $content;
}
public function testScanXmlWithDTD()
{
$xml = <<<XML
<?xml version="1.0"?>
<!DOCTYPE results [
<!ELEMENT results (result+)>
<!ELEMENT result (#PCDATA)>
]>
<results>
<result>test</result>
</results>
XML;
$result = XmlParser::getDomDocument($xml);
$this->assertTrue($result instanceof DOMDocument);
$this->assertTrue($result->validate());
}
