public function testOverrideFilters()
{
$data = '<iframe src="http://www.kickstarter.com/projects/lefnire/habitrpg-mobile/widget/video.html" height="480" width="640" frameborder="0"></iframe>';
$f = new Filter($data, 'http://blabla');
$f->setIframeWhitelist(array('http://www.kickstarter.com'));
$this->assertEquals($data, $f->execute());
$data = '<iframe src="http://www.youtube.com/bla" height="480" width="640" frameborder="0"></iframe>';
$f = new Filter($data, 'http://blabla');
$f->setIframeWhitelist(array('http://www.kickstarter.com'));
$this->assertEmpty($f->execute());
}
/**
* Find the item enclosure
*
* @access public
* @param SimpleXMLElement $entry Feed item
* @param \PicoFeed\Item $item Item object
* @param \PicoFeed\Feed $feed Feed object
*/
public function findItemEnclosure(SimpleXMLElement $entry, Item $item, Feed $feed)
{
foreach ($entry->link as $link) {
if ((string) $link['rel'] === 'enclosure') {
$item->enclosure_url = (string) $link['href'];
$item->enclosure_type = (string) $link['type'];
if (Filter::isRelativePath($item->enclosure_url)) {
$item->enclosure_url = Filter::getAbsoluteUrl($item->enclosure_url, $feed->url);
}
break;
}
}
}
/**
* Parse the HTML content
*
* @access public
* @return bool
*/
public function parse()
{
if ($this->html) {
Logging::setMessage(get_called_class() . ' Fix encoding');
Logging::setMessage(get_called_class() . ': HTTP Encoding "' . $this->encoding . '"');
$this->html = Filter::stripHeadTags($this->html);
$this->html = Encoding::convert($this->html, $this->encoding);
Logging::setMessage(get_called_class() . ' Content length: ' . strlen($this->html) . ' bytes');
$rules = $this->getRules();
if (is_array($rules)) {
Logging::setMessage(get_called_class() . ' Parse content with rules');
$this->parseContentWithRules($rules);
} else {
Logging::setMessage(get_called_class() . ' Parse content with candidates');
$this->parseContentWithCandidates();
}
} else {
Logging::setMessage(get_called_class() . ' No content fetched');
}
Logging::setMessage(get_called_class() . ' Content length: ' . strlen($this->content) . ' bytes');
Logging::setMessage(get_called_class() . ' Grabber done');
return $this->content !== '';
}
/**
* Find the item enclosure
*
* @access public
* @param SimpleXMLElement $entry Feed item
* @param \PicoFeed\Item $item Item object
* @param \PicoFeed\Feed $feed Feed object
*/
public function findItemEnclosure(SimpleXMLElement $entry, Item $item, Feed $feed)
{
if (isset($entry->enclosure)) {
$item->enclosure_url = $this->getNamespaceValue($entry->enclosure, $this->namespaces, 'origEnclosureLink');
if (empty($item->enclosure_url)) {
$item->enclosure_url = isset($entry->enclosure['url']) ? (string) $entry->enclosure['url'] : '';
}
$item->enclosure_type = isset($entry->enclosure['type']) ? (string) $entry->enclosure['type'] : '';
if (Filter::isRelativePath($item->enclosure_url)) {
$item->enclosure_url = Filter::getAbsoluteUrl($item->enclosure_url, $feed->url);
}
}
}
/**
* Get the first XML tag
*
* @access public
* @param string $data Feed content
* @return string
*/
public function getFirstTag($data)
{
// Strip HTML comments (max of 5,000 characters long to prevent crashing)
$data = preg_replace('/<!--(.{0,5000}?)-->/Uis', '', $data);
/* Strip Doctype:
* Doctype needs to be within the first 100 characters. (Ideally the first!)
* If it's not found by then, we need to stop looking to prevent PREG
* from reaching max backtrack depth and crashing.
*/
$data = preg_replace('/^.{0,100}<!DOCTYPE([^>]*)>/Uis', '', $data);
// Strip <?xml version....
$data = Filter::stripXmlTag($data);
// Find the first tag
$open_tag = strpos($data, '<');
$close_tag = strpos($data, '>');
return substr($data, $open_tag, $close_tag);
}
