/**
* Extracts meta tags from the given HTML source.
*
* @param string $html HTML.
* @return array Meta tags.
*/
protected function _extractMetas($html)
{
$Document = $this->_Dom->document($html);
return $Document->tags('meta', function ($Tag) {
return $Tag->matches('property', $this->_metaPattern);
});
}
/**
* Extracts embeddable URLs from an HTML source.
*
* @param string $html The HTML source to be extracted.
* @param string $url URL of the HTML source.
* @return array An array of extracted URLs.
*/
public function crawl($html, $url = '')
{
$Document = $this->_Dom->document($html);
$urls = $this->_extractUrls($Document);
if ($url) {
$urls = $this->_completeUrls($urls, $url);
}
return $this->_filterUrls(array_unique($urls));
}
/**
* Extracts an oEmbed configuration from the given page.
*
* @param string $html HTML page.
* @return array Configuration.
*/
protected function _extractConfig($html)
{
$Document = $this->_Dom->document($html);
$links = $Document->tags('link');
foreach ($links as $Link) {
if ($format = $this->_extractFormat($Link)) {
return new Config($Link->get('href'), $format);
}
}
throw new Exception('Unable to extract any OEmbed endpoint');
}
/**
* Extracts embeddable URLs from an HTML source.
*
* @param string $html The HTML source to be extracted.
* @return array An array of extracted URLs.
*/
public function crawl($html)
{
$Document = $this->_Dom->document($html);
$urls = $this->_extractUrls($Document);
return $this->_filterUrls(array_unique($urls));
}
/**
*
*/
public function testDocument()
{
$Factory = new Native();
$Document = $Factory->document('html');
$this->assertInstanceOf('Essence\\Dom\\Document', $Document);
}
