/**
* @param HtmlDocument $document HTML document to scan for title.
* @return string
*/
protected function getTitle(HtmlDocument $document)
{
$title = $document->getMetaContent('title');
if (strlen($title) > 0) {
return $title;
}
preg_match('/<title>(.+?)<\\/title>/mis', $document->getHtml(), $matches);
return isset($matches[1]) ? $matches[1] : null;
}
/**
* @param HtmlDocument $document
* @return bool
*/
protected function detectOembedLink(HtmlDocument $document)
{
preg_match_all('/<link.+?alternate.+?>/', $document->getHtml(), $linkElements);
foreach ($linkElements[0] as $linkElement) {
$typeJson = strpos($linkElement, $this::OEMBED_TYPE_JSON) !== false;
$typeXml = strpos($linkElement, $this::OEMBED_TYPE_XML) !== false;
if ($typeJson || $typeXml) {
if (preg_match('/href=[\'"](.+?)[\'"]/', $linkElement, $attributes)) {
$this->metadataType = $typeJson ? 'json' : 'xml';
$this->metadataUrl = $attributes[1];
return true;
}
}
}
return false;
}
/**
* @param HtmlDocument $document
* @return bool
*/
protected function initializeDom(HtmlDocument $document)
{
if (!class_exists('DOMDocument')) {
return false;
}
$this->dom = new \DOMDocument();
// Prevents parsing errors bubbling
libxml_use_internal_errors(true);
$result = $this->dom->loadHTML('<?xml encoding="' . $document->getEncoding() . '">' . $document->getHtml());
return $result;
}