/**
* Scrape metatags from DOM
* @return array
*/
public function dom()
{
$url = $this->url->getURL();
$meta['url'] = $url;
$meta['title'] = parse_url($url, PHP_URL_HOST);
$meta['description'] = '';
$html = $this->url->getContent();
$doc = new DOMDocument();
@$doc->loadHTML($html);
// Get document title
$node = $doc->getElementsByTagName('title');
$title = $node->item(0)->nodeValue;
if ($title) {
$meta['title'] = $title;
}
// Get oEmbed content and canonical URLs
$nodes = $doc->getElementsByTagName('link');
foreach ($nodes as $node) {
$rel = $node->getAttribute('rel');
$href = $node->getAttribute('href');
switch ($rel) {
case 'icon':
$meta['icons'][] = $href;
break;
case 'canonical':
$meta['canonical'] = $href;
break;
case 'alternate':
$type = $node->getAttribute('type');
if ($type == 'application/json+oembed' || $type == 'text/json+oembed') {
$oembed_endpoint = $href;
}
break;
}
}
if ($oembed_endpoint) {
$oembed_handler = new UrlHandler($oembed_endpoint);
$json = $oembed_handler->getContent();
if ($json) {
$oembed_params = json_decode($json, true);
if ($oembed_params) {
foreach ($oembed_params as $key => $value) {
if (!$meta[$key]) {
$meta[$key] = $value;
}
if ($key == 'url') {
$meta['oembed_url'] = $value;
}
if ($key == 'thumbnail_url' && !$meta['oembed_url']) {
$meta['oembed_url'] = $value;
}
}
}
}
}
if ($title) {
$meta['title'] = $title;
}
$nodes = $doc->getElementsByTagName('meta');
if (!empty($nodes)) {
foreach ($nodes as $node) {
$name = $node->getAttribute('name');
if (!$name) {
$name = $node->getAttribute('property');
}
if (!$name) {
continue;
}
$name = strtolower($name);
$content = $node->getAttribute('content');
switch ($name) {
default:
if ($name && !$meta[$name]) {
$name = str_replace(':', '_', $name);
$meta[$name] = $content;
}
break;
case 'title':
case 'og:title':
case 'twitter:title':
if (!$meta['title']) {
$meta['title'] = $content;
}
break;
case 'description':
case 'og:description':
case 'twitter:description':
if (!$meta['description']) {
$meta['description'] = $content;
}
break;
case 'keywords':
$meta['keywords'] = $content;
break;
case 'og:site_name':
case 'twitter:site':
if (!$meta['provider_name']) {
$meta['provider_name'] = $content;
}
break;
case 'og:type':
$meta['og_type'] = $content;
break;
case 'og:image':
case 'twitter:image':
$meta['thumbnails'][] = $content;
break;
}
}
}
if (count($meta['thumbnails'])) {
// Display a thumbnail parsed from <meta> tags
$meta['thumbnail_url'] = $meta['thumbnails'][0];
} else {
if (count($meta['icons'])) {
// Display an icon parsed from <link> tags
$meta['thumbnail_url'] = $meta['icons'][0];
} else {
$max_width = 0;
$nodes = $doc->getElementsByTagName('img');
foreach ($nodes as $node) {
$src = $node->getAttribute('src');
$width = $node->getAttribute('width');
if (!$meta['thumbnail_url'] || $width > $max_width) {
$meta['thumbnail_url'] = $src;
$max_width = $width;
}
}
}
}
/**
* @todo: figure out what to do with relative URLs in DOM
*/
if (!Validator::isImage($meta['thumbnail_url'])) {
unset($meta['thumbnail_url']);
}
return $meta;
}
/**
* @covers ::setURL
* @covers ::getURL
*/
public function testSetURL()
{
$url = 'http://example.com';
$this->object->setURL($url);
$this->assertEquals($this->object->getURL(), $url);
}
/**
* Prepare view/hook params
*
* @param array $params Params to pass by default
* @return array
*/
public function prepareParams(array $params = array())
{
$embed_params = array_filter(array('src' => $this->url->getURL(), 'entity' => $this->getEntity(), 'meta' => $this->getMeta()));
$params = array_merge($params, $embed_params);
return $params;
}
