/**
* Creates a WebpageAnalyzer object.
*
* @param string $url The URL of the webpage
* @param \FlameCore\Webtools\HttpClient $http The HttpClient instance to use
* @throws \RuntimeException if the URL could not be loaded.
*/
public function __construct($url, HttpClient $http = null)
{
$this->url = $url;
$this->baseUrl = preg_replace('#^(https?://[^/]+).+$#', '\\1', $url);
$this->localUrl = preg_replace('#^(https?://.+)/[^/]+$#', '\\1', $url);
$http = $http ?: new HttpClient();
$html = HtmlExplorer::fromWeb($url, $http);
$node = $html->findFirstTag('base');
if ($node && ($href = $node->getAttribute('href'))) {
$this->baseUrl = trim($href, ' /');
}
$this->http = $http;
$this->html = $html;
}
/**
* Tries to fetch data for the given content URL using discovery.
*
* @param string $url The content URL
* @param array $args An array of optional extra arguments
* @return object|false Returns the request data as an object, or FALSE on failure.
*/
public function discover($url, array $args = [])
{
$url = trim($url);
$html = HtmlExplorer::fromWeb($url, $this->http);
$typePattern = '#^(' . join('|', array_keys($this->formats)) . ')\\+oembed$#';
$nodes = $html->findTags('link');
foreach ($nodes as $node) {
$rel = strtolower(trim($node->getAttribute('rel')));
$type = strtolower(trim($node->getAttribute('type')));
if (in_array($rel, ['alternate', 'alternative']) && preg_match($typePattern, $type, $matches)) {
$url = trim($node->getAttribute('href'));
$result = $this->queryProvider($url, null, null, $args);
if ($result->success) {
$parser = $this->formats[$matches[1]];
return $parser($result->data);
}
}
}
return false;
}
