Chris@17: <?php
Chris@17: 
Chris@17: namespace Drupal\media\OEmbed;
Chris@17: 
Chris@17: use Drupal\Component\Utility\Html;
Chris@17: use Drupal\Component\Utility\UrlHelper;
Chris@17: use Drupal\Core\Cache\CacheBackendInterface;
Chris@17: use Drupal\Core\Cache\UseCacheBackendTrait;
Chris@17: use Drupal\Core\Extension\ModuleHandlerInterface;
Chris@17: use GuzzleHttp\ClientInterface;
Chris@17: use GuzzleHttp\Exception\RequestException;
Chris@17: 
Chris@17: /**
Chris@17:  * Converts oEmbed media URLs into endpoint-specific resource URLs.
Chris@17:  */
Chris@17: class UrlResolver implements UrlResolverInterface {
Chris@17: 
Chris@17:   use UseCacheBackendTrait;
Chris@17: 
Chris@17:   /**
Chris@17:    * The HTTP client.
Chris@17:    *
Chris@17:    * @var \GuzzleHttp\Client
Chris@17:    */
Chris@17:   protected $httpClient;
Chris@17: 
Chris@17:   /**
Chris@17:    * The OEmbed provider repository service.
Chris@17:    *
Chris@17:    * @var \Drupal\media\OEmbed\ProviderRepositoryInterface
Chris@17:    */
Chris@17:   protected $providers;
Chris@17: 
Chris@17:   /**
Chris@17:    * The OEmbed resource fetcher service.
Chris@17:    *
Chris@17:    * @var \Drupal\media\OEmbed\ResourceFetcherInterface
Chris@17:    */
Chris@17:   protected $resourceFetcher;
Chris@17: 
Chris@17:   /**
Chris@17:    * The module handler service.
Chris@17:    *
Chris@17:    * @var \Drupal\Core\Extension\ModuleHandlerInterface
Chris@17:    */
Chris@17:   protected $moduleHandler;
Chris@17: 
Chris@17:   /**
Chris@17:    * Static cache of discovered oEmbed resource URLs, keyed by canonical URL.
Chris@17:    *
Chris@17:    * A discovered resource URL is the actual endpoint URL for a specific media
Chris@17:    * object, fetched from its canonical URL.
Chris@17:    *
Chris@17:    * @var string[]
Chris@17:    */
Chris@17:   protected $urlCache = [];
Chris@17: 
Chris@17:   /**
Chris@17:    * Constructs a UrlResolver object.
Chris@17:    *
Chris@17:    * @param \Drupal\media\OEmbed\ProviderRepositoryInterface $providers
Chris@17:    *   The oEmbed provider repository service.
Chris@17:    * @param \Drupal\media\OEmbed\ResourceFetcherInterface $resource_fetcher
Chris@17:    *   The OEmbed resource fetcher service.
Chris@17:    * @param \GuzzleHttp\ClientInterface $http_client
Chris@17:    *   The HTTP client.
Chris@17:    * @param \Drupal\Core\Extension\ModuleHandlerInterface $module_handler
Chris@17:    *   The module handler service.
Chris@17:    * @param \Drupal\Core\Cache\CacheBackendInterface $cache_backend
Chris@17:    *   (optional) The cache backend.
Chris@17:    */
Chris@17:   public function __construct(ProviderRepositoryInterface $providers, ResourceFetcherInterface $resource_fetcher, ClientInterface $http_client, ModuleHandlerInterface $module_handler, CacheBackendInterface $cache_backend = NULL) {
Chris@17:     $this->providers = $providers;
Chris@17:     $this->resourceFetcher = $resource_fetcher;
Chris@17:     $this->httpClient = $http_client;
Chris@17:     $this->moduleHandler = $module_handler;
Chris@17:     $this->cacheBackend = $cache_backend;
Chris@17:     $this->useCaches = isset($cache_backend);
Chris@17:   }
Chris@17: 
Chris@17:   /**
Chris@17:    * Runs oEmbed discovery and returns the endpoint URL if successful.
Chris@17:    *
Chris@17:    * @param string $url
Chris@17:    *   The resource's URL.
Chris@17:    *
Chris@17:    * @return string|bool
Chris@17:    *   URL of the oEmbed endpoint, or FALSE if the discovery was unsuccessful.
Chris@17:    *
Chris@17:    * @throws \Drupal\media\OEmbed\ResourceException
Chris@17:    *   If the resource cannot be retrieved.
Chris@17:    */
Chris@17:   protected function discoverResourceUrl($url) {
Chris@17:     try {
Chris@17:       $response = $this->httpClient->get($url);
Chris@17:     }
Chris@17:     catch (RequestException $e) {
Chris@17:       throw new ResourceException('Could not fetch oEmbed resource.', $url, [], $e);
Chris@17:     }
Chris@17: 
Chris@17:     $document = Html::load((string) $response->getBody());
Chris@17:     $xpath = new \DOMXpath($document);
Chris@17: 
Chris@17:     return $this->findUrl($xpath, 'json') ?: $this->findUrl($xpath, 'xml');
Chris@17:   }
Chris@17: 
Chris@17:   /**
Chris@17:    * Tries to find the oEmbed URL in a DOM.
Chris@17:    *
Chris@17:    * @param \DOMXPath $xpath
Chris@17:    *   Page HTML as DOMXPath.
Chris@17:    * @param string $format
Chris@17:    *   Format of oEmbed resource. Possible values are 'json' and 'xml'.
Chris@17:    *
Chris@17:    * @return bool|string
Chris@17:    *   A URL to an oEmbed resource or FALSE if not found.
Chris@17:    */
Chris@17:   protected function findUrl(\DOMXPath $xpath, $format) {
Chris@17:     $result = $xpath->query("//link[@type='application/$format+oembed']");
Chris@17:     return $result->length ? $result->item(0)->getAttribute('href') : FALSE;
Chris@17:   }
Chris@17: 
Chris@17:   /**
Chris@17:    * {@inheritdoc}
Chris@17:    */
Chris@17:   public function getProviderByUrl($url) {
Chris@17:     // Check the URL against every scheme of every endpoint of every provider
Chris@17:     // until we find a match.
Chris@17:     foreach ($this->providers->getAll() as $provider_name => $provider_info) {
Chris@17:       foreach ($provider_info->getEndpoints() as $endpoint) {
Chris@17:         if ($endpoint->matchUrl($url)) {
Chris@17:           return $provider_info;
Chris@17:         }
Chris@17:       }
Chris@17:     }
Chris@17: 
Chris@17:     $resource_url = $this->discoverResourceUrl($url);
Chris@17:     if ($resource_url) {
Chris@17:       return $this->resourceFetcher->fetchResource($resource_url)->getProvider();
Chris@17:     }
Chris@17: 
Chris@17:     throw new ResourceException('No matching provider found.', $url);
Chris@17:   }
Chris@17: 
Chris@17:   /**
Chris@17:    * {@inheritdoc}
Chris@17:    */
Chris@17:   public function getResourceUrl($url, $max_width = NULL, $max_height = NULL) {
Chris@17:     // Try to get the resource URL from the static cache.
Chris@17:     if (isset($this->urlCache[$url])) {
Chris@17:       return $this->urlCache[$url];
Chris@17:     }
Chris@17: 
Chris@17:     // Try to get the resource URL from the persistent cache.
Chris@17:     $cache_id = "media:oembed_resource_url:$url:$max_width:$max_height";
Chris@17: 
Chris@17:     $cached = $this->cacheGet($cache_id);
Chris@17:     if ($cached) {
Chris@17:       $this->urlCache[$url] = $cached->data;
Chris@17:       return $this->urlCache[$url];
Chris@17:     }
Chris@17: 
Chris@17:     $provider = $this->getProviderByUrl($url);
Chris@17:     $endpoints = $provider->getEndpoints();
Chris@17:     $endpoint = reset($endpoints);
Chris@17:     $resource_url = $endpoint->buildResourceUrl($url);
Chris@17: 
Chris@17:     $parsed_url = UrlHelper::parse($resource_url);
Chris@17:     if ($max_width) {
Chris@17:       $parsed_url['query']['maxwidth'] = $max_width;
Chris@17:     }
Chris@17:     if ($max_height) {
Chris@17:       $parsed_url['query']['maxheight'] = $max_height;
Chris@17:     }
Chris@17:     // Let other modules alter the resource URL, because some oEmbed providers
Chris@17:     // provide extra parameters in the query string. For example, Instagram also
Chris@17:     // supports the 'omitscript' parameter.
Chris@17:     $this->moduleHandler->alter('oembed_resource_url', $parsed_url, $provider);
Chris@17:     $resource_url = $parsed_url['path'] . '?' . UrlHelper::buildQuery($parsed_url['query']);
Chris@17: 
Chris@17:     $this->urlCache[$url] = $resource_url;
Chris@17:     $this->cacheSet($cache_id, $resource_url);
Chris@17: 
Chris@17:     return $resource_url;
Chris@17:   }
Chris@17: 
Chris@17: }