3 if (!defined('GNUSOCIAL')) { exit(1); }
5 class OembedPlugin extends Plugin
7 // settings which can be set in config.php with addPlugin('Oembed', array('param'=>'value', ...));
8 // WARNING, these are _regexps_ (slashes added later). Always escape your dots and end your strings
9 public $domain_whitelist = array( // hostname => service provider
10 '^i\d*\.ytimg\.com$' => 'YouTube',
11 '^i\d*\.vimeocdn\.com$' => 'Vimeo',
13 public $append_whitelist = array(); // fill this array as domain_whitelist to add more trusted sources
14 public $check_whitelist = false; // security/abuse precaution
16 protected $imgData = array();
18 // these should be declared protected everywhere
19 public function initialize()
23 $this->domain_whitelist = array_merge($this->domain_whitelist, $this->append_whitelist);
26 public function onCheckSchema()
28 $schema = Schema::get();
29 $schema->ensureTable('file_oembed', File_oembed::schemaDef());
33 public function onRouterInitialized(URLMapper $m)
35 $m->connect('main/oembed', array('action' => 'oembed'));
38 public function onGetRemoteUrlMetadataFromDom($url, DOMDocument $dom, stdClass &$metadata)
41 common_log(LOG_INFO, 'Trying to discover an oEmbed endpoint using link headers.');
42 $api = oEmbedHelper::oEmbedEndpointFromHTML($dom);
43 common_log(LOG_INFO, 'Found oEmbed API endpoint ' . $api . ' for URL ' . $url);
45 'maxwidth' => common_config('thumbnail', 'width'),
46 'maxheight' => common_config('thumbnail', 'height'),
48 $metadata = oEmbedHelper::getOembedFrom($api, $url, $params);
50 // Facebook just gives us javascript in its oembed html,
51 // so use the content of the title element instead
52 if(strpos($url,'https://www.facebook.com/') === 0) {
53 $metadata->html = @$dom->getElementsByTagName('title')->item(0)->nodeValue;
56 // Wordpress sometimes also just gives us javascript, use og:description if it is available
57 $xpath = new DomXpath($dom);
58 $generatorNode = @$xpath->query('//meta[@name="generator"][1]')->item(0);
59 if ($generatorNode instanceof DomElement) {
60 // when wordpress only gives us javascript, the html stripped from tags
61 // is the same as the title, so this helps us to identify this (common) case
62 if(strpos($generatorNode->getAttribute('content'),'WordPress') === 0
63 && trim(strip_tags($metadata->html)) == trim($metadata->title)) {
64 $propertyNode = @$xpath->query('//meta[@property="og:description"][1]')->item(0);
65 if ($propertyNode instanceof DomElement) {
66 $metadata->html = $propertyNode->getAttribute('content');
70 } catch (Exception $e) {
71 common_log(LOG_INFO, 'Could not find an oEmbed endpoint using link headers, trying OpenGraph from HTML.');
73 $metadata = OpenGraphHelper::ogFromHtml($dom);
76 if (isset($metadata->thumbnail_url)) {
77 // sometimes sites serve the path, not the full URL, for images
78 // let's "be liberal in what you accept from others"!
79 // add protocol and host if the thumbnail_url starts with /
80 if(substr($metadata->thumbnail_url,0,1) == '/') {
81 $thumbnail_url_parsed = parse_url($metadata->url);
82 $metadata->thumbnail_url = $thumbnail_url_parsed['scheme']."://".$thumbnail_url_parsed['host'].$metadata->thumbnail_url;
85 // some wordpress opengraph implementations sometimes return a white blank image
86 // no need for us to save that!
87 if($metadata->thumbnail_url == 'https://s0.wp.com/i/blank.jpg') {
88 unset($metadata->thumbnail_url);
94 public function onEndShowHeadElements(Action $action)
96 switch ($action->getActionName()) {
98 $action->element('link',array('rel'=>'alternate',
99 'type'=>'application/json+oembed',
100 'href'=>common_local_url(
103 array('format'=>'json', 'url'=>
104 common_local_url('attachment',
105 array('attachment' => $action->attachment->id)))),
106 'title'=>'oEmbed'),null);
107 $action->element('link',array('rel'=>'alternate',
108 'type'=>'text/xml+oembed',
109 'href'=>common_local_url(
112 array('format'=>'xml','url'=>
113 common_local_url('attachment',
114 array('attachment' => $action->attachment->id)))),
115 'title'=>'oEmbed'),null);
118 if (!$action->notice->isLocal()) {
122 $action->element('link',array('rel'=>'alternate',
123 'type'=>'application/json+oembed',
124 'href'=>common_local_url(
127 array('format'=>'json','url'=>$action->notice->getUrl())),
128 'title'=>'oEmbed'),null);
129 $action->element('link',array('rel'=>'alternate',
130 'type'=>'text/xml+oembed',
131 'href'=>common_local_url(
134 array('format'=>'xml','url'=>$action->notice->getUrl())),
135 'title'=>'oEmbed'),null);
136 } catch (InvalidUrlException $e) {
137 // The notice is probably a share or similar, which don't
138 // have a representational URL of their own.
146 public function onEndShowStylesheets(Action $action) {
147 $action->cssLink($this->path('css/oembed.css'));
152 * Save embedding information for a File, if applicable.
154 * Normally this event is called through File::saveNew()
156 * @param File $file The newly inserted File object.
158 * @return boolean success
160 public function onEndFileSaveNew(File $file)
162 $fo = File_oembed::getKV('file_id', $file->id);
163 if ($fo instanceof File_oembed) {
164 common_log(LOG_WARNING, "Strangely, a File_oembed object exists for new file {$file->id}", __FILE__);
168 if (isset($file->mimetype)
169 && (('text/html' === substr($file->mimetype, 0, 9)
170 || 'application/xhtml+xml' === substr($file->mimetype, 0, 21)))) {
173 $oembed_data = File_oembed::_getOembed($file->url);
174 if ($oembed_data === false) {
175 throw new Exception('Did not get oEmbed data from URL');
177 } catch (Exception $e) {
181 File_oembed::saveNew($oembed_data, $file->id);
186 public function onEndShowAttachmentLink(HTMLOutputter $out, File $file)
188 $oembed = File_oembed::getKV('file_id', $file->id);
189 if (empty($oembed->author_name) && empty($oembed->provider)) {
192 $out->elementStart('div', array('id'=>'oembed_info', 'class'=>'e-content'));
193 if (!empty($oembed->author_name)) {
194 $out->elementStart('div', 'fn vcard author');
195 if (empty($oembed->author_url)) {
196 $out->text($oembed->author_name);
198 $out->element('a', array('href' => $oembed->author_url,
200 $oembed->author_name);
203 if (!empty($oembed->provider)) {
204 $out->elementStart('div', 'fn vcard');
205 if (empty($oembed->provider_url)) {
206 $out->text($oembed->provider);
208 $out->element('a', array('href' => $oembed->provider_url,
213 $out->elementEnd('div');
216 public function onFileEnclosureMetadata(File $file, &$enclosure)
218 // Never treat generic HTML links as an enclosure type!
219 // But if we have oEmbed info, we'll consider it golden.
220 $oembed = File_oembed::getKV('file_id', $file->id);
221 if (!$oembed instanceof File_oembed || !in_array($oembed->type, array('photo', 'video'))) {
225 foreach (array('mimetype', 'url', 'title', 'modified', 'width', 'height') as $key) {
226 if (isset($oembed->{$key}) && !empty($oembed->{$key})) {
227 $enclosure->{$key} = $oembed->{$key};
233 public function onStartShowAttachmentRepresentation(HTMLOutputter $out, File $file)
236 $oembed = File_oembed::getByFile($file);
237 } catch (NoResultException $e) {
241 $out->elementStart('article', ['class'=>'oembed-item']);
242 $out->elementStart('header');
244 $thumb = $file->getThumbnail(128, 128);
245 $out->element('img', $thumb->getHtmlAttrs(['class'=>'oembed-thumb']));
247 } catch (Exception $e) {
248 $out->element('div', ['class'=>'error'], $e->getMessage());
250 $out->elementStart('h5', ['class'=>'oembed-title']);
251 $out->element('a', ['href'=>$file->getUrl()], common_strip_html($oembed->title));
252 $out->elementEnd('h5');
253 $out->elementStart('div', ['class'=>'oembed-source']);
254 if (!empty($oembed->author_name)) {
255 // TRANS: text before the author name of oEmbed attachment representation
256 // FIXME: The whole "By x from y" should be i18n because of different language constructions.
257 $out->text(_('By '));
258 $attrs = ['class'=>'h-card'];
259 if (!empty($oembed->author_url)) {
260 $attrs['href'] = $oembed->author_url;
265 $out->element($tag, $attrs, $oembed->author_name);
267 if (!empty($oembed->provider)) {
268 // TRANS: text between the oEmbed author name and provider url
269 // FIXME: The whole "By x from y" should be i18n because of different language constructions.
270 $out->text(_(' from '));
271 $attrs = ['class'=>'h-card'];
272 if (!empty($oembed->provider_url)) {
273 $attrs['href'] = $oembed->provider_url;
278 $out->element($tag, $attrs, $oembed->provider);
280 $out->elementEnd('div');
281 $out->elementEnd('header');
282 $out->elementStart('div', ['class'=>'oembed-html']);
283 $out->raw(common_purify($oembed->html));
284 $out->elementEnd('div');
285 $out->elementStart('footer');
286 $out->elementEnd('footer');
287 $out->elementEnd('article');
292 public function onShowUnsupportedAttachmentRepresentation(HTMLOutputter $out, File $file)
295 $oembed = File_oembed::getByFile($file);
296 } catch (NoResultException $e) {
300 // the 'photo' type is shown through ordinary means, using StartShowAttachmentRepresentation!
301 switch ($oembed->type) {
304 if (!empty($oembed->html)
305 && (GNUsocial::isAjax() || common_config('attachments', 'show_html'))) {
306 require_once INSTALLDIR.'/extlib/HTMLPurifier/HTMLPurifier.auto.php';
307 $purifier = new HTMLPurifier();
308 // FIXME: do we allow <object> and <embed> here? we did that when we used htmLawed, but I'm not sure anymore...
309 $out->raw($purifier->purify($oembed->html));
318 public function onCreateFileImageThumbnailSource(File $file, &$imgPath, $media=null)
320 // If we are on a private node, we won't do any remote calls (just as a precaution until
321 // we can configure this from config.php for the private nodes)
322 if (common_config('site', 'private')) {
326 // All our remote Oembed images lack a local filename property in the File object
327 if (!is_null($file->filename)) {
332 // If we have proper oEmbed data, there should be an entry in the File_oembed
333 // and File_thumbnail tables respectively. If not, we're not going to do anything.
334 $file_oembed = File_oembed::getByFile($file);
335 $thumbnail = File_thumbnail::byFile($file);
336 } catch (NoResultException $e) {
337 // Not Oembed data, or at least nothing we either can or want to use.
342 $this->storeRemoteFileThumbnail($thumbnail);
343 } catch (AlreadyFulfilledException $e) {
347 $imgPath = $thumbnail->getPath();
353 * @return boolean false on no check made, provider name on success
354 * @throws ServerException if check is made but fails
356 protected function checkWhitelist($url)
358 if (!$this->check_whitelist) {
359 return false; // indicates "no check made"
362 $host = parse_url($url, PHP_URL_HOST);
363 foreach ($this->domain_whitelist as $regex => $provider) {
364 if (preg_match("/$regex/", $host)) {
365 return $provider; // we trust this source, return provider name
369 throw new ServerException(sprintf(_('Domain not in remote thumbnail source whitelist: %s'), $host));
372 protected function storeRemoteFileThumbnail(File_thumbnail $thumbnail)
374 if (!empty($thumbnail->filename) && file_exists($thumbnail->getPath())) {
375 throw new AlreadyFulfilledException(sprintf('A thumbnail seems to already exist for remote file with id==%u', $thumbnail->file_id));
378 $url = $thumbnail->getUrl();
379 $this->checkWhitelist($url);
381 // First we download the file to memory and test whether it's actually an image file
382 // FIXME: To support remote video/whatever files, this needs reworking.
383 common_debug(sprintf('Downloading remote thumbnail for file id==%u with thumbnail URL: %s', $thumbnail->file_id, $url));
384 $imgData = HTTPClient::quickGet($url);
385 $info = @getimagesizefromstring($imgData);
386 if ($info === false) {
387 throw new UnsupportedMediaException(_('Remote file format was not identified as an image.'), $url);
388 } elseif (!$info[0] || !$info[1]) {
389 throw new UnsupportedMediaException(_('Image file had impossible geometry (0 width or height)'));
392 $ext = File::guessMimeExtension($info['mime']);
394 // We'll trust sha256 (File::FILEHASH_ALG) not to have collision issues any time soon :)
395 $filename = 'oembed-'.hash(File::FILEHASH_ALG, $imgData) . ".{$ext}";
396 $fullpath = File_thumbnail::path($filename);
397 // Write the file to disk. Throw Exception on failure
398 if (!file_exists($fullpath) && file_put_contents($fullpath, $imgData) === false) {
399 throw new ServerException(_('Could not write downloaded file to disk.'));
401 // Get rid of the file from memory
404 // Updated our database for the file record
405 $orig = clone($thumbnail);
406 $thumbnail->filename = $filename;
407 $thumbnail->width = $info[0]; // array indexes documented on php.net:
408 $thumbnail->height = $info[1]; // https://php.net/manual/en/function.getimagesize.php
409 // Throws exception on failure.
410 $thumbnail->updateWithKeys($orig);
413 public function onPluginVersion(array &$versions)
415 $versions[] = array('name' => 'Oembed',
416 'version' => GNUSOCIAL_VERSION,
417 'author' => 'Mikael Nordfeldth',
418 'homepage' => 'http://gnu.io/',
420 // TRANS: Plugin description.
421 _m('Plugin for using and representing Oembed data.'));