yt_dlp/extractor/facebook.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import json
   5 import re
   6
   7 from .common import InfoExtractor
   8 from ..compat import (
   9     compat_etree_fromstring,
  10     compat_str,
  11     compat_urllib_parse_unquote,
  12     compat_urllib_parse_unquote_plus,
  13 )
  14 from ..utils import (
  15     clean_html,
  16     error_to_compat_str,
  17     ExtractorError,
  18     float_or_none,
  19     get_element_by_id,
  20     int_or_none,
  21     js_to_json,
  22     limit_length,
  23     merge_dicts,
  24     network_exceptions,
  25     parse_count,
  26     parse_qs,
  27     qualities,
  28     sanitized_Request,
  29     try_get,
  30     url_or_none,
  31     urlencode_postdata,
  32     urljoin,
  33 )
  34
  35
  36 class FacebookIE(InfoExtractor):
  37     _VALID_URL = r'''(?x)
  38                 (?:
  39                     https?://
  40                         (?:[\w-]+\.)?(?:facebook\.com|facebookwkhpilnemxj7asaniu7vnjjbiltxjqhye3mhbshg7kx5tfyd\.onion)/
  41                         (?:[^#]*?\#!/)?
  42                         (?:
  43                             (?:
  44                                 video/video\.php|
  45                                 photo\.php|
  46                                 video\.php|
  47                                 video/embed|
  48                                 story\.php|
  49                                 watch(?:/live)?/?
  50                             )\?(?:.*?)(?:v|video_id|story_fbid)=|
  51                             [^/]+/videos/(?:[^/]+/)?|
  52                             [^/]+/posts/|
  53                             groups/[^/]+/permalink/|
  54                             watchparty/
  55                         )|
  56                     facebook:
  57                 )
  58                 (?P<id>[0-9]+)
  59                 '''
  60     _LOGIN_URL = 'https://www.facebook.com/login.php?next=http%3A%2F%2Ffacebook.com%2Fhome.php&login_attempt=1'
  61     _CHECKPOINT_URL = 'https://www.facebook.com/checkpoint/?next=http%3A%2F%2Ffacebook.com%2Fhome.php&_fb_noscript=1'
  62     _NETRC_MACHINE = 'facebook'
  63     IE_NAME = 'facebook'
  64
  65     _VIDEO_PAGE_TEMPLATE = 'https://www.facebook.com/video/video.php?v=%s'
  66     _VIDEO_PAGE_TAHOE_TEMPLATE = 'https://www.facebook.com/video/tahoe/async/%s/?chain=true&isvideo=true&payloadtype=primary'
  67
  68     _TESTS = [{
  69         'url': 'https://www.facebook.com/video.php?v=637842556329505&fref=nf',
  70         'md5': '6a40d33c0eccbb1af76cf0485a052659',
  71         'info_dict': {
  72             'id': '637842556329505',
  73             'ext': 'mp4',
  74             'title': 're:Did you know Kei Nishikori is the first Asian man to ever reach a Grand Slam',
  75             'uploader': 'Tennis on Facebook',
  76             'upload_date': '20140908',
  77             'timestamp': 1410199200,
  78         },
  79         'skip': 'Requires logging in',
  80     }, {
  81         # data.video
  82         'url': 'https://www.facebook.com/video.php?v=274175099429670',
  83         'info_dict': {
  84             'id': '274175099429670',
  85             'ext': 'mp4',
  86             'title': 'Asif Nawab Butt',
  87             'description': 'Asif Nawab Butt',
  88             'uploader': 'Asif Nawab Butt',
  89             'upload_date': '20140506',
  90             'timestamp': 1399398998,
  91             'thumbnail': r're:^https?://.*',
  92         },
  93         'expected_warnings': [
  94             'title'
  95         ]
  96     }, {
  97         'note': 'Video with DASH manifest',
  98         'url': 'https://www.facebook.com/video.php?v=957955867617029',
  99         'md5': 'b2c28d528273b323abe5c6ab59f0f030',
 100         'info_dict': {
 101             'id': '957955867617029',
 102             'ext': 'mp4',
 103             'title': 'When you post epic content on instagram.com/433 8 million followers, this is ...',
 104             'uploader': 'Demy de Zeeuw',
 105             'upload_date': '20160110',
 106             'timestamp': 1452431627,
 107         },
 108         'skip': 'Requires logging in',
 109     }, {
 110         'url': 'https://www.facebook.com/maxlayn/posts/10153807558977570',
 111         'md5': '037b1fa7f3c2d02b7a0d7bc16031ecc6',
 112         'info_dict': {
 113             'id': '544765982287235',
 114             'ext': 'mp4',
 115             'title': '"What are you doing running in the snow?"',
 116             'uploader': 'FailArmy',
 117         },
 118         'skip': 'Video gone',
 119     }, {
 120         'url': 'https://m.facebook.com/story.php?story_fbid=1035862816472149&id=116132035111903',
 121         'md5': '1deb90b6ac27f7efcf6d747c8a27f5e3',
 122         'info_dict': {
 123             'id': '1035862816472149',
 124             'ext': 'mp4',
 125             'title': 'What the Flock Is Going On In New Zealand  Credit: ViralHog',
 126             'uploader': 'S. Saint',
 127         },
 128         'skip': 'Video gone',
 129     }, {
 130         'note': 'swf params escaped',
 131         'url': 'https://www.facebook.com/barackobama/posts/10153664894881749',
 132         'md5': '97ba073838964d12c70566e0085c2b91',
 133         'info_dict': {
 134             'id': '10153664894881749',
 135             'ext': 'mp4',
 136             'title': 'Average time to confirm recent Supreme Court nominees: 67 days Longest it\'s t...',
 137             'thumbnail': r're:^https?://.*',
 138             'timestamp': 1456259628,
 139             'upload_date': '20160223',
 140             'uploader': 'Barack Obama',
 141         },
 142         'skip': 'Gif on giphy.com gone',
 143     }, {
 144         # have 1080P, but only up to 720p in swf params
 145         # data.video.story.attachments[].media
 146         'url': 'https://www.facebook.com/cnn/videos/10155529876156509/',
 147         'md5': '3f3798adb2b73423263e59376f1f5eb7',
 148         'info_dict': {
 149             'id': '10155529876156509',
 150             'ext': 'mp4',
 151             'title': 'Holocaust survivor becomes US citizen',
 152             'description': 'She survived the holocaust — and years later, she’s getting her citizenship so she can vote for Hillary Clinton http://cnn.it/2eERh5f',
 153             'timestamp': 1477818095,
 154             'upload_date': '20161030',
 155             'uploader': 'CNN',
 156             'thumbnail': r're:^https?://.*',
 157             'view_count': int,
 158         },
 159     }, {
 160         # bigPipe.onPageletArrive ... onPageletArrive pagelet_group_mall
 161         # data.node.comet_sections.content.story.attachments[].style_type_renderer.attachment.media
 162         'url': 'https://www.facebook.com/yaroslav.korpan/videos/1417995061575415/',
 163         'info_dict': {
 164             'id': '1417995061575415',
 165             'ext': 'mp4',
 166             'title': 'Yaroslav Korpan - Довгоочікуване відео',
 167             'description': 'Довгоочікуване відео',
 168             'timestamp': 1486648771,
 169             'upload_date': '20170209',
 170             'uploader': 'Yaroslav Korpan',
 171             'uploader_id': '100000948048708',
 172         },
 173         'params': {
 174             'skip_download': True,
 175         },
 176     }, {
 177         # FIXME
 178         'url': 'https://www.facebook.com/LaGuiaDelVaron/posts/1072691702860471',
 179         'info_dict': {
 180             'id': '1072691702860471',
 181             'ext': 'mp4',
 182             'title': 'md5:ae2d22a93fbb12dad20dc393a869739d',
 183             'timestamp': 1477305000,
 184             'upload_date': '20161024',
 185             'uploader': 'La Guía Del Varón',
 186             'thumbnail': r're:^https?://.*',
 187         },
 188         'params': {
 189             'skip_download': True,
 190         },
 191     }, {
 192         # data.node.comet_sections.content.story.attachments[].style_type_renderer.attachment.media
 193         'url': 'https://www.facebook.com/groups/1024490957622648/permalink/1396382447100162/',
 194         'info_dict': {
 195             'id': '202882990186699',
 196             'ext': 'mp4',
 197             'title': 'Elisabeth Ahtn - Hello? Yes your uber ride is here\n* Jukin...',
 198             'description': 'Hello? Yes your uber ride is here\n* Jukin Media Verified *\nFind this video and others like it by visiting...',
 199             'timestamp': 1486035513,
 200             'upload_date': '20170202',
 201             'uploader': 'Elisabeth Ahtn',
 202             'uploader_id': '100013949973717',
 203         },
 204         'params': {
 205             'skip_download': True,
 206         },
 207     }, {
 208         'url': 'https://www.facebook.com/video.php?v=10204634152394104',
 209         'only_matching': True,
 210     }, {
 211         'url': 'https://www.facebook.com/amogood/videos/1618742068337349/?fref=nf',
 212         'only_matching': True,
 213     }, {
 214         # data.mediaset.currMedia.edges
 215         'url': 'https://www.facebook.com/ChristyClarkForBC/videos/vb.22819070941/10153870694020942/?type=2&theater',
 216         'only_matching': True,
 217     }, {
 218         # data.video.story.attachments[].media
 219         'url': 'facebook:544765982287235',
 220         'only_matching': True,
 221     }, {
 222         # data.node.comet_sections.content.story.attachments[].style_type_renderer.attachment.media
 223         'url': 'https://www.facebook.com/groups/164828000315060/permalink/764967300301124/',
 224         'only_matching': True,
 225     }, {
 226         # data.video.creation_story.attachments[].media
 227         'url': 'https://zh-hk.facebook.com/peoplespower/videos/1135894589806027/',
 228         'only_matching': True,
 229     }, {
 230         # data.video
 231         'url': 'https://www.facebookwkhpilnemxj7asaniu7vnjjbiltxjqhye3mhbshg7kx5tfyd.onion/video.php?v=274175099429670',
 232         'only_matching': True,
 233     }, {
 234         # no title
 235         'url': 'https://www.facebook.com/onlycleverentertainment/videos/1947995502095005/',
 236         'only_matching': True,
 237     }, {
 238         # data.video
 239         'url': 'https://www.facebook.com/WatchESLOne/videos/359649331226507/',
 240         'info_dict': {
 241             'id': '359649331226507',
 242             'ext': 'mp4',
 243             'title': 'Fnatic vs. EG - Group A - Opening Match - ESL One Birmingham Day 1',
 244             'description': '#ESLOne VoD - Birmingham Finals Day#1 Fnatic vs. @Evil Geniuses',
 245             'timestamp': 1527084179,
 246             'upload_date': '20180523',
 247             'uploader': 'ESL One Dota 2',
 248             'uploader_id': '234218833769558',
 249         },
 250         'params': {
 251             'skip_download': True,
 252         },
 253     }, {
 254         # data.node.comet_sections.content.story.attachments[].style_type_renderer.attachment.all_subattachments.nodes[].media
 255         'url': 'https://www.facebook.com/100033620354545/videos/106560053808006/',
 256         'info_dict': {
 257             'id': '106560053808006',
 258         },
 259         'playlist_count': 2,
 260     }, {
 261         # data.video.story.attachments[].media
 262         'url': 'https://www.facebook.com/watch/?v=647537299265662',
 263         'only_matching': True,
 264     }, {
 265         # FIXME: https://github.com/yt-dlp/yt-dlp/issues/542
 266         # data.node.comet_sections.content.story.attachments[].style_type_renderer.attachment.all_subattachments.nodes[].media
 267         'url': 'https://www.facebook.com/PankajShahLondon/posts/10157667649866271',
 268         'info_dict': {
 269             'id': '10157667649866271',
 270         },
 271         'playlist_count': 3,
 272     }, {
 273         # data.nodes[].comet_sections.content.story.attachments[].style_type_renderer.attachment.media
 274         'url': 'https://m.facebook.com/Alliance.Police.Department/posts/4048563708499330',
 275         'info_dict': {
 276             'id': '117576630041613',
 277             'ext': 'mp4',
 278             # TODO: title can be extracted from video page
 279             'title': 'Facebook video #117576630041613',
 280             'uploader_id': '189393014416438',
 281             'upload_date': '20201123',
 282             'timestamp': 1606162592,
 283         },
 284         'skip': 'Requires logging in',
 285     }, {
 286         # node.comet_sections.content.story.attached_story.attachments.style_type_renderer.attachment.media
 287         'url': 'https://www.facebook.com/groups/ateistiskselskab/permalink/10154930137678856/',
 288         'info_dict': {
 289             'id': '211567722618337',
 290             'ext': 'mp4',
 291             'title': 'Facebook video #211567722618337',
 292             'uploader_id': '127875227654254',
 293             'upload_date': '20161122',
 294             'timestamp': 1479793574,
 295         },
 296         'skip': 'No video',
 297     }, {
 298         # data.video.creation_story.attachments[].media
 299         'url': 'https://www.facebook.com/watch/live/?v=1823658634322275',
 300         'only_matching': True,
 301     }, {
 302         'url': 'https://www.facebook.com/watchparty/211641140192478',
 303         'info_dict': {
 304             'id': '211641140192478',
 305         },
 306         'playlist_count': 1,
 307         'skip': 'Requires logging in',
 308     }]
 309     _SUPPORTED_PAGLETS_REGEX = r'(?:pagelet_group_mall|permalink_video_pagelet|hyperfeed_story_id_[0-9a-f]+)'
 310     _api_config = {
 311         'graphURI': '/api/graphql/'
 312     }
 313
 314     @staticmethod
 315     def _extract_urls(webpage):
 316         urls = []
 317         for mobj in re.finditer(
 318                 r'<iframe[^>]+?src=(["\'])(?P<url>https?://www\.facebook\.com/(?:video/embed|plugins/video\.php).+?)\1',
 319                 webpage):
 320             urls.append(mobj.group('url'))
 321         # Facebook API embed
 322         # see https://developers.facebook.com/docs/plugins/embedded-video-player
 323         for mobj in re.finditer(r'''(?x)<div[^>]+
 324                 class=(?P<q1>[\'"])[^\'"]*\bfb-(?:video|post)\b[^\'"]*(?P=q1)[^>]+
 325                 data-href=(?P<q2>[\'"])(?P<url>(?:https?:)?//(?:www\.)?facebook.com/.+?)(?P=q2)''', webpage):
 326             urls.append(mobj.group('url'))
 327         return urls
 328
 329     def _login(self):
 330         useremail, password = self._get_login_info()
 331         if useremail is None:
 332             return
 333
 334         login_page_req = sanitized_Request(self._LOGIN_URL)
 335         self._set_cookie('facebook.com', 'locale', 'en_US')
 336         login_page = self._download_webpage(login_page_req, None,
 337                                             note='Downloading login page',
 338                                             errnote='Unable to download login page')
 339         lsd = self._search_regex(
 340             r'<input type="hidden" name="lsd" value="([^"]*)"',
 341             login_page, 'lsd')
 342         lgnrnd = self._search_regex(r'name="lgnrnd" value="([^"]*?)"', login_page, 'lgnrnd')
 343
 344         login_form = {
 345             'email': useremail,
 346             'pass': password,
 347             'lsd': lsd,
 348             'lgnrnd': lgnrnd,
 349             'next': 'http://facebook.com/home.php',
 350             'default_persistent': '0',
 351             'legacy_return': '1',
 352             'timezone': '-60',
 353             'trynum': '1',
 354         }
 355         request = sanitized_Request(self._LOGIN_URL, urlencode_postdata(login_form))
 356         request.add_header('Content-Type', 'application/x-www-form-urlencoded')
 357         try:
 358             login_results = self._download_webpage(request, None,
 359                                                    note='Logging in', errnote='unable to fetch login page')
 360             if re.search(r'<form(.*)name="login"(.*)</form>', login_results) is not None:
 361                 error = self._html_search_regex(
 362                     r'(?s)<div[^>]+class=(["\']).*?login_error_box.*?\1[^>]*><div[^>]*>.*?</div><div[^>]*>(?P<error>.+?)</div>',
 363                     login_results, 'login error', default=None, group='error')
 364                 if error:
 365                     raise ExtractorError('Unable to login: %s' % error, expected=True)
 366                 self.report_warning('unable to log in: bad username/password, or exceeded login rate limit (~3/min). Check credentials or wait.')
 367                 return
 368
 369             fb_dtsg = self._search_regex(
 370                 r'name="fb_dtsg" value="(.+?)"', login_results, 'fb_dtsg', default=None)
 371             h = self._search_regex(
 372                 r'name="h"\s+(?:\w+="[^"]+"\s+)*?value="([^"]+)"', login_results, 'h', default=None)
 373
 374             if not fb_dtsg or not h:
 375                 return
 376
 377             check_form = {
 378                 'fb_dtsg': fb_dtsg,
 379                 'h': h,
 380                 'name_action_selected': 'dont_save',
 381             }
 382             check_req = sanitized_Request(self._CHECKPOINT_URL, urlencode_postdata(check_form))
 383             check_req.add_header('Content-Type', 'application/x-www-form-urlencoded')
 384             check_response = self._download_webpage(check_req, None,
 385                                                     note='Confirming login')
 386             if re.search(r'id="checkpointSubmitButton"', check_response) is not None:
 387                 self.report_warning('Unable to confirm login, you have to login in your browser and authorize the login.')
 388         except network_exceptions as err:
 389             self.report_warning('unable to log in: %s' % error_to_compat_str(err))
 390             return
 391
 392     def _real_initialize(self):
 393         self._login()
 394
 395     def _extract_from_url(self, url, video_id):
 396         webpage = self._download_webpage(
 397             url.replace('://m.facebook.com/', '://www.facebook.com/'), video_id)
 398
 399         def extract_metadata(webpage):
 400             video_title = self._html_search_regex(
 401                 r'<h2\s+[^>]*class="uiHeaderTitle"[^>]*>([^<]*)</h2>', webpage,
 402                 'title', default=None)
 403             if not video_title:
 404                 video_title = self._html_search_regex(
 405                     r'(?s)<span class="fbPhotosPhotoCaption".*?id="fbPhotoPageCaption"><span class="hasCaption">(.*?)</span>',
 406                     webpage, 'alternative title', default=None)
 407             if not video_title:
 408                 video_title = self._html_search_meta(
 409                     ['og:title', 'twitter:title', 'description'],
 410                     webpage, 'title', default=None)
 411             if video_title:
 412                 video_title = limit_length(video_title, 80)
 413             else:
 414                 video_title = 'Facebook video #%s' % video_id
 415             description = self._html_search_meta(
 416                 ['description', 'og:description', 'twitter:description'],
 417                 webpage, 'description', default=None)
 418             uploader = clean_html(get_element_by_id(
 419                 'fbPhotoPageAuthorName', webpage)) or self._search_regex(
 420                 r'ownerName\s*:\s*"([^"]+)"', webpage, 'uploader',
 421                 default=None) or self._og_search_title(webpage, fatal=False)
 422             timestamp = int_or_none(self._search_regex(
 423                 r'<abbr[^>]+data-utime=["\'](\d+)', webpage,
 424                 'timestamp', default=None))
 425             thumbnail = self._html_search_meta(
 426                 ['og:image', 'twitter:image'], webpage, 'thumbnail', default=None)
 427             # some webpages contain unretrievable thumbnail urls
 428             # like https://lookaside.fbsbx.com/lookaside/crawler/media/?media_id=10155168902769113&get_thumbnail=1
 429             # in https://www.facebook.com/yaroslav.korpan/videos/1417995061575415/
 430             if thumbnail and not re.search(r'\.(?:jpg|png)', thumbnail):
 431                 thumbnail = None
 432             view_count = parse_count(self._search_regex(
 433                 r'\bviewCount\s*:\s*["\']([\d,.]+)', webpage, 'view count',
 434                 default=None))
 435             info_dict = {
 436                 'title': video_title,
 437                 'description': description,
 438                 'uploader': uploader,
 439                 'timestamp': timestamp,
 440                 'thumbnail': thumbnail,
 441                 'view_count': view_count,
 442             }
 443             info_json_ld = self._search_json_ld(webpage, video_id, default={})
 444             if info_json_ld.get('title'):
 445                 info_json_ld['title'] = limit_length(
 446                     re.sub(r'\s*\|\s*Facebook$', '', info_json_ld['title']), 80)
 447             return merge_dicts(info_json_ld, info_dict)
 448
 449         video_data = None
 450
 451         def extract_video_data(instances):
 452             video_data = []
 453             for item in instances:
 454                 if try_get(item, lambda x: x[1][0]) == 'VideoConfig':
 455                     video_item = item[2][0]
 456                     if video_item.get('video_id'):
 457                         video_data.append(video_item['videoData'])
 458             return video_data
 459
 460         server_js_data = self._parse_json(self._search_regex(
 461             [r'handleServerJS\(({.+})(?:\);|,")', r'\bs\.handle\(({.+?})\);'],
 462             webpage, 'server js data', default='{}'), video_id, fatal=False)
 463
 464         if server_js_data:
 465             video_data = extract_video_data(server_js_data.get('instances', []))
 466
 467         def extract_from_jsmods_instances(js_data):
 468             if js_data:
 469                 return extract_video_data(try_get(
 470                     js_data, lambda x: x['jsmods']['instances'], list) or [])
 471
 472         def extract_dash_manifest(video, formats):
 473             dash_manifest = video.get('dash_manifest')
 474             if dash_manifest:
 475                 formats.extend(self._parse_mpd_formats(
 476                     compat_etree_fromstring(compat_urllib_parse_unquote_plus(dash_manifest))))
 477
 478         def process_formats(formats):
 479             # Downloads with browser's User-Agent are rate limited. Working around
 480             # with non-browser User-Agent.
 481             for f in formats:
 482                 f.setdefault('http_headers', {})['User-Agent'] = 'facebookexternalhit/1.1'
 483
 484             self._sort_formats(formats, ('res', 'quality'))
 485
 486         def extract_relay_data(_filter):
 487             return self._parse_json(self._search_regex(
 488                 r'handleWithCustomApplyEach\([^,]+,\s*({.*?%s.*?})\);' % _filter,
 489                 webpage, 'replay data', default='{}'), video_id, fatal=False) or {}
 490
 491         def extract_relay_prefetched_data(_filter):
 492             replay_data = extract_relay_data(_filter)
 493             for require in (replay_data.get('require') or []):
 494                 if require[0] == 'RelayPrefetchedStreamCache':
 495                     return try_get(require, lambda x: x[3][1]['__bbox']['result']['data'], dict) or {}
 496
 497         if not video_data:
 498             server_js_data = self._parse_json(self._search_regex([
 499                 r'bigPipe\.onPageletArrive\(({.+?})\)\s*;\s*}\s*\)\s*,\s*["\']onPageletArrive\s+' + self._SUPPORTED_PAGLETS_REGEX,
 500                 r'bigPipe\.onPageletArrive\(({.*?id\s*:\s*"%s".*?})\);' % self._SUPPORTED_PAGLETS_REGEX
 501             ], webpage, 'js data', default='{}'), video_id, js_to_json, False)
 502             video_data = extract_from_jsmods_instances(server_js_data)
 503
 504         if not video_data:
 505             data = extract_relay_prefetched_data(
 506                 r'"(?:dash_manifest|playable_url(?:_quality_hd)?)"\s*:\s*"[^"]+"')
 507             if data:
 508                 entries = []
 509
 510                 def parse_graphql_video(video):
 511                     formats = []
 512                     q = qualities(['sd', 'hd'])
 513                     for (suffix, format_id) in [('', 'sd'), ('_quality_hd', 'hd')]:
 514                         playable_url = video.get('playable_url' + suffix)
 515                         if not playable_url:
 516                             continue
 517                         formats.append({
 518                             'format_id': format_id,
 519                             'quality': q(format_id),
 520                             'url': playable_url,
 521                         })
 522                     extract_dash_manifest(video, formats)
 523                     process_formats(formats)
 524                     v_id = video.get('videoId') or video.get('id') or video_id
 525                     info = {
 526                         'id': v_id,
 527                         'formats': formats,
 528                         'thumbnail': try_get(video, lambda x: x['thumbnailImage']['uri']),
 529                         'uploader_id': try_get(video, lambda x: x['owner']['id']),
 530                         'timestamp': int_or_none(video.get('publish_time')),
 531                         'duration': float_or_none(video.get('playable_duration_in_ms'), 1000),
 532                     }
 533                     description = try_get(video, lambda x: x['savable_description']['text'])
 534                     title = video.get('name')
 535                     if title:
 536                         info.update({
 537                             'title': title,
 538                             'description': description,
 539                         })
 540                     else:
 541                         info['title'] = description or 'Facebook video #%s' % v_id
 542                     entries.append(info)
 543
 544                 def parse_attachment(attachment, key='media'):
 545                     media = attachment.get(key) or {}
 546                     if media.get('__typename') == 'Video':
 547                         return parse_graphql_video(media)
 548
 549                 nodes = data.get('nodes') or []
 550                 node = data.get('node') or {}
 551                 if not nodes and node:
 552                     nodes.append(node)
 553                 for node in nodes:
 554                     story = try_get(node, lambda x: x['comet_sections']['content']['story'], dict) or {}
 555                     attachments = try_get(story, [
 556                         lambda x: x['attached_story']['attachments'],
 557                         lambda x: x['attachments']
 558                     ], list) or []
 559                     for attachment in attachments:
 560                         attachment = try_get(attachment, lambda x: x['style_type_renderer']['attachment'], dict)
 561                         ns = try_get(attachment, lambda x: x['all_subattachments']['nodes'], list) or []
 562                         for n in ns:
 563                             parse_attachment(n)
 564                         parse_attachment(attachment)
 565
 566                 edges = try_get(data, lambda x: x['mediaset']['currMedia']['edges'], list) or []
 567                 for edge in edges:
 568                     parse_attachment(edge, key='node')
 569
 570                 video = data.get('video') or {}
 571                 if video:
 572                     attachments = try_get(video, [
 573                         lambda x: x['story']['attachments'],
 574                         lambda x: x['creation_story']['attachments']
 575                     ], list) or []
 576                     for attachment in attachments:
 577                         parse_attachment(attachment)
 578                     if not entries:
 579                         parse_graphql_video(video)
 580
 581                 if len(entries) > 1:
 582                     return self.playlist_result(entries, video_id)
 583
 584                 video_info = entries[0]
 585                 webpage_info = extract_metadata(webpage)
 586                 # honor precise duration in video info
 587                 if video_info.get('duration'):
 588                     webpage_info['duration'] = video_info['duration']
 589                 return merge_dicts(webpage_info, video_info)
 590
 591         if not video_data:
 592             m_msg = re.search(r'class="[^"]*uiInterstitialContent[^"]*"><div>(.*?)</div>', webpage)
 593             if m_msg is not None:
 594                 raise ExtractorError(
 595                     'The video is not available, Facebook said: "%s"' % m_msg.group(1),
 596                     expected=True)
 597             elif any(p in webpage for p in (
 598                     '>You must log in to continue',
 599                     'id="login_form"',
 600                     'id="loginbutton"')):
 601                 self.raise_login_required()
 602
 603         if not video_data and '/watchparty/' in url:
 604             post_data = {
 605                 'doc_id': 3731964053542869,
 606                 'variables': json.dumps({
 607                     'livingRoomID': video_id,
 608                 }),
 609             }
 610
 611             prefetched_data = extract_relay_prefetched_data(r'"login_data"\s*:\s*{')
 612             if prefetched_data:
 613                 lsd = try_get(prefetched_data, lambda x: x['login_data']['lsd'], dict)
 614                 if lsd:
 615                     post_data[lsd['name']] = lsd['value']
 616
 617             relay_data = extract_relay_data(r'\[\s*"RelayAPIConfigDefaults"\s*,')
 618             for define in (relay_data.get('define') or []):
 619                 if define[0] == 'RelayAPIConfigDefaults':
 620                     self._api_config = define[2]
 621
 622             living_room = self._download_json(
 623                 urljoin(url, self._api_config['graphURI']), video_id,
 624                 data=urlencode_postdata(post_data))['data']['living_room']
 625
 626             entries = []
 627             for edge in (try_get(living_room, lambda x: x['recap']['watched_content']['edges']) or []):
 628                 video = try_get(edge, lambda x: x['node']['video']) or {}
 629                 v_id = video.get('id')
 630                 if not v_id:
 631                     continue
 632                 v_id = compat_str(v_id)
 633                 entries.append(self.url_result(
 634                     self._VIDEO_PAGE_TEMPLATE % v_id,
 635                     self.ie_key(), v_id, video.get('name')))
 636
 637             return self.playlist_result(entries, video_id)
 638
 639         if not video_data:
 640             # Video info not in first request, do a secondary request using
 641             # tahoe player specific URL
 642             tahoe_data = self._download_webpage(
 643                 self._VIDEO_PAGE_TAHOE_TEMPLATE % video_id, video_id,
 644                 data=urlencode_postdata({
 645                     '__a': 1,
 646                     '__pc': self._search_regex(
 647                         r'pkg_cohort["\']\s*:\s*["\'](.+?)["\']', webpage,
 648                         'pkg cohort', default='PHASED:DEFAULT'),
 649                     '__rev': self._search_regex(
 650                         r'client_revision["\']\s*:\s*(\d+),', webpage,
 651                         'client revision', default='3944515'),
 652                     'fb_dtsg': self._search_regex(
 653                         r'"DTSGInitialData"\s*,\s*\[\]\s*,\s*{\s*"token"\s*:\s*"([^"]+)"',
 654                         webpage, 'dtsg token', default=''),
 655                 }),
 656                 headers={
 657                     'Content-Type': 'application/x-www-form-urlencoded',
 658                 })
 659             tahoe_js_data = self._parse_json(
 660                 self._search_regex(
 661                     r'for\s+\(\s*;\s*;\s*\)\s*;(.+)', tahoe_data,
 662                     'tahoe js data', default='{}'),
 663                 video_id, fatal=False)
 664             video_data = extract_from_jsmods_instances(tahoe_js_data)
 665
 666         if not video_data:
 667             raise ExtractorError('Cannot parse data')
 668
 669         if len(video_data) > 1:
 670             entries = []
 671             for v in video_data:
 672                 video_url = v[0].get('video_url')
 673                 if not video_url:
 674                     continue
 675                 entries.append(self.url_result(urljoin(
 676                     url, video_url), self.ie_key(), v[0].get('video_id')))
 677             return self.playlist_result(entries, video_id)
 678         video_data = video_data[0]
 679
 680         formats = []
 681         subtitles = {}
 682         for f in video_data:
 683             format_id = f['stream_type']
 684             if f and isinstance(f, dict):
 685                 f = [f]
 686             if not f or not isinstance(f, list):
 687                 continue
 688             for quality in ('sd', 'hd'):
 689                 for src_type in ('src', 'src_no_ratelimit'):
 690                     src = f[0].get('%s_%s' % (quality, src_type))
 691                     if src:
 692                         preference = -10 if format_id == 'progressive' else -1
 693                         if quality == 'hd':
 694                             preference += 5
 695                         formats.append({
 696                             'format_id': '%s_%s_%s' % (format_id, quality, src_type),
 697                             'url': src,
 698                             'quality': preference,
 699                             'height': 720 if quality == 'hd' else None
 700                         })
 701             extract_dash_manifest(f[0], formats)
 702             subtitles_src = f[0].get('subtitles_src')
 703             if subtitles_src:
 704                 subtitles.setdefault('en', []).append({'url': subtitles_src})
 705
 706         process_formats(formats)
 707
 708         info_dict = {
 709             'id': video_id,
 710             'formats': formats,
 711             'subtitles': subtitles,
 712         }
 713         info_dict.update(extract_metadata(webpage))
 714
 715         return info_dict
 716
 717     def _real_extract(self, url):
 718         video_id = self._match_id(url)
 719
 720         real_url = self._VIDEO_PAGE_TEMPLATE % video_id if url.startswith('facebook:') else url
 721         return self._extract_from_url(real_url, video_id)
 722
 723
 724 class FacebookPluginsVideoIE(InfoExtractor):
 725     _VALID_URL = r'https?://(?:[\w-]+\.)?facebook\.com/plugins/video\.php\?.*?\bhref=(?P<id>https.+)'
 726
 727     _TESTS = [{
 728         'url': 'https://www.facebook.com/plugins/video.php?href=https%3A%2F%2Fwww.facebook.com%2Fgov.sg%2Fvideos%2F10154383743583686%2F&show_text=0&width=560',
 729         'md5': '5954e92cdfe51fe5782ae9bda7058a07',
 730         'info_dict': {
 731             'id': '10154383743583686',
 732             'ext': 'mp4',
 733             'title': 'What to do during the haze?',
 734             'uploader': 'Gov.sg',
 735             'upload_date': '20160826',
 736             'timestamp': 1472184808,
 737         },
 738         'add_ie': [FacebookIE.ie_key()],
 739     }, {
 740         'url': 'https://www.facebook.com/plugins/video.php?href=https%3A%2F%2Fwww.facebook.com%2Fvideo.php%3Fv%3D10204634152394104',
 741         'only_matching': True,
 742     }, {
 743         'url': 'https://www.facebook.com/plugins/video.php?href=https://www.facebook.com/gov.sg/videos/10154383743583686/&show_text=0&width=560',
 744         'only_matching': True,
 745     }]
 746
 747     def _real_extract(self, url):
 748         return self.url_result(
 749             compat_urllib_parse_unquote(self._match_id(url)),
 750             FacebookIE.ie_key())
 751
 752
 753 class FacebookRedirectURLIE(InfoExtractor):
 754     IE_DESC = False  # Do not list
 755     _VALID_URL = r'https?://(?:[\w-]+\.)?facebook\.com/flx/warn[/?]'
 756     _TESTS = [{
 757         'url': 'https://www.facebook.com/flx/warn/?h=TAQHsoToz&u=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DpO8h3EaFRdo&s=1',
 758         'info_dict': {
 759             'id': 'pO8h3EaFRdo',
 760             'ext': 'mp4',
 761             'title': 'Tripeo Boiler Room x Dekmantel Festival DJ Set',
 762             'description': 'md5:2d713ccbb45b686a1888397b2c77ca6b',
 763             'channel_id': 'UCGBpxWJr9FNOcFYA5GkKrMg',
 764             'playable_in_embed': True,
 765             'categories': ['Music'],
 766             'channel': 'Boiler Room',
 767             'uploader_id': 'brtvofficial',
 768             'uploader': 'Boiler Room',
 769             'tags': 'count:11',
 770             'duration': 3332,
 771             'live_status': 'not_live',
 772             'thumbnail': 'https://i.ytimg.com/vi/pO8h3EaFRdo/maxresdefault.jpg',
 773             'channel_url': 'https://www.youtube.com/channel/UCGBpxWJr9FNOcFYA5GkKrMg',
 774             'availability': 'public',
 775             'uploader_url': 'http://www.youtube.com/user/brtvofficial',
 776             'upload_date': '20150917',
 777             'age_limit': 0,
 778             'view_count': int,
 779             'like_count': int,
 780         },
 781         'add_ie': ['Youtube'],
 782         'params': {'skip_download': 'Youtube'},
 783     }]
 784
 785     def _real_extract(self, url):
 786         redirect_url = url_or_none(parse_qs(url).get('u', [None])[-1])
 787         if not redirect_url:
 788             raise ExtractorError('Invalid facebook redirect URL', expected=True)
 789         return self.url_result(redirect_url)