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