youtube_dl/extractor/brightcove.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import base64
   5 import json
   6 import re
   7 import struct
   8
   9 from .common import InfoExtractor
  10 from .adobepass import AdobePassIE
  11 from ..compat import (
  12     compat_etree_fromstring,
  13     compat_parse_qs,
  14     compat_str,
  15     compat_urllib_parse_urlparse,
  16     compat_urlparse,
  17     compat_xml_parse_error,
  18     compat_HTTPError,
  19 )
  20 from ..utils import (
  21     determine_ext,
  22     ExtractorError,
  23     extract_attributes,
  24     find_xpath_attr,
  25     fix_xml_ampersands,
  26     float_or_none,
  27     js_to_json,
  28     int_or_none,
  29     parse_iso8601,
  30     unescapeHTML,
  31     unsmuggle_url,
  32     update_url_query,
  33     clean_html,
  34     mimetype2ext,
  35 )
  36
  37
  38 class BrightcoveLegacyIE(InfoExtractor):
  39     IE_NAME = 'brightcove:legacy'
  40     _VALID_URL = r'(?:https?://.*brightcove\.com/(services|viewer).*?\?|brightcove:)(?P<query>.*)'
  41     _FEDERATED_URL = 'http://c.brightcove.com/services/viewer/htmlFederated'
  42
  43     _TESTS = [
  44         {
  45             # From http://www.8tv.cat/8aldia/videos/xavier-sala-i-martin-aquesta-tarda-a-8-al-dia/
  46             'url': 'http://c.brightcove.com/services/viewer/htmlFederated?playerID=1654948606001&flashID=myExperience&%40videoPlayer=2371591881001',
  47             'md5': '5423e113865d26e40624dce2e4b45d95',
  48             'note': 'Test Brightcove downloads and detection in GenericIE',
  49             'info_dict': {
  50                 'id': '2371591881001',
  51                 'ext': 'mp4',
  52                 'title': 'Xavier Sala i Martín: “Un banc que no presta és un banc zombi que no serveix per a res”',
  53                 'uploader': '8TV',
  54                 'description': 'md5:a950cc4285c43e44d763d036710cd9cd',
  55                 'timestamp': 1368213670,
  56                 'upload_date': '20130510',
  57                 'uploader_id': '1589608506001',
  58             }
  59         },
  60         {
  61             # From http://medianetwork.oracle.com/video/player/1785452137001
  62             'url': 'http://c.brightcove.com/services/viewer/htmlFederated?playerID=1217746023001&flashID=myPlayer&%40videoPlayer=1785452137001',
  63             'info_dict': {
  64                 'id': '1785452137001',
  65                 'ext': 'flv',
  66                 'title': 'JVMLS 2012: Arrays 2.0 - Opportunities and Challenges',
  67                 'description': 'John Rose speaks at the JVM Language Summit, August 1, 2012.',
  68                 'uploader': 'Oracle',
  69                 'timestamp': 1344975024,
  70                 'upload_date': '20120814',
  71                 'uploader_id': '1460825906',
  72             },
  73         },
  74         {
  75             # From http://mashable.com/2013/10/26/thermoelectric-bracelet-lets-you-control-your-body-temperature/
  76             'url': 'http://c.brightcove.com/services/viewer/federated_f9?&playerID=1265504713001&publisherID=AQ%7E%7E%2CAAABBzUwv1E%7E%2CxP-xFHVUstiMFlNYfvF4G9yFnNaqCw_9&videoID=2750934548001',
  77             'info_dict': {
  78                 'id': '2750934548001',
  79                 'ext': 'mp4',
  80                 'title': 'This Bracelet Acts as a Personal Thermostat',
  81                 'description': 'md5:547b78c64f4112766ccf4e151c20b6a0',
  82                 'uploader': 'Mashable',
  83                 'timestamp': 1382041798,
  84                 'upload_date': '20131017',
  85                 'uploader_id': '1130468786001',
  86             },
  87         },
  88         {
  89             # test that the default referer works
  90             # from http://national.ballet.ca/interact/video/Lost_in_Motion_II/
  91             'url': 'http://link.brightcove.com/services/player/bcpid756015033001?bckey=AQ~~,AAAApYJi_Ck~,GxhXCegT1Dp39ilhXuxMJxasUhVNZiil&bctid=2878862109001',
  92             'info_dict': {
  93                 'id': '2878862109001',
  94                 'ext': 'mp4',
  95                 'title': 'Lost in Motion II',
  96                 'description': 'md5:363109c02998fee92ec02211bd8000df',
  97                 'uploader': 'National Ballet of Canada',
  98             },
  99             'skip': 'Video gone',
 100         },
 101         {
 102             # test flv videos served by akamaihd.net
 103             # From http://www.redbull.com/en/bike/stories/1331655643987/replay-uci-dh-world-cup-2014-from-fort-william
 104             'url': 'http://c.brightcove.com/services/viewer/htmlFederated?%40videoPlayer=ref%3Aevent-stream-356&linkBaseURL=http%3A%2F%2Fwww.redbull.com%2Fen%2Fbike%2Fvideos%2F1331655630249%2Freplay-uci-fort-william-2014-dh&playerKey=AQ%7E%7E%2CAAAApYJ7UqE%7E%2Cxqr_zXk0I-zzNndy8NlHogrCb5QdyZRf&playerID=1398061561001#__youtubedl_smuggle=%7B%22Referer%22%3A+%22http%3A%2F%2Fwww.redbull.com%2Fen%2Fbike%2Fstories%2F1331655643987%2Freplay-uci-dh-world-cup-2014-from-fort-william%22%7D',
 105             # The md5 checksum changes on each download
 106             'info_dict': {
 107                 'id': '3750436379001',
 108                 'ext': 'flv',
 109                 'title': 'UCI MTB World Cup 2014: Fort William, UK - Downhill Finals',
 110                 'uploader': 'RBTV Old (do not use)',
 111                 'description': 'UCI MTB World Cup 2014: Fort William, UK - Downhill Finals',
 112                 'timestamp': 1409122195,
 113                 'upload_date': '20140827',
 114                 'uploader_id': '710858724001',
 115             },
 116             'skip': 'Video gone',
 117         },
 118         {
 119             # playlist with 'videoList'
 120             # from http://support.brightcove.com/en/video-cloud/docs/playlist-support-single-video-players
 121             'url': 'http://c.brightcove.com/services/viewer/htmlFederated?playerID=3550052898001&playerKey=AQ%7E%7E%2CAAABmA9XpXk%7E%2C-Kp7jNgisre1fG5OdqpAFUTcs0lP_ZoL',
 122             'info_dict': {
 123                 'title': 'Sealife',
 124                 'id': '3550319591001',
 125             },
 126             'playlist_mincount': 7,
 127         },
 128         {
 129             # playlist with 'playlistTab' (https://github.com/rg3/youtube-dl/issues/9965)
 130             'url': 'http://c.brightcove.com/services/json/experience/runtime/?command=get_programming_for_experience&playerKey=AQ%7E%7E,AAABXlLMdok%7E,NJ4EoMlZ4rZdx9eU1rkMVd8EaYPBBUlg',
 131             'info_dict': {
 132                 'id': '1522758701001',
 133                 'title': 'Lesson 08',
 134             },
 135             'playlist_mincount': 10,
 136         },
 137         {
 138             # playerID inferred from bcpid
 139             # from http://www.un.org/chinese/News/story.asp?NewsID=27724
 140             'url': 'https://link.brightcove.com/services/player/bcpid1722935254001/?bctid=5360463607001&autoStart=false&secureConnections=true&width=650&height=350',
 141             'only_matching': True,  # Tested in GenericIE
 142         }
 143     ]
 144     FLV_VCODECS = {
 145         1: 'SORENSON',
 146         2: 'ON2',
 147         3: 'H264',
 148         4: 'VP8',
 149     }
 150
 151     @classmethod
 152     def _build_brighcove_url(cls, object_str):
 153         """
 154         Build a Brightcove url from a xml string containing
 155         <object class="BrightcoveExperience">{params}</object>
 156         """
 157
 158         # Fix up some stupid HTML, see https://github.com/rg3/youtube-dl/issues/1553
 159         object_str = re.sub(r'(<param(?:\s+[a-zA-Z0-9_]+="[^"]*")*)>',
 160                             lambda m: m.group(1) + '/>', object_str)
 161         # Fix up some stupid XML, see https://github.com/rg3/youtube-dl/issues/1608
 162         object_str = object_str.replace('<--', '<!--')
 163         # remove namespace to simplify extraction
 164         object_str = re.sub(r'(<object[^>]*)(xmlns=".*?")', r'\1', object_str)
 165         object_str = fix_xml_ampersands(object_str)
 166
 167         try:
 168             object_doc = compat_etree_fromstring(object_str.encode('utf-8'))
 169         except compat_xml_parse_error:
 170             return
 171
 172         fv_el = find_xpath_attr(object_doc, './param', 'name', 'flashVars')
 173         if fv_el is not None:
 174             flashvars = dict(
 175                 (k, v[0])
 176                 for k, v in compat_parse_qs(fv_el.attrib['value']).items())
 177         else:
 178             flashvars = {}
 179
 180         data_url = object_doc.attrib.get('data', '')
 181         data_url_params = compat_parse_qs(compat_urllib_parse_urlparse(data_url).query)
 182
 183         def find_param(name):
 184             if name in flashvars:
 185                 return flashvars[name]
 186             node = find_xpath_attr(object_doc, './param', 'name', name)
 187             if node is not None:
 188                 return node.attrib['value']
 189             return data_url_params.get(name)
 190
 191         params = {}
 192
 193         playerID = find_param('playerID') or find_param('playerId')
 194         if playerID is None:
 195             raise ExtractorError('Cannot find player ID')
 196         params['playerID'] = playerID
 197
 198         playerKey = find_param('playerKey')
 199         # Not all pages define this value
 200         if playerKey is not None:
 201             params['playerKey'] = playerKey
 202         # These fields hold the id of the video
 203         videoPlayer = find_param('@videoPlayer') or find_param('videoId') or find_param('videoID') or find_param('@videoList')
 204         if videoPlayer is not None:
 205             if isinstance(videoPlayer, list):
 206                 videoPlayer = videoPlayer[0]
 207             videoPlayer = videoPlayer.strip()
 208             # UUID is also possible for videoPlayer (e.g.
 209             # http://www.popcornflix.com/hoodies-vs-hooligans/7f2d2b87-bbf2-4623-acfb-ea942b4f01dd
 210             # or http://www8.hp.com/cn/zh/home.html)
 211             if not (re.match(
 212                     r'^(?:\d+|[\da-fA-F]{8}-?[\da-fA-F]{4}-?[\da-fA-F]{4}-?[\da-fA-F]{4}-?[\da-fA-F]{12})$',
 213                     videoPlayer) or videoPlayer.startswith('ref:')):
 214                 return None
 215             params['@videoPlayer'] = videoPlayer
 216         linkBase = find_param('linkBaseURL')
 217         if linkBase is not None:
 218             params['linkBaseURL'] = linkBase
 219         return cls._make_brightcove_url(params)
 220
 221     @classmethod
 222     def _build_brighcove_url_from_js(cls, object_js):
 223         # The layout of JS is as follows:
 224         # customBC.createVideo = function (width, height, playerID, playerKey, videoPlayer, VideoRandomID) {
 225         #   // build Brightcove <object /> XML
 226         # }
 227         m = re.search(
 228             r'''(?x)customBC\.createVideo\(
 229                 .*?                                                  # skipping width and height
 230                 ["\'](?P<playerID>\d+)["\']\s*,\s*                   # playerID
 231                 ["\'](?P<playerKey>AQ[^"\']{48})[^"\']*["\']\s*,\s*  # playerKey begins with AQ and is 50 characters
 232                                                                      # in length, however it's appended to itself
 233                                                                      # in places, so truncate
 234                 ["\'](?P<videoID>\d+)["\']                           # @videoPlayer
 235             ''', object_js)
 236         if m:
 237             return cls._make_brightcove_url(m.groupdict())
 238
 239     @classmethod
 240     def _make_brightcove_url(cls, params):
 241         return update_url_query(cls._FEDERATED_URL, params)
 242
 243     @classmethod
 244     def _extract_brightcove_url(cls, webpage):
 245         """Try to extract the brightcove url from the webpage, returns None
 246         if it can't be found
 247         """
 248         urls = cls._extract_brightcove_urls(webpage)
 249         return urls[0] if urls else None
 250
 251     @classmethod
 252     def _extract_brightcove_urls(cls, webpage):
 253         """Return a list of all Brightcove URLs from the webpage """
 254
 255         url_m = re.search(
 256             r'''(?x)
 257                 <meta\s+
 258                     (?:property|itemprop)=([\'"])(?:og:video|embedURL)\1[^>]+
 259                     content=([\'"])(?P<url>https?://(?:secure|c)\.brightcove.com/(?:(?!\2).)+)\2
 260             ''', webpage)
 261         if url_m:
 262             url = unescapeHTML(url_m.group('url'))
 263             # Some sites don't add it, we can't download with this url, for example:
 264             # http://www.ktvu.com/videos/news/raw-video-caltrain-releases-video-of-man-almost/vCTZdY/
 265             if 'playerKey' in url or 'videoId' in url or 'idVideo' in url:
 266                 return [url]
 267
 268         matches = re.findall(
 269             r'''(?sx)<object
 270             (?:
 271                 [^>]+?class=[\'"][^>]*?BrightcoveExperience.*?[\'"] |
 272                 [^>]*?>\s*<param\s+name="movie"\s+value="https?://[^/]*brightcove\.com/
 273             ).+?>\s*</object>''',
 274             webpage)
 275         if matches:
 276             return list(filter(None, [cls._build_brighcove_url(m) for m in matches]))
 277
 278         matches = re.findall(r'(customBC\.createVideo\(.+?\);)', webpage)
 279         if matches:
 280             return list(filter(None, [
 281                 cls._build_brighcove_url_from_js(custom_bc)
 282                 for custom_bc in matches]))
 283         return [src for _, src in re.findall(
 284             r'<iframe[^>]+src=([\'"])((?:https?:)?//link\.brightcove\.com/services/player/(?!\1).+)\1', webpage)]
 285
 286     def _real_extract(self, url):
 287         url, smuggled_data = unsmuggle_url(url, {})
 288
 289         # Change the 'videoId' and others field to '@videoPlayer'
 290         url = re.sub(r'(?<=[?&])(videoI(d|D)|idVideo|bctid)', '%40videoPlayer', url)
 291         # Change bckey (used by bcove.me urls) to playerKey
 292         url = re.sub(r'(?<=[?&])bckey', 'playerKey', url)
 293         mobj = re.match(self._VALID_URL, url)
 294         query_str = mobj.group('query')
 295         query = compat_urlparse.parse_qs(query_str)
 296
 297         videoPlayer = query.get('@videoPlayer')
 298         if videoPlayer:
 299             # We set the original url as the default 'Referer' header
 300             referer = smuggled_data.get('Referer', url)
 301             if 'playerID' not in query:
 302                 mobj = re.search(r'/bcpid(\d+)', url)
 303                 if mobj is not None:
 304                     query['playerID'] = [mobj.group(1)]
 305             return self._get_video_info(
 306                 videoPlayer[0], query, referer=referer)
 307         elif 'playerKey' in query:
 308             player_key = query['playerKey']
 309             return self._get_playlist_info(player_key[0])
 310         else:
 311             raise ExtractorError(
 312                 'Cannot find playerKey= variable. Did you forget quotes in a shell invocation?',
 313                 expected=True)
 314
 315     def _brightcove_new_url_result(self, publisher_id, video_id):
 316         brightcove_new_url = 'http://players.brightcove.net/%s/default_default/index.html?videoId=%s' % (publisher_id, video_id)
 317         return self.url_result(brightcove_new_url, BrightcoveNewIE.ie_key(), video_id)
 318
 319     def _get_video_info(self, video_id, query, referer=None):
 320         headers = {}
 321         linkBase = query.get('linkBaseURL')
 322         if linkBase is not None:
 323             referer = linkBase[0]
 324         if referer is not None:
 325             headers['Referer'] = referer
 326         webpage = self._download_webpage(self._FEDERATED_URL, video_id, headers=headers, query=query)
 327
 328         error_msg = self._html_search_regex(
 329             r"<h1>We're sorry.</h1>([\s\n]*<p>.*?</p>)+", webpage,
 330             'error message', default=None)
 331         if error_msg is not None:
 332             publisher_id = query.get('publisherId')
 333             if publisher_id and publisher_id[0].isdigit():
 334                 publisher_id = publisher_id[0]
 335             if not publisher_id:
 336                 valid_key = lambda key: key and ',' in key
 337                 player_key = query.get('playerKey')
 338                 if player_key and ',' in player_key[0]:
 339                     player_key = player_key[0]
 340                 else:
 341                     player_id = query.get('playerID')
 342                     if player_id and player_id[0].isdigit():
 343                         player_page = self._download_webpage(
 344                             'http://link.brightcove.com/services/player/bcpid' + player_id[0],
 345                             video_id, headers=headers, fatal=False)
 346                         if player_page:
 347                             player_key = self._search_regex(
 348                                 r'<param\s+name="playerKey"\s+value="([\w~,-]+)"',
 349                                 player_page, 'player key', fatal=False)
 350                 if player_key:
 351                     enc_pub_id = player_key.split(',')[1].replace('~', '=')
 352                     publisher_id = struct.unpack('>Q', base64.urlsafe_b64decode(enc_pub_id))[0]
 353                 if publisher_id:
 354                     return self._brightcove_new_url_result(publisher_id, video_id)
 355             raise ExtractorError(
 356                 'brightcove said: %s' % error_msg, expected=True)
 357
 358         self.report_extraction(video_id)
 359         info = self._search_regex(r'var experienceJSON = ({.*});', webpage, 'json')
 360         info = json.loads(info)['data']
 361         video_info = info['programmedContent']['videoPlayer']['mediaDTO']
 362         video_info['_youtubedl_adServerURL'] = info.get('adServerURL')
 363
 364         return self._extract_video_info(video_info)
 365
 366     def _get_playlist_info(self, player_key):
 367         info_url = 'http://c.brightcove.com/services/json/experience/runtime/?command=get_programming_for_experience&playerKey=%s' % player_key
 368         playlist_info = self._download_webpage(
 369             info_url, player_key, 'Downloading playlist information')
 370
 371         json_data = json.loads(playlist_info)
 372         if 'videoList' in json_data:
 373             playlist_info = json_data['videoList']
 374             playlist_dto = playlist_info['mediaCollectionDTO']
 375         elif 'playlistTabs' in json_data:
 376             playlist_info = json_data['playlistTabs']
 377             playlist_dto = playlist_info['lineupListDTO']['playlistDTOs'][0]
 378         else:
 379             raise ExtractorError('Empty playlist')
 380
 381         videos = [self._extract_video_info(video_info) for video_info in playlist_dto['videoDTOs']]
 382
 383         return self.playlist_result(videos, playlist_id='%s' % playlist_info['id'],
 384                                     playlist_title=playlist_dto['displayName'])
 385
 386     def _extract_video_info(self, video_info):
 387         video_id = compat_str(video_info['id'])
 388         publisher_id = video_info.get('publisherId')
 389         info = {
 390             'id': video_id,
 391             'title': video_info['displayName'].strip(),
 392             'description': video_info.get('shortDescription'),
 393             'thumbnail': video_info.get('videoStillURL') or video_info.get('thumbnailURL'),
 394             'uploader': video_info.get('publisherName'),
 395             'uploader_id': compat_str(publisher_id) if publisher_id else None,
 396             'duration': float_or_none(video_info.get('length'), 1000),
 397             'timestamp': int_or_none(video_info.get('creationDate'), 1000),
 398         }
 399
 400         renditions = video_info.get('renditions', []) + video_info.get('IOSRenditions', [])
 401         if renditions:
 402             formats = []
 403             for rend in renditions:
 404                 url = rend['defaultURL']
 405                 if not url:
 406                     continue
 407                 ext = None
 408                 if rend['remote']:
 409                     url_comp = compat_urllib_parse_urlparse(url)
 410                     if url_comp.path.endswith('.m3u8'):
 411                         formats.extend(
 412                             self._extract_m3u8_formats(
 413                                 url, video_id, 'mp4', 'm3u8_native', m3u8_id='hls', fatal=False))
 414                         continue
 415                     elif 'akamaihd.net' in url_comp.netloc:
 416                         # This type of renditions are served through
 417                         # akamaihd.net, but they don't use f4m manifests
 418                         url = url.replace('control/', '') + '?&v=3.3.0&fp=13&r=FEEFJ&g=RTSJIMBMPFPB'
 419                         ext = 'flv'
 420                 if ext is None:
 421                     ext = determine_ext(url)
 422                 tbr = int_or_none(rend.get('encodingRate'), 1000)
 423                 a_format = {
 424                     'format_id': 'http%s' % ('-%s' % tbr if tbr else ''),
 425                     'url': url,
 426                     'ext': ext,
 427                     'filesize': int_or_none(rend.get('size')) or None,
 428                     'tbr': tbr,
 429                 }
 430                 if rend.get('audioOnly'):
 431                     a_format.update({
 432                         'vcodec': 'none',
 433                     })
 434                 else:
 435                     a_format.update({
 436                         'height': int_or_none(rend.get('frameHeight')),
 437                         'width': int_or_none(rend.get('frameWidth')),
 438                         'vcodec': rend.get('videoCodec'),
 439                     })
 440
 441                 # m3u8 manifests with remote == false are media playlists
 442                 # Not calling _extract_m3u8_formats here to save network traffic
 443                 if ext == 'm3u8':
 444                     a_format.update({
 445                         'format_id': 'hls%s' % ('-%s' % tbr if tbr else ''),
 446                         'ext': 'mp4',
 447                         'protocol': 'm3u8_native',
 448                     })
 449
 450                 formats.append(a_format)
 451             self._sort_formats(formats)
 452             info['formats'] = formats
 453         elif video_info.get('FLVFullLengthURL') is not None:
 454             info.update({
 455                 'url': video_info['FLVFullLengthURL'],
 456                 'vcodec': self.FLV_VCODECS.get(video_info.get('FLVFullCodec')),
 457                 'filesize': int_or_none(video_info.get('FLVFullSize')),
 458             })
 459
 460         if self._downloader.params.get('include_ads', False):
 461             adServerURL = video_info.get('_youtubedl_adServerURL')
 462             if adServerURL:
 463                 ad_info = {
 464                     '_type': 'url',
 465                     'url': adServerURL,
 466                 }
 467                 if 'url' in info:
 468                     return {
 469                         '_type': 'playlist',
 470                         'title': info['title'],
 471                         'entries': [ad_info, info],
 472                     }
 473                 else:
 474                     return ad_info
 475
 476         if not info.get('url') and not info.get('formats'):
 477             uploader_id = info.get('uploader_id')
 478             if uploader_id:
 479                 info.update(self._brightcove_new_url_result(uploader_id, video_id))
 480             else:
 481                 raise ExtractorError('Unable to extract video url for %s' % video_id)
 482         return info
 483
 484
 485 class BrightcoveNewIE(AdobePassIE):
 486     IE_NAME = 'brightcove:new'
 487     _VALID_URL = r'https?://players\.brightcove\.net/(?P<account_id>\d+)/(?P<player_id>[^/]+)_(?P<embed>[^/]+)/index\.html\?.*videoId=(?P<video_id>\d+|ref:[^&]+)'
 488     _TESTS = [{
 489         'url': 'http://players.brightcove.net/929656772001/e41d32dc-ec74-459e-a845-6c69f7b724ea_default/index.html?videoId=4463358922001',
 490         'md5': 'c8100925723840d4b0d243f7025703be',
 491         'info_dict': {
 492             'id': '4463358922001',
 493             'ext': 'mp4',
 494             'title': 'Meet the man behind Popcorn Time',
 495             'description': 'md5:eac376a4fe366edc70279bfb681aea16',
 496             'duration': 165.768,
 497             'timestamp': 1441391203,
 498             'upload_date': '20150904',
 499             'uploader_id': '929656772001',
 500             'formats': 'mincount:20',
 501         },
 502     }, {
 503         # with rtmp streams
 504         'url': 'http://players.brightcove.net/4036320279001/5d112ed9-283f-485f-a7f9-33f42e8bc042_default/index.html?videoId=4279049078001',
 505         'info_dict': {
 506             'id': '4279049078001',
 507             'ext': 'mp4',
 508             'title': 'Titansgrave: Chapter 0',
 509             'description': 'Titansgrave: Chapter 0',
 510             'duration': 1242.058,
 511             'timestamp': 1433556729,
 512             'upload_date': '20150606',
 513             'uploader_id': '4036320279001',
 514             'formats': 'mincount:39',
 515         },
 516         'params': {
 517             # m3u8 download
 518             'skip_download': True,
 519         }
 520     }, {
 521         # ref: prefixed video id
 522         'url': 'http://players.brightcove.net/3910869709001/21519b5c-4b3b-4363-accb-bdc8f358f823_default/index.html?videoId=ref:7069442',
 523         'only_matching': True,
 524     }, {
 525         # non numeric ref: prefixed video id
 526         'url': 'http://players.brightcove.net/710858724001/default_default/index.html?videoId=ref:event-stream-356',
 527         'only_matching': True,
 528     }, {
 529         # unavailable video without message but with error_code
 530         'url': 'http://players.brightcove.net/1305187701/c832abfb-641b-44eb-9da0-2fe76786505f_default/index.html?videoId=4377407326001',
 531         'only_matching': True,
 532     }]
 533
 534     @staticmethod
 535     def _extract_url(ie, webpage):
 536         urls = BrightcoveNewIE._extract_urls(ie, webpage)
 537         return urls[0] if urls else None
 538
 539     @staticmethod
 540     def _extract_urls(ie, webpage):
 541         # Reference:
 542         # 1. http://docs.brightcove.com/en/video-cloud/brightcove-player/guides/publish-video.html#setvideoiniframe
 543         # 2. http://docs.brightcove.com/en/video-cloud/brightcove-player/guides/publish-video.html#tag
 544         # 3. http://docs.brightcove.com/en/video-cloud/brightcove-player/guides/publish-video.html#setvideousingjavascript
 545         # 4. http://docs.brightcove.com/en/video-cloud/brightcove-player/guides/in-page-embed-player-implementation.html
 546         # 5. https://support.brightcove.com/en/video-cloud/docs/dynamically-assigning-videos-player
 547
 548         entries = []
 549
 550         # Look for iframe embeds [1]
 551         for _, url in re.findall(
 552                 r'<iframe[^>]+src=(["\'])((?:https?:)?//players\.brightcove\.net/\d+/[^/]+/index\.html.+?)\1', webpage):
 553             entries.append(url if url.startswith('http') else 'http:' + url)
 554
 555         # Look for <video> tags [2] and embed_in_page embeds [3]
 556         # [2] looks like:
 557         for video, script_tag, account_id, player_id, embed in re.findall(
 558                 r'''(?isx)
 559                     (<video\s+[^>]*\bdata-video-id\s*=\s*['"]?[^>]+>)
 560                     (?:.*?
 561                         (<script[^>]+
 562                             src=["\'](?:https?:)?//players\.brightcove\.net/
 563                             (\d+)/([^/]+)_([^/]+)/index(?:\.min)?\.js
 564                         )
 565                     )?
 566                 ''', webpage):
 567             attrs = extract_attributes(video)
 568
 569             # According to examples from [4] it's unclear whether video id
 570             # may be optional and what to do when it is
 571             video_id = attrs.get('data-video-id')
 572             if not video_id:
 573                 continue
 574
 575             account_id = account_id or attrs.get('data-account')
 576             if not account_id:
 577                 continue
 578
 579             player_id = player_id or attrs.get('data-player') or 'default'
 580             embed = embed or attrs.get('data-embed') or 'default'
 581
 582             bc_url = 'http://players.brightcove.net/%s/%s_%s/index.html?videoId=%s' % (
 583                 account_id, player_id, embed, video_id)
 584
 585             # Some brightcove videos may be embedded with video tag only and
 586             # without script tag or any mentioning of brightcove at all. Such
 587             # embeds are considered ambiguous since they are matched based only
 588             # on data-video-id and data-account attributes and in the wild may
 589             # not be brightcove embeds at all. Let's check reconstructed
 590             # brightcove URLs in case of such embeds and only process valid
 591             # ones. By this we ensure there is indeed a brightcove embed.
 592             if not script_tag and not ie._is_valid_url(
 593                     bc_url, video_id, 'possible brightcove video'):
 594                 continue
 595
 596             entries.append(bc_url)
 597
 598         return entries
 599
 600     def _parse_brightcove_metadata(self, json_data, video_id, headers={}):
 601         title = json_data['name'].strip()
 602
 603         formats = []
 604         for source in json_data.get('sources', []):
 605             container = source.get('container')
 606             ext = mimetype2ext(source.get('type'))
 607             src = source.get('src')
 608             # https://support.brightcove.com/playback-api-video-fields-reference#key_systems_object
 609             if ext == 'ism' or container == 'WVM' or source.get('key_systems'):
 610                 continue
 611             elif ext == 'm3u8' or container == 'M2TS':
 612                 if not src:
 613                     continue
 614                 formats.extend(self._extract_m3u8_formats(
 615                     src, video_id, 'mp4', 'm3u8_native', m3u8_id='hls', fatal=False))
 616             elif ext == 'mpd':
 617                 if not src:
 618                     continue
 619                 formats.extend(self._extract_mpd_formats(src, video_id, 'dash', fatal=False))
 620             else:
 621                 streaming_src = source.get('streaming_src')
 622                 stream_name, app_name = source.get('stream_name'), source.get('app_name')
 623                 if not src and not streaming_src and (not stream_name or not app_name):
 624                     continue
 625                 tbr = float_or_none(source.get('avg_bitrate'), 1000)
 626                 height = int_or_none(source.get('height'))
 627                 width = int_or_none(source.get('width'))
 628                 f = {
 629                     'tbr': tbr,
 630                     'filesize': int_or_none(source.get('size')),
 631                     'container': container,
 632                     'ext': ext or container.lower(),
 633                 }
 634                 if width == 0 and height == 0:
 635                     f.update({
 636                         'vcodec': 'none',
 637                     })
 638                 else:
 639                     f.update({
 640                         'width': width,
 641                         'height': height,
 642                         'vcodec': source.get('codec'),
 643                     })
 644
 645                 def build_format_id(kind):
 646                     format_id = kind
 647                     if tbr:
 648                         format_id += '-%dk' % int(tbr)
 649                     if height:
 650                         format_id += '-%dp' % height
 651                     return format_id
 652
 653                 if src or streaming_src:
 654                     f.update({
 655                         'url': src or streaming_src,
 656                         'format_id': build_format_id('http' if src else 'http-streaming'),
 657                         'source_preference': 0 if src else -1,
 658                     })
 659                 else:
 660                     f.update({
 661                         'url': app_name,
 662                         'play_path': stream_name,
 663                         'format_id': build_format_id('rtmp'),
 664                     })
 665                 formats.append(f)
 666         if not formats:
 667             # for sonyliv.com DRM protected videos
 668             s3_source_url = json_data.get('custom_fields', {}).get('s3sourceurl')
 669             if s3_source_url:
 670                 formats.append({
 671                     'url': s3_source_url,
 672                     'format_id': 'source',
 673                 })
 674
 675         errors = json_data.get('errors')
 676         if not formats and errors:
 677             error = errors[0]
 678             raise ExtractorError(
 679                 error.get('message') or error.get('error_subcode') or error['error_code'], expected=True)
 680
 681         self._sort_formats(formats)
 682
 683         for f in formats:
 684             f.setdefault('http_headers', {}).update(headers)
 685
 686         subtitles = {}
 687         for text_track in json_data.get('text_tracks', []):
 688             if text_track.get('src'):
 689                 subtitles.setdefault(text_track.get('srclang'), []).append({
 690                     'url': text_track['src'],
 691                 })
 692
 693         is_live = False
 694         duration = float_or_none(json_data.get('duration'), 1000)
 695         if duration is not None and duration <= 0:
 696             is_live = True
 697
 698         return {
 699             'id': video_id,
 700             'title': self._live_title(title) if is_live else title,
 701             'description': clean_html(json_data.get('description')),
 702             'thumbnail': json_data.get('thumbnail') or json_data.get('poster'),
 703             'duration': duration,
 704             'timestamp': parse_iso8601(json_data.get('published_at')),
 705             'uploader_id': json_data.get('account_id'),
 706             'formats': formats,
 707             'subtitles': subtitles,
 708             'tags': json_data.get('tags', []),
 709             'is_live': is_live,
 710         }
 711
 712     def _real_extract(self, url):
 713         url, smuggled_data = unsmuggle_url(url, {})
 714         self._initialize_geo_bypass({
 715             'countries': smuggled_data.get('geo_countries'),
 716             'ip_blocks': smuggled_data.get('geo_ip_blocks'),
 717         })
 718
 719         account_id, player_id, embed, video_id = re.match(self._VALID_URL, url).groups()
 720
 721         webpage = self._download_webpage(
 722             'http://players.brightcove.net/%s/%s_%s/index.min.js'
 723             % (account_id, player_id, embed), video_id)
 724
 725         policy_key = None
 726
 727         catalog = self._search_regex(
 728             r'catalog\(({.+?})\);', webpage, 'catalog', default=None)
 729         if catalog:
 730             catalog = self._parse_json(
 731                 js_to_json(catalog), video_id, fatal=False)
 732             if catalog:
 733                 policy_key = catalog.get('policyKey')
 734
 735         if not policy_key:
 736             policy_key = self._search_regex(
 737                 r'policyKey\s*:\s*(["\'])(?P<pk>.+?)\1',
 738                 webpage, 'policy key', group='pk')
 739
 740         api_url = 'https://edge.api.brightcove.com/playback/v1/accounts/%s/videos/%s' % (account_id, video_id)
 741         headers = {
 742             'Accept': 'application/json;pk=%s' % policy_key,
 743         }
 744         referrer = smuggled_data.get('referrer')
 745         if referrer:
 746             headers.update({
 747                 'Referer': referrer,
 748                 'Origin': re.search(r'https?://[^/]+', referrer).group(0),
 749             })
 750         try:
 751             json_data = self._download_json(api_url, video_id, headers=headers)
 752         except ExtractorError as e:
 753             if isinstance(e.cause, compat_HTTPError) and e.cause.code == 403:
 754                 json_data = self._parse_json(e.cause.read().decode(), video_id)[0]
 755                 message = json_data.get('message') or json_data['error_code']
 756                 if json_data.get('error_subcode') == 'CLIENT_GEO':
 757                     self.raise_geo_restricted(msg=message)
 758                 raise ExtractorError(message, expected=True)
 759             raise
 760
 761         errors = json_data.get('errors')
 762         if errors and errors[0].get('error_subcode') == 'TVE_AUTH':
 763             custom_fields = json_data['custom_fields']
 764             tve_token = self._extract_mvpd_auth(
 765                 smuggled_data['source_url'], video_id,
 766                 custom_fields['bcadobepassrequestorid'],
 767                 custom_fields['bcadobepassresourceid'])
 768             json_data = self._download_json(
 769                 api_url, video_id, headers={
 770                     'Accept': 'application/json;pk=%s' % policy_key
 771                 }, query={
 772                     'tveToken': tve_token,
 773                 })
 774
 775         return self._parse_brightcove_metadata(
 776             json_data, video_id, headers=headers)