yt_dlp/extractor/redbulltv.py

   1 from .common import InfoExtractor
   2 from ..networking.exceptions import HTTPError
   3 from ..utils import (
   4     ExtractorError,
   5     float_or_none,
   6 )
   7
   8
   9 class RedBullTVIE(InfoExtractor):
  10     _VALID_URL = r'https?://(?:www\.)?redbull(?:\.tv|\.com(?:/[^/]+)?(?:/tv)?)(?:/events/[^/]+)?/(?:videos?|live|(?:film|episode)s)/(?P<id>AP-\w+)'
  11     _TESTS = [{
  12         # film
  13         'url': 'https://www.redbull.tv/video/AP-1Q6XCDTAN1W11',
  14         'md5': 'fb0445b98aa4394e504b413d98031d1f',
  15         'info_dict': {
  16             'id': 'AP-1Q6XCDTAN1W11',
  17             'ext': 'mp4',
  18             'title': 'ABC of... WRC - ABC of... S1E6',
  19             'description': 'md5:5c7ed8f4015c8492ecf64b6ab31e7d31',
  20             'duration': 1582.04,
  21         },
  22     }, {
  23         # episode
  24         'url': 'https://www.redbull.tv/video/AP-1PMHKJFCW1W11',
  25         'info_dict': {
  26             'id': 'AP-1PMHKJFCW1W11',
  27             'ext': 'mp4',
  28             'title': 'Grime - Hashtags S2E4',
  29             'description': 'md5:5546aa612958c08a98faaad4abce484d',
  30             'duration': 904,
  31         },
  32         'params': {
  33             'skip_download': True,
  34         },
  35     }, {
  36         'url': 'https://www.redbull.com/int-en/tv/video/AP-1UWHCAR9S1W11/rob-meets-sam-gaze?playlist=playlists::3f81040a-2f31-4832-8e2e-545b1d39d173',
  37         'only_matching': True,
  38     }, {
  39         'url': 'https://www.redbull.com/us-en/videos/AP-1YM9QCYE52111',
  40         'only_matching': True,
  41     }, {
  42         'url': 'https://www.redbull.com/us-en/events/AP-1XV2K61Q51W11/live/AP-1XUJ86FDH1W11',
  43         'only_matching': True,
  44     }, {
  45         'url': 'https://www.redbull.com/int-en/films/AP-1ZSMAW8FH2111',
  46         'only_matching': True,
  47     }, {
  48         'url': 'https://www.redbull.com/int-en/episodes/AP-1TQWK7XE11W11',
  49         'only_matching': True,
  50     }]
  51
  52     def extract_info(self, video_id):
  53         session = self._download_json(
  54             'https://api.redbull.tv/v3/session', video_id,
  55             note='Downloading access token', query={
  56                 'category': 'personal_computer',
  57                 'os_family': 'http',
  58             })
  59         if session.get('code') == 'error':
  60             raise ExtractorError('{} said: {}'.format(
  61                 self.IE_NAME, session['message']))
  62         token = session['token']
  63
  64         try:
  65             video = self._download_json(
  66                 'https://api.redbull.tv/v3/products/' + video_id,
  67                 video_id, note='Downloading video information',
  68                 headers={'Authorization': token},
  69             )
  70         except ExtractorError as e:
  71             if isinstance(e.cause, HTTPError) and e.cause.status == 404:
  72                 error_message = self._parse_json(
  73                     e.cause.response.read().decode(), video_id)['error']
  74                 raise ExtractorError(f'{self.IE_NAME} said: {error_message}', expected=True)
  75             raise
  76
  77         title = video['title'].strip()
  78
  79         formats, subtitles = self._extract_m3u8_formats_and_subtitles(
  80             f'https://dms.redbull.tv/v3/{video_id}/{token}/playlist.m3u8',
  81             video_id, 'mp4', entry_protocol='m3u8_native', m3u8_id='hls')
  82
  83         for resource in video.get('resources', []):
  84             if resource.startswith('closed_caption_'):
  85                 splitted_resource = resource.split('_')
  86                 if splitted_resource[2]:
  87                     subtitles.setdefault('en', []).append({
  88                         'url': f'https://resources.redbull.tv/{video_id}/{resource}',
  89                         'ext': splitted_resource[2],
  90                     })
  91
  92         subheading = video.get('subheading')
  93         if subheading:
  94             title += f' - {subheading}'
  95
  96         return {
  97             'id': video_id,
  98             'title': title,
  99             'description': video.get('long_description') or video.get(
 100                 'short_description'),
 101             'duration': float_or_none(video.get('duration'), scale=1000),
 102             'formats': formats,
 103             'subtitles': subtitles,
 104         }
 105
 106     def _real_extract(self, url):
 107         video_id = self._match_id(url)
 108         return self.extract_info(video_id)
 109
 110
 111 class RedBullEmbedIE(RedBullTVIE):  # XXX: Do not subclass from concrete IE
 112     _VALID_URL = r'https?://(?:www\.)?redbull\.com/embed/(?P<id>rrn:content:[^:]+:[\da-f]{8}-[\da-f]{4}-[\da-f]{4}-[\da-f]{4}-[\da-f]{12}:[a-z]{2}-[A-Z]{2,3})'
 113     _TESTS = [{
 114         # HLS manifest accessible only using assetId
 115         'url': 'https://www.redbull.com/embed/rrn:content:episode-videos:f3021f4f-3ed4-51ac-915a-11987126e405:en-INT',
 116         'only_matching': True,
 117     }]
 118     _VIDEO_ESSENSE_TMPL = '''... on %s {
 119       videoEssence {
 120         attributes
 121       }
 122     }'''
 123
 124     def _real_extract(self, url):
 125         rrn_id = self._match_id(url)
 126         asset_id = self._download_json(
 127             'https://edge-graphql.crepo-production.redbullaws.com/v1/graphql',
 128             rrn_id, headers={
 129                 'Accept': 'application/json',
 130                 'API-KEY': 'e90a1ff11335423998b100c929ecc866',
 131             }, query={
 132                 'query': '''{
 133   resource(id: "%s", enforceGeoBlocking: false) {
 134     %s
 135     %s
 136   }
 137 }''' % (rrn_id, self._VIDEO_ESSENSE_TMPL % 'LiveVideo', self._VIDEO_ESSENSE_TMPL % 'VideoResource'),  # noqa: UP031
 138             })['data']['resource']['videoEssence']['attributes']['assetId']
 139         return self.extract_info(asset_id)
 140
 141
 142 class RedBullTVRrnContentIE(InfoExtractor):
 143     _VALID_URL = r'https?://(?:www\.)?redbull\.com/(?P<region>[a-z]{2,3})-(?P<lang>[a-z]{2})/tv/(?:video|live|film)/(?P<id>rrn:content:[^:]+:[\da-f]{8}-[\da-f]{4}-[\da-f]{4}-[\da-f]{4}-[\da-f]{12})'
 144     _TESTS = [{
 145         'url': 'https://www.redbull.com/int-en/tv/video/rrn:content:live-videos:e3e6feb4-e95f-50b7-962a-c70f8fd13c73/mens-dh-finals-fort-william',
 146         'only_matching': True,
 147     }, {
 148         'url': 'https://www.redbull.com/int-en/tv/video/rrn:content:videos:a36a0f36-ff1b-5db8-a69d-ee11a14bf48b/tn-ts-style?playlist=rrn:content:event-profiles:83f05926-5de8-5389-b5e4-9bb312d715e8:extras',
 149         'only_matching': True,
 150     }, {
 151         'url': 'https://www.redbull.com/int-en/tv/film/rrn:content:films:d1f4d00e-4c04-5d19-b510-a805ffa2ab83/follow-me',
 152         'only_matching': True,
 153     }]
 154
 155     def _real_extract(self, url):
 156         region, lang, rrn_id = self._match_valid_url(url).groups()
 157         rrn_id += f':{lang}-{region.upper()}'
 158         return self.url_result(
 159             'https://www.redbull.com/embed/' + rrn_id,
 160             RedBullEmbedIE.ie_key(), rrn_id)
 161
 162
 163 class RedBullIE(InfoExtractor):
 164     _VALID_URL = r'https?://(?:www\.)?redbull\.com/(?P<region>[a-z]{2,3})-(?P<lang>[a-z]{2})/(?P<type>(?:episode|film|(?:(?:recap|trailer)-)?video)s|live)/(?!AP-|rrn:content:)(?P<id>[^/?#&]+)'
 165     _TESTS = [{
 166         'url': 'https://www.redbull.com/int-en/episodes/grime-hashtags-s02-e04',
 167         'md5': 'db8271a7200d40053a1809ed0dd574ff',
 168         'info_dict': {
 169             'id': 'AA-1MT8DQWA91W14',
 170             'ext': 'mp4',
 171             'title': 'Grime - Hashtags S2E4',
 172             'description': 'md5:5546aa612958c08a98faaad4abce484d',
 173         },
 174     }, {
 175         'url': 'https://www.redbull.com/int-en/films/kilimanjaro-mountain-of-greatness',
 176         'only_matching': True,
 177     }, {
 178         'url': 'https://www.redbull.com/int-en/recap-videos/uci-mountain-bike-world-cup-2017-mens-xco-finals-from-vallnord',
 179         'only_matching': True,
 180     }, {
 181         'url': 'https://www.redbull.com/int-en/trailer-videos/kings-of-content',
 182         'only_matching': True,
 183     }, {
 184         'url': 'https://www.redbull.com/int-en/videos/tnts-style-red-bull-dance-your-style-s1-e12',
 185         'only_matching': True,
 186     }, {
 187         'url': 'https://www.redbull.com/int-en/live/mens-dh-finals-fort-william',
 188         'only_matching': True,
 189     }, {
 190         # only available on the int-en website so a fallback is need for the API
 191         # https://www.redbull.com/v3/api/graphql/v1/v3/query/en-GB>en-INT?filter[uriSlug]=fia-wrc-saturday-recap-estonia&rb3Schema=v1:hero
 192         'url': 'https://www.redbull.com/gb-en/live/fia-wrc-saturday-recap-estonia',
 193         'only_matching': True,
 194     }]
 195     _INT_FALLBACK_LIST = ['de', 'en', 'es', 'fr']
 196     _LAT_FALLBACK_MAP = ['ar', 'bo', 'car', 'cl', 'co', 'mx', 'pe']
 197
 198     def _real_extract(self, url):
 199         region, lang, filter_type, display_id = self._match_valid_url(url).groups()
 200         if filter_type == 'episodes':
 201             filter_type = 'episode-videos'
 202         elif filter_type == 'live':
 203             filter_type = 'live-videos'
 204
 205         regions = [region.upper()]
 206         if region != 'int':
 207             if region in self._LAT_FALLBACK_MAP:
 208                 regions.append('LAT')
 209             if lang in self._INT_FALLBACK_LIST:
 210                 regions.append('INT')
 211         locale = '>'.join([f'{lang}-{reg}' for reg in regions])
 212
 213         rrn_id = self._download_json(
 214             'https://www.redbull.com/v3/api/graphql/v1/v3/query/' + locale,
 215             display_id, query={
 216                 'filter[type]': filter_type,
 217                 'filter[uriSlug]': display_id,
 218                 'rb3Schema': 'v1:hero',
 219             })['data']['id']
 220
 221         return self.url_result(
 222             'https://www.redbull.com/embed/' + rrn_id,
 223             RedBullEmbedIE.ie_key(), rrn_id)