youtube_dl/extractor/redbulltv.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5
   6 from .common import InfoExtractor
   7 from ..compat import compat_HTTPError
   8 from ..utils import (
   9     float_or_none,
  10     ExtractorError,
  11 )
  12
  13
  14 class RedBullTVIE(InfoExtractor):
  15     _VALID_URL = r'https?://(?:www\.)?redbull(?:\.tv|\.com(?:/[^/]+)?(?:/tv)?)(?:/events/[^/]+)?/(?:videos?|live|(?:film|episode)s)/(?P<id>AP-\w+)'
  16     _TESTS = [{
  17         # film
  18         'url': 'https://www.redbull.tv/video/AP-1Q6XCDTAN1W11',
  19         'md5': 'fb0445b98aa4394e504b413d98031d1f',
  20         'info_dict': {
  21             'id': 'AP-1Q6XCDTAN1W11',
  22             'ext': 'mp4',
  23             'title': 'ABC of... WRC - ABC of... S1E6',
  24             'description': 'md5:5c7ed8f4015c8492ecf64b6ab31e7d31',
  25             'duration': 1582.04,
  26         },
  27     }, {
  28         # episode
  29         'url': 'https://www.redbull.tv/video/AP-1PMHKJFCW1W11',
  30         'info_dict': {
  31             'id': 'AP-1PMHKJFCW1W11',
  32             'ext': 'mp4',
  33             'title': 'Grime - Hashtags S2E4',
  34             'description': 'md5:5546aa612958c08a98faaad4abce484d',
  35             'duration': 904,
  36         },
  37         'params': {
  38             'skip_download': True,
  39         },
  40     }, {
  41         'url': 'https://www.redbull.com/int-en/tv/video/AP-1UWHCAR9S1W11/rob-meets-sam-gaze?playlist=playlists::3f81040a-2f31-4832-8e2e-545b1d39d173',
  42         'only_matching': True,
  43     }, {
  44         'url': 'https://www.redbull.com/us-en/videos/AP-1YM9QCYE52111',
  45         'only_matching': True,
  46     }, {
  47         'url': 'https://www.redbull.com/us-en/events/AP-1XV2K61Q51W11/live/AP-1XUJ86FDH1W11',
  48         'only_matching': True,
  49     }, {
  50         'url': 'https://www.redbull.com/int-en/films/AP-1ZSMAW8FH2111',
  51         'only_matching': True,
  52     }, {
  53         'url': 'https://www.redbull.com/int-en/episodes/AP-1TQWK7XE11W11',
  54         'only_matching': True,
  55     }]
  56
  57     def extract_info(self, video_id):
  58         session = self._download_json(
  59             'https://api.redbull.tv/v3/session', video_id,
  60             note='Downloading access token', query={
  61                 'category': 'personal_computer',
  62                 'os_family': 'http',
  63             })
  64         if session.get('code') == 'error':
  65             raise ExtractorError('%s said: %s' % (
  66                 self.IE_NAME, session['message']))
  67         token = session['token']
  68
  69         try:
  70             video = self._download_json(
  71                 'https://api.redbull.tv/v3/products/' + video_id,
  72                 video_id, note='Downloading video information',
  73                 headers={'Authorization': token}
  74             )
  75         except ExtractorError as e:
  76             if isinstance(e.cause, compat_HTTPError) and e.cause.code == 404:
  77                 error_message = self._parse_json(
  78                     e.cause.read().decode(), video_id)['error']
  79                 raise ExtractorError('%s said: %s' % (
  80                     self.IE_NAME, error_message), expected=True)
  81             raise
  82
  83         title = video['title'].strip()
  84
  85         formats = self._extract_m3u8_formats(
  86             'https://dms.redbull.tv/v3/%s/%s/playlist.m3u8' % (video_id, token),
  87             video_id, 'mp4', entry_protocol='m3u8_native', m3u8_id='hls')
  88         self._sort_formats(formats)
  89
  90         subtitles = {}
  91         for resource in video.get('resources', []):
  92             if resource.startswith('closed_caption_'):
  93                 splitted_resource = resource.split('_')
  94                 if splitted_resource[2]:
  95                     subtitles.setdefault('en', []).append({
  96                         'url': 'https://resources.redbull.tv/%s/%s' % (video_id, resource),
  97                         'ext': splitted_resource[2],
  98                     })
  99
 100         subheading = video.get('subheading')
 101         if subheading:
 102             title += ' - %s' % subheading
 103
 104         return {
 105             'id': video_id,
 106             'title': title,
 107             'description': video.get('long_description') or video.get(
 108                 'short_description'),
 109             'duration': float_or_none(video.get('duration'), scale=1000),
 110             'formats': formats,
 111             'subtitles': subtitles,
 112         }
 113
 114     def _real_extract(self, url):
 115         video_id = self._match_id(url)
 116         return self.extract_info(video_id)
 117
 118
 119 class RedBullEmbedIE(RedBullTVIE):
 120     _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})'
 121     _TESTS = [{
 122         # HLS manifest accessible only using assetId
 123         'url': 'https://www.redbull.com/embed/rrn:content:episode-videos:f3021f4f-3ed4-51ac-915a-11987126e405:en-INT',
 124         'only_matching': True,
 125     }]
 126     _VIDEO_ESSENSE_TMPL = '''... on %s {
 127       videoEssence {
 128         attributes
 129       }
 130     }'''
 131
 132     def _real_extract(self, url):
 133         rrn_id = self._match_id(url)
 134         asset_id = self._download_json(
 135             'https://edge-graphql.crepo-production.redbullaws.com/v1/graphql',
 136             rrn_id, headers={'API-KEY': 'e90a1ff11335423998b100c929ecc866'},
 137             query={
 138                 'query': '''{
 139   resource(id: "%s", enforceGeoBlocking: false) {
 140     %s
 141     %s
 142   }
 143 }''' % (rrn_id, self._VIDEO_ESSENSE_TMPL % 'LiveVideo', self._VIDEO_ESSENSE_TMPL % 'VideoResource'),
 144             })['data']['resource']['videoEssence']['attributes']['assetId']
 145         return self.extract_info(asset_id)
 146
 147
 148 class RedBullTVRrnContentIE(InfoExtractor):
 149     _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})'
 150     _TESTS = [{
 151         'url': 'https://www.redbull.com/int-en/tv/video/rrn:content:live-videos:e3e6feb4-e95f-50b7-962a-c70f8fd13c73/mens-dh-finals-fort-william',
 152         'only_matching': True,
 153     }, {
 154         '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',
 155         'only_matching': True,
 156     }, {
 157         'url': 'https://www.redbull.com/int-en/tv/film/rrn:content:films:d1f4d00e-4c04-5d19-b510-a805ffa2ab83/follow-me',
 158         'only_matching': True,
 159     }]
 160
 161     def _real_extract(self, url):
 162         region, lang, rrn_id = re.search(self._VALID_URL, url).groups()
 163         rrn_id += ':%s-%s' % (lang, region.upper())
 164         return self.url_result(
 165             'https://www.redbull.com/embed/' + rrn_id,
 166             RedBullEmbedIE.ie_key(), rrn_id)
 167
 168
 169 class RedBullIE(InfoExtractor):
 170     _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>[^/?#&]+)'
 171     _TESTS = [{
 172         'url': 'https://www.redbull.com/int-en/episodes/grime-hashtags-s02-e04',
 173         'md5': 'db8271a7200d40053a1809ed0dd574ff',
 174         'info_dict': {
 175             'id': 'AA-1MT8DQWA91W14',
 176             'ext': 'mp4',
 177             'title': 'Grime - Hashtags S2E4',
 178             'description': 'md5:5546aa612958c08a98faaad4abce484d',
 179         },
 180     }, {
 181         'url': 'https://www.redbull.com/int-en/films/kilimanjaro-mountain-of-greatness',
 182         'only_matching': True,
 183     }, {
 184         'url': 'https://www.redbull.com/int-en/recap-videos/uci-mountain-bike-world-cup-2017-mens-xco-finals-from-vallnord',
 185         'only_matching': True,
 186     }, {
 187         'url': 'https://www.redbull.com/int-en/trailer-videos/kings-of-content',
 188         'only_matching': True,
 189     }, {
 190         'url': 'https://www.redbull.com/int-en/videos/tnts-style-red-bull-dance-your-style-s1-e12',
 191         'only_matching': True,
 192     }, {
 193         'url': 'https://www.redbull.com/int-en/live/mens-dh-finals-fort-william',
 194         'only_matching': True,
 195     }, {
 196         # only available on the int-en website so a fallback is need for the API
 197         # https://www.redbull.com/v3/api/graphql/v1/v3/query/en-GB>en-INT?filter[uriSlug]=fia-wrc-saturday-recap-estonia&rb3Schema=v1:hero
 198         'url': 'https://www.redbull.com/gb-en/live/fia-wrc-saturday-recap-estonia',
 199         'only_matching': True,
 200     }]
 201     _INT_FALLBACK_LIST = ['de', 'en', 'es', 'fr']
 202     _LAT_FALLBACK_MAP = ['ar', 'bo', 'car', 'cl', 'co', 'mx', 'pe']
 203
 204     def _real_extract(self, url):
 205         region, lang, filter_type, display_id = re.search(self._VALID_URL, url).groups()
 206         if filter_type == 'episodes':
 207             filter_type = 'episode-videos'
 208         elif filter_type == 'live':
 209             filter_type = 'live-videos'
 210
 211         regions = [region.upper()]
 212         if region != 'int':
 213             if region in self._LAT_FALLBACK_MAP:
 214                 regions.append('LAT')
 215             if lang in self._INT_FALLBACK_LIST:
 216                 regions.append('INT')
 217         locale = '>'.join(['%s-%s' % (lang, reg) for reg in regions])
 218
 219         rrn_id = self._download_json(
 220             'https://www.redbull.com/v3/api/graphql/v1/v3/query/' + locale,
 221             display_id, query={
 222                 'filter[type]': filter_type,
 223                 'filter[uriSlug]': display_id,
 224                 'rb3Schema': 'v1:hero',
 225             })['data']['id']
 226
 227         return self.url_result(
 228             'https://www.redbull.com/embed/' + rrn_id,
 229             RedBullEmbedIE.ie_key(), rrn_id)