Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/redbulltv.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .common import InfoExtractor
   5 from ..utils import (
   6     float_or_none,
   7     int_or_none,
   8     try_get,
   9     unified_timestamp,
  10 )
  11
  12
  13 class RedBullTVIE(InfoExtractor):
  14     _VALID_URL = r'https?://(?:www\.)?redbull\.tv/(?:video|film)/(?P<id>AP-\w+)'
  15     _TESTS = [{
  16         # film
  17         'url': 'https://www.redbull.tv/video/AP-1Q756YYX51W11/abc-of-wrc',
  18         'md5': '78e860f631d7a846e712fab8c5fe2c38',
  19         'info_dict': {
  20             'id': 'AP-1Q756YYX51W11',
  21             'ext': 'mp4',
  22             'title': 'ABC of...WRC',
  23             'description': 'md5:5c7ed8f4015c8492ecf64b6ab31e7d31',
  24             'duration': 1582.04,
  25             'timestamp': 1488405786,
  26             'upload_date': '20170301',
  27         },
  28     }, {
  29         # episode
  30         'url': 'https://www.redbull.tv/video/AP-1PMT5JCWH1W11/grime?playlist=shows:shows-playall:web',
  31         'info_dict': {
  32             'id': 'AP-1PMT5JCWH1W11',
  33             'ext': 'mp4',
  34             'title': 'Grime - Hashtags S2 E4',
  35             'description': 'md5:334b741c8c1ce65be057eab6773c1cf5',
  36             'duration': 904.6,
  37             'timestamp': 1487290093,
  38             'upload_date': '20170217',
  39             'series': 'Hashtags',
  40             'season_number': 2,
  41             'episode_number': 4,
  42         },
  43     }, {
  44         'url': 'https://www.redbull.tv/film/AP-1MSKKF5T92111/in-motion',
  45         'only_matching': True,
  46     }]
  47
  48     def _real_extract(self, url):
  49         video_id = self._match_id(url)
  50
  51         access_token = self._download_json(
  52             'https://api-v2.redbull.tv/start', video_id,
  53             note='Downloading access token', query={
  54                 'build': '4.0.9',
  55                 'category': 'smartphone',
  56                 'os_version': 23,
  57                 'os_family': 'android',
  58             })['auth']['access_token']
  59
  60         info = self._download_json(
  61             'https://api-v2.redbull.tv/views/%s' % video_id,
  62             video_id, note='Downloading video information',
  63             headers={'Authorization': 'Bearer ' + access_token}
  64         )['blocks'][0]['top'][0]
  65
  66         video = info['video_product']
  67
  68         title = info['title'].strip()
  69         m3u8_url = video['url']
  70
  71         formats = self._extract_m3u8_formats(
  72             m3u8_url, video_id, 'mp4', entry_protocol='m3u8_native',
  73             m3u8_id='hls')
  74
  75         subtitles = {}
  76         for _, captions in (try_get(
  77                 video, lambda x: x['attachments']['captions'],
  78                 dict) or {}).items():
  79             if not captions or not isinstance(captions, list):
  80                 continue
  81             for caption in captions:
  82                 caption_url = caption.get('url')
  83                 if not caption_url:
  84                     continue
  85                 subtitles.setdefault(caption.get('lang') or 'en', []).append({
  86                     'url': caption_url,
  87                     'ext': caption.get('format'),
  88                 })
  89
  90         subheading = info.get('subheading')
  91         if subheading:
  92             title += ' - %s' % subheading
  93
  94         return {
  95             'id': video_id,
  96             'title': title,
  97             'description': info.get('long_description') or info.get(
  98                 'short_description'),
  99             'duration': float_or_none(video.get('duration'), scale=1000),
 100             'timestamp': unified_timestamp(info.get('published')),
 101             'series': info.get('show_title'),
 102             'season_number': int_or_none(info.get('season_number')),
 103             'episode_number': int_or_none(info.get('episode_number')),
 104             'formats': formats,
 105             'subtitles': subtitles,
 106         }