yt_dlp/extractor/rutv.py

   1 import re
   2
   3 from .common import InfoExtractor
   4 from ..utils import (
   5     ExtractorError,
   6     int_or_none,
   7     str_to_int
   8 )
   9
  10
  11 class RUTVIE(InfoExtractor):
  12     IE_DESC = 'RUTV.RU'
  13     _VALID_URL = r'''(?x)
  14                     https?://
  15                         (?:test)?player\.(?:rutv\.ru|vgtrk\.com)/
  16                         (?P<path>
  17                             flash\d+v/container\.swf\?id=|
  18                             iframe/(?P<type>swf|video|live)/id/|
  19                             index/iframe/cast_id/
  20                         )
  21                         (?P<id>\d+)
  22                     '''
  23     _EMBED_URLS = [
  24         r'<iframe[^>]+?src=(["\'])(?P<url>https?://(?:test)?player\.(?:rutv\.ru|vgtrk\.com)/(?:iframe/(?:swf|video|live)/id|index/iframe/cast_id)/.+?)\1',
  25         r'<meta[^>]+?property=(["\'])og:video\1[^>]+?content=(["\'])(?P<url>https?://(?:test)?player\.(?:rutv\.ru|vgtrk\.com)/flash\d+v/container\.swf\?id=.+?\2)',
  26     ]
  27
  28     _TESTS = [
  29         {
  30             'url': 'http://player.rutv.ru/flash2v/container.swf?id=774471&sid=kultura&fbv=true&isPlay=true&ssl=false&i=560&acc_video_id=episode_id/972347/video_id/978186/brand_id/31724',
  31             'info_dict': {
  32                 'id': '774471',
  33                 'ext': 'mp4',
  34                 'title': 'Монологи на все времена',
  35                 'description': 'md5:18d8b5e6a41fb1faa53819471852d5d5',
  36                 'duration': 2906,
  37             },
  38             'params': {
  39                 # m3u8 download
  40                 'skip_download': True,
  41             },
  42         },
  43         {
  44             'url': 'https://player.vgtrk.com/flash2v/container.swf?id=774016&sid=russiatv&fbv=true&isPlay=true&ssl=false&i=560&acc_video_id=episode_id/972098/video_id/977760/brand_id/57638',
  45             'info_dict': {
  46                 'id': '774016',
  47                 'ext': 'mp4',
  48                 'title': 'Чужой в семье Сталина',
  49                 'description': '',
  50                 'duration': 2539,
  51             },
  52             'params': {
  53                 # m3u8 download
  54                 'skip_download': True,
  55             },
  56         },
  57         {
  58             'url': 'http://player.rutv.ru/iframe/swf/id/766888/sid/hitech/?acc_video_id=4000',
  59             'info_dict': {
  60                 'id': '766888',
  61                 'ext': 'mp4',
  62                 'title': 'Вести.net: интернет-гиганты начали перетягивание программных "одеял"',
  63                 'description': 'md5:65ddd47f9830c4f42ed6475f8730c995',
  64                 'duration': 279,
  65             },
  66             'params': {
  67                 # m3u8 download
  68                 'skip_download': True,
  69             },
  70         },
  71         {
  72             'url': 'http://player.rutv.ru/iframe/video/id/771852/start_zoom/true/showZoomBtn/false/sid/russiatv/?acc_video_id=episode_id/970443/video_id/975648/brand_id/5169',
  73             'info_dict': {
  74                 'id': '771852',
  75                 'ext': 'mp4',
  76                 'title': 'Прямой эфир. Жертвы загадочной болезни: смерть от старости в 17 лет',
  77                 'description': 'md5:b81c8c55247a4bd996b43ce17395b2d8',
  78                 'duration': 3096,
  79             },
  80             'params': {
  81                 # m3u8 download
  82                 'skip_download': True,
  83             },
  84         },
  85         {
  86             'url': 'http://player.rutv.ru/iframe/live/id/51499/showZoomBtn/false/isPlay/true/sid/sochi2014',
  87             'info_dict': {
  88                 'id': '51499',
  89                 'ext': 'flv',
  90                 'title': 'Сочи-2014. Биатлон. Индивидуальная гонка. Мужчины ',
  91                 'description': 'md5:9e0ed5c9d2fa1efbfdfed90c9a6d179c',
  92             },
  93             'skip': 'Translation has finished',
  94         },
  95         {
  96             'url': 'http://player.rutv.ru/iframe/live/id/21/showZoomBtn/false/isPlay/true/',
  97             'info_dict': {
  98                 'id': '21',
  99                 'ext': 'mp4',
 100                 'title': 're:^Россия 24. Прямой эфир [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
 101                 'is_live': True,
 102             },
 103             'params': {
 104                 # m3u8 download
 105                 'skip_download': True,
 106             },
 107         },
 108         {
 109             'url': 'https://testplayer.vgtrk.com/iframe/live/id/19201/showZoomBtn/false/isPlay/true/',
 110             'only_matching': True,
 111         },
 112     ]
 113
 114     def _real_extract(self, url):
 115         mobj = self._match_valid_url(url)
 116         video_id = mobj.group('id')
 117         video_path = mobj.group('path')
 118
 119         if re.match(r'flash\d+v', video_path):
 120             video_type = 'video'
 121         elif video_path.startswith('iframe'):
 122             video_type = mobj.group('type')
 123             if video_type == 'swf':
 124                 video_type = 'video'
 125         elif video_path.startswith('index/iframe/cast_id'):
 126             video_type = 'live'
 127
 128         is_live = video_type == 'live'
 129
 130         json_data = self._download_json(
 131             'http://player.vgtrk.com/iframe/data%s/id/%s' % ('live' if is_live else 'video', video_id),
 132             video_id, 'Downloading JSON')
 133
 134         if json_data['errors']:
 135             raise ExtractorError('%s said: %s' % (self.IE_NAME, json_data['errors']), expected=True)
 136
 137         playlist = json_data['data']['playlist']
 138         medialist = playlist['medialist']
 139         media = medialist[0]
 140
 141         if media['errors']:
 142             raise ExtractorError('%s said: %s' % (self.IE_NAME, media['errors']), expected=True)
 143
 144         view_count = playlist.get('count_views')
 145         priority_transport = playlist['priority_transport']
 146
 147         thumbnail = media['picture']
 148         width = int_or_none(media['width'])
 149         height = int_or_none(media['height'])
 150         description = media['anons']
 151         title = media['title']
 152         duration = int_or_none(media.get('duration'))
 153
 154         formats = []
 155
 156         for transport, links in media['sources'].items():
 157             for quality, url in links.items():
 158                 preference = -1 if priority_transport == transport else -2
 159                 if transport == 'rtmp':
 160                     mobj = re.search(r'^(?P<url>rtmp://[^/]+/(?P<app>.+))/(?P<playpath>.+)$', url)
 161                     if not mobj:
 162                         continue
 163                     fmt = {
 164                         'url': mobj.group('url'),
 165                         'play_path': mobj.group('playpath'),
 166                         'app': mobj.group('app'),
 167                         'page_url': 'http://player.rutv.ru',
 168                         'player_url': 'http://player.rutv.ru/flash3v/osmf.swf?i=22',
 169                         'rtmp_live': True,
 170                         'ext': 'flv',
 171                         'vbr': str_to_int(quality),
 172                     }
 173                 elif transport == 'm3u8':
 174                     formats.extend(self._extract_m3u8_formats(
 175                         url, video_id, 'mp4', quality=preference, m3u8_id='hls'))
 176                     continue
 177                 else:
 178                     fmt = {
 179                         'url': url
 180                     }
 181                 fmt.update({
 182                     'width': int_or_none(quality, default=height, invscale=width, scale=height),
 183                     'height': int_or_none(quality, default=height),
 184                     'format_id': '%s-%s' % (transport, quality),
 185                     'source_preference': preference,
 186                 })
 187                 formats.append(fmt)
 188
 189         self._sort_formats(formats)
 190
 191         return {
 192             'id': video_id,
 193             'title': title,
 194             'description': description,
 195             'thumbnail': thumbnail,
 196             'view_count': view_count,
 197             'duration': duration,
 198             'formats': formats,
 199             'is_live': is_live,
 200         }