yt_dlp/extractor/playwire.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     dict_get,
   4     float_or_none,
   5 )
   6
   7
   8 class PlaywireIE(InfoExtractor):
   9     _VALID_URL = r'https?://(?:config|cdn)\.playwire\.com(?:/v2)?/(?P<publisher_id>\d+)/(?:videos/v2|embed|config)/(?P<id>\d+)'
  10     _TESTS = [{
  11         'url': 'http://config.playwire.com/14907/videos/v2/3353705/player.json',
  12         'md5': 'e6398701e3595888125729eaa2329ed9',
  13         'info_dict': {
  14             'id': '3353705',
  15             'ext': 'mp4',
  16             'title': 'S04_RM_UCL_Rus',
  17             'thumbnail': r're:^https?://.*\.png$',
  18             'duration': 145.94,
  19         },
  20     }, {
  21         # m3u8 in f4m
  22         'url': 'http://config.playwire.com/21772/videos/v2/4840492/zeus.json',
  23         'info_dict': {
  24             'id': '4840492',
  25             'ext': 'mp4',
  26             'title': 'ITV EL SHOW FULL',
  27         },
  28         'params': {
  29             # m3u8 download
  30             'skip_download': True,
  31         },
  32     }, {
  33         # Multiple resolutions while bitrates missing
  34         'url': 'http://cdn.playwire.com/11625/embed/85228.html',
  35         'only_matching': True,
  36     }, {
  37         'url': 'http://config.playwire.com/12421/videos/v2/3389892/zeus.json',
  38         'only_matching': True,
  39     }, {
  40         'url': 'http://cdn.playwire.com/v2/12342/config/1532636.json',
  41         'only_matching': True,
  42     }]
  43
  44     def _real_extract(self, url):
  45         mobj = self._match_valid_url(url)
  46         publisher_id, video_id = mobj.group('publisher_id'), mobj.group('id')
  47
  48         player = self._download_json(
  49             'http://config.playwire.com/%s/videos/v2/%s/zeus.json' % (publisher_id, video_id),
  50             video_id)
  51
  52         title = player['settings']['title']
  53         duration = float_or_none(player.get('duration'), 1000)
  54
  55         content = player['content']
  56         thumbnail = content.get('poster')
  57         src = content['media']['f4m']
  58
  59         formats = self._extract_f4m_formats(src, video_id, m3u8_id='hls')
  60         for a_format in formats:
  61             if not dict_get(a_format, ['tbr', 'width', 'height']):
  62                 a_format['quality'] = 1 if '-hd.' in a_format['url'] else 0
  63         self._sort_formats(formats)
  64
  65         return {
  66             'id': video_id,
  67             'title': title,
  68             'thumbnail': thumbnail,
  69             'duration': duration,
  70             'formats': formats,
  71         }