yt_dlp/extractor/playwire.py

   1 from __future__ import unicode_literals
   2
   3
   4 from .common import InfoExtractor
   5 from ..utils import (
   6     dict_get,
   7     float_or_none,
   8 )
   9
  10
  11 class PlaywireIE(InfoExtractor):
  12     _VALID_URL = r'https?://(?:config|cdn)\.playwire\.com(?:/v2)?/(?P<publisher_id>\d+)/(?:videos/v2|embed|config)/(?P<id>\d+)'
  13     _TESTS = [{
  14         'url': 'http://config.playwire.com/14907/videos/v2/3353705/player.json',
  15         'md5': 'e6398701e3595888125729eaa2329ed9',
  16         'info_dict': {
  17             'id': '3353705',
  18             'ext': 'mp4',
  19             'title': 'S04_RM_UCL_Rus',
  20             'thumbnail': r're:^https?://.*\.png$',
  21             'duration': 145.94,
  22         },
  23     }, {
  24         # m3u8 in f4m
  25         'url': 'http://config.playwire.com/21772/videos/v2/4840492/zeus.json',
  26         'info_dict': {
  27             'id': '4840492',
  28             'ext': 'mp4',
  29             'title': 'ITV EL SHOW FULL',
  30         },
  31         'params': {
  32             # m3u8 download
  33             'skip_download': True,
  34         },
  35     }, {
  36         # Multiple resolutions while bitrates missing
  37         'url': 'http://cdn.playwire.com/11625/embed/85228.html',
  38         'only_matching': True,
  39     }, {
  40         'url': 'http://config.playwire.com/12421/videos/v2/3389892/zeus.json',
  41         'only_matching': True,
  42     }, {
  43         'url': 'http://cdn.playwire.com/v2/12342/config/1532636.json',
  44         'only_matching': True,
  45     }]
  46
  47     def _real_extract(self, url):
  48         mobj = self._match_valid_url(url)
  49         publisher_id, video_id = mobj.group('publisher_id'), mobj.group('id')
  50
  51         player = self._download_json(
  52             'http://config.playwire.com/%s/videos/v2/%s/zeus.json' % (publisher_id, video_id),
  53             video_id)
  54
  55         title = player['settings']['title']
  56         duration = float_or_none(player.get('duration'), 1000)
  57
  58         content = player['content']
  59         thumbnail = content.get('poster')
  60         src = content['media']['f4m']
  61
  62         formats = self._extract_f4m_formats(src, video_id, m3u8_id='hls')
  63         for a_format in formats:
  64             if not dict_get(a_format, ['tbr', 'width', 'height']):
  65                 a_format['quality'] = 1 if '-hd.' in a_format['url'] else 0
  66         self._sort_formats(formats)
  67
  68         return {
  69             'id': video_id,
  70             'title': title,
  71             'thumbnail': thumbnail,
  72             'duration': duration,
  73             'formats': formats,
  74         }