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