youtube_dl/extractor/playwire.py

   1 from __future__ import unicode_literals
   2
   3 import re
   4
   5 from .common import InfoExtractor
   6 from ..utils import (
   7     dict_get,
   8     float_or_none,
   9 )
  10
  11
  12 class PlaywireIE(InfoExtractor):
  13     _VALID_URL = r'https?://(?:config|cdn)\.playwire\.com(?:/v2)?/(?P<publisher_id>\d+)/(?:videos/v2|embed|config)/(?P<id>\d+)'
  14     _TESTS = [{
  15         'url': 'http://config.playwire.com/14907/videos/v2/3353705/player.json',
  16         'md5': 'e6398701e3595888125729eaa2329ed9',
  17         'info_dict': {
  18             'id': '3353705',
  19             'ext': 'mp4',
  20             'title': 'S04_RM_UCL_Rus',
  21             'thumbnail': r're:^https?://.*\.png$',
  22             'duration': 145.94,
  23         },
  24     }, {
  25         # m3u8 in f4m
  26         'url': 'http://config.playwire.com/21772/videos/v2/4840492/zeus.json',
  27         'info_dict': {
  28             'id': '4840492',
  29             'ext': 'mp4',
  30             'title': 'ITV EL SHOW FULL',
  31         },
  32         'params': {
  33             # m3u8 download
  34             'skip_download': True,
  35         },
  36     }, {
  37         # Multiple resolutions while bitrates missing
  38         'url': 'http://cdn.playwire.com/11625/embed/85228.html',
  39         'only_matching': True,
  40     }, {
  41         'url': 'http://config.playwire.com/12421/videos/v2/3389892/zeus.json',
  42         'only_matching': True,
  43     }, {
  44         'url': 'http://cdn.playwire.com/v2/12342/config/1532636.json',
  45         'only_matching': True,
  46     }]
  47
  48     def _real_extract(self, url):
  49         mobj = re.match(self._VALID_URL, url)
  50         publisher_id, video_id = mobj.group('publisher_id'), mobj.group('id')
  51
  52         player = self._download_json(
  53             'http://config.playwire.com/%s/videos/v2/%s/zeus.json' % (publisher_id, video_id),
  54             video_id)
  55
  56         title = player['settings']['title']
  57         duration = float_or_none(player.get('duration'), 1000)
  58
  59         content = player['content']
  60         thumbnail = content.get('poster')
  61         src = content['media']['f4m']
  62
  63         formats = self._extract_f4m_formats(src, video_id, m3u8_id='hls')
  64         for a_format in formats:
  65             if not dict_get(a_format, ['tbr', 'width', 'height']):
  66                 a_format['quality'] = 1 if '-hd.' in a_format['url'] else 0
  67         self._sort_formats(formats)
  68
  69         return {
  70             'id': video_id,
  71             'title': title,
  72             'thumbnail': thumbnail,
  73             'duration': duration,
  74             'formats': formats,
  75         }