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     xpath_text,
   8     float_or_none,
   9     int_or_none,
  10 )
  11
  12
  13 class PlaywireIE(InfoExtractor):
  14     _VALID_URL = r'https?://(?:config|cdn)\.playwire\.com(?:/v2)?/(?P<publisher_id>\d+)/(?:videos/v2|embed|config)/(?P<id>\d+)'
  15     _TESTS = [{
  16         'url': 'http://config.playwire.com/14907/videos/v2/3353705/player.json',
  17         'md5': 'e6398701e3595888125729eaa2329ed9',
  18         'info_dict': {
  19             'id': '3353705',
  20             'ext': 'mp4',
  21             'title': 'S04_RM_UCL_Rus',
  22             'thumbnail': 're:^http://.*\.png$',
  23             'duration': 145.94,
  24         },
  25     }, {
  26         'url': 'http://cdn.playwire.com/11625/embed/85228.html',
  27         'only_matching': True,
  28     }, {
  29         'url': 'http://config.playwire.com/12421/videos/v2/3389892/zeus.json',
  30         'only_matching': True,
  31     }, {
  32         'url': 'http://cdn.playwire.com/v2/12342/config/1532636.json',
  33         'only_matching': True,
  34     }]
  35
  36     def _real_extract(self, url):
  37         mobj = re.match(self._VALID_URL, url)
  38         publisher_id, video_id = mobj.group('publisher_id'), mobj.group('id')
  39
  40         player = self._download_json(
  41             'http://config.playwire.com/%s/videos/v2/%s/zeus.json' % (publisher_id, video_id),
  42             video_id)
  43
  44         title = player['settings']['title']
  45         duration = float_or_none(player.get('duration'), 1000)
  46
  47         content = player['content']
  48         thumbnail = content.get('poster')
  49         src = content['media']['f4m']
  50
  51         f4m = self._download_xml(src, video_id)
  52         base_url = xpath_text(f4m, './{http://ns.adobe.com/f4m/1.0}baseURL', 'base url', fatal=True)
  53         formats = []
  54         for media in f4m.findall('./{http://ns.adobe.com/f4m/1.0}media'):
  55             media_url = media.get('url')
  56             if not media_url:
  57                 continue
  58             tbr = int_or_none(media.get('bitrate'))
  59             width = int_or_none(media.get('width'))
  60             height = int_or_none(media.get('height'))
  61             f = {
  62                 'url': '%s/%s' % (base_url, media.attrib['url']),
  63                 'tbr': tbr,
  64                 'width': width,
  65                 'height': height,
  66             }
  67             if not (tbr or width or height):
  68                 f['quality'] = 1 if '-hd.' in media_url else 0
  69             formats.append(f)
  70         self._sort_formats(formats)
  71
  72         return {
  73             'id': video_id,
  74             'title': title,
  75             'thumbnail': thumbnail,
  76             'duration': duration,
  77             'formats': formats,
  78         }