yt_dlp/extractor/vodplatform.py

   1 from .common import InfoExtractor
   2 from ..utils import unescapeHTML
   3
   4
   5 class VODPlatformIE(InfoExtractor):
   6     _VALID_URL = r'https?://(?:(?:www\.)?vod-platform\.net|embed\.kwikmotion\.com)/[eE]mbed/(?P<id>[^/?#]+)'
   7     _EMBED_REGEX = [r'<iframe[^>]+src=(["\'])(?P<url>(?:https?:)?//(?:(?:www\.)?vod-platform\.net|embed\.kwikmotion\.com)/[eE]mbed/.+?)\1']
   8     _TESTS = [{
   9         # from http://www.lbcgroup.tv/watch/chapter/29143/52844/%D8%A7%D9%84%D9%86%D8%B5%D8%B1%D8%A9-%D9%81%D9%8A-%D8%B6%D9%8A%D8%A7%D9%81%D8%A9-%D8%A7%D9%84%D9%80-cnn/ar
  10         'url': 'http://vod-platform.net/embed/RufMcytHDolTH1MuKHY9Fw',
  11         'md5': '1db2b7249ce383d6be96499006e951fc',
  12         'info_dict': {
  13             'id': 'RufMcytHDolTH1MuKHY9Fw',
  14             'ext': 'mp4',
  15             'title': 'LBCi News_ النصرة في ضيافة الـ "سي.أن.أن"',
  16         },
  17     }, {
  18         'url': 'http://embed.kwikmotion.com/embed/RufMcytHDolTH1MuKHY9Fw',
  19         'only_matching': True,
  20     }]
  21
  22     def _real_extract(self, url):
  23         video_id = self._match_id(url)
  24         webpage = self._download_webpage(url, video_id)
  25
  26         title = unescapeHTML(self._og_search_title(webpage))
  27         hidden_inputs = self._hidden_inputs(webpage)
  28
  29         formats = self._extract_wowza_formats(
  30             hidden_inputs.get('HiddenmyhHlsLink') or hidden_inputs['HiddenmyDashLink'], video_id, skip_protocols=['f4m', 'smil'])
  31
  32         return {
  33             'id': video_id,
  34             'title': title,
  35             'thumbnail': hidden_inputs.get('HiddenThumbnail') or self._og_search_thumbnail(webpage),
  36             'formats': formats,
  37         }