youtube_dl/extractor/phoenix.py

   1 from __future__ import unicode_literals
   2
   3 from .zdf import ZDFIE
   4
   5
   6 class PhoenixIE(ZDFIE):
   7     _VALID_URL = r'''(?x)https?://(?:www\.)?phoenix\.de/content/
   8         (?:
   9             phoenix/die_sendungen/(?:[^/]+/)?
  10         )?
  11         (?P<id>[0-9]+)'''
  12     _TESTS = [
  13         {
  14             'url': 'http://www.phoenix.de/content/884301',
  15             'md5': 'ed249f045256150c92e72dbb70eadec6',
  16             'info_dict': {
  17                 'id': '884301',
  18                 'ext': 'mp4',
  19                 'title': 'Michael Krons mit Hans-Werner Sinn',
  20                 'description': 'Im Dialog - Sa. 25.10.14, 00.00 - 00.35 Uhr',
  21                 'upload_date': '20141025',
  22                 'uploader': 'Im Dialog',
  23             }
  24         },
  25         {
  26             'url': 'http://www.phoenix.de/content/phoenix/die_sendungen/869815',
  27             'only_matching': True,
  28         },
  29         {
  30             'url': 'http://www.phoenix.de/content/phoenix/die_sendungen/diskussionen/928234',
  31             'only_matching': True,
  32         },
  33     ]
  34
  35     def _real_extract(self, url):
  36         video_id = self._match_id(url)
  37         webpage = self._download_webpage(url, video_id)
  38
  39         internal_id = self._search_regex(
  40             r'<div class="phx_vod" id="phx_vod_([0-9]+)"',
  41             webpage, 'internal video ID')
  42
  43         api_url = 'http://www.phoenix.de/php/mediaplayer/data/beitrags_details.php?ak=web&id=%s' % internal_id
  44         return self.extract_from_xml_url(video_id, api_url)