youtube_dl/extractor/southparkstudios.py

   1 import re
   2
   3 from .mtv import MTVIE, _media_xml_tag
   4
   5
   6 class SouthParkStudiosIE(MTVIE):
   7     IE_NAME = u'southparkstudios.com'
   8     _VALID_URL = r'(https?://)?(www\.)?(?P<url>southparkstudios\.com/(clips|full-episodes)/(?P<id>.+?)(\?|#|$))'
   9
  10     _FEED_URL = 'http://www.southparkstudios.com/feeds/video-player/mrss'
  11
  12     # Overwrite MTVIE properties we don't want
  13     _TESTS = [{
  14         u'url': u'http://www.southparkstudios.com/clips/104437/bat-daded#tab=featured',
  15         u'file': u'a7bff6c2-ed00-11e0-aca6-0026b9414f30.mp4',
  16         u'info_dict': {
  17             u'title': u'Bat Daded',
  18             u'description': u'Randy disqualifies South Park by getting into a fight with Bat Dad.',
  19         },
  20     }]
  21
  22     def _get_thumbnail_url(self, uri, itemdoc):
  23         search_path = '%s/%s' % (_media_xml_tag('group'), _media_xml_tag('thumbnail'))
  24         thumb_node = itemdoc.find(search_path)
  25         if thumb_node is None:
  26             return None
  27         else:
  28             return thumb_node.attrib['url']
  29
  30     def _real_extract(self, url):
  31         mobj = re.match(self._VALID_URL, url)
  32         url = u'http://www.' + mobj.group(u'url')
  33         video_id = mobj.group('id')
  34         webpage = self._download_webpage(url, video_id)
  35         mgid = self._search_regex(r'swfobject.embedSWF\(".*?(mgid:.*?)"',
  36                                   webpage, u'mgid')
  37         return self._get_videos_info(mgid)
  38
  39 class SouthparkDeIE(SouthParkStudiosIE):
  40     IE_NAME = u'southpark.de'
  41     _VALID_URL = r'(https?://)?(www\.)?(?P<url>southpark\.de/(clips|alle-episoden)/(?P<id>.+?)(\?|#|$))'
  42     _FEED_URL = 'http://www.southpark.de/feeds/video-player/mrss/'
  43
  44     _TESTS = [{
  45         u'url': u'http://www.southpark.de/clips/uygssh/the-government-wont-respect-my-privacy#tab=featured',
  46         u'file': u'85487c96-b3b9-4e39-9127-ad88583d9bf2.mp4',
  47         u'info_dict': {
  48             u'title': u'The Government Won\'t Respect My Privacy',
  49             u'description': u'Cartman explains the benefits of "Shitter" to Stan, Kyle and Craig.',
  50         },
  51     }]