Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/ina.py

   1 import re
   2
   3 from .common import InfoExtractor
   4
   5
   6 class InaIE(InfoExtractor):
   7     """Information Extractor for Ina.fr"""
   8     _VALID_URL = r'(?:http://)?(?:www\.)?ina\.fr/video/(?P<id>I[0-9]+)/.*'
   9
  10     def _real_extract(self,url):
  11         mobj = re.match(self._VALID_URL, url)
  12
  13         video_id = mobj.group('id')
  14         mrss_url='http://player.ina.fr/notices/%s.mrss' % video_id
  15         video_extension = 'mp4'
  16         webpage = self._download_webpage(mrss_url, video_id)
  17
  18         self.report_extraction(video_id)
  19
  20         video_url = self._html_search_regex(r'<media:player url="(?P<mp4url>http://mp4.ina.fr/[^"]+\.mp4)',
  21             webpage, u'video URL')
  22
  23         video_title = self._search_regex(r'<title><!\[CDATA\[(?P<titre>.*?)]]></title>',
  24             webpage, u'title')
  25
  26         return [{
  27             'id':       video_id,
  28             'url':      video_url,
  29             'ext':      video_extension,
  30             'title':    video_title,
  31         }]