_ Git - youtube-dl/blob - youtube_dl/extractor/parliamentliveuk.py

   1 from __future__ import unicode_literals
   2
   3 import re
   4
   5 from .common import InfoExtractor
   6 from ..utils import (
   7     ExtractorError,
   8     unified_strdate,
   9 )
  10
  11
  12 class ParliamentLiveUKIE(InfoExtractor):
  13     IE_NAME = 'parliamentlive.tv'
  14     IE_DESC = 'UK parliament videos'
  15     _VALID_URL = r'https?://www\.parliamentlive\.tv/Main/Player\.aspx\?(?:[^&]+&)*?meetingId=(?P<id>[0-9]+)'
  16
  17     _TEST = {
  18         'url': 'http://www.parliamentlive.tv/Main/Player.aspx?meetingId=15121&player=windowsmedia',
  19         'info_dict': {
  20             'id': '15121',
  21             'ext': 'asf',
  22             'title': 'hoc home affairs committee, 18 mar 2014.pm',
  23             'description': 'md5:033b3acdf83304cd43946b2d5e5798d1',
  24         },
  25         'params': {
  26             'skip_download': True,  # Requires mplayer (mms)
  27         }
  28     }
  29
  30     def _real_extract(self, url):
  31         mobj = re.match(self._VALID_URL, url)
  32         video_id = mobj.group('id')
  33         webpage = self._download_webpage(url, video_id)
  34
  35         asx_url = self._html_search_regex(
  36             r'embed.*?src="([^"]+)" name="MediaPlayer"', webpage,
  37             'metadata URL')
  38         asx = self._download_xml(asx_url, video_id, 'Downloading ASX metadata')
  39         video_url = asx.find('.//REF').attrib['HREF']
  40
  41         title = self._search_regex(
  42             r'''(?x)player\.setClipDetails\(
  43                 (?:(?:[0-9]+|"[^"]+"),\s*){2}
  44                 "([^"]+",\s*"[^"]+)"
  45                 ''',
  46             webpage, 'title').replace('", "', ', ')
  47         description = self._html_search_regex(
  48             r'(?s)<span id="MainContentPlaceHolder_CaptionsBlock_WitnessInfo">(.*?)</span>',
  49             webpage, 'description')
  50
  51         return {
  52             'id': video_id,
  53             'ext': 'asf',
  54             'url': video_url,
  55             'title': title,
  56             'description': description,
  57         }