yt_dlp/extractor/aljazeera.py

   1 from __future__ import unicode_literals
   2
   3 import json
   4
   5 from .common import InfoExtractor
   6
   7
   8 class AlJazeeraIE(InfoExtractor):
   9     _VALID_URL = r'https?://(?:www\.)?aljazeera\.com/(?P<type>program/[^/]+|(?:feature|video)s)/\d{4}/\d{1,2}/\d{1,2}/(?P<id>[^/?&#]+)'
  10
  11     _TESTS = [{
  12         'url': 'https://www.aljazeera.com/program/episode/2014/9/19/deliverance',
  13         'info_dict': {
  14             'id': '3792260579001',
  15             'ext': 'mp4',
  16             'title': 'The Slum - Episode 1: Deliverance',
  17             'description': 'As a birth attendant advocating for family planning, Remy is on the frontline of Tondo\'s battle with overcrowding.',
  18             'uploader_id': '665003303001',
  19             'timestamp': 1411116829,
  20             'upload_date': '20140919',
  21         },
  22         'add_ie': ['BrightcoveNew'],
  23         'skip': 'Not accessible from Travis CI server',
  24     }, {
  25         'url': 'https://www.aljazeera.com/videos/2017/5/11/sierra-leone-709-carat-diamond-to-be-auctioned-off',
  26         'only_matching': True,
  27     }, {
  28         'url': 'https://www.aljazeera.com/features/2017/8/21/transforming-pakistans-buses-into-art',
  29         'only_matching': True,
  30     }]
  31     BRIGHTCOVE_URL_TEMPLATE = 'http://players.brightcove.net/%s/%s_default/index.html?videoId=%s'
  32
  33     def _real_extract(self, url):
  34         post_type, name = self._match_valid_url(url).groups()
  35         post_type = {
  36             'features': 'post',
  37             'program': 'episode',
  38             'videos': 'video',
  39         }[post_type.split('/')[0]]
  40         video = self._download_json(
  41             'https://www.aljazeera.com/graphql', name, query={
  42                 'operationName': 'SingleArticleQuery',
  43                 'variables': json.dumps({
  44                     'name': name,
  45                     'postType': post_type,
  46                 }),
  47             }, headers={
  48                 'wp-site': 'aje',
  49             })['data']['article']['video']
  50         video_id = video['id']
  51         account_id = video.get('accountId') or '665003303001'
  52         player_id = video.get('playerId') or 'BkeSH5BDb'
  53         return self.url_result(
  54             self.BRIGHTCOVE_URL_TEMPLATE % (account_id, player_id, video_id),
  55             'BrightcoveNew', video_id)