yt_dlp/extractor/aljazeera.py

   1 import json
   2
   3 from .common import InfoExtractor
   4 from ..utils import (
   5     try_get,
   6 )
   7
   8
   9 class AlJazeeraIE(InfoExtractor):
  10     _VALID_URL = r'https?://(?P<base>\w+\.aljazeera\.\w+)/(?P<type>programs?/[^/]+|(?:feature|video|new)s)?/\d{4}/\d{1,2}/\d{1,2}/(?P<id>[^/?&#]+)'
  11
  12     _TESTS = [{
  13         'url': 'https://balkans.aljazeera.net/videos/2021/11/6/pojedini-domovi-u-sarajevu-jos-pod-vodom-mjestanima-se-dostavlja-hrana',
  14         'info_dict': {
  15             'id': '6280641530001',
  16             'ext': 'mp4',
  17             'title': 'Pojedini domovi u Sarajevu još pod vodom, mještanima se dostavlja hrana',
  18             'timestamp': 1636219149,
  19             'description': 'U sarajevskim naseljima Rajlovac i Reljevo stambeni objekti, ali i industrijska postrojenja i dalje su pod vodom.',
  20             'upload_date': '20211106',
  21         }
  22     }, {
  23         'url': 'https://balkans.aljazeera.net/videos/2021/11/6/djokovic-usao-u-finale-mastersa-u-parizu',
  24         'info_dict': {
  25             'id': '6280654936001',
  26             'ext': 'mp4',
  27             'title': 'Đoković ušao u finale Mastersa u Parizu',
  28             'timestamp': 1636221686,
  29             'description': 'Novak Đoković je u polufinalu Mastersa u Parizu nakon preokreta pobijedio Poljaka Huberta Hurkacza.',
  30             'upload_date': '20211106',
  31         },
  32     }]
  33     BRIGHTCOVE_URL_RE = r'https?://players.brightcove.net/(?P<account>\d+)/(?P<player_id>[a-zA-Z0-9]+)_(?P<embed>[^/]+)/index.html\?videoId=(?P<id>\d+)'
  34
  35     def _real_extract(self, url):
  36         base, post_type, id = self._match_valid_url(url).groups()
  37         wp = {
  38             'balkans.aljazeera.net': 'ajb',
  39             'chinese.aljazeera.net': 'chinese',
  40             'mubasher.aljazeera.net': 'ajm',
  41         }.get(base) or 'aje'
  42         post_type = {
  43             'features': 'post',
  44             'program': 'episode',
  45             'programs': 'episode',
  46             'videos': 'video',
  47             'news': 'news',
  48         }[post_type.split('/')[0]]
  49         video = self._download_json(
  50             f'https://{base}/graphql', id, query={
  51                 'wp-site': wp,
  52                 'operationName': 'ArchipelagoSingleArticleQuery',
  53                 'variables': json.dumps({
  54                     'name': id,
  55                     'postType': post_type,
  56                 }),
  57             }, headers={
  58                 'wp-site': wp,
  59             })
  60         video = try_get(video, lambda x: x['data']['article']['video']) or {}
  61         video_id = video.get('id')
  62         account = video.get('accountId') or '911432371001'
  63         player_id = video.get('playerId') or 'csvTfAlKW'
  64         embed = 'default'
  65
  66         if video_id is None:
  67             webpage = self._download_webpage(url, id)
  68
  69             account, player_id, embed, video_id = self._search_regex(self.BRIGHTCOVE_URL_RE, webpage, 'video id',
  70                                                                      group=(1, 2, 3, 4), default=(None, None, None, None))
  71
  72             if video_id is None:
  73                 return {
  74                     '_type': 'url_transparent',
  75                     'url': url,
  76                     'ie_key': 'Generic'
  77                 }
  78
  79         return {
  80             '_type': 'url_transparent',
  81             'url': f'https://players.brightcove.net/{account}/{player_id}_{embed}/index.html?videoId={video_id}',
  82             'ie_key': 'BrightcoveNew'
  83         }