yt_dlp/extractor/israelnationalnews.py

   1 from .common import InfoExtractor
   2 from ..utils import ExtractorError, traverse_obj
   3
   4
   5 class IsraelNationalNewsIE(InfoExtractor):
   6     _VALID_URL = r'https?://(?:www\.)?israelnationalnews\.com/news/(?P<id>\d+)'
   7     _TESTS = [{
   8         'url': 'https://www.israelnationalnews.com/news/354520',
   9         'info_dict': {
  10             'id': '354520',
  11         },
  12         'playlist': [{
  13             'info_dict': {
  14                 'id': 'jA84wQhVvg8',
  15                 'title': 'Even CNN Host Is Shocked by How Bad Biden\'s Approval Ratings Have Gotten | DM CLIPS | Rubin Report',
  16                 'ext': 'mp4',
  17                 'description': 'md5:b7325a3d00c7596337dc3ae37e32d35c',
  18                 'channel': 'The Rubin Report',
  19                 'channel_follower_count': int,
  20                 'comment_count': int,
  21                 'categories': ['News & Politics'],
  22                 'like_count': int,
  23                 'uploader_url': 'http://www.youtube.com/user/RubinReport',
  24                 'uploader_id': 'RubinReport',
  25                 'availability': 'public',
  26                 'view_count': int,
  27                 'duration': 240,
  28                 'thumbnail': 'https://i.ytimg.com/vi_webp/jA84wQhVvg8/maxresdefault.webp',
  29                 'live_status': 'not_live',
  30                 'playable_in_embed': True,
  31                 'age_limit': 0,
  32                 'tags': 'count:29',
  33                 'channel_id': 'UCJdKr0Bgd_5saZYqLCa9mng',
  34                 'channel_url': 'https://www.youtube.com/channel/UCJdKr0Bgd_5saZYqLCa9mng',
  35                 'upload_date': '20220606',
  36                 'uploader': 'The Rubin Report',
  37             },
  38         }],
  39     }]
  40
  41     def _real_extract(self, url):
  42         news_article_id = self._match_id(url)
  43         article_json = self._download_json(
  44             f'https://www.israelnationalnews.com/Generic/NewAPI/Item?type=0&Item={news_article_id}', news_article_id)
  45
  46         urls = traverse_obj(article_json, ('Content2', ..., 'content', ..., 'attrs', 'src'))
  47         if not urls:
  48             raise ExtractorError('This article does not have any videos', expected=True)
  49
  50         return self.playlist_from_matches(urls, news_article_id, ie='Youtube')