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

   1 from .common import InfoExtractor
   2 from ..utils import get_element_by_id
   3
   4 class ShahidIE(InfoExtractor):
   5     _VALID_URL = r'https?://shahid\.mbc\.net/ar/episode/(?P<id>\d+)/?'
   6     _TESTS = [
   7         {
   8             'url': 'https://shahid.mbc.net/ar/episode/108084/%D8%AE%D9%88%D8%A7%D8%B7%D8%B1-%D8%A7%D9%84%D9%85%D9%88%D8%B3%D9%85-11-%D8%A7%D9%84%D8%AD%D9%84%D9%82%D8%A9-1.html',
   9             'info_dict': {
  10                 'id': '108084',
  11                 'ext': 'm3u8',
  12                 'title': 'بسم الله',
  13                 'description': 'بسم الله'
  14             },
  15             'params': {
  16                 # m3u8 download
  17                 'skip_download': True,
  18             }
  19         },
  20         {
  21             #shahid plus subscriber only
  22             'url': 'https://shahid.mbc.net/ar/series/90497/%D9%85%D8%B1%D8%A7%D9%8A%D8%A7-2011.html',
  23             'only_matching': True
  24         }
  25     ]
  26
  27     def _real_extract(self, url):
  28         video_id = self._match_id(url)
  29         webpage = self._download_webpage(url, video_id)
  30         json_data = self._parse_json(
  31             get_element_by_id('jsonld', webpage),
  32             video_id
  33         )
  34         title = json_data['name']
  35         thumbnail = json_data['image']
  36         categories = json_data['genre']
  37         description = json_data['description']
  38         player_json_data = self._download_json(
  39             'https://shahid.mbc.net/arContent/getPlayerContent-param-.id-'+video_id+'.type-player.html',
  40             video_id
  41         )['data']
  42         if 'url' in player_json_data:
  43             m3u8_url = player_json_data['url']
  44         else:
  45             for error in json_data['error'].values():
  46                 self.report_warning(error)
  47             return
  48         formats = self._extract_m3u8_formats(m3u8_url, video_id)
  49         return {
  50             'id': video_id,
  51             'title': title,
  52             'thumbnail': thumbnail,
  53             'categories': categories,
  54             'description': description,
  55             'formats': formats,
  56         }