yt_dlp/extractor/yapfiles.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     ExtractorError,
   4     int_or_none,
   5     qualities,
   6     url_or_none,
   7 )
   8
   9
  10 class YapFilesIE(InfoExtractor):
  11     _WORKING = False
  12     _YAPFILES_URL = r'//(?:(?:www|api)\.)?yapfiles\.ru/get_player/*\?.*?\bv=(?P<id>\w+)'
  13     _VALID_URL = rf'https?:{_YAPFILES_URL}'
  14     _EMBED_REGEX = [rf'<iframe\b[^>]+\bsrc=(["\'])(?P<url>(?:https?:)?{_YAPFILES_URL}.*?)\1']
  15     _TESTS = [{
  16         # with hd
  17         'url': 'http://www.yapfiles.ru/get_player/?v=vMDE1NjcyNDUt0413',
  18         'md5': '2db19e2bfa2450568868548a1aa1956c',
  19         'info_dict': {
  20             'id': 'vMDE1NjcyNDUt0413',
  21             'ext': 'mp4',
  22             'title': 'Самый худший пароль WIFI',
  23             'thumbnail': r're:^https?://.*\.jpg$',
  24             'duration': 72,
  25         },
  26     }, {
  27         # without hd
  28         'url': 'https://api.yapfiles.ru/get_player/?uid=video_player_1872528&plroll=1&adv=1&v=vMDE4NzI1Mjgt690b',
  29         'only_matching': True,
  30     }]
  31
  32     def _real_extract(self, url):
  33         video_id = self._match_id(url)
  34
  35         webpage = self._download_webpage(url, video_id, fatal=False)
  36
  37         player_url = None
  38         query = {}
  39         if webpage:
  40             player_url = self._search_regex(
  41                 r'player\.init\s*\(\s*(["\'])(?P<url>(?:(?!\1).)+)\1', webpage,
  42                 'player url', default=None, group='url')
  43
  44         if not player_url:
  45             player_url = f'http://api.yapfiles.ru/load/{video_id}/'
  46             query = {
  47                 'md5': 'ded5f369be61b8ae5f88e2eeb2f3caff',
  48                 'type': 'json',
  49                 'ref': url,
  50             }
  51
  52         player = self._download_json(
  53             player_url, video_id, query=query)['player']
  54
  55         playlist_url = player['playlist']
  56         title = player['title']
  57         thumbnail = player.get('poster')
  58
  59         if title == 'Ролик удален' or 'deleted.jpg' in (thumbnail or ''):
  60             raise ExtractorError(
  61                 f'Video {video_id} has been removed', expected=True)
  62
  63         playlist = self._download_json(
  64             playlist_url, video_id)['player']['main']
  65
  66         hd_height = int_or_none(player.get('hd'))
  67
  68         QUALITIES = ('sd', 'hd')
  69         quality_key = qualities(QUALITIES)
  70         formats = []
  71         for format_id in QUALITIES:
  72             is_hd = format_id == 'hd'
  73             format_url = url_or_none(playlist.get(
  74                 'file%s' % ('_hd' if is_hd else '')))
  75             if not format_url:
  76                 continue
  77             formats.append({
  78                 'url': format_url,
  79                 'format_id': format_id,
  80                 'quality': quality_key(format_id),
  81                 'height': hd_height if is_hd else None,
  82             })
  83
  84         return {
  85             'id': video_id,
  86             'title': title,
  87             'thumbnail': thumbnail,
  88             'duration': int_or_none(player.get('length')),
  89             'formats': formats,
  90         }