yt_dlp/extractor/stripchat.py

   1 from .common import InfoExtractor
   2 from ..utils import ExtractorError, lowercase_escape, traverse_obj
   3
   4
   5 class StripchatIE(InfoExtractor):
   6     _VALID_URL = r'https?://stripchat\.com/(?P<id>[^/?#]+)'
   7     _TESTS = [{
   8         'url': 'https://stripchat.com/Joselin_Flower',
   9         'info_dict': {
  10             'id': 'Joselin_Flower',
  11             'ext': 'mp4',
  12             'title': 're:^Joselin_Flower [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
  13             'description': str,
  14             'is_live': True,
  15             'age_limit': 18,
  16         },
  17         'skip': 'Room is offline',
  18     }, {
  19         'url': 'https://stripchat.com/Rakhijaan@xh',
  20         'only_matching': True
  21     }]
  22
  23     def _real_extract(self, url):
  24         video_id = self._match_id(url)
  25         webpage = self._download_webpage(url, video_id, headers=self.geo_verification_headers())
  26
  27         data = self._parse_json(
  28             self._search_regex(
  29                 r'<script\b[^>]*>\s*window\.__PRELOADED_STATE__\s*=(?P<value>.*?)<\/script>',
  30                 webpage, 'data', default='{}', group='value'),
  31             video_id, transform_source=lowercase_escape, fatal=False)
  32         if not data:
  33             raise ExtractorError('Unable to find configuration for stream.')
  34
  35         if traverse_obj(data, ('viewCam', 'show'), expected_type=dict):
  36             raise ExtractorError('Model is in private show', expected=True)
  37         elif not traverse_obj(data, ('viewCam', 'model', 'isLive'), expected_type=bool):
  38             raise ExtractorError('Model is offline', expected=True)
  39
  40         server = traverse_obj(data, ('viewCam', 'viewServers', 'flashphoner-hls'), expected_type=str)
  41         model_id = traverse_obj(data, ('viewCam', 'model', 'id'), expected_type=int)
  42
  43         formats = []
  44         for host in traverse_obj(data, (
  45                 'config', 'data', (('featuresV2', 'hlsFallback', 'fallbackDomains', ...), 'hlsStreamHost'))):
  46             formats = self._extract_m3u8_formats(
  47                 f'https://b-{server}.{host}/hls/{model_id}/{model_id}.m3u8',
  48                 video_id, ext='mp4', m3u8_id='hls', fatal=False, live=True)
  49             if formats:
  50                 break
  51         if not formats:
  52             self.raise_no_formats('No active streams found', expected=True)
  53
  54         return {
  55             'id': video_id,
  56             'title': video_id,
  57             'description': self._og_search_description(webpage),
  58             'is_live': True,
  59             'formats': formats,
  60             # Stripchat declares the RTA meta-tag, but in an non-standard format so _rta_search() can't be used
  61             'age_limit': 18,
  62         }