yt_dlp/extractor/stripchat.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     ExtractorError,
   4     UserNotLive,
   5     lowercase_escape,
   6     traverse_obj,
   7 )
   8
   9
  10 class StripchatIE(InfoExtractor):
  11     _VALID_URL = r'https?://stripchat\.com/(?P<id>[^/?#]+)'
  12     _TESTS = [{
  13         'url': 'https://stripchat.com/Joselin_Flower',
  14         'info_dict': {
  15             'id': 'Joselin_Flower',
  16             'ext': 'mp4',
  17             'title': 're:^Joselin_Flower [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
  18             'description': str,
  19             'is_live': True,
  20             'age_limit': 18,
  21         },
  22         'skip': 'Room is offline',
  23     }, {
  24         'url': 'https://stripchat.com/Rakhijaan@xh',
  25         'only_matching': True
  26     }]
  27
  28     def _real_extract(self, url):
  29         video_id = self._match_id(url)
  30         webpage = self._download_webpage(url, video_id, headers=self.geo_verification_headers())
  31
  32         data = self._parse_json(
  33             self._search_regex(
  34                 r'<script\b[^>]*>\s*window\.__PRELOADED_STATE__\s*=(?P<value>.*?)<\/script>',
  35                 webpage, 'data', default='{}', group='value'),
  36             video_id, transform_source=lowercase_escape, fatal=False)
  37         if not data:
  38             raise ExtractorError('Unable to find configuration for stream.')
  39
  40         if traverse_obj(data, ('viewCam', 'show'), expected_type=dict):
  41             raise ExtractorError('Model is in private show', expected=True)
  42         elif not traverse_obj(data, ('viewCam', 'model', 'isLive'), expected_type=bool):
  43             raise UserNotLive(video_id=video_id)
  44
  45         model_id = traverse_obj(data, ('viewCam', 'model', 'id'), expected_type=int)
  46
  47         formats = []
  48         for host in traverse_obj(data, ('config', 'data', (
  49                 (('features', 'featuresV2'), 'hlsFallback', 'fallbackDomains', ...), 'hlsStreamHost'))):
  50             formats = self._extract_m3u8_formats(
  51                 f'https://edge-hls.{host}/hls/{model_id}/master/{model_id}_auto.m3u8',
  52                 video_id, ext='mp4', m3u8_id='hls', fatal=False, live=True)
  53             if formats:
  54                 break
  55         if not formats:
  56             self.raise_no_formats('No active streams found', expected=True)
  57
  58         return {
  59             'id': video_id,
  60             'title': video_id,
  61             'description': self._og_search_description(webpage),
  62             'is_live': True,
  63             'formats': formats,
  64             # Stripchat declares the RTA meta-tag, but in an non-standard format so _rta_search() can't be used
  65             'age_limit': 18,
  66         }