yt_dlp/extractor/stripchat.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .common import InfoExtractor
   5 from ..compat import (
   6     compat_str,
   7 )
   8 from ..utils import (
   9     ExtractorError,
  10     lowercase_escape,
  11     try_get,
  12 )
  13
  14
  15 class StripchatIE(InfoExtractor):
  16     _VALID_URL = r'https?://stripchat\.com/(?P<id>[0-9A-Za-z-_]+)'
  17     _TESTS = [{
  18         'url': 'https://stripchat.com/feel_me',
  19         'info_dict': {
  20             'id': 'feel_me',
  21             'ext': 'mp4',
  22             'title': 're:^feel_me [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
  23             'description': str,
  24             'is_live': True,
  25             'age_limit': 18,
  26         },
  27         'skip': 'Room is offline',
  28     }]
  29
  30     def _real_extract(self, url):
  31         video_id = self._match_id(url)
  32         webpage = self._download_webpage(
  33             'https://stripchat.com/%s/' % video_id, video_id,
  34             headers=self.geo_verification_headers())
  35
  36         data = self._parse_json(
  37             self._search_regex(
  38                 r'<script\b[^>]*>\s*window\.__PRELOADED_STATE__\s*=(?P<value>.*?)<\/script>',
  39                 webpage, 'data', default='{}', group='value'),
  40             video_id, transform_source=lowercase_escape, fatal=False)
  41         if not data:
  42             raise ExtractorError('Unable to find configuration for stream.')
  43
  44         if try_get(data, lambda x: x['viewCam']['show'], dict):
  45             raise ExtractorError('Model is in private show', expected=True)
  46         elif not try_get(data, lambda x: x['viewCam']['model']['isLive'], bool):
  47             raise ExtractorError('Model is offline', expected=True)
  48
  49         server = try_get(data, lambda x: x['viewCam']['viewServers']['flashphoner-hls'], compat_str)
  50         host = try_get(data, lambda x: x['config']['data']['hlsStreamHost'], compat_str)
  51         model_id = try_get(data, lambda x: x['viewCam']['model']['id'], int)
  52
  53         formats = self._extract_m3u8_formats(
  54             'https://b-%s.%s/hls/%d/%d.m3u8' % (server, host, model_id, model_id),
  55             video_id, ext='mp4', m3u8_id='hls', fatal=False, live=True)
  56         self._sort_formats(formats)
  57
  58         return {
  59             'id': video_id,
  60             'title': self._live_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         }