]> jfr.im git - yt-dlp.git/blame - yt_dlp/extractor/stripchat.py
[extractor] Deprecate `_sort_formats`
[yt-dlp.git] / yt_dlp / extractor / stripchat.py
CommitLineData
c6118ca2 1from .common import InfoExtractor
a349d4d6 2from ..utils import ExtractorError, lowercase_escape, traverse_obj
c6118ca2 3
4
5class StripchatIE(InfoExtractor):
befcac11 6 _VALID_URL = r'https?://stripchat\.com/(?P<id>[^/?#]+)'
c6118ca2 7 _TESTS = [{
a349d4d6 8 'url': 'https://stripchat.com/Joselin_Flower',
c6118ca2 9 'info_dict': {
a349d4d6 10 'id': 'Joselin_Flower',
c6118ca2 11 'ext': 'mp4',
a349d4d6 12 'title': 're:^Joselin_Flower [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
c6118ca2 13 'description': str,
14 'is_live': True,
15 'age_limit': 18,
16 },
17 'skip': 'Room is offline',
befcac11
AM
18 }, {
19 'url': 'https://stripchat.com/Rakhijaan@xh',
20 'only_matching': True
c6118ca2 21 }]
22
23 def _real_extract(self, url):
24 video_id = self._match_id(url)
76f2bb17 25 webpage = self._download_webpage(url, video_id, headers=self.geo_verification_headers())
c6118ca2 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
a349d4d6 35 if traverse_obj(data, ('viewCam', 'show'), expected_type=dict):
c6118ca2 36 raise ExtractorError('Model is in private show', expected=True)
a349d4d6 37 elif not traverse_obj(data, ('viewCam', 'model', 'isLive'), expected_type=bool):
c6118ca2 38 raise ExtractorError('Model is offline', expected=True)
39
a349d4d6 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
3b87f4d9 43 formats = []
a349d4d6 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
3b87f4d9
AM
51 if not formats:
52 self.raise_no_formats('No active streams found', expected=True)
c6118ca2 53
c6118ca2 54 return {
55 'id': video_id,
39ca3b5c 56 'title': video_id,
c6118ca2 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 }