[extractor/rokfin] Re-construct manifest url (#6507)

Authored by: vampirefrog
This commit is contained in:
vampirefrog 2023-03-11 18:52:36 +02:00 committed by GitHub
parent 89dbf08483
commit 7a6c8a0807
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23

View file

@ -45,6 +45,7 @@ class RokfinIE(InfoExtractor):
'live_status': 'not_live', 'live_status': 'not_live',
'dislike_count': int, 'dislike_count': int,
'like_count': int, 'like_count': int,
'duration': 213,
} }
}, { }, {
'url': 'https://rokfin.com/post/223/Julian-Assange-Arrested-Streaming-In-Real-Time', 'url': 'https://rokfin.com/post/223/Julian-Assange-Arrested-Streaming-In-Real-Time',
@ -72,7 +73,7 @@ class RokfinIE(InfoExtractor):
'title': '"It\'s A Crazy Mess" Regional Director Blows Whistle On Pfizer\'s Vaccine Trial Data', 'title': '"It\'s A Crazy Mess" Regional Director Blows Whistle On Pfizer\'s Vaccine Trial Data',
'thumbnail': r're:https://img\.production\.rokfin\.com/.+', 'thumbnail': r're:https://img\.production\.rokfin\.com/.+',
'description': 'md5:324ce2d3e3b62e659506409e458b9d8e', 'description': 'md5:324ce2d3e3b62e659506409e458b9d8e',
'channel': 'Ryan Cristián', 'channel': 'TLAVagabond',
'channel_id': 53856, 'channel_id': 53856,
'channel_url': 'https://rokfin.com/TLAVagabond', 'channel_url': 'https://rokfin.com/TLAVagabond',
'availability': 'public', 'availability': 'public',
@ -86,6 +87,47 @@ class RokfinIE(InfoExtractor):
'dislike_count': int, 'dislike_count': int,
'like_count': int, 'like_count': int,
'tags': ['FreeThinkingMedia^'], 'tags': ['FreeThinkingMedia^'],
'duration': None,
}
}, {
'url': 'https://rokfin.com/post/126703/Brave-New-World--Aldous-Huxley-DEEPDIVE--Chpts-13--Quite-Frankly--Jay-Dyer',
'info_dict': {
'id': 'post/126703',
'ext': 'mp4',
'title': 'Brave New World - Aldous Huxley DEEPDIVE! (Chpts 1-3) - Quite Frankly & Jay Dyer',
'thumbnail': r're:https://img\.production\.rokfin\.com/.+',
'channel': 'Jay Dyer',
'channel_id': 186881,
'channel_url': 'https://rokfin.com/jaydyer',
'availability': 'premium_only',
'live_status': 'not_live',
'dislike_count': int,
'like_count': int,
'timestamp': 1678213357,
'upload_date': '20230307',
'tags': ['FreeThinkingMedia^', 'OpenMind^'],
'description': 'md5:cb04e32e68326c9b2b251b297bacff35',
'duration': 3100,
}
}, {
'url': 'https://rokfin.com/stream/31332/The-Grayzone-live-on-Nordstream-blame-game',
'info_dict': {
'id': 'stream/31332',
'ext': 'mp4',
'title': 'The Grayzone live on Nordstream blame game',
'thumbnail': r're:https://image\.v\.rokfin\.com/.+',
'channel': 'Max Blumenthal',
'channel_id': 248902,
'channel_url': 'https://rokfin.com/MaxBlumenthal',
'availability': 'premium_only',
'live_status': 'was_live',
'dislike_count': int,
'like_count': int,
'timestamp': 1678475166,
'release_timestamp': 1678475166.0,
'release_date': '20230310',
'upload_date': '20230310',
'tags': ['FreeThinkingMedia^'],
} }
}] }]
@ -100,6 +142,12 @@ def _real_extract(self, url):
else 'not_live') else 'not_live')
video_url = traverse_obj(metadata, 'url', ('content', 'contentUrl'), expected_type=url_or_none) video_url = traverse_obj(metadata, 'url', ('content', 'contentUrl'), expected_type=url_or_none)
if video_url in (None, 'fake.m3u8'):
video_url = format_field(self._search_regex(
r'https?://[^/]+/([^/]+)/storyboard.vtt',
traverse_obj(metadata, 'timelineUrl', ('content', 'timelineUrl'), expected_type=url_or_none),
video_id, default=None), None, 'https://stream.v.rokfin.com/%s.m3u8')
formats, subtitles = [{'url': video_url}] if video_url else [], {} formats, subtitles = [{'url': video_url}] if video_url else [], {}
if determine_ext(video_url) == 'm3u8': if determine_ext(video_url) == 'm3u8':
formats, subtitles = self._extract_m3u8_formats_and_subtitles( formats, subtitles = self._extract_m3u8_formats_and_subtitles(