projects / yt-dlp.git / blame
| Commit | Line | Data |
|---|---|---|
| f8c7ba99 TA |
1 | from .common import InfoExtractor |
| 2 | ||
| 3 | ||
| 4 | class EpochIE(InfoExtractor): | |
| 5 | _VALID_URL = r'https?://www.theepochtimes\.com/[\w-]+_(?P<id>\d+).html' | |
| 6 | _TESTS = [ | |
| 7 | { | |
| 8 | 'url': 'https://www.theepochtimes.com/they-can-do-audio-video-physical-surveillance-on-you-24h-365d-a-year-rex-lee-on-intrusive-apps_4661688.html', | |
| 9 | 'info_dict': { | |
| 10 | 'id': 'a3dd732c-4750-4bc8-8156-69180668bda1', | |
| 11 | 'ext': 'mp4', | |
| 12 | 'title': '‘They Can Do Audio, Video, Physical Surveillance on You 24H/365D a Year’: Rex Lee on Intrusive Apps', | |
| 13 | } | |
| 14 | }, | |
| 15 | { | |
| 16 | 'url': 'https://www.theepochtimes.com/the-communist-partys-cyberattacks-on-america-explained-rex-lee-talks-tech-hybrid-warfare_4342413.html', | |
| 17 | 'info_dict': { | |
| 18 | 'id': '276c7f46-3bbf-475d-9934-b9bbe827cf0a', | |
| 19 | 'ext': 'mp4', | |
| 20 | 'title': 'The Communist Party’s Cyberattacks on America Explained; Rex Lee Talks Tech Hybrid Warfare', | |
| 21 | } | |
| 22 | }, | |
| 23 | { | |
| 24 | 'url': 'https://www.theepochtimes.com/kash-patel-a-6-year-saga-of-government-corruption-from-russiagate-to-mar-a-lago_4690250.html', | |
| 25 | 'info_dict': { | |
| 26 | 'id': 'aa9ceecd-a127-453d-a2de-7153d6fd69b6', | |
| 27 | 'ext': 'mp4', | |
| 28 | 'title': 'Kash Patel: A ‘6-Year-Saga’ of Government Corruption, From Russiagate to Mar-a-Lago', | |
| 29 | } | |
| 30 | }, | |
| 31 | ] | |
| 32 | ||
| 33 | def _real_extract(self, url): | |
| 34 | video_id = self._match_id(url) | |
| 35 | webpage = self._download_webpage(url, video_id) | |
| 36 | ||
| 37 | youmaker_video_id = self._search_regex(r'data-trailer="[\w-]+" data-id="([\w-]+)"', webpage, 'url') | |
| 38 | formats, subtitles = self._extract_m3u8_formats_and_subtitles( | |
| 39 | f'http://vs1.youmaker.com/assets/{youmaker_video_id}/playlist.m3u8', video_id, 'mp4', m3u8_id='hls') | |
| 40 | ||
| 41 | return { | |
| 42 | 'id': youmaker_video_id, | |
| 43 | 'formats': formats, | |
| 44 | 'subtitles': subtitles, | |
| 45 | 'title': self._html_extract_title(webpage) | |
| 46 | } |