]>
Commit | Line | Data |
---|---|---|
20a7304e H |
1 | from .common import InfoExtractor |
2 | from ..utils import parse_duration, traverse_obj | |
3 | ||
4 | ||
5 | class UnscriptedNewsVideoIE(InfoExtractor): | |
6 | _VALID_URL = r'https?://www\.unscripted\.news/videos/(?P<id>[\w-]+)' | |
7 | _TESTS = [{ | |
8 | 'url': 'https://www.unscripted.news/videos/a-day-at-the-farmers-protest', | |
9 | 'info_dict': { | |
10 | 'id': '60c0a55cd1e99b1079918a57', | |
11 | 'display_id': 'a-day-at-the-farmers-protest', | |
12 | 'ext': 'mp4', | |
13 | 'title': 'A Day at the Farmers\' Protest', | |
14 | 'description': 'md5:4b3df22747a03e8f14f746dd72190384', | |
15 | 'thumbnail': 'https://s3.unscripted.news/anj2/60c0a55cd1e99b1079918a57/5f199a65-c803-4a5c-8fce-2077359c3b72.jpg', | |
16 | 'duration': 2251.0, | |
17 | 'series': 'Ground Reports', | |
18 | } | |
19 | }, { | |
20 | 'url': 'https://www.unscripted.news/videos/you-get-the-politicians-you-deserve-ft-shashi-tharoor', | |
21 | 'info_dict': { | |
22 | 'id': '5fb3afbf18ac817d341a74d8', | |
23 | 'display_id': 'you-get-the-politicians-you-deserve-ft-shashi-tharoor', | |
24 | 'ext': 'mp4', | |
25 | 'cast': ['Avalok Langer', 'Ashwin Mehta'], | |
26 | 'thumbnail': 'https://s3.unscripted.news/anj2/5fb3afbf18ac817d341a74d8/82bd7942-4f20-4cd8-98ae-83f9e814f998.jpg', | |
27 | 'description': 'md5:1e91b069238a705ca3a40f87e6f1182c', | |
28 | 'duration': 1046.0, | |
29 | 'series': 'Dumb Questions Only', | |
30 | 'title': 'You Get The Politicians You Deserve! ft. Shashi Tharoor', | |
31 | } | |
32 | }] | |
33 | ||
34 | def _real_extract(self, url): | |
35 | display_id = self._match_id(url) | |
36 | webpage = self._download_webpage(url, display_id) | |
37 | nextjs_data = self._search_nextjs_data(webpage, display_id)['props']['pageProps']['dataLocal'] | |
38 | ||
39 | # TODO: get subtitle from srt key | |
40 | formats, subtitles = self._extract_m3u8_formats_and_subtitles(nextjs_data['alt_content'], display_id) | |
41 | ||
42 | return { | |
43 | 'id': nextjs_data['_id'], | |
44 | 'display_id': display_id, | |
45 | 'title': nextjs_data.get('title') or self._og_search_title(webpage), | |
46 | 'description': nextjs_data.get('sh_heading') or self._og_search_description(webpage), | |
47 | 'formats': formats, | |
48 | 'subtitles': subtitles, | |
49 | 'thumbnail': self._og_search_thumbnail(webpage), | |
50 | 'duration': parse_duration(nextjs_data.get('duration')), | |
51 | 'series': traverse_obj(nextjs_data, ('show', 'topic')), | |
52 | 'cast': traverse_obj(nextjs_data, ('cast_crew', ..., 'displayname')), | |
53 | } |