]> jfr.im git - yt-dlp.git/blame - yt_dlp/extractor/tempo.py
[extractor/youtube] Handle `consent.youtube`
[yt-dlp.git] / yt_dlp / extractor / tempo.py
CommitLineData
67685a54
H
1from .common import InfoExtractor
2from ..utils import int_or_none, parse_iso8601, str_or_none, traverse_obj
3
4
5class TempoIE(InfoExtractor):
6 _VALID_URL = r'https?://video\.tempo\.co/\w+/\d+/(?P<id>[\w-]+)'
7 _TESTS = [{
8 'url': 'https://video.tempo.co/read/30058/anies-baswedan-ajukan-banding-putusan-ptun-batalkan-ump-dki',
9 'info_dict': {
10 'id': '2144438',
11 'ext': 'mp4',
12 'title': 'Anies Baswedan Ajukan Banding Putusan PTUN Batalkan UMP DKI',
13 'display_id': 'anies-baswedan-ajukan-banding-putusan-ptun-batalkan-ump-dki',
14 'duration': 84,
15 'description': 'md5:a6822b7c4c874fa7e5bd63e96a387b66',
16 'thumbnail': 'https://statik.tempo.co/data/2022/07/27/id_1128287/1128287_720.jpg',
17 'timestamp': 1658911277,
18 'upload_date': '20220727',
19 'tags': ['Anies Baswedan', ' PTUN', ' PTUN | Pengadilan Tata Usaha Negara', ' PTUN Batalkan UMP DKI', ' UMP DKI'],
20 }
21 }]
22
23 def _real_extract(self, url):
24 display_id = self._match_id(url)
25 webpage = self._download_webpage(url, display_id)
26
27 player_key, widget_id = self._search_regex(
28 r'<ivs-player\s*[^>]+data-ivs-key\s*=\s*"(?P<player_key>[\w]+)[^>]+\bdata-ivs-wid="(?P<widget_id>[\w-]+)',
29 webpage, 'player_key, widget_id', group=('player_key', 'widget_id'))
30
31 json_ld_data = self._search_json_ld(webpage, display_id)
32
33 json_data = self._download_json(
34 f'https://ivxplayer.ivideosmart.com/prod/widget/{widget_id}',
35 display_id, query={'key': player_key})
36 formats, subtitles = self._extract_m3u8_formats_and_subtitles(
37 json_data['player']['video_url'], display_id, ext='mp4')
38
39 return {
40 'id': str(json_data['ivx']['id']),
41 'display_id': display_id,
42 'formats': formats,
43 'subtitles': subtitles,
44 'title': (self._html_search_meta('twitter:title', webpage) or self._og_search_title(webpage)
45 or traverse_obj(json_data, ('ivx', 'name'))),
46 'duration': int_or_none(traverse_obj(json_data, ('ivx', 'duration'))),
47 'thumbnail': (self._html_search_meta('twitter:image:src', webpage) or self._og_search_thumbnail(webpage)
48 or traverse_obj(json_data, ('ivx', 'thumbnail_url'))),
49 'description': (json_ld_data.get('description') or self._html_search_meta(['description', 'twitter:description'], webpage)
50 or self._og_search_description(webpage)),
51 'timestamp': parse_iso8601(traverse_obj(json_data, ('ivx', 'created_at'))),
52 'tags': str_or_none(self._html_search_meta('keywords', webpage), '').split(','),
53 }