]>
jfr.im git - yt-dlp.git/blob - yt_dlp/extractor/ina.py
1 from .common
import InfoExtractor
2 from ..utils
import unified_strdate
5 class InaIE(InfoExtractor
):
6 _VALID_URL
= r
'https?://(?:(?:www|m)\.)?ina\.fr/(?:[^/]+/)?(?:video|audio)/(?P<id>\w+)'
8 'url': 'https://www.ina.fr/video/I12055569/francois-hollande-je-crois-que-c-est-clair-video.html',
9 'md5': 'c5a09e5cb5604ed10709f06e7a377dda',
13 'title': 'François Hollande "Je crois que c\'est clair"',
14 'description': 'md5:08201f1c86fb250611f0ba415d21255a',
15 'upload_date': '20070712',
16 'thumbnail': 'https://cdn-hub.ina.fr/notice/690x517/3c4/I12055569.jpeg',
19 'url': 'https://www.ina.fr/video/S806544_001/don-d-organes-des-avancees-mais-d-importants-besoins-video.html',
20 'only_matching': True,
22 'url': 'https://www.ina.fr/audio/P16173408',
23 'only_matching': True,
25 'url': 'https://www.ina.fr/video/P16173408-video.html',
26 'only_matching': True,
28 'url': 'http://m.ina.fr/video/I12055569',
29 'only_matching': True,
31 'url': 'https://www.ina.fr/ina-eclaire-actu/video/cpb8205116303/les-jeux-electroniques',
32 'md5': '4b8284a9a3a184fdc7e744225b8251e7',
34 'id': 'CPB8205116303',
36 'title': 'Les jeux électroniques',
37 'description': 'md5:e09f7683dad1cc60b74950490127d233',
38 'upload_date': '19821204',
40 'thumbnail': 'https://cdn-hub.ina.fr/notice/690x517/203/CPB8205116303.jpeg',
44 def _real_extract(self
, url
):
45 video_id
= self
._match
_id
(url
).upper()
46 webpage
= self
._download
_webpage
(url
, video_id
)
48 api_url
= self
._html
_search
_regex
(
49 r
'asset-details-url\s*=\s*["\'](?P
<api_url
>[^
"\']+)',
50 webpage, 'api_url').replace(video_id, f'{video_id}.json')
52 api_response = self._download_json(api_url, video_id)
56 'url': api_response['resourceUrl'],
57 'ext': {'video': 'mp4', 'audio': 'mp3'}.get(api_response.get('type')),
58 'title': api_response.get('title'),
59 'description': api_response.get('description'),
60 'upload_date': unified_strdate(api_response.get('dateOfBroadcast')),
61 'duration': api_response.get('duration'),
62 'thumbnail': api_response.get('resourceThumbnail'),