]> jfr.im git - yt-dlp.git/blob - yt_dlp/extractor/clubic.py
403e44aafd248dcc71ecc4cd2457e1968e517cfd
[yt-dlp.git] / yt_dlp / extractor / clubic.py
1 from .common import InfoExtractor
2 from ..utils import (
3 clean_html,
4 qualities,
5 )
6
7
8 class ClubicIE(InfoExtractor):
9 _VALID_URL = r'https?://(?:www\.)?clubic\.com/video/(?:[^/]+/)*video.*-(?P<id>[0-9]+)\.html'
10
11 _TESTS = [{
12 'url': 'http://www.clubic.com/video/clubic-week/video-clubic-week-2-0-le-fbi-se-lance-dans-la-photo-d-identite-448474.html',
13 'md5': '1592b694ba586036efac1776b0b43cd3',
14 'info_dict': {
15 'id': '448474',
16 'ext': 'mp4',
17 'title': 'Clubic Week 2.0 : le FBI se lance dans la photo d\u0092identité',
18 'description': 're:Gueule de bois chez Nokia. Le constructeur a indiqué cette.*',
19 'thumbnail': r're:^http://img\.clubic\.com/.*\.jpg$',
20 }
21 }, {
22 'url': 'http://www.clubic.com/video/video-clubic-week-2-0-apple-iphone-6s-et-plus-mais-surtout-le-pencil-469792.html',
23 'only_matching': True,
24 }]
25
26 def _real_extract(self, url):
27 video_id = self._match_id(url)
28
29 player_url = 'http://player.m6web.fr/v1/player/clubic/%s.html' % video_id
30 player_page = self._download_webpage(player_url, video_id)
31
32 config = self._parse_json(self._search_regex(
33 r'(?m)M6\.Player\.config\s*=\s*(\{.+?\});$', player_page,
34 'configuration'), video_id)
35
36 video_info = config['videoInfo']
37 sources = config['sources']
38 quality_order = qualities(['sd', 'hq'])
39
40 formats = [{
41 'format_id': src['streamQuality'],
42 'url': src['src'],
43 'quality': quality_order(src['streamQuality']),
44 } for src in sources]
45
46 return {
47 'id': video_id,
48 'title': video_info['title'],
49 'formats': formats,
50 'description': clean_html(video_info.get('description')),
51 'thumbnail': config.get('poster'),
52 }