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     _WORKING = False
  10     _VALID_URL = r'https?://(?:www\.)?clubic\.com/video/(?:[^/]+/)*video.*-(?P<id>[0-9]+)\.html'
  11
  12     _TESTS = [{
  13         'url': 'http://www.clubic.com/video/clubic-week/video-clubic-week-2-0-le-fbi-se-lance-dans-la-photo-d-identite-448474.html',
  14         'md5': '1592b694ba586036efac1776b0b43cd3',
  15         'info_dict': {
  16             'id': '448474',
  17             'ext': 'mp4',
  18             'title': 'Clubic Week 2.0 : le FBI se lance dans la photo d\u0092identité',
  19             'description': 're:Gueule de bois chez Nokia. Le constructeur a indiqué cette.*',
  20             'thumbnail': r're:^http://img\.clubic\.com/.*\.jpg$',
  21         },
  22     }, {
  23         'url': 'http://www.clubic.com/video/video-clubic-week-2-0-apple-iphone-6s-et-plus-mais-surtout-le-pencil-469792.html',
  24         'only_matching': True,
  25     }]
  26
  27     def _real_extract(self, url):
  28         video_id = self._match_id(url)
  29
  30         player_url = f'http://player.m6web.fr/v1/player/clubic/{video_id}.html'
  31         player_page = self._download_webpage(player_url, video_id)
  32
  33         config = self._parse_json(self._search_regex(
  34             r'(?m)M6\.Player\.config\s*=\s*(\{.+?\});$', player_page,
  35             'configuration'), video_id)
  36
  37         video_info = config['videoInfo']
  38         sources = config['sources']
  39         quality_order = qualities(['sd', 'hq'])
  40
  41         formats = [{
  42             'format_id': src['streamQuality'],
  43             'url': src['src'],
  44             'quality': quality_order(src['streamQuality']),
  45         } for src in sources]
  46
  47         return {
  48             'id': video_id,
  49             'title': video_info['title'],
  50             'formats': formats,
  51             'description': clean_html(video_info.get('description')),
  52             'thumbnail': config.get('poster'),
  53         }