Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/clubic.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .common import InfoExtractor
   5 from ..utils import (
   6     clean_html,
   7     qualities,
   8 )
   9
  10
  11 class ClubicIE(InfoExtractor):
  12     _VALID_URL = r'https?://(?:www\.)?clubic\.com/video/(?:[^/]+/)*video.*-(?P<id>[0-9]+)\.html'
  13
  14     _TESTS = [{
  15         'url': 'http://www.clubic.com/video/clubic-week/video-clubic-week-2-0-le-fbi-se-lance-dans-la-photo-d-identite-448474.html',
  16         'md5': '1592b694ba586036efac1776b0b43cd3',
  17         'info_dict': {
  18             'id': '448474',
  19             'ext': 'mp4',
  20             'title': 'Clubic Week 2.0 : le FBI se lance dans la photo d\u0092identité',
  21             'description': 're:Gueule de bois chez Nokia. Le constructeur a indiqué cette.*',
  22             'thumbnail': 're:^http://img\.clubic\.com/.*\.jpg$',
  23         }
  24     }, {
  25         'url': 'http://www.clubic.com/video/video-clubic-week-2-0-apple-iphone-6s-et-plus-mais-surtout-le-pencil-469792.html',
  26         'only_matching': True,
  27     }]
  28
  29     def _real_extract(self, url):
  30         video_id = self._match_id(url)
  31
  32         player_url = 'http://player.m6web.fr/v1/player/clubic/%s.html' % video_id
  33         player_page = self._download_webpage(player_url, video_id)
  34
  35         config = self._parse_json(self._search_regex(
  36             r'(?m)M6\.Player\.config\s*=\s*(\{.+?\});$', player_page,
  37             'configuration'), video_id)
  38
  39         video_info = config['videoInfo']
  40         sources = config['sources']
  41         quality_order = qualities(['sd', 'hq'])
  42
  43         formats = [{
  44             'format_id': src['streamQuality'],
  45             'url': src['src'],
  46             'quality': quality_order(src['streamQuality']),
  47         } for src in sources]
  48         self._sort_formats(formats)
  49
  50         return {
  51             'id': video_id,
  52             'title': video_info['title'],
  53             'formats': formats,
  54             'description': clean_html(video_info.get('description')),
  55             'thumbnail': config.get('poster'),
  56         }