_ Git - youtube-dl/blob - youtube_dl/extractor/ultimedia.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5
   6 from .common import InfoExtractor
   7 from ..utils import int_or_none
   8
   9
  10 class UltimediaIE(InfoExtractor):
  11     _VALID_URL = r'https?://(?:www\.)?ultimedia\.com/deliver/(?P<type>generic|musique)(?:/[^/]+)*/(?:src|article)/(?P<id>[\d+a-z]+)'
  12     _TESTS = [{
  13         # news
  14         'url': 'https://www.ultimedia.com/deliver/generic/iframe/mdtk/01601930/zone/1/src/s8uk0r/autoplay/yes/ad/no/width/714/height/435',
  15         'md5': '276a0e49de58c7e85d32b057837952a2',
  16         'info_dict': {
  17             'id': 's8uk0r',
  18             'ext': 'mp4',
  19             'title': 'Loi sur la fin de vie: le texte prévoit un renforcement des directives anticipées',
  20             'thumbnail': 're:^https?://.*\.jpg',
  21             'duration': 74,
  22             'upload_date': '20150317',
  23             'timestamp': 1426604939,
  24             'uploader_id': '3fszv',
  25         },
  26     }, {
  27         # music
  28         'url': 'https://www.ultimedia.com/deliver/musique/iframe/mdtk/01601930/zone/1/article/xvpfp8/autoplay/yes/ad/no/width/714/height/435',
  29         'md5': '2ea3513813cf230605c7e2ffe7eca61c',
  30         'info_dict': {
  31             'id': 'xvpfp8',
  32             'ext': 'mp4',
  33             'title': 'Two - C\'est La Vie (clip)',
  34             'thumbnail': 're:^https?://.*\.jpg',
  35             'duration': 233,
  36             'upload_date': '20150224',
  37             'timestamp': 1424760500,
  38             'uploader_id': '3rfzk',
  39         },
  40     }]
  41
  42     @staticmethod
  43     def _extract_url(webpage):
  44         mobj = re.search(
  45             r'<(?:iframe|script)[^>]+src=["\'](?P<url>(?:https?:)?//(?:www\.)?ultimedia\.com/deliver/(?:generic|musique)(?:/[^/]+)*/(?:src|article)/[\d+a-z]+)',
  46             webpage)
  47         if mobj:
  48             return mobj.group('url')
  49
  50     def _real_extract(self, url):
  51         video_type, video_id = re.match(self._VALID_URL, url).groups()
  52
  53         deliver_info = self._download_json(
  54             'http://www.ultimedia.com/deliver/video?video=%s&topic=%s' % (video_id, video_type),
  55             video_id)
  56
  57         yt_id = deliver_info.get('yt_id')
  58         if yt_id:
  59             return self.url_result(yt_id, 'Youtube')
  60
  61         jwconf = deliver_info['jwconf']
  62
  63         formats = []
  64         for source in jwconf['playlist'][0]['sources']:
  65             formats.append({
  66                 'url': source['file'],
  67                 'format_id': source.get('label'),
  68             })
  69
  70         self._sort_formats(formats)
  71
  72         title = deliver_info['title']
  73         thumbnail = jwconf.get('image')
  74         duration = int_or_none(deliver_info.get('duration'))
  75         timestamp = int_or_none(deliver_info.get('release_time'))
  76         uploader_id = deliver_info.get('owner_id')
  77
  78         return {
  79             'id': video_id,
  80             'title': title,
  81             'thumbnail': thumbnail,
  82             'duration': duration,
  83             'timestamp': timestamp,
  84             'uploader_id': uploader_id,
  85             'formats': formats,
  86         }