yt_dlp/extractor/digiteka.py

   1 from .common import InfoExtractor
   2 from ..utils import int_or_none
   3
   4
   5 class DigitekaIE(InfoExtractor):
   6     _VALID_URL = r'''(?x)
   7         https?://(?:www\.)?(?:digiteka\.net|ultimedia\.com)/
   8         (?:
   9             deliver/
  10             (?P<embed_type>
  11                 generic|
  12                 musique
  13             )
  14             (?:/[^/]+)*/
  15             (?:
  16                 src|
  17                 article
  18             )|
  19             default/index/video
  20             (?P<site_type>
  21                 generic|
  22                 music
  23             )
  24             /id
  25         )/(?P<id>[\d+a-z]+)'''
  26     _EMBED_REGEX = [r'<(?:iframe|script)[^>]+src=["\'](?P<url>(?:https?:)?//(?:www\.)?ultimedia\.com/deliver/(?:generic|musique)(?:/[^/]+)*/(?:src|article)/[\d+a-z]+)']
  27     _TESTS = [{
  28         # news
  29         'url': 'https://www.ultimedia.com/default/index/videogeneric/id/s8uk0r',
  30         'md5': '276a0e49de58c7e85d32b057837952a2',
  31         'info_dict': {
  32             'id': 's8uk0r',
  33             'ext': 'mp4',
  34             'title': 'Loi sur la fin de vie: le texte prévoit un renforcement des directives anticipées',
  35             'thumbnail': r're:^https?://.*\.jpg',
  36             'duration': 74,
  37             'upload_date': '20150317',
  38             'timestamp': 1426604939,
  39             'uploader_id': '3fszv',
  40         },
  41     }, {
  42         # music
  43         'url': 'https://www.ultimedia.com/default/index/videomusic/id/xvpfp8',
  44         'md5': '2ea3513813cf230605c7e2ffe7eca61c',
  45         'info_dict': {
  46             'id': 'xvpfp8',
  47             'ext': 'mp4',
  48             'title': 'Two - C\'est La Vie (clip)',
  49             'thumbnail': r're:^https?://.*\.jpg',
  50             'duration': 233,
  51             'upload_date': '20150224',
  52             'timestamp': 1424760500,
  53             'uploader_id': '3rfzk',
  54         },
  55     }, {
  56         'url': 'https://www.digiteka.net/deliver/generic/iframe/mdtk/01637594/src/lqm3kl/zone/1/showtitle/1/autoplay/yes',
  57         'only_matching': True,
  58     }]
  59
  60     def _real_extract(self, url):
  61         mobj = self._match_valid_url(url)
  62         video_id = mobj.group('id')
  63         video_type = mobj.group('embed_type') or mobj.group('site_type')
  64         if video_type == 'music':
  65             video_type = 'musique'
  66
  67         deliver_info = self._download_json(
  68             f'http://www.ultimedia.com/deliver/video?video={video_id}&topic={video_type}',
  69             video_id)
  70
  71         yt_id = deliver_info.get('yt_id')
  72         if yt_id:
  73             return self.url_result(yt_id, 'Youtube')
  74
  75         jwconf = deliver_info['jwconf']
  76
  77         formats = []
  78         for source in jwconf['playlist'][0]['sources']:
  79             formats.append({
  80                 'url': source['file'],
  81                 'format_id': source.get('label'),
  82             })
  83
  84         title = deliver_info['title']
  85         thumbnail = jwconf.get('image')
  86         duration = int_or_none(deliver_info.get('duration'))
  87         timestamp = int_or_none(deliver_info.get('release_time'))
  88         uploader_id = deliver_info.get('owner_id')
  89
  90         return {
  91             'id': video_id,
  92             'title': title,
  93             'thumbnail': thumbnail,
  94             'duration': duration,
  95             'timestamp': timestamp,
  96             'uploader_id': uploader_id,
  97             'formats': formats,
  98         }