yt_dlp/extractor/digiteka.py

   1 import re
   2
   3 from .common import InfoExtractor
   4 from ..utils import int_or_none
   5
   6
   7 class DigitekaIE(InfoExtractor):
   8     _VALID_URL = r'''(?x)
   9         https?://(?:www\.)?(?:digiteka\.net|ultimedia\.com)/
  10         (?:
  11             deliver/
  12             (?P<embed_type>
  13                 generic|
  14                 musique
  15             )
  16             (?:/[^/]+)*/
  17             (?:
  18                 src|
  19                 article
  20             )|
  21             default/index/video
  22             (?P<site_type>
  23                 generic|
  24                 music
  25             )
  26             /id
  27         )/(?P<id>[\d+a-z]+)'''
  28     _TESTS = [{
  29         # news
  30         'url': 'https://www.ultimedia.com/default/index/videogeneric/id/s8uk0r',
  31         'md5': '276a0e49de58c7e85d32b057837952a2',
  32         'info_dict': {
  33             'id': 's8uk0r',
  34             'ext': 'mp4',
  35             'title': 'Loi sur la fin de vie: le texte prévoit un renforcement des directives anticipées',
  36             'thumbnail': r're:^https?://.*\.jpg',
  37             'duration': 74,
  38             'upload_date': '20150317',
  39             'timestamp': 1426604939,
  40             'uploader_id': '3fszv',
  41         },
  42     }, {
  43         # music
  44         'url': 'https://www.ultimedia.com/default/index/videomusic/id/xvpfp8',
  45         'md5': '2ea3513813cf230605c7e2ffe7eca61c',
  46         'info_dict': {
  47             'id': 'xvpfp8',
  48             'ext': 'mp4',
  49             'title': 'Two - C\'est La Vie (clip)',
  50             'thumbnail': r're:^https?://.*\.jpg',
  51             'duration': 233,
  52             'upload_date': '20150224',
  53             'timestamp': 1424760500,
  54             'uploader_id': '3rfzk',
  55         },
  56     }, {
  57         'url': 'https://www.digiteka.net/deliver/generic/iframe/mdtk/01637594/src/lqm3kl/zone/1/showtitle/1/autoplay/yes',
  58         'only_matching': True,
  59     }]
  60
  61     @staticmethod
  62     def _extract_url(webpage):
  63         mobj = re.search(
  64             r'<(?:iframe|script)[^>]+src=["\'](?P<url>(?:https?:)?//(?:www\.)?ultimedia\.com/deliver/(?:generic|musique)(?:/[^/]+)*/(?:src|article)/[\d+a-z]+)',
  65             webpage)
  66         if mobj:
  67             return mobj.group('url')
  68
  69     def _real_extract(self, url):
  70         mobj = self._match_valid_url(url)
  71         video_id = mobj.group('id')
  72         video_type = mobj.group('embed_type') or mobj.group('site_type')
  73         if video_type == 'music':
  74             video_type = 'musique'
  75
  76         deliver_info = self._download_json(
  77             'http://www.ultimedia.com/deliver/video?video=%s&topic=%s' % (video_id, video_type),
  78             video_id)
  79
  80         yt_id = deliver_info.get('yt_id')
  81         if yt_id:
  82             return self.url_result(yt_id, 'Youtube')
  83
  84         jwconf = deliver_info['jwconf']
  85
  86         formats = []
  87         for source in jwconf['playlist'][0]['sources']:
  88             formats.append({
  89                 'url': source['file'],
  90                 'format_id': source.get('label'),
  91             })
  92
  93         self._sort_formats(formats)
  94
  95         title = deliver_info['title']
  96         thumbnail = jwconf.get('image')
  97         duration = int_or_none(deliver_info.get('duration'))
  98         timestamp = int_or_none(deliver_info.get('release_time'))
  99         uploader_id = deliver_info.get('owner_id')
 100
 101         return {
 102             'id': video_id,
 103             'title': title,
 104             'thumbnail': thumbnail,
 105             'duration': duration,
 106             'timestamp': timestamp,
 107             'uploader_id': uploader_id,
 108             'formats': formats,
 109         }