yt_dlp/extractor/myvi.py

   1 from .common import InfoExtractor
   2 from .vimple import SprutoBaseIE
   3
   4
   5 class MyviIE(SprutoBaseIE):
   6     _VALID_URL = r'''(?x)
   7                         (?:
   8                             https?://
   9                                 (?:www\.)?
  10                                 myvi\.
  11                                 (?:
  12                                     (?:ru/player|tv)/
  13                                     (?:
  14                                         (?:
  15                                             embed/html|
  16                                             flash|
  17                                             api/Video/Get
  18                                         )/|
  19                                         content/preloader\.swf\?.*\bid=
  20                                     )|
  21                                     ru/watch/
  22                                 )|
  23                             myvi:
  24                         )
  25                         (?P<id>[\da-zA-Z_-]+)
  26                     '''
  27     _EMBED_REGEX = [r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//myvi\.(?:ru/player|tv)/(?:embed/html|flash)/[^"]+)\1']
  28     _TESTS = [{
  29         'url': 'http://myvi.ru/player/embed/html/oOy4euHA6LVwNNAjhD9_Jq5Ha2Qf0rtVMVFMAZav8wObeRTZaCATzucDQIDph8hQU0',
  30         'md5': '571bbdfba9f9ed229dc6d34cc0f335bf',
  31         'info_dict': {
  32             'id': 'f16b2bbd-cde8-481c-a981-7cd48605df43',
  33             'ext': 'mp4',
  34             'title': 'хозяин жизни',
  35             'thumbnail': r're:^https?://.*\.jpg$',
  36             'duration': 25,
  37         },
  38     }, {
  39         'url': 'http://myvi.ru/player/content/preloader.swf?id=oOy4euHA6LVwNNAjhD9_Jq5Ha2Qf0rtVMVFMAZav8wOYf1WFpPfc_bWTKGVf_Zafr0',
  40         'only_matching': True,
  41     }, {
  42         'url': 'http://myvi.ru/player/api/Video/Get/oOy4euHA6LVwNNAjhD9_Jq5Ha2Qf0rtVMVFMAZav8wObeRTZaCATzucDQIDph8hQU0',
  43         'only_matching': True,
  44     }, {
  45         'url': 'http://myvi.tv/embed/html/oTGTNWdyz4Zwy_u1nraolwZ1odenTd9WkTnRfIL9y8VOgHYqOHApE575x4_xxS9Vn0?ap=0',
  46         'only_matching': True,
  47     }, {
  48         'url': 'http://myvi.ru/player/flash/ocp2qZrHI-eZnHKQBK4cZV60hslH8LALnk0uBfKsB-Q4WnY26SeGoYPi8HWHxu0O30',
  49         'only_matching': True,
  50     }, {
  51         'url': 'https://www.myvi.ru/watch/YwbqszQynUaHPn_s82sx0Q2',
  52         'only_matching': True,
  53     }, {
  54         'url': 'myvi:YwbqszQynUaHPn_s82sx0Q2',
  55         'only_matching': True,
  56     }]
  57
  58     def _real_extract(self, url):
  59         video_id = self._match_id(url)
  60
  61         spruto = self._download_json(
  62             'http://myvi.ru/player/api/Video/Get/%s?sig' % video_id, video_id)['sprutoData']
  63
  64         return self._extract_spruto(spruto, video_id)
  65
  66
  67 class MyviEmbedIE(InfoExtractor):
  68     _VALID_URL = r'https?://(?:www\.)?myvi\.tv/(?:[^?]+\?.*?\bv=|embed/)(?P<id>[\da-z]+)'
  69     _TESTS = [{
  70         'url': 'https://www.myvi.tv/embed/ccdqic3wgkqwpb36x9sxg43t4r',
  71         'info_dict': {
  72             'id': 'b3ea0663-3234-469d-873e-7fecf36b31d1',
  73             'ext': 'mp4',
  74             'title': 'Твоя (original song).mp4',
  75             'thumbnail': r're:^https?://.*\.jpg$',
  76             'duration': 277,
  77         },
  78         'params': {
  79             'skip_download': True,
  80         },
  81     }, {
  82         'url': 'https://www.myvi.tv/idmi6o?v=ccdqic3wgkqwpb36x9sxg43t4r#watch',
  83         'only_matching': True,
  84     }]
  85
  86     @classmethod
  87     def suitable(cls, url):
  88         return False if MyviIE.suitable(url) else super(MyviEmbedIE, cls).suitable(url)
  89
  90     def _real_extract(self, url):
  91         video_id = self._match_id(url)
  92
  93         webpage = self._download_webpage(
  94             'https://www.myvi.tv/embed/%s' % video_id, video_id)
  95
  96         myvi_id = self._search_regex(
  97             r'CreatePlayer\s*\(\s*["\'].*?\bv=([\da-zA-Z_]+)',
  98             webpage, 'video id')
  99
 100         return self.url_result('myvi:%s' % myvi_id, ie=MyviIE.ie_key())