yt_dlp/extractor/myvi.py

   1 import re
   2
   3 from .common import InfoExtractor
   4 from .vimple import SprutoBaseIE
   5
   6
   7 class MyviIE(SprutoBaseIE):
   8     _VALID_URL = r'''(?x)
   9                         (?:
  10                             https?://
  11                                 (?:www\.)?
  12                                 myvi\.
  13                                 (?:
  14                                     (?:ru/player|tv)/
  15                                     (?:
  16                                         (?:
  17                                             embed/html|
  18                                             flash|
  19                                             api/Video/Get
  20                                         )/|
  21                                         content/preloader\.swf\?.*\bid=
  22                                     )|
  23                                     ru/watch/
  24                                 )|
  25                             myvi:
  26                         )
  27                         (?P<id>[\da-zA-Z_-]+)
  28                     '''
  29     _TESTS = [{
  30         'url': 'http://myvi.ru/player/embed/html/oOy4euHA6LVwNNAjhD9_Jq5Ha2Qf0rtVMVFMAZav8wObeRTZaCATzucDQIDph8hQU0',
  31         'md5': '571bbdfba9f9ed229dc6d34cc0f335bf',
  32         'info_dict': {
  33             'id': 'f16b2bbd-cde8-481c-a981-7cd48605df43',
  34             'ext': 'mp4',
  35             'title': 'хозяин жизни',
  36             'thumbnail': r're:^https?://.*\.jpg$',
  37             'duration': 25,
  38         },
  39     }, {
  40         'url': 'http://myvi.ru/player/content/preloader.swf?id=oOy4euHA6LVwNNAjhD9_Jq5Ha2Qf0rtVMVFMAZav8wOYf1WFpPfc_bWTKGVf_Zafr0',
  41         'only_matching': True,
  42     }, {
  43         'url': 'http://myvi.ru/player/api/Video/Get/oOy4euHA6LVwNNAjhD9_Jq5Ha2Qf0rtVMVFMAZav8wObeRTZaCATzucDQIDph8hQU0',
  44         'only_matching': True,
  45     }, {
  46         'url': 'http://myvi.tv/embed/html/oTGTNWdyz4Zwy_u1nraolwZ1odenTd9WkTnRfIL9y8VOgHYqOHApE575x4_xxS9Vn0?ap=0',
  47         'only_matching': True,
  48     }, {
  49         'url': 'http://myvi.ru/player/flash/ocp2qZrHI-eZnHKQBK4cZV60hslH8LALnk0uBfKsB-Q4WnY26SeGoYPi8HWHxu0O30',
  50         'only_matching': True,
  51     }, {
  52         'url': 'https://www.myvi.ru/watch/YwbqszQynUaHPn_s82sx0Q2',
  53         'only_matching': True,
  54     }, {
  55         'url': 'myvi:YwbqszQynUaHPn_s82sx0Q2',
  56         'only_matching': True,
  57     }]
  58
  59     @classmethod
  60     def _extract_url(cls, webpage):
  61         mobj = re.search(
  62             r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//myvi\.(?:ru/player|tv)/(?:embed/html|flash)/[^"]+)\1', webpage)
  63         if mobj:
  64             return mobj.group('url')
  65
  66     def _real_extract(self, url):
  67         video_id = self._match_id(url)
  68
  69         spruto = self._download_json(
  70             'http://myvi.ru/player/api/Video/Get/%s?sig' % video_id, video_id)['sprutoData']
  71
  72         return self._extract_spruto(spruto, video_id)
  73
  74
  75 class MyviEmbedIE(InfoExtractor):
  76     _VALID_URL = r'https?://(?:www\.)?myvi\.tv/(?:[^?]+\?.*?\bv=|embed/)(?P<id>[\da-z]+)'
  77     _TESTS = [{
  78         'url': 'https://www.myvi.tv/embed/ccdqic3wgkqwpb36x9sxg43t4r',
  79         'info_dict': {
  80             'id': 'b3ea0663-3234-469d-873e-7fecf36b31d1',
  81             'ext': 'mp4',
  82             'title': 'Твоя (original song).mp4',
  83             'thumbnail': r're:^https?://.*\.jpg$',
  84             'duration': 277,
  85         },
  86         'params': {
  87             'skip_download': True,
  88         },
  89     }, {
  90         'url': 'https://www.myvi.tv/idmi6o?v=ccdqic3wgkqwpb36x9sxg43t4r#watch',
  91         'only_matching': True,
  92     }]
  93
  94     @classmethod
  95     def suitable(cls, url):
  96         return False if MyviIE.suitable(url) else super(MyviEmbedIE, cls).suitable(url)
  97
  98     def _real_extract(self, url):
  99         video_id = self._match_id(url)
 100
 101         webpage = self._download_webpage(
 102             'https://www.myvi.tv/embed/%s' % video_id, video_id)
 103
 104         myvi_id = self._search_regex(
 105             r'CreatePlayer\s*\(\s*["\'].*?\bv=([\da-zA-Z_]+)',
 106             webpage, 'video id')
 107
 108         return self.url_result('myvi:%s' % myvi_id, ie=MyviIE.ie_key())