yt_dlp/extractor/vbox7.py

   1 import re
   2
   3 from .common import InfoExtractor
   4 from ..utils import ExtractorError
   5
   6
   7 class Vbox7IE(InfoExtractor):
   8     _VALID_URL = r'''(?x)
   9                     https?://
  10                         (?:[^/]+\.)?vbox7\.com/
  11                         (?:
  12                             play:|
  13                             (?:
  14                                 emb/external\.php|
  15                                 player/ext\.swf
  16                             )\?.*?\bvid=
  17                         )
  18                         (?P<id>[\da-fA-F]+)
  19                     '''
  20     _GEO_COUNTRIES = ['BG']
  21     _TESTS = [{
  22         'url': 'http://vbox7.com/play:0946fff23c',
  23         'md5': 'a60f9ab3a3a2f013ef9a967d5f7be5bf',
  24         'info_dict': {
  25             'id': '0946fff23c',
  26             'ext': 'mp4',
  27             'title': 'Борисов: Притеснен съм за бъдещето на България',
  28             'description': 'По думите му е опасно страната ни да бъде обявена за "сигурна"',
  29             'thumbnail': r're:^https?://.*\.jpg$',
  30             'timestamp': 1470982814,
  31             'upload_date': '20160812',
  32             'uploader': 'zdraveibulgaria',
  33         },
  34         'params': {
  35             'proxy': '127.0.0.1:8118',
  36         },
  37     }, {
  38         'url': 'http://vbox7.com/play:249bb972c2',
  39         'md5': '99f65c0c9ef9b682b97313e052734c3f',
  40         'info_dict': {
  41             'id': '249bb972c2',
  42             'ext': 'mp4',
  43             'title': 'Смях! Чудо - чист за секунди - Скрита камера',
  44         },
  45         'skip': 'georestricted',
  46     }, {
  47         'url': 'http://vbox7.com/emb/external.php?vid=a240d20f9c&autoplay=1',
  48         'only_matching': True,
  49     }, {
  50         'url': 'http://i49.vbox7.com/player/ext.swf?vid=0946fff23c&autoplay=1',
  51         'only_matching': True,
  52     }]
  53
  54     @staticmethod
  55     def _extract_url(webpage):
  56         mobj = re.search(
  57             r'<iframe[^>]+src=(?P<q>["\'])(?P<url>(?:https?:)?//vbox7\.com/emb/external\.php.+?)(?P=q)',
  58             webpage)
  59         if mobj:
  60             return mobj.group('url')
  61
  62     def _real_extract(self, url):
  63         video_id = self._match_id(url)
  64
  65         response = self._download_json(
  66             'https://www.vbox7.com/ajax/video/nextvideo.php?vid=%s' % video_id,
  67             video_id)
  68
  69         if 'error' in response:
  70             raise ExtractorError(
  71                 '%s said: %s' % (self.IE_NAME, response['error']), expected=True)
  72
  73         video = response['options']
  74
  75         title = video['title']
  76         video_url = video['src']
  77
  78         if '/na.mp4' in video_url:
  79             self.raise_geo_restricted(countries=self._GEO_COUNTRIES)
  80
  81         uploader = video.get('uploader')
  82
  83         webpage = self._download_webpage(
  84             'http://vbox7.com/play:%s' % video_id, video_id, fatal=None)
  85
  86         info = {}
  87
  88         if webpage:
  89             info = self._search_json_ld(
  90                 webpage.replace('"/*@context"', '"@context"'), video_id,
  91                 fatal=False)
  92
  93         info.update({
  94             'id': video_id,
  95             'title': title,
  96             'url': video_url,
  97             'uploader': uploader,
  98             'thumbnail': self._proto_relative_url(
  99                 info.get('thumbnail') or self._og_search_thumbnail(webpage),
 100                 'http:'),
 101         })
 102         return info