yt_dlp/extractor/tunepk.py

   1 from .common import InfoExtractor
   2 from ..compat import compat_str
   3 from ..utils import (
   4     int_or_none,
   5     try_get,
   6     unified_timestamp,
   7 )
   8
   9
  10 class TunePkIE(InfoExtractor):
  11     _VALID_URL = r'''(?x)
  12                     https?://
  13                         (?:
  14                             (?:www\.)?tune\.pk/(?:video/|player/embed_player.php?.*?\bvid=)|
  15                             embed\.tune\.pk/play/
  16                         )
  17                         (?P<id>\d+)
  18                     '''
  19     _TESTS = [{
  20         'url': 'https://tune.pk/video/6919541/maudie-2017-international-trailer-1-ft-ethan-hawke-sally-hawkins',
  21         'md5': '0c537163b7f6f97da3c5dd1e3ef6dd55',
  22         'info_dict': {
  23             'id': '6919541',
  24             'ext': 'mp4',
  25             'title': 'Maudie (2017) | International Trailer # 1 ft Ethan Hawke, Sally Hawkins',
  26             'description': 'md5:eb5a04114fafef5cec90799a93a2d09c',
  27             'thumbnail': r're:^https?://.*\.jpg$',
  28             'timestamp': 1487327564,
  29             'upload_date': '20170217',
  30             'uploader': 'Movie Trailers',
  31             'duration': 107,
  32             'view_count': int,
  33         }
  34     }, {
  35         'url': 'https://tune.pk/player/embed_player.php?vid=6919541&folder=2017/02/17/&width=600&height=350&autoplay=no',
  36         'only_matching': True,
  37     }, {
  38         'url': 'https://embed.tune.pk/play/6919541?autoplay=no&ssl=yes&inline=true',
  39         'only_matching': True,
  40     }]
  41
  42     def _real_extract(self, url):
  43         video_id = self._match_id(url)
  44
  45         webpage = self._download_webpage(
  46             'https://tune.pk/video/%s' % video_id, video_id)
  47
  48         details = self._parse_json(
  49             self._search_regex(
  50                 r'new\s+TunePlayer\(({.+?})\)\s*;\s*\n', webpage, 'tune player'),
  51             video_id)['details']
  52
  53         video = details['video']
  54         title = video.get('title') or self._og_search_title(
  55             webpage, default=None) or self._html_search_meta(
  56             'title', webpage, 'title', fatal=True)
  57
  58         formats = self._parse_jwplayer_formats(
  59             details['player']['sources'], video_id)
  60
  61         description = self._og_search_description(
  62             webpage, default=None) or self._html_search_meta(
  63             'description', webpage, 'description')
  64
  65         thumbnail = video.get('thumb') or self._og_search_thumbnail(
  66             webpage, default=None) or self._html_search_meta(
  67             'thumbnail', webpage, 'thumbnail')
  68
  69         timestamp = unified_timestamp(video.get('date_added'))
  70         uploader = try_get(
  71             video, lambda x: x['uploader']['name'],
  72             compat_str) or self._html_search_meta('author', webpage, 'author')
  73
  74         duration = int_or_none(video.get('duration'))
  75         view_count = int_or_none(video.get('views'))
  76
  77         return {
  78             'id': video_id,
  79             'title': title,
  80             'description': description,
  81             'thumbnail': thumbnail,
  82             'timestamp': timestamp,
  83             'uploader': uploader,
  84             'duration': duration,
  85             'view_count': view_count,
  86             'formats': formats,
  87         }