Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/tunepk.py

   1 from __future__ import unicode_literals
   2
   3 from .common import InfoExtractor
   4 from ..compat import compat_str
   5 from ..utils import (
   6     int_or_none,
   7     try_get,
   8     unified_timestamp,
   9 )
  10
  11
  12 class TunePkIE(InfoExtractor):
  13     _VALID_URL = r'''(?x)
  14                     https?://
  15                         (?:
  16                             (?:www\.)?tune\.pk/(?:video/|player/embed_player.php?.*?\bvid=)|
  17                             embed\.tune\.pk/play/
  18                         )
  19                         (?P<id>\d+)
  20                     '''
  21     _TESTS = [{
  22         'url': 'https://tune.pk/video/6919541/maudie-2017-international-trailer-1-ft-ethan-hawke-sally-hawkins',
  23         'md5': '0c537163b7f6f97da3c5dd1e3ef6dd55',
  24         'info_dict': {
  25             'id': '6919541',
  26             'ext': 'mp4',
  27             'title': 'Maudie (2017) | International Trailer # 1 ft Ethan Hawke, Sally Hawkins',
  28             'description': 'md5:eb5a04114fafef5cec90799a93a2d09c',
  29             'thumbnail': r're:^https?://.*\.jpg$',
  30             'timestamp': 1487327564,
  31             'upload_date': '20170217',
  32             'uploader': 'Movie Trailers',
  33             'duration': 107,
  34             'view_count': int,
  35         }
  36     }, {
  37         'url': 'https://tune.pk/player/embed_player.php?vid=6919541&folder=2017/02/17/&width=600&height=350&autoplay=no',
  38         'only_matching': True,
  39     }, {
  40         'url': 'https://embed.tune.pk/play/6919541?autoplay=no&ssl=yes&inline=true',
  41         'only_matching': True,
  42     }]
  43
  44     def _real_extract(self, url):
  45         video_id = self._match_id(url)
  46
  47         webpage = self._download_webpage(
  48             'https://tune.pk/video/%s' % video_id, video_id)
  49
  50         details = self._parse_json(
  51             self._search_regex(
  52                 r'new\s+TunePlayer\(({.+?})\)\s*;\s*\n', webpage, 'tune player'),
  53             video_id)['details']
  54
  55         video = details['video']
  56         title = video.get('title') or self._og_search_title(
  57             webpage, default=None) or self._html_search_meta(
  58             'title', webpage, 'title', fatal=True)
  59
  60         formats = self._parse_jwplayer_formats(
  61             details['player']['sources'], video_id)
  62         self._sort_formats(formats)
  63
  64         description = self._og_search_description(
  65             webpage, default=None) or self._html_search_meta(
  66             'description', webpage, 'description')
  67
  68         thumbnail = video.get('thumb') or self._og_search_thumbnail(
  69             webpage, default=None) or self._html_search_meta(
  70             'thumbnail', webpage, 'thumbnail')
  71
  72         timestamp = unified_timestamp(video.get('date_added'))
  73         uploader = try_get(
  74             video, lambda x: x['uploader']['name'],
  75             compat_str) or self._html_search_meta('author', webpage, 'author')
  76
  77         duration = int_or_none(video.get('duration'))
  78         view_count = int_or_none(video.get('views'))
  79
  80         return {
  81             'id': video_id,
  82             'title': title,
  83             'description': description,
  84             'thumbnail': thumbnail,
  85             'timestamp': timestamp,
  86             'uploader': uploader,
  87             'duration': duration,
  88             'view_count': view_count,
  89             'formats': formats,
  90         }