yt_dlp/extractor/myspass.py

   1 from .common import InfoExtractor
   2 from ..compat import compat_str
   3 from ..utils import (
   4     int_or_none,
   5     parse_duration,
   6     xpath_text,
   7 )
   8
   9
  10 class MySpassIE(InfoExtractor):
  11     _VALID_URL = r'https?://(?:www\.)?myspass\.de/(?:[^/]+/)*(?P<id>\d+)/?[^/]*$'
  12     _TESTS = [{
  13         'url': 'http://www.myspass.de/myspass/shows/tvshows/absolute-mehrheit/Absolute-Mehrheit-vom-17022013-Die-Highlights-Teil-2--/11741/',
  14         'md5': '0b49f4844a068f8b33f4b7c88405862b',
  15         'info_dict': {
  16             'id': '11741',
  17             'ext': 'mp4',
  18             'description': 'md5:9f0db5044c8fe73f528a390498f7ce9b',
  19             'title': '17.02.2013 - Die Highlights, Teil 2',
  20             'thumbnail': r're:.*\.jpg',
  21             'duration': 323.0,
  22             'episode': '17.02.2013 - Die Highlights, Teil 2',
  23             'season_id': '544',
  24             'episode_number': 1,
  25             'series': 'Absolute Mehrheit',
  26             'season_number': 2,
  27             'season': 'Season 2',
  28         },
  29     },
  30         {
  31         'url': 'https://www.myspass.de/shows/tvshows/tv-total/Novak-Puffovic-bei-bester-Laune--/44996/',
  32         'md5': 'eb28b7c5e254192046e86ebaf7deac8f',
  33         'info_dict': {
  34             'id': '44996',
  35             'ext': 'mp4',
  36             'description': 'md5:74c7f886e00834417f1e427ab0da6121',
  37             'title': 'Novak Puffovic bei bester Laune',
  38             'thumbnail': r're:.*\.jpg',
  39             'episode_number': 8,
  40             'episode': 'Novak Puffovic bei bester Laune',
  41             'series': 'TV total',
  42             'season': 'Season 19',
  43             'season_id': '987',
  44             'duration': 2941.0,
  45             'season_number': 19,
  46         },
  47     },
  48         {
  49         'url': 'https://www.myspass.de/channels/tv-total-raabigramm/17033/20831/',
  50         'md5': '7b293a6b9f3a7acdd29304c8d0dbb7cc',
  51         'info_dict': {
  52             'id': '20831',
  53             'ext': 'mp4',
  54             'description': 'Gefühle pur: Schaut euch die ungeschnittene Version von Stefans Liebesbeweis an die Moderationsgrazie von Welt, Verona Feldbusch, an.',
  55             'title': 'Raabigramm Verona Feldbusch',
  56             'thumbnail': r're:.*\.jpg',
  57             'episode_number': 6,
  58             'episode': 'Raabigramm Verona Feldbusch',
  59             'series': 'TV total',
  60             'season': 'Season 1',
  61             'season_id': '34',
  62             'duration': 105.0,
  63             'season_number': 1,
  64         },
  65     }]
  66
  67     def _real_extract(self, url):
  68         video_id = self._match_id(url)
  69
  70         metadata = self._download_xml('http://www.myspass.de/myspass/includes/apps/video/getvideometadataxml.php?id=' + video_id, video_id)
  71
  72         title = xpath_text(metadata, 'title', fatal=True)
  73         video_url = xpath_text(metadata, 'url_flv', 'download url', True)
  74         video_id_int = int(video_id)
  75         for group in self._search_regex(r'/myspass2009/\d+/(\d+)/(\d+)/(\d+)/', video_url, 'myspass', group=(1, 2, 3), default=[]):
  76             group_int = int(group)
  77             if group_int > video_id_int:
  78                 video_url = video_url.replace(group, compat_str(group_int // video_id_int))
  79
  80         return {
  81             'id': video_id,
  82             'url': video_url,
  83             'title': title,
  84             'thumbnail': xpath_text(metadata, 'imagePreview'),
  85             'description': xpath_text(metadata, 'description'),
  86             'duration': parse_duration(xpath_text(metadata, 'duration')),
  87             'series': xpath_text(metadata, 'format'),
  88             'season_number': int_or_none(xpath_text(metadata, 'season')),
  89             'season_id': xpath_text(metadata, 'season_id'),
  90             'episode': title,
  91             'episode_number': int_or_none(xpath_text(metadata, 'episode')),
  92         }