yt_dlp/extractor/myspass.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .common import InfoExtractor
   5 from ..compat import compat_str
   6 from ..utils import (
   7     int_or_none,
   8     parse_duration,
   9     xpath_text,
  10 )
  11
  12
  13 class MySpassIE(InfoExtractor):
  14     _VALID_URL = r'https?://(?:www\.)?myspass\.de/(?:[^/]+/)*(?P<id>\d+)/?[^/]*$'
  15     _TESTS = [{
  16         'url': 'http://www.myspass.de/myspass/shows/tvshows/absolute-mehrheit/Absolute-Mehrheit-vom-17022013-Die-Highlights-Teil-2--/11741/',
  17         'md5': '0b49f4844a068f8b33f4b7c88405862b',
  18         'info_dict': {
  19             'id': '11741',
  20             'ext': 'mp4',
  21             'description': 'md5:9f0db5044c8fe73f528a390498f7ce9b',
  22             'title': '17.02.2013 - Die Highlights, Teil 2',
  23             'thumbnail': r're:.*\.jpg',
  24             'duration': 323.0,
  25             'episode': '17.02.2013 - Die Highlights, Teil 2',
  26             'season_id': '544',
  27             'episode_number': 1,
  28             'series': 'Absolute Mehrheit',
  29             'season_number': 2,
  30             'season': 'Season 2',
  31         },
  32     },
  33         {
  34         'url': 'https://www.myspass.de/shows/tvshows/tv-total/Novak-Puffovic-bei-bester-Laune--/44996/',
  35         'md5': 'eb28b7c5e254192046e86ebaf7deac8f',
  36         'info_dict': {
  37             'id': '44996',
  38             'ext': 'mp4',
  39             'description': 'md5:74c7f886e00834417f1e427ab0da6121',
  40             'title': 'Novak Puffovic bei bester Laune',
  41             'thumbnail': r're:.*\.jpg',
  42             'episode_number': 8,
  43             'episode': 'Novak Puffovic bei bester Laune',
  44             'series': 'TV total',
  45             'season': 'Season 19',
  46             'season_id': '987',
  47             'duration': 2941.0,
  48             'season_number': 19,
  49         },
  50     },
  51         {
  52         'url': 'https://www.myspass.de/channels/tv-total-raabigramm/17033/20831/',
  53         'md5': '7b293a6b9f3a7acdd29304c8d0dbb7cc',
  54         'info_dict': {
  55             'id': '20831',
  56             'ext': 'mp4',
  57             'description': 'Gefühle pur: Schaut euch die ungeschnittene Version von Stefans Liebesbeweis an die Moderationsgrazie von Welt, Verona Feldbusch, an.',
  58             'title': 'Raabigramm Verona Feldbusch',
  59             'thumbnail': r're:.*\.jpg',
  60             'episode_number': 6,
  61             'episode': 'Raabigramm Verona Feldbusch',
  62             'series': 'TV total',
  63             'season': 'Season 1',
  64             'season_id': '34',
  65             'duration': 105.0,
  66             'season_number': 1,
  67         },
  68     }]
  69
  70     def _real_extract(self, url):
  71         video_id = self._match_id(url)
  72
  73         metadata = self._download_xml('http://www.myspass.de/myspass/includes/apps/video/getvideometadataxml.php?id=' + video_id, video_id)
  74
  75         title = xpath_text(metadata, 'title', fatal=True)
  76         video_url = xpath_text(metadata, 'url_flv', 'download url', True)
  77         video_id_int = int(video_id)
  78         for group in self._search_regex(r'/myspass2009/\d+/(\d+)/(\d+)/(\d+)/', video_url, 'myspass', group=(1, 2, 3), default=[]):
  79             group_int = int(group)
  80             if group_int > video_id_int:
  81                 video_url = video_url.replace(group, compat_str(group_int // video_id_int))
  82
  83         return {
  84             'id': video_id,
  85             'url': video_url,
  86             'title': title,
  87             'thumbnail': xpath_text(metadata, 'imagePreview'),
  88             'description': xpath_text(metadata, 'description'),
  89             'duration': parse_duration(xpath_text(metadata, 'duration')),
  90             'series': xpath_text(metadata, 'format'),
  91             'season_number': int_or_none(xpath_text(metadata, 'season')),
  92             'season_id': xpath_text(metadata, 'season_id'),
  93             'episode': title,
  94             'episode_number': int_or_none(xpath_text(metadata, 'episode')),
  95         }