yt_dlp/extractor/germanupa.py

   1 from .common import InfoExtractor
   2 from .vimeo import VimeoIE
   3 from ..utils import (
   4     parse_qs,
   5     traverse_obj,
   6     url_or_none,
   7 )
   8
   9
  10 class GermanupaIE(InfoExtractor):
  11     IE_DESC = 'germanupa.de'
  12     _VALID_URL = r'https?://germanupa\.de/mediathek/(?P<id>[\w-]+)'
  13     _TESTS = [{
  14         'url': 'https://germanupa.de/mediathek/4-figma-beratung-deine-sprechstunde-fuer-figma-fragen',
  15         'info_dict': {
  16             'id': '909179246',
  17             'title': 'Tutorial: #4 Figma Beratung - Deine Sprechstunde für Figma-Fragen',
  18             'ext': 'mp4',
  19             'uploader': 'German UPA',
  20             'uploader_id': 'germanupa',
  21             'thumbnail': 'https://i.vimeocdn.com/video/1792564420-7415283ccef8bf8702dab8c6b7515555ceeb7a1c11371ffcc133b8e887dbf70e-d_1280',
  22             'uploader_url': 'https://vimeo.com/germanupa',
  23             'duration': 3987,
  24         },
  25         'expected_warnings': ['Failed to parse XML: not well-formed'],
  26         'params': {'skip_download': 'm3u8'},
  27     }, {
  28         'note': 'audio, uses GenericIE',
  29         'url': 'https://germanupa.de/mediathek/live-vom-ux-festival-neuigkeiten-von-figma-jobmarkt-agenturszene-interview-zu-sustainable',
  30         'info_dict': {
  31             'id': '1867346676',
  32             'title': 'Live vom UX Festival: Neuigkeiten von Figma, Jobmarkt, Agenturszene & Interview zu Sustainable UX',
  33             'ext': 'opus',
  34             'timestamp': 1720545088,
  35             'upload_date': '20240709',
  36             'duration': 3910.557,
  37             'like_count': int,
  38             'description': 'md5:db2aed5ff131e177a7b33901e9a8db05',
  39             'uploader': 'German UPA',
  40             'repost_count': int,
  41             'genres': ['Science'],
  42             'license': 'all-rights-reserved',
  43             'uploader_url': 'https://soundcloud.com/user-80097677',
  44             'uploader_id': '471579486',
  45             'view_count': int,
  46             'comment_count': int,
  47             'thumbnail': 'https://i1.sndcdn.com/artworks-oCti2e9GhaZFWBqY-48ybGw-original.jpg',
  48         },
  49     }, {
  50         'note': 'Nur für Mitglieder/Just for members',
  51         'url': 'https://germanupa.de/mediathek/ux-festival-2024-usability-tests-und-ai',
  52         'info_dict': {
  53             'id': '986994430',
  54             'title': 'UX Festival 2024 "Usability Tests und AI" von Lennart Weber',
  55             'ext': 'mp4',
  56             'release_date': '20240719',
  57             'uploader_url': 'https://vimeo.com/germanupa',
  58             'timestamp': 1721373980,
  59             'license': 'by-sa',
  60             'like_count': int,
  61             'thumbnail': 'https://i.vimeocdn.com/video/1904187064-2a672630c30f9ad787bd390bff3f51d7506a3e8416763ba6dbf465732b165c5c-d_1280',
  62             'duration': 2146,
  63             'release_timestamp': 1721373980,
  64             'uploader': 'German UPA',
  65             'uploader_id': 'germanupa',
  66             'upload_date': '20240719',
  67             'comment_count': int,
  68         },
  69         'expected_warnings': ['Failed to parse XML: not well-formed'],
  70         'skip': 'login required',
  71     }]
  72
  73     def _real_extract(self, url):
  74         video_id = self._match_id(url)
  75         webpage = self._download_webpage(url, video_id)
  76
  77         param_url = traverse_obj(
  78             self._search_regex(
  79                 r'<iframe[^>]+data-src\s*?=\s*?([\'"])(?P<url>https://germanupa\.de/media/oembed\?url=(?:(?!\1).)+)\1',
  80                 webpage, 'embedded video', default=None, group='url'),
  81             ({parse_qs}, 'url', 0, {url_or_none}))
  82
  83         if not param_url:
  84             if self._search_regex(
  85                     r'<div[^>]+class\s*?=\s*?([\'"])(?:(?!\1).)*login-wrapper(?:(?!\1).)*\1',
  86                     webpage, 'login wrapper', default=None):
  87                 self.raise_login_required('This video is only available for members')
  88             return self.url_result(url, 'Generic')  # Fall back to generic to extract audio
  89
  90         real_url = param_url.replace('https://vimeo.com/', 'https://player.vimeo.com/video/')
  91         return self.url_result(VimeoIE._smuggle_referrer(real_url, url), VimeoIE, video_id)