yt_dlp/extractor/videocampus_sachsen.py

   1 import re
   2
   3 from .common import InfoExtractor
   4 from ..compat import compat_HTTPError
   5 from ..utils import ExtractorError
   6
   7
   8 class VideocampusSachsenIE(InfoExtractor):
   9     IE_NAME = 'Vimp'
  10     _INSTANCES = (
  11         'campus.demo.vimp.com',
  12         'corporate.demo.vimp.com',
  13         'dancehalldatabase.com',
  14         'educhannel.hs-gesundheit.de',
  15         'emedia.ls.haw-hamburg.de',
  16         'globale-evolution.net',
  17         'k210039.vimp.mivitec.net',
  18         'media.cmslegal.com',
  19         'media.hs-furtwangen.de',
  20         'media.hwr-berlin.de',
  21         'mediathek.dkfz.de',
  22         'mediathek.htw-berlin.de',
  23         'mediathek.polizei-bw.de',
  24         'medien.hs-merseburg.de',
  25         'mportal.europa-uni.de',
  26         'pacific.demo.vimp.com',
  27         'slctv.com',
  28         'tube.isbonline.cn',
  29         'univideo.uni-kassel.de',
  30         'ursula2.genetics.emory.edu',
  31         'ursulablicklevideoarchiv.com',
  32         'v.agrarumweltpaedagogik.at',
  33         'video.eplay-tv.de',
  34         'video.fh-dortmund.de',
  35         'video.hs-offenburg.de',
  36         'video.hs-pforzheim.de',
  37         'video.hspv.nrw.de',
  38         'video.irtshdf.fr',
  39         'video.pareygo.de',
  40         'video.tu-freiberg.de',
  41         'videocampus.sachsen.de',
  42         'videoportal.uni-freiburg.de',
  43         'videoportal.vm.uni-freiburg.de',
  44         'videos.duoc.cl',
  45         'videos.uni-paderborn.de',
  46         'vimp-bemus.udk-berlin.de',
  47         'vimp.aekwl.de',
  48         'vimp.hs-mittweida.de',
  49         'vimp.oth-regensburg.de',
  50         'vimp.ph-heidelberg.de',
  51         'vimp.sma-events.com',
  52         'vimp.weka-fachmedien.de',
  53         'webtv.univ-montp3.fr',
  54         'www.b-tu.de/media',
  55         'www.bigcitytv.de',
  56         'www.cad-videos.de',
  57         'www.fh-bielefeld.de/medienportal',
  58         'www.orvovideo.com',
  59         'www.rwe.tv',
  60         'www.wenglor-media.com',
  61         'www2.univ-sba.dz',
  62     )
  63     _VALID_URL = r'''(?x)https?://(?P<host>%s)/(?:
  64         m/(?P<tmp_id>[0-9a-f]+)|
  65         (?:category/)?video/(?P<display_id>[\w-]+)/(?P<id>[0-9a-f]{32})|
  66         media/embed.*(?:\?|&)key=(?P<embed_id>[0-9a-f]{32}&?)
  67     )''' % ('|'.join(map(re.escape, _INSTANCES)))
  68
  69     _TESTS = [
  70         {
  71             'url': 'https://videocampus.sachsen.de/m/e0d6c8ce6e394c188f1342f1ab7c50ed6fc4490b808699801def5cb2e46d76ca7367f622a9f516c542ffb805b24d6b643bd7c81f385acaac4c59081b87a2767b',
  72             'info_dict': {
  73                 'id': 'e6b9349905c1628631f175712250f2a1',
  74                 'title': 'Konstruktiver Entwicklungsprozess Vorlesung 7',
  75                 'description': 'Konstruktiver Entwicklungsprozess Vorlesung 7',
  76                 'ext': 'mp4',
  77             },
  78         },
  79         {
  80             'url': 'https://videocampus.sachsen.de/video/Was-ist-selbstgesteuertes-Lernen/fc99c527e4205b121cb7c74433469262',
  81             'info_dict': {
  82                 'id': 'fc99c527e4205b121cb7c74433469262',
  83                 'title': 'Was ist selbstgesteuertes Lernen?',
  84                 'description': 'md5:196aa3b0509a526db62f84679522a2f5',
  85                 'display_id': 'Was-ist-selbstgesteuertes-Lernen',
  86                 'ext': 'mp4',
  87             },
  88         },
  89         {
  90             'url': 'https://videocampus.sachsen.de/category/video/Tutorial-zur-Nutzung-von-Adobe-Connect-aus-Veranstalter-Sicht/09d4ed029002eb1bdda610f1103dd54c/100',
  91             'info_dict': {
  92                 'id': '09d4ed029002eb1bdda610f1103dd54c',
  93                 'title': 'Tutorial zur Nutzung von Adobe Connect aus Veranstalter-Sicht',
  94                 'description': 'md5:3d379ca3cc17b9da6784d7f58cca4d58',
  95                 'display_id': 'Tutorial-zur-Nutzung-von-Adobe-Connect-aus-Veranstalter-Sicht',
  96                 'ext': 'mp4',
  97             },
  98         },
  99         {
 100             'url': 'https://www2.univ-sba.dz/video/Presentation-de-la-Faculte-de-droit-et-des-sciences-politiques-Journee-portes-ouvertes-202122/0183356e41af7bfb83d7667b20d9b6a3',
 101             'info_dict': {
 102                 'url': 'https://www2.univ-sba.dz/getMedium/0183356e41af7bfb83d7667b20d9b6a3.mp4',
 103                 'id': '0183356e41af7bfb83d7667b20d9b6a3',
 104                 'title': 'Présentation de la Faculté de droit et des sciences politiques - Journée portes ouvertes 2021/22',
 105                 'description': 'md5:508958bd93e0ca002ac731d94182a54f',
 106                 'display_id': 'Presentation-de-la-Faculte-de-droit-et-des-sciences-politiques-Journee-portes-ouvertes-202122',
 107                 'ext': 'mp4',
 108             }
 109         },
 110         {
 111             'url': 'https://vimp.weka-fachmedien.de/video/Preisverleihung-Produkte-des-Jahres-2022/c8816f1cc942c12b6cce57c835cffd7c',
 112             'info_dict': {
 113                 'id': 'c8816f1cc942c12b6cce57c835cffd7c',
 114                 'title': 'Preisverleihung »Produkte des Jahres 2022«',
 115                 'description': 'md5:60c347568ca89aa25b772c4ea564ebd3',
 116                 'display_id': 'Preisverleihung-Produkte-des-Jahres-2022',
 117                 'ext': 'mp4',
 118             },
 119         },
 120         {
 121             'url': 'https://videocampus.sachsen.de/media/embed?key=fc99c527e4205b121cb7c74433469262',
 122             'info_dict': {
 123                 'id': 'fc99c527e4205b121cb7c74433469262',
 124                 'title': 'Was ist selbstgesteuertes Lernen?',
 125                 'ext': 'mp4',
 126             },
 127         }
 128     ]
 129
 130     def _real_extract(self, url):
 131         host, video_id, tmp_id, display_id, embed_id = self._match_valid_url(url).group(
 132             'host', 'id', 'tmp_id', 'display_id', 'embed_id')
 133         webpage = self._download_webpage(url, video_id or tmp_id, fatal=False) or ''
 134
 135         if not video_id:
 136             video_id = embed_id or self._html_search_regex(
 137                 rf'src="https?://{host}/media/embed.*(?:\?|&)key=([0-9a-f]+)&?',
 138                 webpage, 'video_id')
 139
 140         if not (display_id or tmp_id):
 141             # Title, description from embedded page's meta wouldn't be correct
 142             title = self._html_search_regex(r'<img[^>]* title="([^"<]+)"', webpage, 'title', fatal=False)
 143             description = None
 144         else:
 145             title = self._html_search_meta(('og:title', 'twitter:title', 'title'), webpage, fatal=False)
 146             description = self._html_search_meta(
 147                 ('og:description', 'twitter:description', 'description'), webpage, default=None)
 148
 149         formats, subtitles = [], {}
 150         try:
 151             formats, subtitles = self._extract_m3u8_formats_and_subtitles(
 152                 f'https://{host}/media/hlsMedium/key/{video_id}/format/auto/ext/mp4/learning/0/path/m3u8',
 153                 video_id, 'mp4', m3u8_id='hls', fatal=True)
 154         except ExtractorError as e:
 155             if not isinstance(e.cause, compat_HTTPError) or e.cause.code not in (404, 500):
 156                 raise
 157
 158         formats.append({'url': f'https://{host}/getMedium/{video_id}.mp4'})
 159         self._sort_formats(formats)
 160
 161         return {
 162             'id': video_id,
 163             'title': title,
 164             'description': description,
 165             'display_id': display_id,
 166             'formats': formats,
 167             'subtitles': subtitles
 168         }