yt_dlp/extractor/videocampus_sachsen.py

   1 import functools
   2 import re
   3
   4 from .common import InfoExtractor
   5 from ..networking.exceptions import HTTPError
   6 from ..utils import ExtractorError, OnDemandPagedList, urlencode_postdata
   7
   8
   9 class VideocampusSachsenIE(InfoExtractor):
  10     IE_NAME = 'ViMP'
  11     _INSTANCES = (
  12         'bergauf.tv',
  13         'campus.demo.vimp.com',
  14         'corporate.demo.vimp.com',
  15         'dancehalldatabase.com',
  16         'drehzahl.tv',
  17         'educhannel.hs-gesundheit.de',
  18         'emedia.ls.haw-hamburg.de',
  19         'globale-evolution.net',
  20         'hohu.tv',
  21         'htvideos.hightechhigh.org',
  22         'k210039.vimp.mivitec.net',
  23         'media.cmslegal.com',
  24         'media.hs-furtwangen.de',
  25         'media.hwr-berlin.de',
  26         'mediathek.dkfz.de',
  27         'mediathek.htw-berlin.de',
  28         'mediathek.polizei-bw.de',
  29         'medien.hs-merseburg.de',
  30         'mportal.europa-uni.de',
  31         'pacific.demo.vimp.com',
  32         'slctv.com',
  33         'streaming.prairiesouth.ca',
  34         'tube.isbonline.cn',
  35         'univideo.uni-kassel.de',
  36         'ursula2.genetics.emory.edu',
  37         'ursulablicklevideoarchiv.com',
  38         'v.agrarumweltpaedagogik.at',
  39         'video.eplay-tv.de',
  40         'video.fh-dortmund.de',
  41         'video.hs-offenburg.de',
  42         'video.hs-pforzheim.de',
  43         'video.hspv.nrw.de',
  44         'video.irtshdf.fr',
  45         'video.pareygo.de',
  46         'video.tu-freiberg.de',
  47         'videocampus.sachsen.de',
  48         'videoportal.uni-freiburg.de',
  49         'videoportal.vm.uni-freiburg.de',
  50         'videos.duoc.cl',
  51         'videos.uni-paderborn.de',
  52         'vimp-bemus.udk-berlin.de',
  53         'vimp.aekwl.de',
  54         'vimp.hs-mittweida.de',
  55         'vimp.oth-regensburg.de',
  56         'vimp.ph-heidelberg.de',
  57         'vimp.sma-events.com',
  58         'vimp.weka-fachmedien.de',
  59         'webtv.univ-montp3.fr',
  60         'www.b-tu.de/media',
  61         'www.bergauf.tv',
  62         'www.bigcitytv.de',
  63         'www.cad-videos.de',
  64         'www.drehzahl.tv',
  65         'www.fh-bielefeld.de/medienportal',
  66         'www.hohu.tv',
  67         'www.orvovideo.com',
  68         'www.rwe.tv',
  69         'www.salzi.tv',
  70         'www.wenglor-media.com',
  71         'www2.univ-sba.dz',
  72     )
  73     _VALID_URL = r'''(?x)https?://(?P<host>{})/(?:
  74         m/(?P<tmp_id>[0-9a-f]+)|
  75         (?:category/)?video/(?P<display_id>[\w-]+)/(?P<id>[0-9a-f]{{32}})|
  76         media/embed.*(?:\?|&)key=(?P<embed_id>[0-9a-f]{{32}}&?)
  77     )'''.format('|'.join(map(re.escape, _INSTANCES)))
  78
  79     _TESTS = [
  80         {
  81             'url': 'https://videocampus.sachsen.de/m/e0d6c8ce6e394c188f1342f1ab7c50ed6fc4490b808699801def5cb2e46d76ca7367f622a9f516c542ffb805b24d6b643bd7c81f385acaac4c59081b87a2767b',
  82             'info_dict': {
  83                 'id': 'e6b9349905c1628631f175712250f2a1',
  84                 'title': 'Konstruktiver Entwicklungsprozess Vorlesung 7',
  85                 'description': 'Konstruktiver Entwicklungsprozess Vorlesung 7',
  86                 'thumbnail': 'https://videocampus.sachsen.de/cache/1a985379ad3aecba8097a6902c7daa4e.jpg',
  87                 'ext': 'mp4',
  88             },
  89         },
  90         {
  91             'url': 'https://videocampus.sachsen.de/video/Was-ist-selbstgesteuertes-Lernen/fc99c527e4205b121cb7c74433469262',
  92             'info_dict': {
  93                 'id': 'fc99c527e4205b121cb7c74433469262',
  94                 'title': 'Was ist selbstgesteuertes Lernen?',
  95                 'description': 'md5:196aa3b0509a526db62f84679522a2f5',
  96                 'thumbnail': 'https://videocampus.sachsen.de/cache/6f4a85096ba24cb398e6ce54446b57ae.jpg',
  97                 'display_id': 'Was-ist-selbstgesteuertes-Lernen',
  98                 'ext': 'mp4',
  99             },
 100         },
 101         {
 102             'url': 'https://videocampus.sachsen.de/category/video/Tutorial-zur-Nutzung-von-Adobe-Connect-aus-Veranstalter-Sicht/09d4ed029002eb1bdda610f1103dd54c/100',
 103             'info_dict': {
 104                 'id': '09d4ed029002eb1bdda610f1103dd54c',
 105                 'title': 'Tutorial zur Nutzung von Adobe Connect aus Veranstalter-Sicht',
 106                 'description': 'md5:3d379ca3cc17b9da6784d7f58cca4d58',
 107                 'thumbnail': 'https://videocampus.sachsen.de/cache/2452498fe8c2d5a7dc79a05d30f407b6.jpg',
 108                 'display_id': 'Tutorial-zur-Nutzung-von-Adobe-Connect-aus-Veranstalter-Sicht',
 109                 'ext': 'mp4',
 110             },
 111         },
 112         {
 113             'url': 'https://www2.univ-sba.dz/video/Presentation-de-la-Faculte-de-droit-et-des-sciences-politiques-Journee-portes-ouvertes-202122/0183356e41af7bfb83d7667b20d9b6a3',
 114             'info_dict': {
 115                 'url': 'https://www2.univ-sba.dz/getMedium/0183356e41af7bfb83d7667b20d9b6a3.mp4',
 116                 'id': '0183356e41af7bfb83d7667b20d9b6a3',
 117                 'title': 'Présentation de la Faculté de droit et des sciences politiques - Journée portes ouvertes 2021/22',
 118                 'description': 'md5:508958bd93e0ca002ac731d94182a54f',
 119                 'thumbnail': 'https://www2.univ-sba.dz/cache/4d5d4a0b4189271a8cc6cb5328e14769.jpg',
 120                 'display_id': 'Presentation-de-la-Faculte-de-droit-et-des-sciences-politiques-Journee-portes-ouvertes-202122',
 121                 'ext': 'mp4',
 122             },
 123         },
 124         {
 125             'url': 'https://vimp.weka-fachmedien.de/video/Preisverleihung-Produkte-des-Jahres-2022/c8816f1cc942c12b6cce57c835cffd7c',
 126             'info_dict': {
 127                 'id': 'c8816f1cc942c12b6cce57c835cffd7c',
 128                 'title': 'Preisverleihung »Produkte des Jahres 2022«',
 129                 'description': 'md5:60c347568ca89aa25b772c4ea564ebd3',
 130                 'thumbnail': 'https://vimp.weka-fachmedien.de/cache/da9f3090e9227b25beacf67ccf94de14.png',
 131                 'display_id': 'Preisverleihung-Produkte-des-Jahres-2022',
 132                 'ext': 'mp4',
 133             },
 134         },
 135         {
 136             'url': 'https://videocampus.sachsen.de/media/embed?key=fc99c527e4205b121cb7c74433469262',
 137             'info_dict': {
 138                 'id': 'fc99c527e4205b121cb7c74433469262',
 139                 'title': 'Was ist selbstgesteuertes Lernen?',
 140                 'ext': 'mp4',
 141             },
 142         },
 143     ]
 144
 145     def _real_extract(self, url):
 146         host, video_id, tmp_id, display_id, embed_id = self._match_valid_url(url).group(
 147             'host', 'id', 'tmp_id', 'display_id', 'embed_id')
 148         webpage = self._download_webpage(url, video_id or tmp_id, fatal=False) or ''
 149
 150         if not video_id:
 151             video_id = embed_id or self._html_search_regex(
 152                 rf'src="https?://{host}/media/embed.*(?:\?|&)key=([0-9a-f]+)&?',
 153                 webpage, 'video_id')
 154
 155         if not (display_id or tmp_id):
 156             # Title, description from embedded page's meta wouldn't be correct
 157             title = self._html_search_regex(r'<video-js[^>]* data-piwik-title="([^"<]+)"', webpage, 'title', fatal=False)
 158             description = None
 159             thumbnail = None
 160         else:
 161             title = self._html_search_meta(('og:title', 'twitter:title', 'title'), webpage, fatal=False)
 162             description = self._html_search_meta(
 163                 ('og:description', 'twitter:description', 'description'), webpage, fatal=False)
 164             thumbnail = self._html_search_meta(('og:image', 'twitter:image'), webpage, fatal=False)
 165
 166         formats, subtitles = [], {}
 167         try:
 168             formats, subtitles = self._extract_m3u8_formats_and_subtitles(
 169                 f'https://{host}/media/hlsMedium/key/{video_id}/format/auto/ext/mp4/learning/0/path/m3u8',
 170                 video_id, 'mp4', m3u8_id='hls', fatal=True)
 171         except ExtractorError as e:
 172             if not isinstance(e.cause, HTTPError) or e.cause.status not in (404, 500):
 173                 raise
 174
 175         formats.append({'url': f'https://{host}/getMedium/{video_id}.mp4'})
 176
 177         return {
 178             'id': video_id,
 179             'title': title,
 180             'description': description,
 181             'thumbnail': thumbnail,
 182             'display_id': display_id,
 183             'formats': formats,
 184             'subtitles': subtitles,
 185         }
 186
 187
 188 class ViMPPlaylistIE(InfoExtractor):
 189     IE_NAME = 'ViMP:Playlist'
 190     _VALID_URL = r'''(?x)(?P<host>https?://(?:{}))/(?:
 191         album/view/aid/(?P<album_id>[0-9]+)|
 192         (?P<mode>category|channel)/(?P<name>[\w-]+)/(?P<id>[0-9]+)
 193     )'''.format('|'.join(map(re.escape, VideocampusSachsenIE._INSTANCES)))
 194
 195     _TESTS = [{
 196         'url': 'https://vimp.oth-regensburg.de/channel/Designtheorie-1-SoSe-2020/3',
 197         'info_dict': {
 198             'id': 'channel-3',
 199             'title': 'Designtheorie 1 SoSe 2020 :: Channels :: ViMP OTH Regensburg',
 200         },
 201         'playlist_mincount': 9,
 202     }, {
 203         'url': 'https://www.fh-bielefeld.de/medienportal/album/view/aid/208',
 204         'info_dict': {
 205             'id': 'album-208',
 206             'title': 'KG Praktikum ABT/MEC :: Playlists :: FH-Medienportal',
 207         },
 208         'playlist_mincount': 4,
 209     }, {
 210         'url': 'https://videocampus.sachsen.de/category/online-tutorials-onyx/91',
 211         'info_dict': {
 212             'id': 'category-91',
 213             'title': 'Online-Seminare ONYX - BPS - Bildungseinrichtungen - VCS',
 214         },
 215         'playlist_mincount': 7,
 216     }]
 217     _PAGE_SIZE = 10
 218
 219     def _fetch_page(self, host, url_part, playlist_id, data, page):
 220         webpage = self._download_webpage(
 221             f'{host}/media/ajax/component/boxList/{url_part}', playlist_id,
 222             query={'page': page, 'page_only': 1}, data=urlencode_postdata(data))
 223         urls = re.findall(r'"([^"]+/video/[^"]+)"', webpage)
 224
 225         for url in urls:
 226             yield self.url_result(host + url, VideocampusSachsenIE)
 227
 228     def _real_extract(self, url):
 229         host, album_id, mode, name, playlist_id = self._match_valid_url(url).group(
 230             'host', 'album_id', 'mode', 'name', 'id')
 231
 232         webpage = self._download_webpage(url, album_id or playlist_id, fatal=False) or ''
 233         title = (self._html_search_meta('title', webpage, fatal=False)
 234                  or self._html_extract_title(webpage))
 235
 236         url_part = (f'aid/{album_id}' if album_id
 237                     else f'category/{name}/category_id/{playlist_id}' if mode == 'category'
 238                     else f'title/{name}/channel/{playlist_id}')
 239
 240         mode = mode or 'album'
 241         data = {
 242             'vars[mode]': mode,
 243             f'vars[{mode}]': album_id or playlist_id,
 244             'vars[context]': '4' if album_id else '1' if mode == 'category' else '3',
 245             'vars[context_id]': album_id or playlist_id,
 246             'vars[layout]': 'thumb',
 247             'vars[per_page][thumb]': str(self._PAGE_SIZE),
 248         }
 249
 250         return self.playlist_result(
 251             OnDemandPagedList(functools.partial(
 252                 self._fetch_page, host, url_part, album_id or playlist_id, data), self._PAGE_SIZE),
 253             playlist_title=title, id=f'{mode}-{album_id or playlist_id}')