[VideocampusSachsen] Improve extractor (#3604)

Authored by: FestplattenSchnitzel
This commit is contained in:
FestplattenSchnitzel 2022-05-05 19:31:54 +02:00 committed by GitHub
parent ff4d7860d5
commit 10fa2471fc
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
2 changed files with 117 additions and 47 deletions

View File

@ -1899,10 +1899,7 @@
from .vidbit import VidbitIE from .vidbit import VidbitIE
from .viddler import ViddlerIE from .viddler import ViddlerIE
from .videa import VideaIE from .videa import VideaIE
from .videocampus_sachsen import ( from .videocampus_sachsen import VideocampusSachsenIE
VideocampusSachsenIE,
VideocampusSachsenEmbedIE,
)
from .videodetective import VideoDetectiveIE from .videodetective import VideoDetectiveIE
from .videofyme import VideofyMeIE from .videofyme import VideofyMeIE
from .videomore import ( from .videomore import (

View File

@ -1,11 +1,70 @@
import re
from .common import InfoExtractor from .common import InfoExtractor
from ..compat import compat_HTTPError
from ..utils import ExtractorError
class VideocampusSachsenIE(InfoExtractor): class VideocampusSachsenIE(InfoExtractor):
_VALID_URL = r'''(?x)https?://videocampus\.sachsen\.de/(?: IE_NAME = 'Vimp'
_INSTANCES = (
'campus.demo.vimp.com',
'corporate.demo.vimp.com',
'dancehalldatabase.com',
'educhannel.hs-gesundheit.de',
'emedia.ls.haw-hamburg.de',
'globale-evolution.net',
'k210039.vimp.mivitec.net',
'media.cmslegal.com',
'media.hs-furtwangen.de',
'media.hwr-berlin.de',
'mediathek.dkfz.de',
'mediathek.htw-berlin.de',
'mediathek.polizei-bw.de',
'medien.hs-merseburg.de',
'mportal.europa-uni.de',
'pacific.demo.vimp.com',
'slctv.com',
'tube.isbonline.cn',
'univideo.uni-kassel.de',
'ursula2.genetics.emory.edu',
'ursulablicklevideoarchiv.com',
'v.agrarumweltpaedagogik.at',
'video.eplay-tv.de',
'video.fh-dortmund.de',
'video.hs-offenburg.de',
'video.hs-pforzheim.de',
'video.hspv.nrw.de',
'video.irtshdf.fr',
'video.pareygo.de',
'video.tu-freiberg.de',
'videocampus.sachsen.de',
'videoportal.uni-freiburg.de',
'videoportal.vm.uni-freiburg.de',
'videos.duoc.cl',
'videos.uni-paderborn.de',
'vimp-bemus.udk-berlin.de',
'vimp.aekwl.de',
'vimp.hs-mittweida.de',
'vimp.oth-regensburg.de',
'vimp.ph-heidelberg.de',
'vimp.sma-events.com',
'vimp.weka-fachmedien.de',
'webtv.univ-montp3.fr',
'www.b-tu.de/media',
'www.bigcitytv.de',
'www.cad-videos.de',
'www.fh-bielefeld.de/medienportal',
'www.orvovideo.com',
'www.rwe.tv',
'www.wenglor-media.com',
'www2.univ-sba.dz',
)
_VALID_URL = r'''(?x)https?://(?P<host>%s)/(?:
m/(?P<tmp_id>[0-9a-f]+)| m/(?P<tmp_id>[0-9a-f]+)|
(?:category/)?video/(?P<display_id>[\w-]+)/(?P<id>[0-9a-f]{32}) (?:category/)?video/(?P<display_id>[\w-]+)/(?P<id>[0-9a-f]{32})|
)''' media/embed.*(?:\?|&)key=(?P<embed_id>[0-9a-f]{32}&?)
)''' % ('|'.join(map(re.escape, _INSTANCES)))
_TESTS = [ _TESTS = [
{ {
@ -13,6 +72,7 @@ class VideocampusSachsenIE(InfoExtractor):
'info_dict': { 'info_dict': {
'id': 'e6b9349905c1628631f175712250f2a1', 'id': 'e6b9349905c1628631f175712250f2a1',
'title': 'Konstruktiver Entwicklungsprozess Vorlesung 7', 'title': 'Konstruktiver Entwicklungsprozess Vorlesung 7',
'description': 'Konstruktiver Entwicklungsprozess Vorlesung 7',
'ext': 'mp4', 'ext': 'mp4',
}, },
}, },
@ -21,6 +81,7 @@ class VideocampusSachsenIE(InfoExtractor):
'info_dict': { 'info_dict': {
'id': 'fc99c527e4205b121cb7c74433469262', 'id': 'fc99c527e4205b121cb7c74433469262',
'title': 'Was ist selbstgesteuertes Lernen?', 'title': 'Was ist selbstgesteuertes Lernen?',
'description': 'md5:196aa3b0509a526db62f84679522a2f5',
'display_id': 'Was-ist-selbstgesteuertes-Lernen', 'display_id': 'Was-ist-selbstgesteuertes-Lernen',
'ext': 'mp4', 'ext': 'mp4',
}, },
@ -30,43 +91,32 @@ class VideocampusSachsenIE(InfoExtractor):
'info_dict': { 'info_dict': {
'id': '09d4ed029002eb1bdda610f1103dd54c', 'id': '09d4ed029002eb1bdda610f1103dd54c',
'title': 'Tutorial zur Nutzung von Adobe Connect aus Veranstalter-Sicht', 'title': 'Tutorial zur Nutzung von Adobe Connect aus Veranstalter-Sicht',
'description': 'md5:3d379ca3cc17b9da6784d7f58cca4d58',
'display_id': 'Tutorial-zur-Nutzung-von-Adobe-Connect-aus-Veranstalter-Sicht', 'display_id': 'Tutorial-zur-Nutzung-von-Adobe-Connect-aus-Veranstalter-Sicht',
'ext': 'mp4', 'ext': 'mp4',
}, },
}, },
] {
'url': 'https://www2.univ-sba.dz/video/Presentation-de-la-Faculte-de-droit-et-des-sciences-politiques-Journee-portes-ouvertes-202122/0183356e41af7bfb83d7667b20d9b6a3',
def _real_extract(self, url): 'info_dict': {
video_id, tmp_id, display_id = self._match_valid_url(url).group('id', 'tmp_id', 'display_id') 'url': 'https://www2.univ-sba.dz/getMedium/0183356e41af7bfb83d7667b20d9b6a3.mp4',
webpage = self._download_webpage(url, video_id or tmp_id, fatal=False) or '' 'id': '0183356e41af7bfb83d7667b20d9b6a3',
'title': 'Présentation de la Faculté de droit et des sciences politiques - Journée portes ouvertes 2021/22',
if not tmp_id: 'description': 'md5:508958bd93e0ca002ac731d94182a54f',
video_id = self._html_search_regex( 'display_id': 'Presentation-de-la-Faculte-de-droit-et-des-sciences-politiques-Journee-portes-ouvertes-202122',
r'src="https?://videocampus\.sachsen\.de/media/embed\?key=([0-9a-f]+)&', 'ext': 'mp4',
webpage, 'video_id')
title = self._html_search_regex(
(r'<h1>(?P<content>[^<]+)</h1>', *self._meta_regex('title')),
webpage, 'title', group='content', fatal=False)
formats, subtitles = self._extract_m3u8_formats_and_subtitles(
f'https://videocampus.sachsen.de/media/hlsMedium/key/{video_id}/format/auto/ext/mp4/learning/0/path/m3u8',
video_id, 'mp4', 'm3u8_native', m3u8_id='hls')
self._sort_formats(formats)
return {
'id': video_id,
'title': title,
'display_id': display_id,
'formats': formats,
'subtitles': subtitles
} }
},
{
class VideocampusSachsenEmbedIE(InfoExtractor): 'url': 'https://vimp.weka-fachmedien.de/video/Preisverleihung-Produkte-des-Jahres-2022/c8816f1cc942c12b6cce57c835cffd7c',
_VALID_URL = r'https?://videocampus.sachsen.de/media/embed\?key=(?P<id>[0-9a-f]+)' 'info_dict': {
'id': 'c8816f1cc942c12b6cce57c835cffd7c',
_TESTS = [ 'title': 'Preisverleihung »Produkte des Jahres 2022«',
'description': 'md5:60c347568ca89aa25b772c4ea564ebd3',
'display_id': 'Preisverleihung-Produkte-des-Jahres-2022',
'ext': 'mp4',
},
},
{ {
'url': 'https://videocampus.sachsen.de/media/embed?key=fc99c527e4205b121cb7c74433469262', 'url': 'https://videocampus.sachsen.de/media/embed?key=fc99c527e4205b121cb7c74433469262',
'info_dict': { 'info_dict': {
@ -78,18 +128,41 @@ class VideocampusSachsenEmbedIE(InfoExtractor):
] ]
def _real_extract(self, url): def _real_extract(self, url):
video_id = self._match_id(url) host, video_id, tmp_id, display_id, embed_id = self._match_valid_url(url).group(
'host', 'id', 'tmp_id', 'display_id', 'embed_id')
webpage = self._download_webpage(url, video_id or tmp_id, fatal=False) or ''
webpage = self._download_webpage(url, video_id) if not video_id:
video_id = embed_id or self._html_search_regex(
rf'src="https?://{host}/media/embed.*(?:\?|&)key=([0-9a-f]+)&?',
webpage, 'video_id')
if not (display_id or tmp_id):
# Title, description from embedded page's meta wouldn't be correct
title = self._html_search_regex(r'<img[^>]* title="([^"<]+)"', webpage, 'title', fatal=False) title = self._html_search_regex(r'<img[^>]* title="([^"<]+)"', webpage, 'title', fatal=False)
description = None
else:
title = self._html_search_meta(('og:title', 'twitter:title', 'title'), webpage, fatal=False)
description = self._html_search_meta(
('og:description', 'twitter:description', 'description'), webpage, default=None)
formats, subtitles = [], {}
try:
formats, subtitles = self._extract_m3u8_formats_and_subtitles( formats, subtitles = self._extract_m3u8_formats_and_subtitles(
f'https://videocampus.sachsen.de/media/hlsMedium/key/{video_id}/format/auto/ext/mp4/learning/0/path/m3u8', f'https://{host}/media/hlsMedium/key/{video_id}/format/auto/ext/mp4/learning/0/path/m3u8',
video_id, 'mp4', 'm3u8_native', m3u8_id='hls') video_id, 'mp4', m3u8_id='hls', fatal=True)
except ExtractorError as e:
if not isinstance(e.cause, compat_HTTPError) or e.cause.code not in (404, 500):
raise
formats.append({'url': f'https://{host}/getMedium/{video_id}.mp4'})
self._sort_formats(formats) self._sort_formats(formats)
return { return {
'id': video_id, 'id': video_id,
'title': title, 'title': title,
'description': description,
'display_id': display_id,
'formats': formats, 'formats': formats,
'subtitles': subtitles, 'subtitles': subtitles
} }