Fix not adding two of the same manifest values

(Sometimes I hate it when technology is implemented like this in the first place: Link to a "alternative", but it's just the original)
This commit is contained in:
spookyahell 2023-03-16 00:59:46 +01:00 committed by GitHub
parent 071670cbea
commit 1a6fdcef9c
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23

View file

@ -60,6 +60,9 @@ def _real_extract(self, url):
formats = []
subtitles = {}
# list to track the urls and ensure that not a second manifest url with the same value is added
avoid_duplicate_manifest_urls = []
# check if the metadata contains a direct URL to a file
for kind, media in media_resource.items():
if kind == 'captionsHash':
@ -79,6 +82,11 @@ def _real_extract(self, url):
if tag_name not in ('videoURL', 'audioURL'):
continue
if medium_url not in avoid_duplicate_manifest_urls:
avoid_duplicate_manifest_urls.append(medium_url)
else:
continue
ext = determine_ext(medium_url)
if ext == 'm3u8':
formats.extend(self._extract_m3u8_formats(
@ -164,7 +172,7 @@ class WDRPageIE(WDRIE): # XXX: Do not subclass from concrete IE
'ext': 'mp3',
'display_id': 'wdr3-gespraech-am-samstag/audio-schriftstellerin-juli-zeh-100',
'title': 'Schriftstellerin Juli Zeh',
'alt_title': 'WDR 3 Gespräch am Samstag',
'alt_title': 'WDR 3 Gespräch am Samstag',
'upload_date': '20160312',
'description': 'md5:e127d320bc2b1f149be697ce044a3dd7',
'is_live': False,
@ -232,7 +240,7 @@ class WDRPageIE(WDRIE): # XXX: Do not subclass from concrete IE
'info_dict': {
'id': 'mdb-1556012',
'ext': 'mp4',
'title': 'DHB-Vizepräsident Bob Hanning - "Die Weltspitze ist extrem breit"',
'title': 'DHB-Vizepräsident Bob Hanning - "Die Weltspitze ist extrem breit"',
'upload_date': '20180111',
},
'params': {