From fe515e5c75bf8be6024893a78ce6c24364bbc1dd Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Sergey=20M=E2=80=A4?= Date: Fri, 29 May 2020 02:01:51 +0700 Subject: [PATCH] [ard:beta] Extend _VALID_URL (closes #25405) --- youtube_dl/extractor/ard.py | 19 ++++++++++++++++--- 1 file changed, 16 insertions(+), 3 deletions(-) diff --git a/youtube_dl/extractor/ard.py b/youtube_dl/extractor/ard.py index e23b71466..5b7b2dd6d 100644 --- a/youtube_dl/extractor/ard.py +++ b/youtube_dl/extractor/ard.py @@ -313,9 +313,9 @@ class ARDIE(InfoExtractor): class ARDBetaMediathekIE(ARDMediathekBaseIE): - _VALID_URL = r'https://(?:beta|www)\.ardmediathek\.de/(?P[^/]+)/(?:player|live)/(?P[a-zA-Z0-9]+)(?:/(?P[^/?#]+))?' + _VALID_URL = r'https://(?:(?:beta|www)\.)?ardmediathek\.de/(?P[^/]+)/(?:player|live|video)/(?P(?:[^/]+/)*)(?P[a-zA-Z0-9]+)' _TESTS = [{ - 'url': 'https://beta.ardmediathek.de/ard/player/Y3JpZDovL2Rhc2Vyc3RlLmRlL3RhdG9ydC9mYmM4NGM1NC0xNzU4LTRmZGYtYWFhZS0wYzcyZTIxNGEyMDE/die-robuste-roswita', + 'url': 'https://ardmediathek.de/ard/video/die-robuste-roswita/Y3JpZDovL2Rhc2Vyc3RlLmRlL3RhdG9ydC9mYmM4NGM1NC0xNzU4LTRmZGYtYWFhZS0wYzcyZTIxNGEyMDE', 'md5': 'dfdc87d2e7e09d073d5a80770a9ce88f', 'info_dict': { 'display_id': 'die-robuste-roswita', @@ -328,6 +328,15 @@ class ARDBetaMediathekIE(ARDMediathekBaseIE): 'upload_date': '20191222', 'ext': 'mp4', }, + }, { + 'url': 'https://beta.ardmediathek.de/ard/video/Y3JpZDovL2Rhc2Vyc3RlLmRlL3RhdG9ydC9mYmM4NGM1NC0xNzU4LTRmZGYtYWFhZS0wYzcyZTIxNGEyMDE', + 'only_matching': True, + }, { + 'url': 'https://ardmediathek.de/ard/video/saartalk/saartalk-gesellschaftsgift-haltung-gegen-hass/sr-fernsehen/Y3JpZDovL3NyLW9ubGluZS5kZS9TVF84MTY4MA/', + 'only_matching': True, + }, { + 'url': 'https://www.ardmediathek.de/ard/video/trailer/private-eyes-s01-e01/one/Y3JpZDovL3dkci5kZS9CZWl0cmFnLTE1MTgwYzczLWNiMTEtNGNkMS1iMjUyLTg5MGYzOWQxZmQ1YQ/', + 'only_matching': True, }, { 'url': 'https://www.ardmediathek.de/ard/player/Y3JpZDovL3N3ci5kZS9hZXgvbzEwNzE5MTU/', 'only_matching': True, @@ -339,7 +348,11 @@ class ARDBetaMediathekIE(ARDMediathekBaseIE): def _real_extract(self, url): mobj = re.match(self._VALID_URL, url) video_id = mobj.group('video_id') - display_id = mobj.group('display_id') or video_id + display_id = mobj.group('display_id') + if display_id: + display_id = display_id.rstrip('/') + if not display_id: + display_id = video_id player_page = self._download_json( 'https://api.ardmediathek.de/public-gateway',