[ARDBetaMediathek] Handle new URLs

Adapted from https://github.com/ytdl-org/youtube-dl/commit/8562218350a79d4709da8593bb0c538aa0824acf
Closes #1601

diff --git a/yt_dlp/extractor/ard.py b/yt_dlp/extractor/ard.py
index 048d30f27dda1d2343621d0bce200dbed1f5390d..f8d57109e1c6ee810acb19c0b2154b591b44411c 100644 (file)
--- a/yt_dlp/extractor/ard.py
+++ b/yt_dlp/extractor/ard.py
@@ -388,7 +388,13 @@ def _real_extract(self, url):
 
 
 class ARDBetaMediathekIE(ARDMediathekBaseIE):
-    _VALID_URL = r'https://(?:(?:beta|www)\.)?ardmediathek\.de/(?P<client>[^/]+)/(?P<mode>player|live|video|sendung|sammlung)/(?P<display_id>(?:[^/]+/)*)(?P<video_id>[a-zA-Z0-9]+)'
+    _VALID_URL = r'''(?x)https://
+        (?:(?:beta|www)\.)?ardmediathek\.de/
+        (?:(?P<client>[^/]+)/)?
+        (?:player|live|video|(?P<playlist>sendung|sammlung))/
+        (?:(?P<display_id>[^?#]+)/)?
+        (?P<id>(?(playlist)|Y3JpZDovL)[a-zA-Z0-9]+)'''
+
     _TESTS = [{
         'url': 'https://www.ardmediathek.de/mdr/video/die-robuste-roswita/Y3JpZDovL21kci5kZS9iZWl0cmFnL2Ntcy84MWMxN2MzZC0wMjkxLTRmMzUtODk4ZS0wYzhlOWQxODE2NGI/',
         'md5': 'a1dc75a39c61601b980648f7c9f9f71d',
@@ -403,6 +409,18 @@ class ARDBetaMediathekIE(ARDMediathekBaseIE):
             'upload_date': '20200805',
             'ext': 'mp4',
         },
+        'skip': 'Error',
+    }, {
+        'url': 'https://www.ardmediathek.de/video/tagesschau-oder-tagesschau-20-00-uhr/das-erste/Y3JpZDovL2Rhc2Vyc3RlLmRlL3RhZ2Vzc2NoYXUvZmM4ZDUxMjgtOTE0ZC00Y2MzLTgzNzAtNDZkNGNiZWJkOTll',
+        'md5': 'f1837e563323b8a642a8ddeff0131f51',
+        'info_dict': {
+            'id': '10049223',
+            'ext': 'mp4',
+            'title': 'tagesschau, 20:00 Uhr',
+            'timestamp': 1636398000,
+            'description': 'md5:39578c7b96c9fe50afdf5674ad985e6b',
+            'upload_date': '20211108',
+        },
     }, {
         'url': 'https://beta.ardmediathek.de/ard/video/Y3JpZDovL2Rhc2Vyc3RlLmRlL3RhdG9ydC9mYmM4NGM1NC0xNzU4LTRmZGYtYWFhZS0wYzcyZTIxNGEyMDE',
         'only_matching': True,
@@ -426,6 +444,12 @@ class ARDBetaMediathekIE(ARDMediathekBaseIE):
         # playlist of type 'sammlung'
         'url': 'https://www.ardmediathek.de/ard/sammlung/team-muenster/5JpTzLSbWUAK8184IOvEir/',
         'only_matching': True,
+    }, {
+        'url': 'https://www.ardmediathek.de/video/coronavirus-update-ndr-info/astrazeneca-kurz-lockdown-und-pims-syndrom-81/ndr/Y3JpZDovL25kci5kZS84NzE0M2FjNi0wMWEwLTQ5ODEtOTE5NS1mOGZhNzdhOTFmOTI/',
+        'only_matching': True,
+    }, {
+        'url': 'https://www.ardmediathek.de/ard/player/Y3JpZDovL3dkci5kZS9CZWl0cmFnLWQ2NDJjYWEzLTMwZWYtNGI4NS1iMTI2LTU1N2UxYTcxOGIzOQ/tatort-duo-koeln-leipzig-ihr-kinderlein-kommet',
+        'only_matching': True,
     }]
 
     def _ARD_load_playlist_snipped(self, playlist_id, display_id, client, mode, pageNumber):
@@ -525,20 +549,12 @@ def _ARD_extract_playlist(self, url, playlist_id, display_id, client, mode):
         return self.playlist_result(entries, playlist_title=display_id)
 
     def _real_extract(self, url):
-        mobj = self._match_valid_url(url)
-        video_id = mobj.group('video_id')
-        display_id = mobj.group('display_id')
-        if display_id:
-            display_id = display_id.rstrip('/')
-        if not display_id:
-            display_id = video_id
-
-        if mobj.group('mode') in ('sendung', 'sammlung'):
-            # this is a playlist-URL
-            return self._ARD_extract_playlist(
-                url, video_id, display_id,
-                mobj.group('client'),
-                mobj.group('mode'))
+        video_id, display_id, playlist_type, client = self._match_valid_url(url).group(
+            'id', 'display_id', 'playlist', 'client')
+        display_id, client = display_id or video_id, client or 'ard'
+
+        if playlist_type:
+            return self._ARD_extract_playlist(url, video_id, display_id, client, playlist_type)
 
         player_page = self._download_json(
             'https://api.ardmediathek.de/public-gateway',
@@ -574,7 +590,7 @@ def _real_extract(self, url):
       }
     }
   }
-}''' % (mobj.group('client'), video_id),
+}''' % (client, video_id),
             }).encode(), headers={
                 'Content-Type': 'application/json'
             })['data']['playerPage']