yt_dlp/extractor/rtve.py

@@ -9,7 +9,6 @@ from ..utils import (
     clean_html,
     determine_ext,
     float_or_none,
-    make_archive_id,
     parse_iso8601,
     qualities,
     url_or_none,
@@ -122,7 +121,7 @@ class RTVEALaCartaIE(RTVEBaseIE):
     IE_NAME = 'rtve.es:alacarta'
     IE_DESC = 'RTVE a la carta and Play'
     _VALID_URL = [
-        r'https?://(?:www\.)?rtve\.es/(?:m/)?(?:(?:alacarta|play)/videos|filmoteca)/(?!directo)(?:[^/?#]+/){2}(?P<id>\d+)',
+        r'https?://(?:www\.)?rtve\.es/(m/)?(?:(?:alacarta|play)/videos|filmoteca)/[^/?#]+/[^/?#]+/(?P<id>\d+)',
         r'https?://(?:www\.)?rtve\.es/infantil/serie/[^/?#]+/video/[^/?#]+/(?P<id>\d+)',
     ]
 
@@ -219,14 +218,14 @@ class RTVEALaCartaIE(RTVEBaseIE):
             'formats': formats,
             'subtitles': subtitles,
             **self._parse_metadata(metadata),
-            '_old_archive_ids': [make_archive_id('rtveinfantil', video_id)] if is_infantil else None,
+            '_old_archive_ids': [f'rtveinfantil {video_id}'] if is_infantil else None,
         }
 
 
 class RTVEAudioIE(RTVEBaseIE):
     IE_NAME = 'rtve.es:audio'
     IE_DESC = 'RTVE audio'
-    _VALID_URL = r'https?://(?:www\.)?rtve\.es/(alacarta|play)/audios/(?:[^/?#]+/){2}(?P<id>\d+)'
+    _VALID_URL = r'https?://(?:www\.)?rtve\.es/(alacarta|play)/audios/[^/?#]+/[^/?#]+/(?P<id>\d+)'
 
     _TESTS = [{
         'url': 'https://www.rtve.es/alacarta/audios/a-hombros-de-gigantes/palabra-ingeniero-codigos-informaticos-27-04-21/5889192/',
@@ -287,40 +286,19 @@ class RTVEAudioIE(RTVEBaseIE):
 class RTVELiveIE(RTVEBaseIE):
     IE_NAME = 'rtve.es:live'
     IE_DESC = 'RTVE.es live streams'
-    _VALID_URL = [
+    _VALID_URL = r'https?://(?:www\.)?rtve\.es/directo/(?P<id>[a-zA-Z0-9-]+)'
-        r'https?://(?:www\.)?rtve\.es/directo/(?P<id>[a-zA-Z0-9-]+)',
-        r'https?://(?:www\.)?rtve\.es/play/videos/directo/[^/?#]+/(?P<id>[a-zA-Z0-9-]+)',
-    ]
 
     _TESTS = [{
         'url': 'http://www.rtve.es/directo/la-1/',
         'info_dict': {
             'id': 'la-1',
             'ext': 'mp4',
-            'live_status': 'is_live',
             'title': str,
-            'description': str,
-            'thumbnail': r're:https://img\d\.rtve\.es/resources/thumbslive/\d+\.jpg',
-            'timestamp': int,
-            'upload_date': str,
-        },
-        'params': {'skip_download': 'live stream'},
-    }, {
-        'url': 'https://www.rtve.es/play/videos/directo/deportes/tdp/',
-        'info_dict': {
-            'id': 'tdp',
-            'ext': 'mp4',
             'live_status': 'is_live',
-            'title': str,
-            'description': str,
-            'thumbnail': r're:https://img2\d\.rtve\.es/resources/thumbslive/\d+\.jpg',
-            'timestamp': int,
-            'upload_date': str,
         },
-        'params': {'skip_download': 'live stream'},
+        'params': {
-    }, {
+            'skip_download': 'live stream',
-        'url': 'http://www.rtve.es/play/videos/directo/canales-lineales/la-1/',
+        },
-        'only_matching': True,
     }]
 
     def _real_extract(self, url):
@@ -335,7 +313,6 @@ class RTVELiveIE(RTVEBaseIE):
 
         return {
             'id': video_id,
-            **self._search_json_ld(webpage, video_id, fatal=False),
             'title': self._html_extract_title(webpage),
             'formats': formats,
             'subtitles': subtitles,