yt_dlp/extractor/rinsefm.py

@@ -1,9 +1,12 @@
+import re
+import json
+from datetime import datetime, timezone
+
 from .common import InfoExtractor
-from ..utils import format_field, parse_iso8601
 
 
 class RinseFMIE(InfoExtractor):
-    _VALID_URL = r'https?://(?:www\.)?rinse\.fm/episodes/(?P<id>[^/?#]+)'
+    _VALID_URL = r'https?://(?:www\.)?rinse\.fm/episodes/(?P<id>.+)'
     _TESTS = [{
         'url': 'https://rinse.fm/episodes/cameo-blush-01-09-2023-2300/',
         'md5': '9284abbd785e6b86e67d1cdca6224feb',
@@ -12,8 +15,6 @@ class RinseFMIE(InfoExtractor):
             'ext': 'mp3',
             'title': 'Cameo Blush - 01/09/2023 - 23:00',
             'thumbnail': r're:^https?://.*\.JPG$',
-            'release_timestamp': 1693522800,
-            'release_date': '20230831'
         }
     }]
 
@@ -21,13 +22,24 @@ class RinseFMIE(InfoExtractor):
         video_id = self._match_id(url)
         webpage = self._download_webpage(url, video_id)
 
-        entry = self._search_nextjs_data(webpage, video_id)['props']['pageProps']['entry']
+        response = re.findall(r'<script[^>]*id\s*=\s*[\'"]__NEXT_DATA__[\'"][^>]*>(.*?)<\/script>', webpage, re.DOTALL)
+        info = json.loads(response[0])
+
+        entry = info['props']['pageProps']['entry']
+
+        track_id = entry['id']
+        title = entry['title']
+
+        date_obj = datetime.fromisoformat(entry['episodeDate'])
+        unix_timestamp = int(date_obj.replace(tzinfo=timezone.utc).timestamp())
+
+        url = entry['fileUrl']
+        thumbnail = "https://rinse.imgix.net/media/" + entry['featuredImage'][0]['filename']
+
         return {
-            'id': entry['id'],
+            'id': track_id,
-            'title': entry.get('title'),
+            'title': title,
-            'url': entry['fileUrl'],
+            'url': url,
-            'vcodec': 'none',
+            'release_timestamp': unix_timestamp,
-            'release_timestamp': parse_iso8601(entry.get('episodeDate')),
+            'thumbnail': thumbnail
-            'thumbnail': format_field(
-                entry, [('featuredImage', 0, 'filename')], 'https://rinse.imgix.net/media/%s', default=None),
         }