[extractor/common] Fix `json_ld` type checks (#5145)

Closes #5144, #5143
Authored by: Grub4K

diff --git a/yt_dlp/extractor/common.py b/yt_dlp/extractor/common.py
index 31a45b37a22ffd3992e7408c3e29185157f261f2..18a52a855922f47a70a24546961c4e65c2c9f55d 100644 (file)
--- a/yt_dlp/extractor/common.py
+++ b/yt_dlp/extractor/common.py
@@ -1467,10 +1467,6 @@ def _json_ld(self, json_ld, video_id, fatal=True, expected_type=None):
         if not json_ld:
             return {}
         info = {}
-        if not isinstance(json_ld, (list, tuple, dict)):
-            return info
-        if isinstance(json_ld, dict):
-            json_ld = [json_ld]
 
         INTERACTION_TYPE_MAP = {
             'CommentAction': 'comment',
@@ -1570,11 +1566,13 @@ def extract_video_object(e):
             extract_chapter_information(e)
 
         def traverse_json_ld(json_ld, at_top_level=True):
-            for e in json_ld:
+            for e in variadic(json_ld):
+                if not isinstance(e, dict):
+                    continue
                 if at_top_level and '@context' not in e:
                     continue
                 if at_top_level and set(e.keys()) == {'@context', '@graph'}:
-                    traverse_json_ld(variadic(e['@graph'], allowed_types=(dict,)), at_top_level=False)
+                    traverse_json_ld(e['@graph'], at_top_level=False)
                     break
                 if expected_type is not None and not is_type(e, expected_type):
                     continue
@@ -1629,8 +1627,8 @@ def traverse_json_ld(json_ld, at_top_level=True):
                     continue
                 else:
                     break
-        traverse_json_ld(json_ld)
 
+        traverse_json_ld(json_ld)
         return filter_dict(info)
 
     def _search_nextjs_data(self, webpage, video_id, *, transform_source=None, fatal=True, **kw):

diff --git a/yt_dlp/extractor/generic.py b/yt_dlp/extractor/generic.py
index 73422f937c35f639244c440d0d63b90101961fde..92390a387508c4bc4ea946076360818615daf70f 100644 (file)
--- a/yt_dlp/extractor/generic.py
+++ b/yt_dlp/extractor/generic.py
@@ -2463,6 +2463,21 @@ class GenericIE(InfoExtractor):
                 'duration': 111.0,
             }
         },
+        {
+            'note': 'JSON LD with unexpected data type',
+            'url': 'https://www.autoweek.nl/autotests/artikel/porsche-911-gt3-rs-rij-impressie-2/',
+            'info_dict': {
+                'id': 'porsche-911-gt3-rs-rij-impressie-2',
+                'ext': 'mp4',
+                'title': 'Test: Porsche 911 GT3 RS',
+                'description': 'Je ziet het niet, maar het is er wel. Downforce, hebben we het dan over. En in de nieuwe Porsche 911 GT3 RS is er zelfs heel veel downforce.',
+                'timestamp': 1664920902,
+                'upload_date': '20221004',
+                'thumbnail': r're:^https://media.autoweek.nl/m/.+\.jpg$',
+                'age_limit': 0,
+                'direct': True,
+            }
+        }
     ]
 
     def report_following_redirect(self, new_url):