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

Closes #5144, #5143
Authored by: Grub4K
This commit is contained in:
Simon Sawicki 2022-10-09 05:17:58 +02:00 committed by GitHub
parent f324fe8c59
commit 1d55ebabc9
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
2 changed files with 20 additions and 7 deletions

View File

@ -1467,10 +1467,6 @@ class InfoExtractor:
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 @@ class InfoExtractor:
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 @@ class InfoExtractor:
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):

View File

@ -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):