[extractor/common] Add validation for JSON-LD URLs
This commit is contained in:
parent
4c237ab787
commit
bebef10909
1 changed files with 3 additions and 2 deletions
|
@ -69,6 +69,7 @@ from ..utils import (
|
||||||
update_url_query,
|
update_url_query,
|
||||||
urljoin,
|
urljoin,
|
||||||
url_basename,
|
url_basename,
|
||||||
|
url_or_none,
|
||||||
xpath_element,
|
xpath_element,
|
||||||
xpath_text,
|
xpath_text,
|
||||||
xpath_with_ns,
|
xpath_with_ns,
|
||||||
|
@ -1213,10 +1214,10 @@ class InfoExtractor(object):
|
||||||
def extract_video_object(e):
|
def extract_video_object(e):
|
||||||
assert e['@type'] == 'VideoObject'
|
assert e['@type'] == 'VideoObject'
|
||||||
info.update({
|
info.update({
|
||||||
'url': e.get('contentUrl'),
|
'url': url_or_none(e.get('contentUrl')),
|
||||||
'title': unescapeHTML(e.get('name')),
|
'title': unescapeHTML(e.get('name')),
|
||||||
'description': unescapeHTML(e.get('description')),
|
'description': unescapeHTML(e.get('description')),
|
||||||
'thumbnail': e.get('thumbnailUrl') or e.get('thumbnailURL'),
|
'thumbnail': url_or_none(e.get('thumbnailUrl') or e.get('thumbnailURL')),
|
||||||
'duration': parse_duration(e.get('duration')),
|
'duration': parse_duration(e.get('duration')),
|
||||||
'timestamp': unified_timestamp(e.get('uploadDate')),
|
'timestamp': unified_timestamp(e.get('uploadDate')),
|
||||||
'filesize': float_or_none(e.get('contentSize')),
|
'filesize': float_or_none(e.get('contentSize')),
|
||||||
|
|
Loading…
Reference in a new issue