[extractor/common] Strip src attribute for HTML5 entries code (closes #18485, closes #21169)

This commit is contained in:
Sergey M․ 2019-05-23 23:52:11 +07:00
parent 8af49fc276
commit f856816b94
No known key found for this signature in database
GPG Key ID: 2C393E0F18A9236D

View File

@ -67,6 +67,7 @@ from ..utils import (
sanitized_Request, sanitized_Request,
sanitize_filename, sanitize_filename,
str_or_none, str_or_none,
strip_or_none,
unescapeHTML, unescapeHTML,
unified_strdate, unified_strdate,
unified_timestamp, unified_timestamp,
@ -2480,7 +2481,7 @@ class InfoExtractor(object):
'subtitles': {}, 'subtitles': {},
} }
media_attributes = extract_attributes(media_tag) media_attributes = extract_attributes(media_tag)
src = media_attributes.get('src') src = strip_or_none(media_attributes.get('src'))
if src: if src:
_, formats = _media_formats(src, media_type) _, formats = _media_formats(src, media_type)
media_info['formats'].extend(formats) media_info['formats'].extend(formats)
@ -2490,7 +2491,7 @@ class InfoExtractor(object):
s_attr = extract_attributes(source_tag) s_attr = extract_attributes(source_tag)
# data-video-src and data-src are non standard but seen # data-video-src and data-src are non standard but seen
# several times in the wild # several times in the wild
src = dict_get(s_attr, ('src', 'data-video-src', 'data-src')) src = strip_or_none(dict_get(s_attr, ('src', 'data-video-src', 'data-src')))
if not src: if not src:
continue continue
f = parse_content_type(s_attr.get('type')) f = parse_content_type(s_attr.get('type'))
@ -2533,7 +2534,7 @@ class InfoExtractor(object):
track_attributes = extract_attributes(track_tag) track_attributes = extract_attributes(track_tag)
kind = track_attributes.get('kind') kind = track_attributes.get('kind')
if not kind or kind in ('subtitles', 'captions'): if not kind or kind in ('subtitles', 'captions'):
src = track_attributes.get('src') src = strip_or_none(track_attributes.get('src'))
if not src: if not src:
continue continue
lang = track_attributes.get('srclang') or track_attributes.get('lang') or track_attributes.get('label') lang = track_attributes.get('srclang') or track_attributes.get('lang') or track_attributes.get('label')