[extractor/common] Support HTML media elements without child nodes
parent
55642487f0
commit
cea364f70c
|
@ -1,6 +1,7 @@
|
||||||
version <unreleased>
|
version <unreleased>
|
||||||
|
|
||||||
Core
|
Core
|
||||||
|
+ Support HTML media elements without child nodes
|
||||||
* [Makefile] Support for GNU make < 4 is fixed; BSD make dropped (#9387)
|
* [Makefile] Support for GNU make < 4 is fixed; BSD make dropped (#9387)
|
||||||
|
|
||||||
Extractors
|
Extractors
|
||||||
|
|
|
@ -1802,7 +1802,11 @@ class InfoExtractor(object):
|
||||||
return is_plain_url, formats
|
return is_plain_url, formats
|
||||||
|
|
||||||
entries = []
|
entries = []
|
||||||
for media_tag, media_type, media_content in re.findall(r'(?s)(<(?P<tag>video|audio)[^>]*>)(.*?)</(?P=tag)>', webpage):
|
media_tags = [(media_tag, media_type, '')
|
||||||
|
for media_tag, media_type
|
||||||
|
in re.findall(r'(?s)(<(video|audio)[^>]*/>)', webpage)]
|
||||||
|
media_tags.extend(re.findall(r'(?s)(<(?P<tag>video|audio)[^>]*>)(.*?)</(?P=tag)>', webpage))
|
||||||
|
for media_tag, media_type, media_content in media_tags:
|
||||||
media_info = {
|
media_info = {
|
||||||
'formats': [],
|
'formats': [],
|
||||||
'subtitles': {},
|
'subtitles': {},
|
||||||
|
|
Loading…
Reference in New Issue