[rule34us] make 'file_url' extraction more robust (#6289)
update to b64bede9ae
This commit is contained in:
@@ -36,10 +36,8 @@ class Rule34usExtractor(BooruExtractor):
|
|||||||
"score" : text.extract(extr('Score: ', '> - <'), ">", "<")[0],
|
"score" : text.extract(extr('Score: ', '> - <'), ">", "<")[0],
|
||||||
"width" : extr('Size: ', 'w'),
|
"width" : extr('Size: ', 'w'),
|
||||||
"height" : extr(' x ', 'h'),
|
"height" : extr(' x ', 'h'),
|
||||||
"" : extr('</script>', '</div></center>'),
|
"file_url": extr('<source src="', '"') or extr('<img src="', '"'),
|
||||||
"file_url": extr(' src="', '"'),
|
|
||||||
}
|
}
|
||||||
del post[""]
|
|
||||||
|
|
||||||
url = post["file_url"]
|
url = post["file_url"]
|
||||||
if "//video-cdn1." in url:
|
if "//video-cdn1." in url:
|
||||||
|
|||||||
Reference in New Issue
Block a user