mirror of
https://github.com/internetarchive/brozzler.git
synced 2025-02-24 08:39:59 -05:00
skip more exts, plus chrome-error
This commit is contained in:
parent
737770e3ba
commit
f2c89d1c18
@ -34,7 +34,10 @@ thread_local = threading.local()
|
|||||||
|
|
||||||
|
|
||||||
def is_html_maybe(url):
|
def is_html_maybe(url):
|
||||||
skip_url_exts = ["pdf", "jpg", "jpeg", "png", "gif", "mp4", "mpeg"]
|
if "chrome-error:" in url:
|
||||||
|
return False
|
||||||
|
|
||||||
|
skip_url_exts = ["pdf", "jpg", "jpeg", "png", "gif", "mp3", "mp4", "mpeg", "css", "js"]
|
||||||
|
|
||||||
parsed_url = urlparse(url)
|
parsed_url = urlparse(url)
|
||||||
base_url, ext = os.path.splitext(parsed_url.path)
|
base_url, ext = os.path.splitext(parsed_url.path)
|
||||||
|
Loading…
x
Reference in New Issue
Block a user