Seems to be a bug when paring multiple '@'s ?

[YuanchengJiang]

Hi,

I notice one difference by applying differential testing with urllib.parse.

Case 1:

URL = "https://@david:passwd@testing.xyz"
f = furl(URL)

Error: ValueError: Invalid port 'passwd@testing.xyz'. However, it should access the host 'testing.xyz' with username '@david' and password 'passwd'.

Case 2:

URL = "https://aa@bb@testing.xyz"
f = furl(URL)

Error: ValueError: Invalid host 'bb@testing.xyz'. However, it should access the host 'testing.xyz' with username 'aa@bb'.

When parsing multiple '@'s, I think the domain should be sliced by the last '@' rather than the first '@'.

Thanks,
Yuancheng