python-validators/validators

Localhost no longer recognized

23pointsNorth opened this issue · 5 comments

23pointsNorth commented

Setup:
python3 -m pip install --upgrade validators==0.22.0

Code to reproduce:

import validators
validators.url("http://127.0.0.1:80/")
validators.url("http://localhost:80/")

Expected outcome:
Both calls to return true.

Similar to the other issue, this commit c43826c seems to comment the tests for localhost

23pointsNorth commented

Right, thanks for pointing it out. I remember the documentation noted that it was introduced as of 0.11.x as a new feature.

A problem with the suggested solution is that simple_host=True also allows for

>>> validators.url("http://l:8000/", simple_host=True)
True

which is not a valid "default" hostname. So in essence, we still need to regex localhost in the target URL?

yozachar commented

It is true, that l or 13 are not common hostnames, but they are valid though.

In a (W)LAN, I could have my hosts named as a, b, c, ... or even 1, 2, 3, ... etc. It's not a clever naming scheme (and not recommended) but definitely legal. So a GET http://13:8080 from host c to host 13 is quite valid indeed.

Currently URL validation accepts all of these parameters:

def url(
    value: str,
    /,
    *,
    skip_ipv6_addr: bool = False,
    skip_ipv4_addr: bool = False,
    may_have_port: bool = True,
    simple_host: bool = False,
    strict_query: bool = True,
    rfc_1034: bool = False,
    rfc_2782: bool = False,
):

It is possible to add another boolean parameter say only_localhost, but I think it might not be good for the following reasons:

  • simple_host must be True for only_localhost to work. - complexity
  • should only_localhost refer to literal localhost string or all hostnames defined in /etc/hosts? - confusion
  • AFAIK, hostnames such as abc, xyz, etc. are very uncommon. - viability

Instead, if an user want to ensure that an URL is strictly localhost, one could:

from validators import url as is_valid_url
from urllib.parse import urlsplit

if is_valid_url(given_url, simple_host=False) and urlsplit(given_url).netloc == 'localhost' :
    # do something
23pointsNorth commented

Thanks, I understand the perspective, just assumed some stability of the API.

Happy for you to close the issues as resolved.

yozachar commented

Thanks for using validators. Feel free to ping this thread if any ideas pop up.

  • ❤️1