Add support of IDN domains for URLField validator.

[Nikolay]

With Django trunk (revision 12617), I have the following:

>>> from django.forms import URLField
>>> URLField().clean('http://xn--e1afmkfd.xn--80akhbyknj4f/')
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/niksite/webapps/django/datamining/dist/django/forms/fields.py", line 156, in clean
    self.run_validators(value)
  File "/home/niksite/webapps/django/datamining/dist/django/forms/fields.py", line 145, in run_validators
    raise ValidationError(errors)
django.core.exceptions.ValidationError: [u'Enter a valid URL.']
>>> URLField().clean('http://пример.испытание/')
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/niksite/webapps/django/datamining/dist/django/forms/fields.py", line 156, in clean
    self.run_validators(value)
  File "/home/niksite/webapps/django/datamining/dist/django/forms/fields.py", line 145, in run_validators
    raise ValidationError(errors)
django.core.exceptions.ValidationError: [u'Enter a valid URL.']

BTW, these URLs are quit valid, moreover -- there is a wiki page after these URLs.

[Ulrich Petri]

I am the author of the patch on #9764. 

It seems however that the problem you are describing has nothing to do with IDN handling pre-se but rather that the TLD in your example
(which is one of the ​IANA IDN test TLDs (click on IDNs)) has "too many" characters for the URLValidator regex which assumes that TLDs have 6 characters at most.

So the solution could be to either lift the restriction on TLD length or to add special handling for IDN TLDs (which are easily enough to distinguish, as they begin with "xn--" in IDNA encoded form) to the regex.

[James Bennett]

This is a feature request, and as such cannot happen for 1.2, which is feature-frozen.

[fwenzel]

We've been bitten by this in the Firefox Input application as well. If people want to report IDNs like ​http://www.правительство.рф/ == ​http://www.xn--80aealotwbjpid2k.xn--p1ai/ , the URLField will throw a ValidationError. Our downstream bug is: ​https://bugzilla.mozilla.org/show_bug.cgi?id=573105 .

We'd appreciate if this made it into a dot-release of Django 1.2!

[Jannis Leidel]

In [15502]:

Fixed #12988 -- Extended regular expression of the URLValidator to fully support IDN-URLs, especially the long TLDs.

[Jannis Leidel]

In [15509]:

[1.2.X] Fixed #12988 -- Extended regular expression of the URLValidator to fully support IDN-URLs, especially the long TLDs.

Backport from trunk (r15502).

[Jacob]

Milestone 1.3 deleted