This will be my first post here and I would like to double check a potential bug I’ve found in django.contrib.postgres.forms SplitArrayField before actually reporting it.
I created an ArrayField(models.CharField(), size=3)
on one of my models and created a form for it using the SplitArrayField
. My requirements where that the array field was optional to fill in and that you could also just fill in one of the CharFields and leave the others blank. So according to the docs I was to set the CharField on my form to required=True
and remove_trailing_nulls=True
.
SplitArrayField(
forms.CharField(required=True, max_length=148),
size=3,
widget=SplitArrayWidget(forms.TextInput(attrs={'size': '148'}), 3),
remove_trailing_nulls=True,
required=False,
)
To my surprise the form did not validate properly and I was able to submit inputs that exceeded the max_length specified in my form.
Looking at the clean
method of the SplitArrayField
I suspect a problem there.
try:
cleaned_data.append(self.base_field.clean(item))
except ValidationError as error:
errors.append(
prefix_validation_error(
error,
self.error_messages["item_invalid"],
code="item_invalid",
params={"nth": index + 1},
)
)
cleaned_data.append(None)
If there is any ValidationError the error is added to the errors
, but None
is appended to the cleaned_data.
This causes issues with the self._remove_trailing_nulls
later on
cleaned_data, null_index = self._remove_trailing_nulls(cleaned_data)
if null_index is not None:
errors = errors[:null_index]
errors = list(filter(None, errors))
cleaned_data
is passed to self._remove_trailing_nulls
and because remove_trailing_nulls=True
null_index
will be 0
thus errors is reassigned to an empty list.
I wrote a test case showcasing this issue:
from django import forms
from django.contrib.postgres.forms import SplitArrayField
class TestForm1(forms.Form):
field = SplitArrayField(
forms.CharField(max_length=5, required=True),
size=3,
remove_trailing_nulls=True,
required=False,
)
class TestForm2(forms.Form):
field = forms.CharField(max_length=5, required=True)
class SplitArrayFieldTest(TestCase):
def test_validation_error(self):
invalid_value = 'invalid'
form1 = TestForm1(data={'field': [invalid_value, '', '']})
form2 = TestForm2(data={'field': invalid_value})
self.assertFalse(form1.is_valid()) #is_valid() should return False, but returns True
self.assertFalse(form2.is_valid())
Is my suspicion correct? Or am I missing something?