Starting using Hypothesis with Django, having a few problems...

[Etienne Desautels]

Hi,

I'm starting using Hypothesis with Django and my first try is creating some tests for this Django model:

class Period(models.Model):
    name = models.CharField(max_length=10, unique=True)
    start_date = models.DateField(
        validators=[
            MinValueValidator(datetime.date(2025, 4, 1)),
            MaxValueValidator(datetime.date(2055, 1, 1)),
        ]
    )
    end_date = models.DateField(
        validators=[
            MinValueValidator(datetime.date(2025, 12, 31)),
            MaxValueValidator(datetime.date(2055, 12, 31)),
        ]
    )

    def __str__(self) -> str:
        return self.name

    def save(self, *args, **kwargs):
        # Forces validation during saving. Note that validation
        # will still not be called for 'update' and 'bulk_create'.
        self.full_clean()
        return super().save(*args, **kwargs)

    #def other_methods(self):
    #   pass


    class Meta:
        constraints = [
            models.UniqueConstraint(
                fields=['start_date', 'end_date'], name='unique_start_end'
            ),
            models.CheckConstraint(
                condition=Q(date_debut__lt=F('end_date')),
                name='start_lt_end',
            ),
        ]


First try:

from hypothesis import given
from hypothesis.extra.django import TestCase, from_model

from .models import Period


class PeriodModelTests(TestCase):
    period_strategy = from_model(Period)

    @given(period=period_strategy)
    def test_str_method(self, period):
        self.assertEqual(str(period), period.name)

This gives me the filter_too_much error. Is this normal?

I was thinking that the too much filtering was probably related to the two date fields, so, to circumvent the problem, I tried specifying the date fields like this:

import datetime

from hypothesis import strategies as st

class PeriodModelTests(TestCase):
    period_strategy = from_model(
        Period,
        start_date=st.dates(
            min_value=datetime.date(2025, 4, 1),
            max_value=datetime.date(2055, 1, 1),
        ),
        end_date=st.dates(
            min_value=datetime.date(2025, 12, 31),
            max_value=datetime.date(2055, 12, 31),
        ),
    )

Now the test failed because when trying to save period instances to the DB the data did not validate (validation is called by full_clean() in the save() method) because, in the strategy, nothing enforce that start_date should be smaller then end_date.

I then tried enforcing this with @combine like this:

class PeriodeModelTests(TestCase):
    @st.composite
    def period_strategy(draw):
        start_date = draw(
            st.dates(
                min_value=datetime.date(2025, 4, 1),
                max_value=datetime.date(2055, 1, 1),
            )
        )
        end_date = draw(
            st.dates(
                min_value=start_date + datetime.timedelta(days=1),
                max_value=datetime.date(2055, 12, 31),
            )
        )

        return from_model(Period, start_date=start_date, end_date=end_date)

    @given(period=period_strategy())
    def test_str_method(self, period):
        self.assertEqual(str(period), period.name)

This gives me the error:
AttributeError: 'LazyStrategy' object has no attribute 'name'

From here I out of ideas. If someone can help me with this, it will be appreciated.

Also, during my different tests, when the system is creating entries in the DB (postgres) I'm receiving the too_slow error. I added this setting to shutdown this healt check:
@settings(suppress_health_check=[HealthCheck.too_slow])

Is this normal when creating entries in the DB?

Thanks!

Etienne

[Etienne Desautels]

OK, I found a solution. The problem was that I needed to compare the two dates before the creation of the Period object (as it call save() and then full_clean() that raised a validation error). My solution, that it's probably not ideal, is this:

@st.composite
def period_strategy(draw):
    start_date = draw(
        st.dates(
            min_value=datetime.date(2025, 4, 1),
            max_value=datetime.date(2055, 1, 1),
        )
    )
    end_date = draw(
        st.dates(

            min_value=datetime.date(2025, 12, 31),
            max_value=datetime.date(2055, 12, 31),
        )

    )
    assume(start_date < end_date)
    period = draw(
        from_model(
            Period,
            start_date=st.dates(min_value=start_date, max_value=start_date),
            end_date=st.dates(min_value=end_date, max_value=end_date),
        )
    )
    return period

If you have any commentaries or suggestions go ahead!

Thanks

Etienne