Question

I have some models that are laid out like so (with a lot more fields):

class A(models.Model):
    name = models.
class B(models.Model):
    foo = models.ForeignKey(A)

and I have a search system that uses Q objects; 0 and 1 are primary keys, obtained by a ModelMultipleChoiceField in a form, and both of their foos point to the same A.

So for or logic, I can use this

>>> A.objects.filter(Q(b__pk=0) | Q(b__pk=1))
[<A: A object>, <A: A object>]

and I can fix the duplication by using

A.objects.filter(Q(b__pk=0) | Q(b__pk=1)).distinct()

However, I also want to do and logic. Ideally it would work by using & instead of |, but that doesn’t work.

>>> A.objects.filter(Q(b__pk=0) & Q(b__pk=1))
[]

So, what is the simplest way to do this?

Answer 1

Guess you want A instances that having at least one B() w/ primary key 1 and one B() w/ primary key 0

Then you need

(A.objects.filter(b__pk=0) & A.objects.filter(b__pk=1)).distinct()

Or the following if join hurts

A.objects.filter(
    models.Q(pk__in=B.objects.filter(pk=0).values('foo_id')) &
    models.Q(pk__in=B.objects.filter(pk=1).values('foo_id')))

If you need A() that having only two relative B() one w/ pk=0 and another pk=1, limitation should be applied to the above queryset by

queryset.annotate(b_count=models.Count('b')).filter(b_count=2)