Thanks for post. But i have a question - how do you test these methods?

simple example:

def filter_by_published(self):
     return self.filter(is_published=True)


Thats ok, i just create instance with is_published==True,  is_pubslished==False and then check, which of instance in these queryset method response.

more complecated

def filter_by_ready_for_sending(self):
    return self.filter_by_published().filter(date_send=now())


How i do it now. I just mock filter_by_published with QuerySet.all method and testing second filter.

class TestQuerySetMethods(TestCase):
    def test_filter_by_ready_for_sending(self):
        with patch('filter_by_published', QuerySet.all):
            # test that filtering by date works right
            # i dont care, how filter_by_published works,
            # because i've already tested it before


even more complecated

def filter_by_ready_for_buying(self):
    return (self.filter_by_published()
                .filter(price__isnull=False)
                .filter_by_delivered())


And now i have to mock default .filter, and custom filter_by_published, filter_by_delivered:

with patch('filter_by_published', QuerySet.all):
    with patch('filter', QuerySet.all):
        with patch('filter_by_delivered', Mock(return_value=SomeExpectedSet)):
            response = Books.objects.filter_by_ready_for_buying()
            self.assertEqual(set(response), set(SomeExpectedSet))

I create 3 test methods to test every queryset method, which appeares in filter_by_ready_for_buying.
But what if i'll add new filtering to this method? I'll have to add new mock for every test method.

What if it will be more than 1 default filter method? My mock will mock them all and SomeExpectedSet will not
be representative.