I try to combine AND and OR in a filter using Q objects. It looks like that the | behave like an AND. This is related to the previous annotate which is run in the same query and not as a subquery.
What is the correct way to handle this with Django?
models.py
class Type(models.Model):
name = models.CharField(_('name'), max_length=100)
stock = models.BooleanField(_('in stock'), default=True)
hide = models.BooleanField(_('hide'), default=False)
deleted = models.BooleanField(_('deleted'), default=False)
class Item(models.Model):
barcode = models.CharField(_('barcode'), max_length=100, blank=True)
quantity = models.IntegerField(_('quantity'), default=1)
type = models.ForeignKey('Type', related_name='items', verbose_name=_('type'))
views.py
def hire(request):
categories_list = Category.objects.all().order_by('sorting')
types_list = Type.objects.annotate(quantity=Sum('items__quantity')).filter(
Q(hide=False) & Q(deleted=False),
Q(stock=False) | Q(quantity__gte=1))
return render_to_response('equipment/hire.html', {
'categories_list': categories_list,
'types_list': types_list,
}, context_instance=RequestContext(request))
resulting SQL query
SELECT "equipment_type"."id" [...] FROM "equipment_type" LEFT OUTER JOIN
"equipment_subcategory" ON ("equipment_type"."subcategory_id" =
"equipment_subcategory"."id") LEFT OUTER JOIN "equipment_item" ON
("equipment_type"."id" = "equipment_item"."type_id") WHERE
("equipment_type"."hide" = False AND "equipment_type"."deleted" = False )
AND ("equipment_type"."stock" = False )) GROUP BY "equipment_type"."id"
[...] HAVING SUM("equipment_item"."quantity") >= 1
expected SQL query
SELECT
*
FROM
equipment_type
LEFT JOIN (
SELECT type_id, SUM(quantity) AS qty
FROM equipment_item
GROUP BY type_id
) T1
ON id = T1.type_id
WHERE hide=0 AND deleted=0 AND (T1.qty > 0 OR stock=0)
EDIT: I added the expected SQL query (without the join on equipment_subcategory)
Try adding parentheses to explicitly specify your grouping? As you already figured out, multiple params to filter() are just joined via AND in the underlying SQL.
Originally you had this for the filter:
If you wanted (A & B) & (C | D) then this should work:
This answer is late but could be helpful to a lot of guys out there.
This will generate something similar to
OK, no success here or on #django. So I choose to use a raw SQL query to solve this problem...
Here the working code:
Here's my example of complicated query, I hope you find it helpful
The problem with this method is that you get an
empty (AND: )
case in youror_condition
. It doesn't affect the query at all, but it annoys me! My current solution is as follows