Conditional Expressions

Conditional expressions let you use ifelifelse logic within filters, annotations, aggregations, and updates. A conditional expression evaluates a series of conditions for each row of a table and returns the matching result expression. Conditional expressions can also be combined and nested like other expressions.

The conditional expression classes

We’ll be using the following model in the subsequent examples:

  1. from django.db import models
  2. class Client(models.Model):
  3. REGULAR = 'R'
  4. GOLD = 'G'
  5. PLATINUM = 'P'
  6. ACCOUNT_TYPE_CHOICES = [
  7. (REGULAR, 'Regular'),
  8. (GOLD, 'Gold'),
  9. (PLATINUM, 'Platinum'),
  10. ]
  11. name = models.CharField(max_length=50)
  12. registered_on = models.DateField()
  13. account_type = models.CharField(
  14. max_length=1,
  15. choices=ACCOUNT_TYPE_CHOICES,
  16. default=REGULAR,
  17. )

When

class When(condition=None, then=None, \*lookups*)

A When() object is used to encapsulate a condition and its result for use in the conditional expression. Using a When() object is similar to using the filter() method. The condition can be specified using field lookups, Q objects, or Expression objects that have an output_field that is a BooleanField. The result is provided using the then keyword.

Changed in Django 4.0:

Support for lookup expressions was added.

Some examples:

  1. >>> from django.db.models import F, Q, When
  2. >>> # String arguments refer to fields; the following two examples are equivalent:
  3. >>> When(account_type=Client.GOLD, then='name')
  4. >>> When(account_type=Client.GOLD, then=F('name'))
  5. >>> # You can use field lookups in the condition
  6. >>> from datetime import date
  7. >>> When(registered_on__gt=date(2014, 1, 1),
  8. ... registered_on__lt=date(2015, 1, 1),
  9. ... then='account_type')
  10. >>> # Complex conditions can be created using Q objects
  11. >>> When(Q(name__startswith="John") | Q(name__startswith="Paul"),
  12. ... then='name')
  13. >>> # Condition can be created using boolean expressions.
  14. >>> from django.db.models import Exists, OuterRef
  15. >>> non_unique_account_type = Client.objects.filter(
  16. ... account_type=OuterRef('account_type'),
  17. ... ).exclude(pk=OuterRef('pk')).values('pk')
  18. >>> When(Exists(non_unique_account_type), then=Value('non unique'))
  19. >>> # Condition can be created using lookup expressions.
  20. >>> from django.db.models.lookups import GreaterThan, LessThan
  21. >>> When(
  22. ... GreaterThan(F('registered_on'), date(2014, 1, 1)) &
  23. ... LessThan(F('registered_on'), date(2015, 1, 1)),
  24. ... then='account_type',
  25. ... )

Keep in mind that each of these values can be an expression.

Note

Since the then keyword argument is reserved for the result of the When(), there is a potential conflict if a Model has a field named then. This can be resolved in two ways:

  1. >>> When(then__exact=0, then=1)
  2. >>> When(Q(then=0), then=1)

Changed in Django 3.2:

Support for using the condition argument with lookups was added.

Case

class Case(\cases, **extra*)

A Case() expression is like the ifelifelse statement in Python. Each condition in the provided When() objects is evaluated in order, until one evaluates to a truthful value. The result expression from the matching When() object is returned.

An example:

  1. >>>
  2. >>> from datetime import date, timedelta
  3. >>> from django.db.models import Case, Value, When
  4. >>> Client.objects.create(
  5. ... name='Jane Doe',
  6. ... account_type=Client.REGULAR,
  7. ... registered_on=date.today() - timedelta(days=36))
  8. >>> Client.objects.create(
  9. ... name='James Smith',
  10. ... account_type=Client.GOLD,
  11. ... registered_on=date.today() - timedelta(days=5))
  12. >>> Client.objects.create(
  13. ... name='Jack Black',
  14. ... account_type=Client.PLATINUM,
  15. ... registered_on=date.today() - timedelta(days=10 * 365))
  16. >>> # Get the discount for each Client based on the account type
  17. >>> Client.objects.annotate(
  18. ... discount=Case(
  19. ... When(account_type=Client.GOLD, then=Value('5%')),
  20. ... When(account_type=Client.PLATINUM, then=Value('10%')),
  21. ... default=Value('0%'),
  22. ... ),
  23. ... ).values_list('name', 'discount')
  24. <QuerySet [('Jane Doe', '0%'), ('James Smith', '5%'), ('Jack Black', '10%')]>

Case() accepts any number of When() objects as individual arguments. Other options are provided using keyword arguments. If none of the conditions evaluate to TRUE, then the expression given with the default keyword argument is returned. If a default argument isn’t provided, None is used.

If we wanted to change our previous query to get the discount based on how long the Client has been with us, we could do so using lookups:

  1. >>> a_month_ago = date.today() - timedelta(days=30)
  2. >>> a_year_ago = date.today() - timedelta(days=365)
  3. >>> # Get the discount for each Client based on the registration date
  4. >>> Client.objects.annotate(
  5. ... discount=Case(
  6. ... When(registered_on__lte=a_year_ago, then=Value('10%')),
  7. ... When(registered_on__lte=a_month_ago, then=Value('5%')),
  8. ... default=Value('0%'),
  9. ... )
  10. ... ).values_list('name', 'discount')
  11. <QuerySet [('Jane Doe', '5%'), ('James Smith', '0%'), ('Jack Black', '10%')]>

Note

Remember that the conditions are evaluated in order, so in the above example we get the correct result even though the second condition matches both Jane Doe and Jack Black. This works just like an ifelifelse statement in Python.

Case() also works in a filter() clause. For example, to find gold clients that registered more than a month ago and platinum clients that registered more than a year ago:

  1. >>> a_month_ago = date.today() - timedelta(days=30)
  2. >>> a_year_ago = date.today() - timedelta(days=365)
  3. >>> Client.objects.filter(
  4. ... registered_on__lte=Case(
  5. ... When(account_type=Client.GOLD, then=a_month_ago),
  6. ... When(account_type=Client.PLATINUM, then=a_year_ago),
  7. ... ),
  8. ... ).values_list('name', 'account_type')
  9. <QuerySet [('Jack Black', 'P')]>

Advanced queries

Conditional expressions can be used in annotations, aggregations, filters, lookups, and updates. They can also be combined and nested with other expressions. This allows you to make powerful conditional queries.

Conditional update

Let’s say we want to change the account_type for our clients to match their registration dates. We can do this using a conditional expression and the update() method:

  1. >>> a_month_ago = date.today() - timedelta(days=30)
  2. >>> a_year_ago = date.today() - timedelta(days=365)
  3. >>> # Update the account_type for each Client from the registration date
  4. >>> Client.objects.update(
  5. ... account_type=Case(
  6. ... When(registered_on__lte=a_year_ago,
  7. ... then=Value(Client.PLATINUM)),
  8. ... When(registered_on__lte=a_month_ago,
  9. ... then=Value(Client.GOLD)),
  10. ... default=Value(Client.REGULAR)
  11. ... ),
  12. ... )
  13. >>> Client.objects.values_list('name', 'account_type')
  14. <QuerySet [('Jane Doe', 'G'), ('James Smith', 'R'), ('Jack Black', 'P')]>

Conditional aggregation

What if we want to find out how many clients there are for each account_type? We can use the filter argument of aggregate functions to achieve this:

  1. >>> # Create some more Clients first so we can have something to count
  2. >>> Client.objects.create(
  3. ... name='Jean Grey',
  4. ... account_type=Client.REGULAR,
  5. ... registered_on=date.today())
  6. >>> Client.objects.create(
  7. ... name='James Bond',
  8. ... account_type=Client.PLATINUM,
  9. ... registered_on=date.today())
  10. >>> Client.objects.create(
  11. ... name='Jane Porter',
  12. ... account_type=Client.PLATINUM,
  13. ... registered_on=date.today())
  14. >>> # Get counts for each value of account_type
  15. >>> from django.db.models import Count
  16. >>> Client.objects.aggregate(
  17. ... regular=Count('pk', filter=Q(account_type=Client.REGULAR)),
  18. ... gold=Count('pk', filter=Q(account_type=Client.GOLD)),
  19. ... platinum=Count('pk', filter=Q(account_type=Client.PLATINUM)),
  20. ... )
  21. {'regular': 2, 'gold': 1, 'platinum': 3}

This aggregate produces a query with the SQL 2003 FILTER WHERE syntax on databases that support it:

  1. SELECT count('id') FILTER (WHERE account_type=1) as regular,
  2. count('id') FILTER (WHERE account_type=2) as gold,
  3. count('id') FILTER (WHERE account_type=3) as platinum
  4. FROM clients;

On other databases, this is emulated using a CASE statement:

  1. SELECT count(CASE WHEN account_type=1 THEN id ELSE null) as regular,
  2. count(CASE WHEN account_type=2 THEN id ELSE null) as gold,
  3. count(CASE WHEN account_type=3 THEN id ELSE null) as platinum
  4. FROM clients;

The two SQL statements are functionally equivalent but the more explicit FILTER may perform better.

Conditional filter

When a conditional expression returns a boolean value, it is possible to use it directly in filters. This means that it will not be added to the SELECT columns, but you can still use it to filter results:

  1. >>> non_unique_account_type = Client.objects.filter(
  2. ... account_type=OuterRef('account_type'),
  3. ... ).exclude(pk=OuterRef('pk')).values('pk')
  4. >>> Client.objects.filter(~Exists(non_unique_account_type))

In SQL terms, that evaluates to:

  1. SELECT ...
  2. FROM client c0
  3. WHERE NOT EXISTS (
  4. SELECT c1.id
  5. FROM client c1
  6. WHERE c1.account_type = c0.account_type AND NOT c1.id = c0.id
  7. )