Skip to main content

Improved API for aggregating using Subquery

Project description

https://travis-ci.org/martsberger/django-sql-utils.svg?branch=master

Django SQL Utils

This package provides utilities for working with Django querysets so that you can generate the SQL that you want, with an API you enjoy.

Subquery Aggregates

The Count aggregation in Django:

Parent.objects.annotate(child_count=Count('child'))

generates SQL like the following:

SELECT parent.*, Count(child.id) as child_count
FROM parent
JOIN child on child.parent_id = parent.id
GROUP BY parent.id

In many cases, this is not as performant as doing the count in a SUBQUERY instead of with a JOIN:

SELECT parent.*,
       (SELECT Count(id)
        FROM child
        WHERE parent_id = parent.id) as child_count
FROM parent

Django allows us to generate this SQL using The Subquery and OuterRef classes:

subquery = Subquery(Child.objects.filter(parent_id=OuterRef('id')).order_by()
                    .values('parent').annotate(count=Count('pk'))
                    .values('count'), output_field=IntegerField())
Parent.objects.annotate(child_count=Coalesce(subquery, 0))

Holy cow! It’s not trivial to figure what everything is doing in the above code and it’s not particularly good for maintenance. SubqueryAggregates allow you to forget all that complexity and generate the subquery count like this:

Parent.objects.annotate(child_count=SubqueryCount('child'))

Phew! Much easier to read and understand. It’s the same API as the original Count just specifying the Subquery version.

Easier API for Exists

If you have a Parent/Child relationship (Child has a ForeignKey to Parent), you can annotate a queryset of Parent objects with a boolean indicating whether or not the parent has children:

from django.db.models import Exists

parents = Parent.objects.annotate(
    has_children=Exists(Child.objects.filter(parent=OuterRef('pk'))
)

That’s a bit more boilerplate than should be necessary, so we provide a simpler API for Exists:

from sql_util.utils import Exists

parents = Parent.objects.annotate(
    has_children=Exists('child')
)

The child queryset can be filtered with the keyword argument filter. E.g.,:

parents = Parent.objects.annotate(
    has_child_named_John = Exists('child', filter=Q(name='John'))
)

The sql_util version of Exists can also take a queryset as the first parameter and behave just like the Django Exists class, so you are able to use it everywhere without worrying about name confusion.

Installation and Usage

Install from PyPI:

pip install django-sql-utils

Then you can:

from sql_util.utils import SubqueryCount

And use that as shown above.

In addition to SubqueryCount, this package provides

  • SubqueryMin
  • SubqueryMax
  • SubquerySum
  • SubqueryAvg

If you want to use other aggregates, you can use the generic SubqueryAggregate class. For example, if you want to use Postgres’ ArrayAgg to get an array of Child.name for each Parent:

from django.contrib.postgres.aggregates import ArrayAgg

aggregate = SubqueryAggregate('child__name', aggregate=ArrayAgg)
Parent.objects.annotate(child_names=aggregate)

Or subclass SubqueryAggregate:

from django.contrib.postgres.aggregates import ArrayAgg

class SubqueryArrayAgg(SubqueryAggregate)
    aggregate = ArrayAgg
    unordered = True

Parent.objects.annotate(avg_child_age=SubqueryArrayAgg('child__age'))

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Filename, size & hash SHA256 hash help File type Python version Upload date
django_sql_utils-0.4.0-py3-none-any.whl (13.1 kB) Copy SHA256 hash SHA256 Wheel py3
django-sql-utils-0.4.0.tar.gz (9.8 kB) Copy SHA256 hash SHA256 Source None

Supported by

Elastic Elastic Search Pingdom Pingdom Monitoring Google Google BigQuery Sentry Sentry Error logging AWS AWS Cloud computing DataDog DataDog Monitoring Fastly Fastly CDN SignalFx SignalFx Supporter DigiCert DigiCert EV certificate StatusPage StatusPage Status page