Group by arbitrary model fields
Project description
This package provides a mixin for Django QuerySets that adds a method group_by that behaves mostly like the values method, but with one difference: its iterator does not return dictionaries, but a model-like object with model instances for related values.
Installation
Install from PyPI:
pip install django-group-by
Compatibility
This package is compatible with Django 1.8, 1.9 and 1.10, and Python versions 2.7, 3.4, 3.5 and 3.6. Probably others, but those 12 combinations are the ones against which we build (by Travis CI).
Usage
Create a QuerySet subclass with the GroupByMixin to use in a model’s manager:
# models.py
from django.db import models
from django.db.models.query import QuerySet
from django_group_by import GroupByMixin
class BookQuerySet(QuerySet, GroupByMixin):
pass
class Book(Model):
objects = BookQuerySet.as_manager()
title = models.CharField(max_length=100)
author = models.ForeignKey('another_app.Author')
publication_date = models.DateField()
...
Then use it just like values, and you’ll get a similar query set:
>>> some_rows = Book.objects.group_by('title', 'author', 'author__nationality').distinct()
>>> some_rows.count()
4
The difference is that every row is not a dictionary but an AggregatedGroup instance, with only the grouped fields:
>>> row = some_rows[0] >>> row <AggregatedGroup for Book> >>> row.title The Colour of Magic >>> row.publication_date *** AttributeError: 'AggregatedGroup' object has no attribute 'publication_date'
But of course the main advantage is that you also get related model instances, as far as you want:
>>> row.author <Author: Terry Pratchett> >>> row.author_nationality <Nation: Great Britain>
