Navigation Menu

Skip to content


Repository files navigation


docs Documentation Status
Travis-CI Build Status Requirements Status
Coverage Status Coverage Status
Scrutinizer Status Codacy Code Quality Status CodeClimate Quality Status
PyPI Package latest release PyPI Wheel Supported versions Supported implementations
Commits since latest release

Simple and generic model related data prefetch framework for Django solving the "1+N queries" problem that happens when you need related data for your objects.

In most of the cases you'll have forward relations (foreign keys to something) and can use select_related to fetch that data on the same query. However, in some cases you cannot design your models that way and need data from reverse relations (models that have foreign keys to your objects).

Django has prefetch_related for this, however, this framework provides greater flexibility than Django's prefetch_related queryset method at the cost of writting the mapping and query functions for the data. This has the advantage that you can do things prefetch_related cannot (see the latest_book example below).

  • Free software: BSD license

Installation guide

Install it:

pip install django-prefetch

Use it as your model's default manager (or as a base class if you have custom manager).





Python 2.7, 3.3+ or PyPy




Here's a simple example of models and prefetch setup:

from django.db import models
from prefetch import PrefetchManager, Prefetcher

class Author(models.Model):
    name = models.CharField(max_length=100)

    objects = PrefetchManager(
        books = Prefetcher(
            filter = lambda ids: Book.objects.filter(author__in=ids),
            reverse_mapper = lambda book: [book.author_id],
            decorator = lambda author, books=(): setattr(author, 'books', books)
        latest_book = Prefetcher(
            filter = lambda ids: Book.objects.filter(author__in=ids),
            reverse_mapper = lambda book: [book.author_id],
            decorator = lambda author, books=(): setattr(
                max(books, key=lambda book: book.created) if books else None

class Book(models.Model):
    class Meta:
        get_latest_by = 'created'

    name = models.CharField(max_length=100)
    created = models.DateTimeField(auto_now_add=True)
    author = models.ForeignKey(Author)

Use it like this:

for a in Author.objects.prefetch('books', 'latest_book'):
    print a.books
    print a.latest_book

Prefetcher arguments

Example models:

class LatestNBooks(Prefetcher):
    def __init__(self, count=2):
        self.count = count

    def filter(self, ids):
        return Book.objects.filter(author__in=ids)

    def reverse_mapper(self, book):
        return [book.author_id]

    def decorator(self, author, books=()):
        books = sorted(books, key=lambda book: book.created, reverse=True)
                'latest_%s_books' % self.count,

class Author(models.Model):
    name = models.CharField(max_length=100)

    objects = PrefetchManager(
        latest_n_books = LatestNBooks

Use it like this:

from prefetch import P

for a in Author.objects.prefetch(P('latest_n_books', count=5)):
    print a.latest_5_book


P is optional and you can only use for prefetch definitions that are Prefetcher subclasses. You can't use it with prefetcher-instance style definitions like in the first example. Don't worry, if you do, you will get an exception explaining what's wrong.

Other examples

Check out the tests for more examples.


  • Document collect option of Prefetcher
  • Create tests covering custom collect and mapper


To run all the tests run:


Note, to combine the coverage data from all the tox environments run:

set PYTEST_ADDOPTS=--cov-append
PYTEST_ADDOPTS=--cov-append tox