Skip to main content

Detect N+1s in your Django app

Project description

zeal

Catch N+1 queries in your Django project.

Static Badge PyPI - Version

Features

  • Detects N+1s from missing prefetches and from use of .defer()/.only()
  • Friendly error messages like N+1 detected on User.followers at myapp/views.py:25 in get_user
  • Configurable thresholds
  • Allow-list
  • Well-tested
  • No dependencies

Acknowledgements

This library draws heavily from jmcarp's nplusone. It's not exactly a fork, but not far from it.

Installation

First:

pip install zeal

Then, add zeal to your INSTALLED_APPS and MIDDLEWARE. You probably don't want to run it in production: I haven't profiled it but it will have a performance impact.

if DEBUG:
    INSTALLED_APPS.append("zeal")
    MIDDLEWARE.append("zeal.middleware.zeal_middleware")

This will detect N+1s that happen in web requests. To catch N+1s in more places, read on!

Celery

If you use Celery, you can configure this using signals:

from celery.signals import task_prerun, task_postrun
from zeal import setup, teardown
from django.conf import settings

@task_prerun.connect()
def setup_zeal(*args, **kwargs):
    setup()

@task_postrun.connect()
def teardown_zeal(*args, **kwargs):
    teardown()

Tests

Django runs tests with DEBUG=False, so to run zeal in your tests, you'll first need to ensure it's added to your INSTALLED_APPS and MIDDLEWARE. You could do something like:

import sys

TEST = "test" in sys.argv
if DEBUG or TEST:
    INSTALLED_APPS.append("zeal")
    MIDDLEWARE.append("zeal.middleware.zeal_middleware")

This will enable zeal in any tests that go through your middleware. If you want to enable it in all tests, you need to do a bit more work.

If you use pytest, use a fixture in your conftest.py:

import pytest
from zeal import zeal_context

@pytest.fixture(scope="function", autouse=True)
def use_zeal():
    with zeal_context():
        yield

If you use unittest, add a custom test runner:

# In e.g. `myapp/testing/test_runners.py`
from zeal import setup as zeal_setup, teardown as zeal_teardown
from django.test.runner import DiscoverRunner
from unittest.runner import TextTestResult

class ZealTestResult(TextTestResult):
    def startTest(self, test):
        zeal_setup()
        return super().startTest(test)

    def addError(self, test, err) -> None:
        zeal_teardown()
        return super().addError(test, err)

    def addFailure(self, test, err) -> None:
        zeal_teardown()
        return super().addFailure(test, err)

    def addSuccess(self, test):
        zeal_teardown()
        return super().addSuccess(test)

class ZealTestRunner(DiscoverRunner):
    def get_resultclass(self):
        return ZealTestResult


# And in your settings:
TEST_RUNNER = (
    "myapp.testing.test_runners.ZealTestRunner"
)

Generic setup

If you also want to detect N+1s in other places not covered here, you can use the setup and teardown functions, or the zeal_context context manager:

from zeal import setup, teardown, zeal_context


def foo():
    setup()
    try:
        # your code goes here
    finally:
        teardown()


@zeal_context()
def bar():
    # your code goes here


def baz():
    with zeal_context():
        # your code goes here

Configuration

By default, any issues detected by zeal will raise a ZealError. If you'd rather log any detected N+1s, you can set:

ZEAL_RAISE = False

N+1s will be reported when the same query is executed twice. To configure this threshold, set the following in your Django settings.

ZEAL_NPLUSONE_THRESHOLD = 3

To handle false positives, you can temporarily disable zeal in parts of your code using a context manager:

from zeal import zeal_ignore

with zeal_ignore():
    # code in this block will not log/raise zeal errors

Finally, if you want to ignore N+1 alerts from a specific model/field globally, you can add it to your settings:

ZEAL_ALLOWLIST = [
    {"model": "polls.Question", "field": "options"},

    # you can use fnmatch syntax in the model/field, too
    {"model": "polls.*", "field": "options"},

    # if you don't pass in a field, all N+1s arising from the model will be ignored
    {"model": "polls.Question"},
]

Comparison to nplusone

zeal borrows heavily from nplusone, but has some differences:

  • zeal also detects N+1 caused by using .only() and .defer()
  • it lets you configure your own threshold for what constitutes an N+1
  • it has slightly more helpful error messages that tell you where the N+1 occurred
  • nplusone patches the Django ORM even in production when it's not enabled. zeal does not!
  • nplusone appears to be abandoned at this point.
  • however, zeal only works with Django, whereas nplusone can also be used with SQLAlchemy.

Contributing

  1. First, install uv.
  2. Create a virtual env using uv venv and activate it with source .venv/bin/activate.
  3. Run make install to install dev dependencies.
  4. To run tests, run make test.

Project details


Download files

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

Source Distributions

No source distribution files available for this release.See tutorial on generating distribution archives.

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

django_zeal-1.0.0-py3-none-any.whl (10.4 kB view details)

Uploaded Python 3

File details

Details for the file django_zeal-1.0.0-py3-none-any.whl.

File metadata

  • Download URL: django_zeal-1.0.0-py3-none-any.whl
  • Upload date:
  • Size: 10.4 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/5.1.1 CPython/3.9.19

File hashes

Hashes for django_zeal-1.0.0-py3-none-any.whl
Algorithm Hash digest
SHA256 30c1b47ea714d0fe3cbf442a7bf9c0b2ce1796c441674a5714b9f51ba1d7b0bc
MD5 8bd1bdad105597c7efbc68932aecfeab
BLAKE2b-256 20c1a1c1d0ec99a89cb2be565f821c302b9de5c8594ceadb1bf489b869a173a5

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page