pylint-django 2.6.1

A Pylint plugin to help Pylint understand the Django web framework

About

pylint-django is a Pylint plugin for improving code analysis when analysing code using Django. It is also used by the Prospector tool.

Installation

To install:

pip install pylint-django

WARNING: pylint-django will not install Django by default because this causes more trouble than good, see discussion. If you wish to automatically install the latest version of Django then:

pip install pylint-django[with-django]

otherwise sort out your testing environment and please DO NOT report issues about missing Django!

Usage

Ensure pylint-django is installed and on your path. In order to access some of the internal Django features to improve pylint inspections, you should also provide a Django settings module appropriate to your project. This can be done either with an environment variable:

DJANGO_SETTINGS_MODULE=your.app.settings pylint --load-plugins pylint_django [..other options..] <path_to_your_sources>

Alternatively, this can be passed in as a commandline flag:

pylint --load-plugins pylint_django --django-settings-module=your.app.settings [..other options..] <path_to_your_sources>

If you do not configure Django, default settings will be used but this will not include, for example, which applications to include in INSTALLED_APPS and so the linting and type inference will be less accurate. It is recommended to specify a settings module.

Prospector

If you have prospector installed, then pylint-django will already be installed as a dependency, and will be activated automatically if Django is detected:

prospector [..other options..]

Features

• Prevents warnings about Django-generated attributes such as Model.objects or Views.request.

• Prevents warnings when using ForeignKey attributes (“Instance of ForeignKey has no <x> member”).

• Fixes pylint’s knowledge of the types of Model and Form field attributes

• Validates Model.__unicode__ methods.

• Meta informational classes on forms and models do not generate errors.

• Flags dangerous use of the exclude attribute in ModelForm.Meta.

• Uses Django’s internal machinery to try and resolve models referenced as strings in ForeignKey fields. That relies on django.setup() which needs the appropriate project settings defined!

Additional plugins

pylint_django.checkers.migrations looks for migrations which:

• add new model fields and these fields have a default value. According to Django docs this may have performance penalties especially on large tables. The preferred way is to add a new DB column with null=True because it will be created instantly and then possibly populate the table with the desired default values. Only the last migration from a sub-directory will be examined;

• are migrations.RunPython() without a reverse callable - these will result in non reversible data migrations;

This plugin is disabled by default! To enable it:

pylint --load-plugins pylint_django --load-plugins pylint_django.checkers.migrations

Contributing

Please feel free to add your name to the CONTRIBUTORS.rst file if you want to be credited when pull requests get merged. You can also add to the CHANGELOG.rst file if you wish, although we’ll also do that when merging.

Tests

The structure of the test package follows that from pylint itself.

It is fairly simple: create a module starting with func_ followed by a test name, and insert into it some code. The tests will run pylint against these modules. If the idea is that no messages now occur, then that is fine, just check to see if it works by running scripts/test.sh.

Any command line argument passed to scripts/test.sh will be passed to the internal invocation of pytest. For example if you want to debug the tests you can execute scripts/test.sh --capture=no. A specific test case can be run by filtering based on the file name of the test case ./scripts/test.sh -k 'func_noerror_views'.

Ideally, add some pylint error suppression messages to the file to prevent spurious warnings, since these are all tiny little modules not designed to do anything so there’s no need to be perfect.

It is possible to make tests with expected error output, for example, if adding a new message or simply accepting that pylint is supposed to warn. A test_file_name.txt file contains a list of expected error messages in the format error-type:line number:class name or empty:1st line of detailed error text:confidence or empty.

License

pylint-django is available under the GPLv2 license.