Skip to content

psrb/pylint-django

 
 

Repository files navigation

pylint-django

image

image

image

image

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 and then execute:

pylint --load-plugins pylint_django [..other options..] <path_to_your_sources>

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.

Additional plugins

pylint_django.checkers.db_performance 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 prefered 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!

This plugin is disabled by default! To enable it:

pylint --load-plugins pylint_django --load-plugins pylint_django.checkers.db_performance

Known issues

If you reference foreign-key models by their name (as string) pylint-django may not be able to find the model and will report issues because it has no idea what the underlying type of this field is. Supported options are:

- ``self`` and ``Model`` - look for this class in the current module which is being examined
- ``app.Model`` - try loading ``app.models`` into the AST parser and look for ``Model`` there

If your models.py itself is not importing the foreign-key class there's probably some import problem (likely circular dependencies) preventing referencing the foreign-key class directly. In this case pylint-django can't do much about it. We always recommend referencing foreign-key models by their classes.

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.

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.

License

pylint-django is available under the GPLv2 license.

About

Pylint plugin for improving code analysis for when using Django

Resources

License

Security policy

Stars

Watchers

Forks

Packages

No packages published

Languages

  • Python 97.3%
  • Shell 2.7%