boa-restrictor 1.3.0

A custom Python linter from Ambient

Welcome to the boa-restrictor - a custom Python linter from Ambient

• PyPI
• GitHub
• Full documentation
• Creator & Maintainer: Ambient Digital

Rules

Positional arguments not allowed (PBR001)

This rule enforces that functions and methods don't contain any positional arguments.

This will make refactorings easier, is more explicit, and you avoid the boolean bug trap.

Wrong:

def my_func(a, b):
    pass

Correct:

def my_func(*, a, b):
    pass

Return type hints required if a return statement exists (PBR002)

This rule will enforce that you add a return type-hint to all methods and functions that contain a return statement. This way we can be more explicit and let the IDE help the next developer because it will add warnings if you use wrong types.

Wrong:

def my_func(a, b):
    return a * b

Correct:

def my_func(a, b) -> int:
    return a * b

Avoid nested import of datetime module (PBR003)

This rule will enforce that you never import a datetime object from the datetime module, but instead import the datetime module and get the object from there.

Since you can't distinguish in the code between a datetime module and datetime object without looking at the imports, this leads to inconsistent and unclear code.

Importing the date object can cause a namespace conflict with the Django template tag date, therefore this is not allowed as well.

Wrong:

from datetime import datetime

my_datetime = datetime(2024, 9, 19)

Correct:

import datetime

my_datetime = datetime.datetime(2024, 9, 19)

Note, that other imports from the datetime module like UTC are allowed since there are no known conflicts.

Use dataclasses with "kw_only" (PBR004)

This rule will enforce that you use the kw_only parameter in every dataclass decorator.

This will force the developer to set all dataclass attributes as kwargs instead of args, which is more explicit and easier to refactor.

Wrong:

from dataclasses import dataclass


@dataclass
class MyDataClass:
    pass

Correct:

from dataclasses import dataclass


@dataclass(kw_only=True)
class MyDataClass:
    pass

Service classes have one public method called "process" (PBR005)

Putting business logic in classes called "service" is a well-known and widely used pattern. To hide the inner workings of this logic, it's recommended to prefix all methods with an underscore ("_") to mark them as protected. The single entrypoint should be a public method called "process".

Abstract classes inherit from "abc.ABC" (PBR006)

Python provides a base class for abstract classes. If a class is named "abstract", it should therefore inherit from the abc.ABC class.

Installation

Add the following to your .pre-commit-config.yaml file:

  - repo: https://github.com/ambient-innovation/boa-restrictor
    rev: v1.2.1
    hooks:
      - id: boa-restrictor
        args: [ --config=pyproject.toml ]

Now you can run the linter manually:

pre-commit run --all-files boa-restrictor

Configuration

Exclude certain files

You can easily exclude certain files, for example, your tests, by using the exclude parameter from pre-commit:

  - repo: https://github.com/ambient-innovation/boa-restrictor
    rev: v1.2.1
    hooks:
      - id: boa-restrictor
        ...
        exclude: |
          (?x)^(
            /.*/tests/.*
            |.*/test_.*\.py
          )$

Globally exclude configuration rule

You can disable any rule in your pyproject.toml file as follows:

[tool.boa-restrictor]
exclude = [
    "PBR001",
    "PBR002",
]

Per-file exclusion of configuration rule

You can disable rules on a per-file-basis in your pyproject.toml file as follows:

[tool.boa-restrictor.per-file-excludes]
"*/tests/*" = [
    "PBR001",
    "PBR002",
]
"scripts/*" = [
    "PBR003",
]

Take care that the path is relative to the location of your pyproject.toml. This means that example two targets all files living in a scripts/ directory on the projects top level.

Ruff support

If you are using ruff, you need to tell it about our linting rules. Otherwise, ruff will remove all # noqa statements from your codebase.

[tool.ruff.lint]
# Avoiding flagging (and removing) any codes starting with `PBR` from any
# `# noqa` directives, despite Ruff's lack of support for `boa-restrictor`.
external = ["PBR"]

https://docs.astral.sh/ruff/settings/#lint_extend-unsafe-fixes

Contribute

Setup package for development

• Create a Python virtualenv and activate it
• Install "pip-tools" with pip install -U pip-tools
• Compile the requirements with pip-compile --extra dev, -o requirements.txt pyproject.toml --resolver=backtracking
• Sync the dependencies with your virtualenv with pip-sync

Add functionality

• Create a new branch for your feature
• Change the dependency in your requirements.txt to a local (editable) one that points to your local file system: -e /Users/workspace/boa-restrictor or via pip pip install -e /Users/workspace/boa-restrictor
• Ensure the code passes the tests
• Create a pull request

Run tests

• Run tests

  pytest --ds settings tests
  

• Check coverage

  coverage run -m pytest tests
  coverage report -m
  

Git hooks (via pre-commit)

We use pre-push hooks to ensure that only linted code reaches our remote repository and pipelines aren't triggered in vain.

To enable the configured pre-push hooks, you need to install pre-commit and run once:

pre-commit install -t pre-push -t pre-commit --install-hooks

This will permanently install the git hooks for both, frontend and backend, in your local .git/hooks folder. The hooks are configured in the .pre-commit-config.yaml.

You can check whether hooks work as intended using the run command:

pre-commit run [hook-id] [options]

Example: run single hook

pre-commit run ruff --all-files --hook-stage push

Example: run all hooks of pre-push stage

pre-commit run --all-files --hook-stage push

Update documentation

• To build the documentation, run: sphinx-build docs/ docs/_build/html/.
• Open docs/_build/html/index.html to see the documentation.

Publish to ReadTheDocs.io

• Fetch the latest changes in GitHub mirror and push them
• Trigger new build at ReadTheDocs.io (follow instructions in admin panel at RTD) if the GitHub webhook is not yet set up.

Publish to PyPi

• Update documentation about new/changed functionality

• Update the Changelog

• Increment version in main __init__.py

• Create pull request / merge to main

• This project uses the flit package to publish to PyPI. Thus, publishing should be as easy as running:

  flit publish
  

  To publish to TestPyPI use the following to ensure that you have set up your .pypirc as shown here and use the following command:

  flit publish --repository testpypi
  

Create new version for pre-commit

To be able to use the latest version in pre-commit, you have to create a git tag for the current commit. So please tag your commit and push it to GitHub.

Maintenance

Please note that this package supports the ambient-package-update. So you don't have to worry about the maintenance of this package. This updater is rendering all important configuration and setup files. It works similar to well-known updaters like pyupgrade or django-upgrade.

To run an update, refer to the documentation page of the "ambient-package-update".