Python Package Template

Project template for a Python Package using Copier.

Features

Project for Python 3.8+.
Testing with Pytest using GitHub actions.
Packaging powered by poetry.
Optionally generates a CLI entry point powered by Typer and Rich.
Optionally makes it a Django package.
Uses Ruff for formatting and linting.
Comes with pre-commit hook config for Ruff.
Style guide enforced on CI.
Dependencies kept up to date by Renovate.
Follow the all-contributors specification.
Follow to the conventional commits specification.
Automated releasing using python-semantic-release.
Documentation configured with Sphinx and MyST Parser.
Follows the contributor covenant code of conduct.
Standardised list of GitHub labels synchronised on push to master using the labels CLI.

Usage

Generate a new project with:

copier copy --trust "gh:browniebroke/pypackage-template" path-to-project

This will prompt you for a few questions and create new directory with the name you used as project slug.

Note: the --trust option is required because this template may execute some tasks after generating the project, like initialising the git repo, installing dependencies and so forth. These are all listed in the copier.yml of this repo, under the _tasks key. They are all optional and safe to run. You can take my word for it, or better, check the code yourself!

Start developing

The project uses Poetry for dependencies management and packaging. Make sure you have it installed in your development machine. To install the development dependencies in a virtual environment, type:

poetry install

This will also generate a poetry.lock file, you should track this file in version control. To execute the test suite, call pytest inside Poetry's virtual environment via poetry run:

poetry run pytest

Check out the Poetry documentation for more information on the available commands.

GitHub Actions

When you first push to GitHub, it'll start a ci GitHub workflow that you can see in the "Actions" tab of your repository. This workflow runs a couple of jobs:

The test job will run your test suite with Pytest against all Python version from 3.8 to 3.11
A few things will run in the lint job:
- Ruff format
- Ruff lint with several flake8, isort and pyupgrade plugins.

A labels workflow will also run and synchronise the GitHub labels based on the .github/labels.toml file.

Secrets

The workflows need a few secrets to be setup in your GitHub repository:

GH_PAT a personal access token (PAT) with the repo scope for opening pull requests and updating the repository topics. This is used by the poetry-upgrade and labels workflows.
CODECOV_TOKEN to upload coverage data to codecov.io in the Test workflow.

If you have the GitHub CLI installed and chose to set up GitHub, they will be created with a dummy value (changeme).

Automated release

By following the conventional commits specification, we're able to completely automate versioning and releasing to PyPI. It runs on every push to your main branch, as part of the release job of the ci.yml workflow. You shouldn't need to create a token, but you'll need to setup trusted publisher for the project.

Trusted publisher setup

The first time you push, the workflow will try to create a release in PyPI, however the project doesn't exist there yet, which seems like a chicken and egg situation. Luckily, you can add a trusted publisher before creating the PyPI project here: https://pypi.org/manage/account/publishing/. Here are the infos that you should use:

PyPI project name: what you've entered as "project slug"
Owner: your GitHub username
Repository name: what you've entered as "project slug"
Workflow name: ci.yml
Environment name: release

If the release phase failed the first time, you might have to remove the release and tag from GitHub, and perhaps tidy up the changelog.

How it works

Here is an overview of what it's doing:

Check the commit log since the last release, and determine the next version to be released.
If no significant change detected, stop here (e.g. just dependencies update).
Otherwise, bump the version in code locations specified in setup.cfg.
Update the CHANGELOG.md file.
Commit changes.
Create a git tag.
Push to GitHub.
Create a release in GitHub with the changes as release notes.
Build the source and binary distribution (wheel).
Upload the sources to PyPI and attach them to the Github release, using trusted publisher.

For more details, check out the conventional commits website and Python semantic release Github action.

Pre-commit

The project comes with the config for pre-commit. If you're not familiar with it, follow their documentation on how to install it and set it up.

Documentation

The project assumes that the documentation will be hosted on Read the Docs and written in Markdown with the MyST parser for Sphinx.

To enable it, you might need to go into your dashboard and import the project from Github. Everything else should work out of the box.

Dependencies update

The project dependencies are kept up to date with Renovate which requires the Github app to be installed.

The main advantage of Renovate over Dependabot is the auto-merge option, which is configured to automatically merge minor/patch updates with all the CI checks passing. It supports a variety of package managers, including Poetry, GitHub actions and pre-commit hooks which are used by default.

All contributors

This is a specification that help you highlight every open source contribution in your README. This is easy to maintain as it comes with a GitHub bot to do the updates for you, so more manual updates on the contributors file.

If you never used it before, you will have to install the Github app and give it access to your repo.