CLI

Our CLI is written in Python. It has two modules: quick_client and quick. The quick_client module is the REST API client for the manager. It's automatically generated as explained in Development - OpenAPI The quick module is a CLI wrapper for the REST API client.

Development

Setup

We use poetry as build system. You can run poetry install in the project directory. This command creates a new virtual environment with all runtime and dev dependencies installed by default.

Note

You can run poetry config virtualenvs.path <PATH> to set the path where virtualenvs are created.

To create a new version and install it, run:

poetry build -f wheel && \
pip install ./dist/quick_cli-0.1.3-py3-none-any.whl --force-reinstall

If you want to add a new dependency, make sure to commit the updated poetry.lock file.

Code quality

We use flake8, isort, and black for formatting and linting the project. You can use the provided pre-commit-config to make sure your PRs pass. Install the hooks with pre-commit install (making them run before each commit). You can also run them manually with pre-commit run --all-files. Currently, mypy is excluded from pre-commit. However, feel free to run mypy . and fix possible errors.

Documentation

The CLI comes with an argparse-to-markdown script for generating the CLI's documentation. The created file is commands.md. After updating commands, update it with the following command:

python quick/generate_docs.py > commands.md

Architecture

The CLI uses Python's ArgParse. It follows an OOP approach for defining new subcommands. There are two base classes:

Group

A Group is a subcommand that has one more child subcommands. The subcommands are defined in its sub_parser field.

Attention

A new top-level group (i.e., used as quick <new-group>) must be added the COMMANDS list in commands/__init__.py.

Command

A Command is a subcommand that executes an action. You can override the add_args method to add new required and optional arguments. The execute method contains the logic for the command action.

This approach using base classes lets us define common behavior, like recurring arguments (for example, --debug) and setup configurations.