Data Science Cookie Cutter Template

This repository is a Cookie Cutter template for spinning up a data science project in which you'd also like to quickly develop and refactor Python code into a separate package.

This template uses poetry to install dependencies, and will install the package that's being created in editable mode into the current virtual environment.

This template:

• creates a data folder
• creates a notebooks folder
• creates a Python package folder with name you passed on the command line
• creates scripts folder for global scripts to be run
• creates a tests folder to test your package
• creates a pyproject.toml file with common dev dependencies, default tool configurations and Python package configurations
• creates a .git-dev folder:
  • hooks/prepare-commit-msg - a git hook that appends your ticket number (pulled from your active branch name) and inserts it at the end of your commit message
  • .gitmessage - a default template for your commit messages
  • config - additional project-specific configurations for git that get appended to your .git/config file
    • configure git to use VSCode as the commit message editor
    • configure git to use the gitmessage template inside .git/.gitmessage as the default commit message template
• creates a .vscode-dev folder:
  • it gets renamed to .vscode during installation so it's used by your VSCode IDE
  • it contains settings.json with:
    • a configuration to use pytest on the tests directory in the root of the repo
    • your current active poetry Python interpreter gets injected into the settings.json file at installation time
    • mypy validation is enabled by default
• creates a .pre-commit-config.yaml file:
  • it's configured so that you just run pre-commit install and it installs pre-commit and pre-push hooks
• creates a .sourcery.yaml file to configure sourcery to refactor your code; it comes with the Google Style Guide rules applied
• creates a .gitignore file with a lot of default ignorable file patterns
• creates a Makefile with some helper PHONY targets

Usage

This is a cookie cutter template. So the steps to using this repo are to:

1. clone this repo
2. edit the cookiecutter.json file to reflect your project's details
3. run cookiecutter <path-to-this-repo> from wherever you'd like to create your new project folder

You can also run cookiecutter directly from GitHub with this template by running one of:

cookiecutter https://github.com/jcorrado76/data-science-cookie-cutter-with-package.git
cookiecutter git+ssh://[email protected]:jcorrado76/data-science-cookie-cutter-with-package.git
cookiecutter gh:jcorrado76/data-science-cookie-cutter-with-package

Read more about using cookiecutter from here.

You can find the command line options here.

After running these commands, you should have your project folder, by the name of {{ cookiecutter.project_name }}.

Inside that project folder, you'll have the above named resources, plus a folder with the value of {{ cookiecutter.package_name }} that you supplied.

The idea is that we have tests in the root of the project, so it's easy for developers to run testing from the root of the repo.

This facilitates:

1. not including tests in your deployed package
2. installing your python package locally in editable mode, so you can use it easier

The .pre-commit-config starts with only the isort step uncommented, since it's relatively non-controversial, and doesn't make large edits to your code (it just sorts your imports).

The inspiration for the directory structure comes from the Prefect repo.

FAQ

If you get this error when the pre-commit hooks are installing:

ImportError: dlopen(/Users/josephcorrado/anaconda3/lib/python3.10/lib-dynload/_sqlite3.cpython-310-darwin.so, 0x0002): Symbol not found: _sqlite3_enable_load_extension
  Referenced from: <D8F5DD12-3FC5-3F3D-BF16-05B2C1D7FD2B> /Users/josephcorrado/anaconda3/lib/python3.10/lib-dynload/_sqlite3.cpython-310-darwin.so
  Expected in:     <710E4989-2656-3774-8114-0C2936A8FC33> /usr/lib/libsqlite3.dylib

It's because the files from an older installation of the poetry virtual environment are cached and being used here, so there's a version mismatch. To fix it, just figure out which poetry virtual environment you're trying to install to:

poetry env list

Then remove it:

poetry env remove <env-name>

Then try again.