Python Packaging: Key Points

Python Scripts and Modules

Any Python file can be considered a ‘module’, and can be import-ed. This just runs the code in the file, and makes the names of any objects in the module accessible using dot-notation.
If we bundle our Python scripts into a collection of functions, we can reuse those functions in other modules or in the Python interpretter.
After turning our scripts into reusable modules, we can maintain script-like behaviour using the idiom if __name__ == "__main__".
argparse can be used to create sophisticated command line interfaces.

Packages can be used to better organise our code as it becomes more complex
Packages are defined by a directory structure and the presence of __init__.py
By controlling what is or isn’t imported in __init__.py, we can define a public API for our project.
A package can be run as a script using the -m flag, provided it contains a file __main__.py.
Through clever use of argparse, we can provide a rich scripting interface to a whole package.

We can configure our projects for pip installation by setting up a pyproject.toml file.
Simple projects require very little to be added to this file, but there are many optional extras we can add.

The ‘best practices’ for Python packaging has changed a number of times, and there are now many competing tools for accomplishing the same task.
Be wary of information on this topic in online tutorials – not all guides have been updated to use the most recent methods, and some advice may no longer be relevant.

Versioning our projects is important so that our users know what’s compatible.
GitHub is a powerful service for hosting our projects and managing their development.
Each new release of our packages should be uploaded to PyPI using build and twine.