Packages and Setup

We have seen how you can write your own libraries and packages of code and import them when they are in the same folder as a script. What if you want to be able to import your code from any location by installing it to your Python packages?

import mypack

You can do all this using a file called setup.py and the setuptools Python package.

Package Structure¶

There is a bit of variation in how Python packages are structured in practice. We’ll talk about one option, which is fairly standard for low-complexity projects. Let’s say your Python package is called mypack. All the source materials for your package (including code, documentation, tests...) should be contained in a single folder, ideally a git repository. One example is the package in the python-packages repository.

Inside this top-level folder, you will have several sub-folders and files:

• mypack - the folder containing __init__.py which is your Python package source code.

• doc - folder which holds documentation for your package.

• tests folder, or test.py - code for testing your package (unit tests).

• requirements.txt - dependencies of your package (e.g. numpy, scipy, etc.)

• setup.py - a Python script used to configure and automate your package installation

• LICENSE - a license for people who use your code. See choosealicence.com for help choosing a license for your original work. If your code is based on someone else’s you will probably need to keep their license.

• README.md - a file that contains all the information people should read before installing/using your package.

There may be additional files as well (e.g. .git, files for CI, etc.) the above list is not exhaustive.

requirements.txt¶

This is a file which lists dependencies of your package. You can list packages without version numbers, or use some version specifiers

matplotlib # no version number
numpy >= 1.18 # must be version 1.18 or later
scipy >= 1.5 # must be version 1.5 or later

You can install requirements using conda:

conda install --file requirements.txt

or pip:

pip install -r requirements.txt

setup.py¶

This is a file that is used to configure and install your repository to your Python path. From the root of your repository, you should be able to run

pip install .

to install the code to your Python path. You can then use mypack anywhere - not just in the directory that contains the source code.

In order to uninstall your package, you can run

pip uninstall mypack

You can also run setup.py as a script:

python setup.py install

In general, you’ll want to use the setuptools package. Basic useage is to use setuptools.setup, which accepts a variety of keyword arguments that help build and describe your package. See here for more details.

• name: your package name (name='mypack' in this case)

• packages: the actual package source code location (packages=['mypack'])