CODEWITHSUNDEEP
pythonimportsetup-project
Razer
Razer

Reputation: 8201

Python folder structure for larger projects

I'm new to using python for larger projects. I figured out following folder structure for my python project:

project
├── doc
├── src
│   ├── hardware
│   ├── devices
│   │   ├── device1
│   │   └── device2
│   ├── measurement
│   ├── ui
│   └── util
└── tests
    ├── hardware
    ├── devices
    │   ├── device1
    │   └── device2
    ├── measurement
    ├── ui
    └── util

The tests folder contains unittests for modules in the soure folder. Is this the right approach for such a program?

How should I do the importing? I thought to add the root folder to the python path and import every module absolute from the root path. Should I do this so?

This program is used on several coimputers. How should I deploy it? Currently I use SVN and every computer gets the whole project and starts it from there. There is one disadvantage, I have to add the prject path to the PYTHONPATH on every computer.

Upvotes: 2

Views: 438

Answers (2)

lbolla
lbolla

Reputation: 5411

Your best option is to use virtualenv, and have a setup.py to describe the package.

Then, for development, inside the virtual environment, just do python setup.py develop to install the egg in development mode and you are ready to go: http://packages.python.org/distribute/setuptools.html#develop-deploy-the-project-source-in-development-mode

In production, just use the usual: python setup.py install (always inside a virtualenv).

Upvotes: 0

Nicolas Barbey
Nicolas Barbey

Reputation: 6797

The global structure looks good to me. You need a module to write a setup.py script to install your package such as setuptools. This will allow installation and distribution of the package. See also Pypi for distribution of the package. Typically, you can :

sudo setup.py install

to install your package at the system level. Or you can :

sudo setup.py install --prefix=~/local

to install at the user level. But then you have to add ~/local/lib/python/site-packages to your PYTHONPATH. But it has to be done only once for all packages installed by this user.

If your package is stored on PyPI, you can on any machine do :

easy_install mypackage

to auto-magically install the package.

Even better you can make your packages only visible in isolated python "virtual environment" using virtualenv. This allow for testing of various packages and versions on the same computer a breeze.

You can also add README files and such at the package root level.

Upvotes: 2

Related Questions