pundler

Bundler inspired requirements pinning for Python

MIT License

Downloads
61
Stars
27
Committers
4

Pundler_

.. image:: https://travis-ci.org/steder/pundler.png

About

Pundler is an attempt to better manage python requirements files.

Pundler is inspired by Ruby's Gem Bundler_.

Specifically the goal is to process requirements.yml or requirements.in into a frozen lock file requirements.txt similar to the way Gemfile and Gemfile.lock are related in the ruby world.

The advantage of doing something like this is that your requirements file specifies only versions of things that you specifically depend on and dependencies brought in by those dependencies can be easily identified and separated out.

For example, if I install the requirement jinja2==2.7 I don't actually care about also installing jinja2's dependency markupsafe but it will be installed. By running Pundler I end up with a nicely pinned requirements.txt that I can just regenerate from my "real" requirements.in whenever requirements I actually care about change.

For example, say I have in requirements.in::

a==1.0 b==2.0 c==3.0

And if we include the dependencies of those packages we have::

a==1.0 adep1==1.0 adep2==1.0 b==2.0 bdep1==2.0 c==3.0

Say we eventually upgrade a to version 2.0::

a==2.0 adep1==1.0 adep2==1.0 b==2.0 bdep1==2.0 c==3.0

With version 2.0 of package a the dependency adep1==1.0 is no longer needed. If we have one requirements file with all versions pinned it isn't clear that that dependency can now be removed.

If instead we simply updated the original requirements.in we could regenerate the full requirements (as a requirements.txt) and it would be clear that adep1==1.0 was no longer required.

Usage

Simply run pundler in a directory with your requirements.in or requirements.yml::

pundler install

If requirements.txt doesn't exist Pundler will process your requirements.yml or requirements.in file and create a requirements.txt that has all packages pinned to specific versions and identifies clearly what depends on what packages depend on what.

(TODO) If requirements.txt exists than pundler will pass args through to pip install, essentially::

pip install -r requirements.txt


Updating (TODO)

To update all your dependencies::

pundler update

This should update all unpinned dependencies to the latest version and appropriately update your generated requirements.txt.


Virtualenv

By default Pundler operates on the current environment (whatever pip is pointed at the moment.)

If you have a virtualenv enabled when you run pundler install it will be used.

Example

Given the following requirements.in::

pyramid==1.4.2 jinja2 txtemplate

Pundler will generate the this requirements.txt::

requirement 'pyramid==1.4.2' depends on:

WebOb==1.2.3 pyramid==1.4.2 translationstring==1.1 repoze.lru==0.6 Mako==0.8.1 MarkupSafe==0.18 PasteDeploy==1.5.0 Chameleon==2.11 venusian==1.0a8 zope.deprecation==4.0.2 zope.interface==4.0.5 setuptools==0.6c11

requirement 'jinja2' depends on:

jinja2==2.7 markupsafe==0.18

requirement 'txtemplate' depends on:

genshi==0.7 #jinja2==2.7 twisted==13.0.0 #markupsafe==0.18 txtemplate==1.0.2 #zope.interface==4.0.5 #setuptools==0.6c11

Advanced Configuration (TODO)

An alternative to requirements.in files is a simple requirements.yml configuration file.

The above example would look like::

sources:

Above, sources is optional.

A more interesting configuration with multiple groups like development and production would look like this::

sources: - https://pypi.python.org/simple/ groups: development: - nose production: - pyramid==1.4.2 - jinja2 - txtemplate

by default packages from all groups are installed

but you can customize this so that you can install

only specific things by defining targets and default

targets: # select a target with pundler install <target> development: # targets are a list of groups to install - production - development production: - production default: production # what happens if you just do pundler install

.. _pundler: http://github.com/steder/pundler .. _bundler: https://github.com/bundler/bundler .. _Michael Steder: http://mikesteder.com

Package Rankings
Top 21.3% on Pypi.org
Related Projects