Tutorial
How To Manage Python with Pyenv and Direnv
While this tutorial has content that we believe is of great benefit to our community, we have not yet tested or edited it to ensure you have an error-free learning experience. It's on our list, and we're working on it! You can help us out by using the "report an issue" button at the bottom of the tutorial.
Introduction
Whether you’re just getting started or you’re a seasoned Python developer, you may have found managing your Python environments to be tedious and painful. Managing Python versions, libraries, and various dependencies is like playing shuffleboard where the object is to not hit any other puck; if you do, the probability of a cascading effect of pucks flying everywhere you don’t want them to be will soon follow.
In this tutorial, you’ll install pyenv for managing python environments, install direnv to auto configure and source the virtualenv for projects, set a global Python version and a local Python version for projects, and configure a virtualenv and auto source the venv when moving into your project directories. By the end of this tutorial, you will be able to install any valid version of python, set up and configure virtual environments for each project, and bring sanity from chaos.
Prerequisites
We’re going to install both pyenv
and direnv
via homebrew
to ease the install process. Before getting started, install homebrew if you don’t have it already.
Working with pyenv
To start, you’ll see how to work with pyenv.
Installing pyenv
To get things started, we’re going to install pyenv
with homebrew and add the required pyenv
init to our ~/.bashrc
file.
$ brew install pyenv
$ echo 'eval "$(pyenv init -)"' >> ~/.bashrc # initialize pyenv on new shells
$ source ~/.bashrc # reinitialize bashrc to reflect changes in your current shell
Once it’s installed, take a moment to examine what our environment looks like. On a fresh system, you’ll see something similar to this:
$ which pyenv
/usr/local/bin/pyenv
$ pyenv versions
* system
$ which pip
/usr/local/bin/pip
$ which python
/usr/local/bin/python
This is a pretty standard snapshot. When you execute python version
, you’ll notice * system
, as the reference implies, this is your system’s python version. The asterisk denotes the current python binary sourced in your ${PATH}
. A good rule of thumb is to leave the system binary alone. If you want to see the list of python versions that pyenv
can install for you, use pyenv install --list
.
$ pyenv install 2.7.15
$ pyenv install 3.7.0
$ pyenv versions
* system (set by /Users/iamjohnnym/.pyenv/version)
2.7.15
3.7.0
Configuring the Base Requirements
Now, let’s upgrade pip
since chances are, it installed an older version. The following command will loop through your installed versions and update pip
to the latest.
for VERSION in $(pyenv versions --bare) ; do
pyenv shell ${VERSION} ;
pip install --upgrade pip ;
done
For the sake of our desired workflow, we want to install py2venv
for our Python 2.x
versions so we can mimic how python 3.x
installs virtualenvs. python -m venv .venv
.
for VERSION in $(pyenv versions --bare | egrep '^2.') ; do
pyenv shell ${VERSION} ;
pip install py2venv ;
done
Setting the Global Python Version
Even though we have pyenv
installed, it will still default to system
. To change this, set python 3.7.0
as our global version:
$ pyenv global 3.7.0
$ pyenv versions
system
2.7.15
* 3.7.0 (set by /Users/iamjohnnym/.pyenv/version)
$ which python
/Users/iamjohnnym/.pyenv/shims/python
We’ve got pyenv
setup and functional. Let’s move on to direnv
.
Working with direnv
direnv
is a handy utility that allows you to create a file that’s placed in any directory that you want that functions like .bashrc
. Whenever you enter the directory with this file, your shell will automatically execute it. The capabilities are endless but for the purpose of this post, we’re going to use it to configure a python virtualenv based on the file .python-version
and then activate it for us. The purpose is to create a seamless development flow. No need to worry about manually configuring or activating virtualenvs; let your computer do the work for you.
Installing direnv
Installation is straightforward with homebrew:
$ brew install direnv
direnv
is now installed, and ready for exploitation. Let’s try a sample project.
Creating a New Project
Let’s start up a project. We’re going to create a new directory, set a local python version, set up our .envrc
file, and activate it.
$ cd -p ~/python-projects/pyenv-tutorial
$ cd $_ # if you're unaware, $_ will execute the last argument of your command
$ pwd
/Users/iamjohnnym/ python-projects/pyenv-tutorial
$ pyenv local 3.7.0
$ cat .python-version
3.7.0
Configuring the Environment File
At this point, we’re ready to create our .envrc
file. With your favorite editor, create that file and add the following contents:
# check if python version is set in current dir
if [ -f ".python-version" ] ; then
if [ ! -d ".venv" ] ; then
echo "Installing virtualenv for $(python -V)"
# if we didn't install `py2venv` for python 2.x, we would need to use
# `virtualenv`, which you would have to install separately.
python -m venv .venv
fi
echo "Activating $(python -V) virtualenv"
source .venv/bin/activate
fi
# announce python version and show the path of the current python in ${PATH}
echo "Virtualenv has been activated for $(python -V)"
echo "$(which python)"
Save the file. If you did this via a shell editor, such as vim
. You’ll see the following message, direnv: error .envrc is blocked. Run direnv allow to approve its content
. Don’t be alarmed as this is a security feature to prevent auto-execution of the file. Whenever this file is changed, it requires manual approval before it will auto-execute again. To activate it, simply type direnv allow
from the project dir.
$ direnv allow
direnv: loading .envrc
Installing virtualenv for Python 3.7.0
Activating Python 3.7.0 virtualenv
Virtualenv has been activated for Python 3.7.0
/Users/iamjohnnym/.personal/tutorials/pyenv-direnv/.venv/bin/python
direnv: export +VIRTUAL_ENV ~PATH
Conclusion
You now have the tools you need to manage different python versions and project dependencies!