Create and use Python virtual environments.
From the Julia REPL, press ] to enter the package manager, then run
pkg> add https://github.com/cjdoris/VEnv.jl
The following script creates a temporary directory containing a virtual environment, installs cowsay into it, then calls it to print a message to the terminal.
using VEnv
venv = mktempdir()
VEnv.create(venv)
VEnv.run(`pip install cowsay`, venv);
VEnv.run(`python -m cowsay --character tux 'Hello, world!'`, venv); _____________
| Hello, world! |
=============
\
\
\
.--.
|o_o |
|:_/ |
// \ \
(| | )
/'\_ _/`\
\___)=(___/
Here is a minimal example of a package providing an interface to the Python cowsay package.
The _venv() function returns the path to a virtual environment with the cowsay package
installed.
The virtual environment is installed into a scratch space using Scratch.jl. This is a Julia-managed directory which will be automatically deleted when the package itself is deleted.
When VEnv.create is given the version argument, it will skip creating the environment if
it already exists and the version has not changed. This means that subsequent calls to
get_venv() are fast, even after restarting Julia.
The body of the do block is also called when the environment is created. Use this to
install packages. Increment the version whenever the create call changes to ensure the
environment is recreated - for example when changing the packages installed (or the versions
of packages).
The cowsay() function is an example of how to use the virtual environment. In this case,
we use VEnv.run to run the cowsay command in the environment.
using VEnv, Scratch
function get_venv()
venv = Scratch.@get_scratch!("venv")
VEnv.create(venv, version="1") do venv
VEnv.run(`pip install cowsay`, venv)
end
return venv
end
function cowsay(message; character="cow")
VEnv.run(`python -m cowsay --character $character $message`, get_venv())
endSee the docstrings for keyword arguments and detailed usage.
VEnv.create([func], venv): Create a venv.VEnv.run(cmd, venv): Run a command in the venv.VEnv.addenv(cmd, venv): Return the given command with environment variables set so it will run in the venv.VEnv.getenv(venv): Return the environment variables for the venv.
By default, Python is bundled with this package (specifically by Python_jll.jl) so you do not need to install it yourself.
If your platform is not supported by Python_jll (e.g. Windows) then you will need it to install it yourself.