pipx

The pipx tool allows to install and run Python command-line applications with pip in isolated virtual environments. As such it fulfils a rather precise goal and should not be confused with the general-purpose installations of pip and development tools like pipenv or poetry.

Motivation

Besides the Python packages that install some libraries to be used inside your Python code, there are others that install command-line tools. Noteworthy examples are black and flake8, but also projects like ansible (to automate the configuration of a computer) or cookiecutter (to generate a project’s file structure from a template) . These tools are often used outside of a given Python programming project and its dependency management system.

The naive approach to make them available globally across projects is to install them with pip. But if multiple packages are installed, you may run into issues of dependencies with mutually incompatible versions. Similarly to our previous discussion of virtual environments, a possible solution is to install each application into a dedicated venv. This is exactly what pipx does.

Usage

The following command installs a given package with pipx.

$ pipx install <package>

Behind the scenes pipx creates a folder structure inside ~/.local/pipx/ to contain the virtual environments for each installed application. The executables are linked into ~/.local/bin/. To make the tools available in the shell, add this directory manually to your $PATH or issue the pipx ensurepath helper command.

The already installed packages and associated command-line tools can be listed.

$ pipx list
venvs are in ~/.local/pipx/venvs
apps are exposed on your $PATH at ~/.local/bin
   package black 23.3.0, Python 3.11.2
    - black
    - black-primer
    - blackd

Other commands include

$ pipx uninstall <package>      # remove an installed package
$ pipx upgrade <package>        # upgrade an installed package
$ pipx upgrade-all              # upgrade all installed packages
$ pipx reinstall-all            # reinstall all packages (eg to fix issues after python upgrades)
$ pipx inject <package> <pkg>   # install additional pkg into existing venv of installed package