Warning

This project is no longer maintained. Please use uv instead.

pie

A command-line tool to manage Python virtual environments

Important

This tool is currently in alpha. There maybe breaking changes, although I'll try to avoid it as much as possible and if I can't, I'll provide clear instructions to migrate in the release notes. These changes will only be limited to a minor release.

The tool was previously named pyvenv and got renamed to pie in v0.9.0. This virtual environments created using pyvenv should not be migrated as is because the information stored by each virtual environment still refers to the name pyvenv.

Overview

pie is a command-line tool to create and manage multiple Python virtual environments. Some of the features are listed below:

• Single tool to manage all the environments
• Specify a Python version to create an environment
• Uses the venv module included in the standard library to create an environment
• Keep track of the project which uses the virtualenv
• Windows support

Installation

Using Homebrew/Linuxbrew on macOS/Linux:

brew tap dhruvmanila/pie https://github.com/dhruvmanila/pie
brew install pie

Using scoop on Windows:

scoop install https://raw.githubusercontent.com/dhruvmanila/pie/main/scoop/pie.json

Build from source

Recent Go toolchain is necessary to build from source. Use Go 1.16 or later.

# Install the latest stable version
go install github.com/dhruvmanila/pie@latest

# Install the head of main branch
go install github.com/dhruvmanila/pie

You can also download a binary from the releases page.

Usage

To create an environment using the default Python version:

pie create

To create an environment using a specific Python version:

pie create --python 3.11.0

List out all the managed environments:

pie list  # or pie ls

Use the --verbose flag to show information such as the Python version and project path.

Delete a virtualenv for the current project:

pie remove  # or pie rm

The remove command will only work from within the project directory for which the virtual envionment was created for. The confirmation prompt can be skipped with the --yes flag.

Activating a virtual environment

The tool itself cannot activate a virtual environment as execution of the binary takes place in a subshell and the virtual environment should be activated in the current shell.

A workaround is to define a shell function to activate the virtualenv. To support this with minimal shell scripting, a flag is provided which will print out the absolute path to the virtual environment directory based on the current working directory. In case there's no environment associated with the current working directory, it will exit with code 1 without printing anything.

pie --venv

The tool keeps track of the project for which a specific virtual environment was created for. This helps in finding out the venv directory based on the current directory even if it's inside the root project directory.

Based on your preferred shell, you can use either of the following:

PowerShell

# Activate a Python virtual environment built using the `pie` command.
# https://github.com/dhruvmanila/pie
function Pie-Activate() {
  if (Get-Command -Name pie -ErrorAction SilentlyContinue) {
    $VenvDir = (pie --venv)
    if ($VenvDir) {
      Invoke-Expression -Command "$VenvDir/Scripts/Activate.ps1"
    }
  }
}

# Alias it for quick activation
Set-Alias a Pie-Activate

bash/zsh

pie_activate() {
  # Activate a Python virtual environment built using the `pie` command.
  # https://github.com/dhruvmanila/pie
  #
  # The activation part cannot be a script as that is executed in a subshell
  # and so the `source` part will also be executed in the subshell instead of
  # the current shell.
  if command -v pie > /dev/null 2>&1; then
    VENV_DIR=$(pie --venv 2> /dev/null)
    if [ -n "$VENV_DIR" ]; then
      . "$VENV_DIR/bin/activate"
    fi
  fi
}

# Alias it for quick activation
alias a="pie_activate"

fish

Development

Building the project

The executable can be built using either the make command or using the go build command directly. Refer to make help for more info.

To build a local snapshot using goreleaser use the following command:

goreleaser build --single-target --rm-dist --snapshot

This will create the release binary for your system in a dist/ folder.

Release process

The release is automated using goreleaser and GitHub Actions using this workflow. It can be triggered by pushing the next version tag on GitHub. This will also update the Homebrew and scoop package specification.

This project follows Semantic Versioning.

License

This project is licensed under the MIT License.

See LICENSE for details.