virtualenv-auto-activate with support for zsh and virtualenvwrapper

virtualenv-auto-activate.sh

#!/bin/bash
# virtualenv-auto-activate.sh
# 
# Installation:
#   Add this line to your .bashrc or .bash-profile:
#
#       source /path/to/virtualenv-auto-activate.sh
#
#   Go to your project folder, run "virtualenv .venv", so your project folder
#   has a .venv folder at the top level, next to your version control directory.
#   For example:
#   .
#   ├── .git
#   │   ├── HEAD
#   │   ├── config
#   │   ├── description
#   │   ├── hooks
#   │   ├── info
#   │   ├── objects
#   │   └── refs
#   └── .venv
#       ├── bin
#       ├── include
#       └── lib
#
#   The virtualenv will be activated automatically when you enter the directory.

# Check for virtualenvwrapper
if type workon >/dev/null 2>&1; then
  VENV_WRAPPER=true
else
  VENV_WRAPPER=false
fi

function _virtualenv_auto_activate() {
    if [ -e ".venv" ]; then
        # Check for symlink pointing to virtualenv
        if [ -L ".venv" ]; then
          _VENV_PATH=$(readlink .venv)
          _VENV_WRAPPER_ACTIVATE=false
        # Check for directory containing virtualenv
        elif [ -d ".venv" ]; then
          _VENV_PATH=$(pwd -P)/.venv
          _VENV_WRAPPER_ACTIVATE=false
        # Check for file containing name of virtualenv
        elif [ -f ".venv" -a $VENV_WRAPPER = "true" ]; then
          _VENV_PATH=$WORKON_HOME/$(cat .venv)
          _VENV_WRAPPER_ACTIVATE=true
        else
          return
        fi

        # Check to see if already activated to avoid redundant activating
        if [ "$VIRTUAL_ENV" != $_VENV_PATH ]; then
            if $_VENV_WRAPPER_ACTIVATE; then
              _VENV_NAME=$(basename $_VENV_PATH)
              workon $_VENV_NAME
            else
              _VENV_NAME=$(basename `pwd`)
              VIRTUAL_ENV_DISABLE_PROMPT=1
              source .venv/bin/activate
              _OLD_VIRTUAL_PS1="$PS1"
              PS1="($_VENV_NAME)$PS1"
              export PS1
            fi
            echo Activated virtualenv \"$_VENV_NAME\".
        fi
    fi
}

export PROMPT_COMMAND=_virtualenv_auto_activate
if [ -n "$ZSH_VERSION" ]; then
  function chpwd() {
    _virtualenv_auto_activate
  }
fi

EDIT: changed to only use virtualenvwrapper for files containing name of venv, not for directories/symlinks

Any idea how I could make this work when I source it from my ~/.bash_aliases file? The script runs and exports PROMPT_COMMAND, but the _virtualenv_auto_activate function doesn't get defined for some reason.

@coreybrett I've forked a fork that's based off this that works on mac/linux using either bash or zsh, but more importantly, it fits the oh-my-zsh plugin architecture so you can add it as a plugin if you give oh-my-zsh and zsh a try.

• oh-my-zsh
• virtualenv-autodetect
• adding custom plugins

what would one need to change in this script to make the venv automatically activate after running virtualenv .venv? Right now, I need to cd out of and back into the project directory to activate it.

what would one need to change in this script to make the venv automatically activate after running virtualenv .venv? Right now, I need to cd out of and back into the project directory to activate it.

If you use my fork, https://github.com/RobertDeRose/virtualenv-autodetect.git you just have to do cd .