subprocess.Popen: mkvirtualenv not found

Question

I'm using virtualenvwrapper in my deployment. To setup new environments, I'm running a python script, which contains all needed steps.

The setupscript includes:

cmd = 'mkvirtualenv %s --no-site-packages'%('testname')
head = subprocess.Popen(cmd,shell=True, stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
for line in head.stdout.read().splitlines():
    print line

The output is:

/bin/sh: mkvirtualenv: not found

How can I correctly use virtualenvwrapper within my python script?

EDIT:

The following code works for me:

cmd = 'source /usr/local/bin/virtualenvwrapper.sh && mkvirtualenv %s --no-site-packages'%('testname')
head = subprocess.Popen(cmd, executable='bash', shell=True, stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
for line in head.stdout.read().splitlines():
    print line

Thanks for all answers.

Answer 1

Edit:

I learned that mkvirtualenv is a shell function. In this case the question becomes how to run shell function in python. My answer below can be applied to standalone binaries. For your question, please look at the answer: https://stackoverflow.com/a/5826523/1906700 You can indirectly call mkvirtualenv() function from the script that defines it.

Set PATH variable correctly

The problem seems to be caused by your $PATH variable. You need to correctly set this variable so that mkvirtualenv executable can be found. For example, if you have mkvirtualenv executable in `/home/thore/scripts, you need to set your $PATH in .bashrc or .zshrc (depending on your shell) as follows:

 export PATH=$PATH:/home/thore/scripts

In that way, mkvirtualenv script will be found correctly and run.

Another Solution

The other solution to the problem would be using the exact path for the script in question. In that case, you can give /home/thore/scripts/mkvirtualenv as a parameter to subprocess.

Answer 2

mkvirtualenv might be a shell function that is added to your environment by sourcing virtualenvwrapper.sh script from your shell's startup file. The default command invoked on shell=True (e.g., /bin/sh -c ...) might not read it.

You could source the file explicitly:

import pipes
from subprocess import check_call

check_call("""source /path/to/virtualenvwrapper.sh &&
    mkvirtualenv --no-site-packages """ + pipes.quote(envname),
    executable='bash', shell=True)