run_cmd

run_cmd.py

import shlex
import subprocess

def run_cmd(cmd, stdout=subprocess.PIPE, stderr=subprocess.STDOUT, block=True, *args, **kwargs):
    """
    Run a command using the subprocess module.

    :param cmd: command to be executed
    :type cmd: str or list
    :return: process object
    :rtype: subprocess.Process
    """
    if isinstance(cmd, str):
        cmdargs = shlex.split(cmd)
    elif isinstance(cmd, list):
        cmdargs = cmd
    else:
        raise ValueError('command must be either a string or a list')
    proc = subprocess.Popen(cmdargs, stdout=stdout, stderr=stderr, *args, **kwargs)
    if block:
        proc.wait()
        # FIXME: this will block if you use pipes and the process produces too much output!
        # TODO: for that case, we should use proc.communicate()!
        # The question now is: if we don't use pipes, is it still safe to use proc.communicate() with the assurance that
        # it will not read the stdout/stderr?
    else:
        # If neither wait nor poll are executed, returncode is not set
        # (i.e. it will be None until wait/poll is executed)
        proc.poll()
    return proc