How do i run the python 'sdist' command from within a python automated script without using subprocess?

Question

I am writing a script to automate the packaging of a 'home-made' python module and distributing it on a remote machine.

i am using Pip and have created a setup.py file but i then have to call the subprocess module to call the "python setup.py sdist" command.

i have looked at the "run_setup" method in distutils.core but i am trying to avoid using the subprocess module alltogether. (i see no point in opening a shell to run a python command if i am already in python...)

is there a way to import the distutils module into my script and pass the setup information directly to one of its methods and avoid using the shell command entirely? or any other suggestions that may help me

thanks

Answer 1

Just for the sake of completeness, I wanted to answer this since I came across it trying to find out how to do this myself. In my case, I wanted to be sure that the same python version was being used to execute the command, which is why using subprocess was not a good option. (Edit: as pointed out in comment, I could use sys.executable with subprocess, though programmatic execution is IMO still a cleaner approah -- and obviously pretty straightforward.)

(Using distutils.core.run_setup does not call subprocess, but uses exec in a controlled scope/environment.)

from distutils.core import run_setup

run_setup('setup.py', script_args=['sdist'])

Another option, may be to use the setuptools commands, though I have not explored this to completion. Obviously, you still have to figure out how to avoid duplicating your proj metadata.

from setuptools.dist import Distribution
from setuptools.command.sdist import sdist

dist = Distribution({'name': 'my-project', 'version': '1.0.0'}) # etc.
dist.script_name = 'setup.py'
cmd = sdist(dist)
cmd.ensure_finalized()
cmd.run()  # TODO: error handling

Anyway, hopefully that will help someone in the right direction. There are plenty of valid reasons to want to perform packaging operations programmatically, after all.