I'm interested in using a

docker_image

as a dependency in a python script. I've written a script for deploying my application and I would like to use pants to build and run the deployment script in one step. At the moment I'm looking at

runtime_package_dependencies

in

python_test

but there doesn't seem to be an equivalent for

python_source

unless I'm missing something. Additionally, it's not clear to me how to obtain the digest of the image built by the corresponding

pants publish

command. I'll potentially be running multiple of these in parallel and don't want to rely on fixed tags.

Additionally, it's not clear to me how to obtain the digest of the image built by the corresponding

pants publish

command

Metadata about the image is output as a JSON file, if you run

pants package path/to:image

you can see it in your

dist/

folder I think.

At the moment I'm looking at

runtime_package_dependencies

in

python_test

but there doesn't seem to be an equivalent for

python_source

unless I'm missing something

Unfortunately I don't think you're missing something. I believe one thing people do is build an orchestration binary that itself runs Pants, e.g.

pants package path/to:run-my-deploy && dist/path.to/run-my-deploy.pex

or similar schemes along those lines. Might that work for you?

This is my temporary solution, in my BUILD I've added:

docker_image(
    ...
    image_tags=["{build_args.INVOCATION_UUID}"]
)

And in my python run script:

# Build the docker image for this run.
  with tempfile.NamedTemporaryFile() as publish:
    await execute_and_stream(
        f"pants publish {_docker_target.value} --publish-output={publish.name}",
        env=os.environ | {
            "PANTS_CONCURRENT": "True",
            "INVOCATION_UUID": uuid.uuid4().hex,
        },
        cwd=os.environ["PWD"])
    build_result = json.load(publish)