openapi-python-client doesn't find autoflake even when on $PATH

[Julian]

Describe the bug
Even when autoflake is on the PATH, openapi-python-client seems to claim it isn't.

To Reproduce
Steps to reproduce the behavior:

⊙  python3.10 -m venv venv && venv/bin/python -m pip install --quiet openapi-python-client

⊙  ls venv/bin/
Activate.ps1  activate.csh   autoflake  blackd  isort                   openapi-python-client  pip3     pyflakes  python3
activate      activate.fish  black      httpx   isort-identify-imports  pip                    pip3.10  python    python3.10

⊙  PATH=venv/bin:$PATH

⊙  type -a openapi-python-client
openapi-python-client is venv/bin/openapi-python-client

⊙  type -a autoflake
autoflake is venv/bin/autoflake

⊙  autoflake --help
<help output>

⊙  openapi-python-client generate --path openapi.yaml
Generating foo-client
Error(s) encountered while generating, client was not created

autoflake failed

/bin/sh: autoflake: command not found



If you believe this was a mistake or this tool is missing a feature you need, please open an issue at https://github.com/openapi-generators/openapi-python-client/issues/new/choose

Expected behavior
autoflake is found

OpenAPI Spec File

openapi: 3.0.1
info:
  title: Foo
  description: bar`
  contact:
    email: [email protected]
  version: 0.1.0
servers:
- url: example.com
tags:
- name: foo
  description: |
      bar
paths:
  "/baz":
    responses:
      get:
        tags:
        - foo
        summary: List foo
        operationId: listFoo
        parameters:
        responses:
          "200":
            description: Foo
            content:
              application/json:
                schema:
                  type: array
                  items:
                    type: number

Desktop (please complete the following information):

⊙  uname -a                                                                                                                                                                         julian@Airm
Darwin Airm.local 21.6.0 Darwin Kernel Version 21.6.0: Wed Aug 10 14:28:35 PDT 2022; root:xnu-8020.141.5~2/RELEASE_ARM64_T8101 arm64 arm Darwin

~/Desktop/test 
⊙  venv/bin/python --version                                                                                                                                                        julian@Airm
Python 3.10.7

~/Desktop/test 
⊙  venv/bin/openapi-python-client --version                                                                                                                                         julian@Airm
openapi-python-client version: 0.11.6

Additional context
I haven't investigated carefully yet, though I'm possibly happy to do so after getting my client working, but I suspect the issue is here:

openapi-python-client/openapi_python_client/__init__.py

Line 157 in 9cea5af

cmd, cwd=self.project_dir, shell=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE, check=True

where commands are executed in a subshell.

As I say I definitely don't fully understand the internals of the project, but generally I think using shell=True is worth trying to avoid as often as possible. I suspect here from what I see, someone can specify arbitrary commands to run post-generation and is doing so as a str in the config? If it's not appealing to change that instead to be a list (and then you don't need shell=True), you could also use shlex.split, which is equally a bit fiddly but would mean you don't spawn a subshell. Or alternatively you could assume the things are installed to the same Python interpreter as openapi-python-generator, and invoke python -m black or python -m autoflake by using sys.executable on the same Python openapi-python-generator is run with, without needing to search $PATH.