pepython 0.2.1

Simple task automation for projects.

NOTE: It happens that there is a much more mature project with active maintenance that is surprisingly similar to this one: pyinvoke/invoke. I'd be using it so not maintaining this one.

PePython

Need help organizing your project's tasks? Just call PePython!

Install

pip install pepython

Quick start

Define your tasks in a tasks.py file your root directory. Note how dependencies are just method calls in the body of the tasks themselves. This allow params passing/forwarding and a more flexible schema than just pre/post dependencies.

from pepython.task_def import task, s

@task
def clean(*args):
    s("rm -rf build dist *.egg-info")

@task
def build(*args):
    clean()
    s("pipenv run python setup.py sdist bdist_wheel")

@task
def publish(*args):
    s("pipenv run twine upload dist/*", interactive=True)

@task
def build_and_publish(*args):
    build()
    publish()

Then just call for PePython!

$ pepython --help
Usage:
        pepython [--defs-path path-to-your-tasks-definitions.py] task [task params ...]

$ pepython clean
Executed task 'clean' successfuly.

$ pepython build_and_publish
Enter your username: your-name-here
Enter your password: ************
Uploading distributions to https://test.pypi.org/legacy/
Uploading pepython-0.1.0-py2-none-any.whl
100%|█████████████████████████████████████████████████████| 7.77k/7.77k [00:00<00:00, 42.6kB/s]
Uploading pepython-0.1.0.tar.gz
100%|█████████████████████████████████████████████████████| 5.65k/5.65k [00:01<00:00, 4.74kB/s]

Executed task 'build_and_publish' successfuly.

If you need task arguments, they are defined naturally:

@task
def task_args(first_arg, second_arg, *rest_of_args):
    print(
        "You've passed {} and {} as the 2 first parms, then passed:\n{}."
        .format(first_arg, second_arg, rest_of_args)
    )

$ pepython task_args a b c d e
You've passed a and b as the 2 first parms, then passed:
('c', 'd', 'e').

Executed task 'task_args' successfuly.

The main idea is to use Python for your task automation, with all of it's flexibility but without loosing the shell. This is why there is s() (short for shell).

Apart from the interactive param shown above (see the password prompt in the output), it has some other possibilities:

Interactive processes are easily left piped to the user:

@task
def change_pass():
    s("passwd", interactive=True)

$ pepython change_pass
Changing password for nando.
(current) UNIX password:
Enter new UNIX password:
Retype new UNIX password:
passwd: password updated successfully

Executed task 'change_pass' successfuly.

While other non interactive processes can be left to be managed by the task:

@task
def process_python_files():
    ls_result = s("ls *py", fail_fast=False, verbose=True)

    if not ls_result.ok:
        exit("No python files here")

    python_files_raw = ls_result.value['out'].split("\n")
    python_files = [f.strip() for f in python_files_raw if f.strip()]

    print(
        "These are the python files in this directory:\n{}"
        .format(python_files)
    )

$ pepython process_python_files
Executed shell command 'ls *py'
exit code was: 0
stdout was:
setup.py
tasks.py

stderr was:

These are the python files in this directory:
[u'setup.py', u'tasks.py']

Executed task 'shell_returned_values' successfuly.

Development

Make sure you have the lastest pip and pipenv versions:

pip install --update pip pipenv

To start developing, start the environment by:

pipenv shell
pipenv install -d

This tool uses both pipenv for development and setuptools for packaging and distribution. To this date there is not a 100% community-accepted best practice so I've taken this approach. In summary:

To add an application dependency, add it in setup.py and leave it with a loose version definition. Then, just do pipenv install -e . to install the dependency. Pipenv locking mecanism will work as expected, since pepython itself in in the [packages] section of Pipfile (check Pipfile.lock and you'll find the deps there).

To add a development dependency, add it to Pipfile via pipenv install -d <my-dependency>.

This way there's a single source of truth for package definition. No need to repeat the deps in setup.py and Pipfile*.

License

This project is licensed under the MIT License - see the LICENSE file for details.