Metadata-Version: 2.1
Name: popl
Version: 0.1.1
Summary: A pip wrapper behaving like npm
Home-page: https://github.com/emdiet/popl
Author: emdiet
Classifier: Programming Language :: Python :: 3.7
Classifier: License :: OSI Approved :: MIT License
Classifier: Operating System :: OS Independent
Requires-Python: >=3.7
Description-Content-Type: text/markdown
License-File: LICENSE

# popl
popl: pip local projects. a pip wrapper, but behaving more like npm

## What is this?

This is a pip wrapper that keeps all dependencies in your project direcory, and manages virtual enviroments behind the scenes so you don't have to deal with any of that noise.

It's like npm(lite), but for python.

## Usage

1) have python/pip installed
2) copy popl.py into your project directory
3) inside your project directory run `python ./popl.py init`
4) install packages like you would with pip: `python ./popl.py install <packages, args, etc.>`
5) run your python scripts with `python ./popl.py run <script-name.py>`

e.g.: 

```powershell
python ./popl.py init
python ./popl.py install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu124
python ./popl.py run my_inference_script.py
```

 just works. (in windows). no need to activate virtual environments, no need to worry about global packages, no need to worry about system packages.

## Motivation

I wanted to install pytorch cuda (https://pytorch.org/get-started/locally/), but 
1) installing it with pip is a bad idea, according to everyone
2) I'm afraid of forgetting to activate/deactivate virtual environments
3) the poetry maintainers are jerks (https://github.com/python-poetry/poetry/issues/7685)
4) activating virtual environments on windows doesn't actually work unless you set the execution policy to unrestricted (https://stackoverflow.com/questions/18713086/virtualenv-wont-activate-on-windows). How about... no?

So that's how bob's your uncle.

## How it works

1) `init` creates a virtual environment called `.venv` (your "node_modules") and a `popl.json` (your "package.json") file in the current directory
2) `install` installs the package into the `.venv` virtual environment, and adds it to the `popl.json` file, and uses the standard `requirements.txt` as your lock file, meaning your pip friends or your docker build or whatever can just use that if they don't want to use popl.
3) `run` ensures that the script is run in the `.venv` virtual environment. 

## Notes on: requirements.txt

`install` will honor the requirements.txt file if it exists, even if the dependencies are not in popl.py. popl will still properly install everything in the appropriate virtual environment, so you can use popl with your existing projects.

## Notes on: popl.json

`popl.json` is where your core dependencies could live, if you wanted to reinstall the project. It's like a lightweight version of the `package.json` file. You can ignore it if you want, and use the requirements.txt to port your dependencies. It might be useful however if you want to clean your dependencies at some point.

## Notes on: Using this collabortively with git:

add `.venv` to your `.gitignore` file

your buddies can just work with the requirements.txt file if they don't want to use popl.

## Notes on: uninstalling everything

just delete the project directory. packages might still temporarily reside in the *python download cache* (https://pip.pypa.io/en/stable/topics/caching/) (could be cleaned with `pip cache purge` (note: pip, not popl!)), but it should expire on its own.

## Dependencies

- python 3.7+

## Feature Requests

please open an issue (https://github.com/emdiet/popl/issues), thx :)
