Docstring Formatting Options Would Be Useful!

[bkusuma]

There are many Docstring formats available so it would be great to have some options to select from when generating documentation. At the very least, the 4 most common formats would be lovely:

1. Sphinx Style: Often used in large projects where comprehensive documentation is required. It integrates well with Sphinx's documentation generator.
2. Google Style docstrings: Ideal for projects where simplicity and readability are prioritized, especially within organizations using Google's style guides.
3. NumPy/SciPy docstrings: Best suited for projects requiring detailed and extensive documentation, common in data science and scientific computing.
4. PyDoc: Useful for generating text and HTML documentation, but lacks the structured format of others.

[sumansaurabh]

Thank you for the suggestion @bkusuma. Yes, we are working on it. The architecture supports it, though prompt might not be perfectly aligned with the respective Docstrings formt. We'll try to add these styles Sphinx, Google, NumPy/SciPy, and PyDoc asap. Appreciate you pointing this out!

[bkusuma]

I came across this parser: https://github.com/rr-/docstring_parser

Perhaps it could be useful to integrate somehow? Or as a final pass to convert after the AI has finished documenting?

[sumansaurabh]

Hi @bkusuma ! Sorry for the delay in response.

Thanks for the link! Our backend already supported it, but it wasn't surfacing on the frontend — we’ve now added support for that. Here’s a screenshot showing it in action.

Additionally, we’ve created a new library - PyDocSmith, which is a fork of docstring_parser, that handles all the bad inputs LLMs can sometimes produce much more reliably.