A GraphRAG Without The DataShaper Dependency

[darien-schettler]

Hi there.

First off, thanks for this incredible library. It's easy to read and easy to use and I'm really digging into it and enjoying it.
I plan on many future experiments and investigations and can't wait to share with the community.

During my investigation I did my best to learn what I could about the Microsoft Datashaper library.

This was not a fun time.

The datashaper library is an incredibly difficult to understand library with nearly 0 working examples or documentation to support general Python usage in pandas. (PS: Feel free to correct me if I'm wrong on this and point me to it. The main example here doesn't work and I can't even find when it EVER worked (no trace of the pipeline module).

---

Can someone help me understand the core things that Datashaper helps with? I would love to attempt an implementation that completely removes the complexities introduced by datashaper. Alternatively, if the documentation and examples and codebase could be improved for datashaper that would also help.

---

My current understanding of where Datashaper is used:

• Execution of flows (Python functions doing pandas stuff run through the datashaper library)
• Callback management
• Progress Bar stuff
• Typing
• Table management (although I might consider this in the execution of the flows part)

Anything else I'm missing?

---

Can you help me understand as well what the benefit of using datashaper over something else would be? Is it faster? Better at Async? etc.

---

Thanks for humouring my questions and I look forward to the answers. If this gets any positive feedback I will 100% look into this seriously.

Thanks again for this amazing library.

Darien