ERRORS with memgpt load dirctory --input-files ./MemGPT-WhitePaper.pdf --name MemGPT-Whitepaper

[agiletechnologist]

Describe the bug
Perform a load of MemGPT-whitepaper and name it MemGPT-Whitepaper:
memgpt load directory --input-files ./MemGPT-WhitePaper.pdf --name MemGPT-Whitepaper

1. table name created during load: memgpt_memgpt_whitepaper it appears all "memgpt"s are stripped, including the name. as seen in agent1/config.json
   "data_sources": [ "whitepaper" ],

To Reproduce
Steps to reproduce the behavior:

1. On the CLI, execute memgpt load directory --input-files ./MemGPT-WhitePaper.pdf --name MemGPT-Whitepaper
2. Look in the Postgresql database for table name: memgpt_memgpt_whitepaper
3. On CLI, execute memgpt run --agent agent1
4. /attach it lists only whitepaper select it.
5. Look in agent1/config.json
6. See error: "data_sources": [ "whitepaper" ],

Expected behavior
It should use memgpt_whitepaper as this is a legitimate name, and the memgpt should not be stripped of the name.
I should only strip the first memgpt. the second one is an actual name starting with memgpt_whitepaper

---

How did you install MemGPT?

• pip install -U pymemgpt

Your setup (please complete the following information)

• Your OS: Linux
• Where you're trying to run MemGPT from
  • Terminal
• Your python version: Python 3.11.2
• pymemgpt version: pymemgpt 0.2.4`

[agiletechnologist]

Oh, and I am using Postgres with vector extensions for archival storage.

[sarahwooders]

Thanks for reporting this issue! This was indeed a bug with the database table listing, and should be fixed in #535.