Add MongoDB local vectorstore integration LABELS: Feature, Help Wanted

[NikhilMathursProjects]

Hi Langchain team 👋,
Ive added a new local MongoDB based vectorstore implementation that supports FAISS indexing (Flat and HNSW).This is my first contribution to Langchain community and Id appreciate any guidance on improving the test cases.

Here i use MongoDB for persistence and FAISS for similarity search.

Features:

• Local FAISS index (supports Flat and HNSW)
• MongoDB-backed metadata & embeddings
• Add, delete, update, and similarity search
• Save/load FAISS and ID mappings
• Support for reusing deleted FAISS IDs
• Designed to work offline or on-prem

This includes:

• A new class in langchain_community/vectorstores/mongodb_local.py
• Integration tests in tests/integration_tests/vectorstores/test_mongodb_local.py (This has not yet been updated with the right tests, its basically a copy of MongoAtlasVectorSearch, i will update the tests once i finish them :) )
  This is version 1 and would love getting all feedback to make it the best it can possibly be.

I would love your feedback and help with:

• Writing tests (especially for FAISS index handling)
• Making sure the structure aligns with other vectorstore implementations like MongoAtlas

Labels Requested: feature, help wanted

Author: Nikhil Mathur (@NikhilMathursProjects)

[NikhilMathursProjects]

Hi maintainers,
This is my first contribution to LangChain
I’ve implemented MongoDBLocalVectorStore (similar to MongoDB Atlas VS)

The CI still shows as pending — could someone please approve the workflow run?
Also, I’d love feedback on the code and guidance on writing better tests if possible, thank you!