Modernize 2-retrieval-qa-finish for LangChain v1 compatibility

[Flcookie]

Summary

This PR updates the 2-retrieval-qa-finish branch to be compatible with the latest LangChain v1 ecosystem.

Changes

• ingestion.py

  • Updated imports to langchain_text_splitters, langchain_openai, and langchain_pinecone.
  • Added safe metadata handling and clearer argument naming.

• backend/core.py

  • Updated imports to use langchain_classic for create_retrieval_chain and create_stuff_documents_chain.
  • Preserved all educational structure and function names for the course.

Motivation

Keep the course code runnable under LangChain v1 while maintaining its original teaching structure and clarity.

Notes

If no local folder langchain-docs/... exists, ingestion.py will load 0 documents — this is expected.

---

🙏 Acknowledgment
A huge thanks to the course author and contributors for creating such a clear and practical learning project.
It has been incredibly helpful for understanding how to build retrieval-augmented generation pipelines with LangChain.

[Flcookie]

Thanks for this amazing course and project! It really helped me understand RAG and LangChain in depth 🙏