Related Issues

N/A

Purpose

This PR restores missing chatbot support modules and improves Retrieval-Augmented Generation (RAG) integration for the eSim Copilot.

The changes ensure that chatbot functionality is available, knowledge-base retrieval is properly integrated into chatbot responses, and relevant documentation can be retrieved more reliably. Additionally, the Ubuntu setup workflow has been improved by automating RAG ingestion and simplifying installation.

Approach

Restored chatbot support modules:

• chatbot_core.py
• image_handler.py
• ollama_runner.py
• knowledge_base.py
• error_solutions.py
• stt_handler.py

RAG improvements:

• Added RAG retrieval integration to chatbot responses through chatbot_thread.py.
• Fixed RAG relevance filtering in knowledge_base.py to improve retrieval quality and prevent relevant documentation from being filtered out.

Setup improvements:

Enhanced setup_copilot_ubuntu.sh by:

• Automating RAG ingestion during installation.
• Adding fallback guidance for hdlparse installation failures.
• Updating PaddlePaddle installation.
• Improving startup instructions.
• Preserving existing Vosk setup support.

Testing

• Created a fresh virtual environment.
• Executed scripts/setup_copilot_ubuntu.sh from scratch.
• Verified dependency installation.
• Verified Ollama model downloads.
• Verified Vosk model installation.
• Verified ChromaDB creation and RAG ingestion.
• Verified knowledge retrieval functionality.