Claude RAG MCP Pipeline

import streamlit as st import os import sys # Add src directory to path sys.path.append(os.path.join(os.path.dirname(__file__), 'src')) from src.rag_system import ConversationalRAGSystem # Page configuration st.set_page_config( page_title="Conversational Claude RAG", page_icon="💬", layout="wide" ) # Initialize RAG system @st.cache_resource def load_rag_system(): return ConversationalRAGSystem(embedding_provider="openai") def main(): st.title("💬 Conversational Claude RAG Pipeline") st.write("Chat with Claude about your documents with full conversation memory!") rag = load_rag_system() # Initialize session state for conversation if 'messages' not in st.session_state: st.session_state.messages = [] # Sidebar for system management st.sidebar.header("🔧 System Management") # Document ingestion (same as before) st.sidebar.subheader("📄 Document Management") docs_directory = st.sidebar.text_input("Documents Directory Path", "./documents") col1, col2 = st.sidebar.columns(2) with col1: if st.button("📥 Ingest Documents"): if os.path.exists(docs_directory): with st.spinner("Processing documents..."): success = rag.ingest_documents(docs_directory) if success: st.sidebar.success("Documents ingested successfully!") st.rerun() else: st.sidebar.error("No documents found or processing failed!") else: st.sidebar.error("Directory not found!") with col2: if st.button("🗑️ Clear Database"): rag.clear_database() st.sidebar.success("Database cleared!") st.rerun() # NEW: Conversation management st.sidebar.subheader("💬 Conversation") col3, col4 = st.sidebar.columns(2) with col3: if st.button("🔄 Clear Chat"): st.session_state.messages = [] rag.clear_conversation() st.rerun() with col4: conversation_summary = rag.get_conversation_summary() st.write(f"📊 {conversation_summary}") # System stats stats = rag.get_system_stats() st.sidebar.metric("Total Documents", stats['total_documents']) # Main chat interface st.header("💬 Chat with Your Documents") # Display conversation history for message in st.session_state.messages: with st.chat_message(message["role"]): st.write(message["content"]) if message["role"] == "assistant" and "sources" in message: if message["sources"]: with st.expander("📚 Sources Used"): for i, source in enumerate(message["sources"], 1): st.write(f"📄 {i}. {source['source']}") # Replace the query processing section with: if prompt := st.chat_input("Ask any question..."): # Add user message to chat history st.session_state.messages.append({"role": "user", "content": prompt}) with st.chat_message("user"): st.write(prompt) # Generate assistant response with st.chat_message("assistant"): with st.spinner("🔍 Searching and thinking..."): if stats['total_documents'] == 0: # No documents - use general knowledge response_text = rag.llm_service.generate_simple_response(prompt) sources = [] mode = "general_knowledge" else: # Hybrid mode result = rag.query_hybrid(prompt, n_results=5) response_text = result['answer'] sources = result['sources'] mode = result['mode'] st.write(response_text) # Show mode indicator if mode == "document_based": st.info("📚 Answer based on your documents") if sources: with st.expander("📄 Sources Used"): for i, source in enumerate(sources, 1): st.write(f"{i}. {source['source']}") else: st.info("🧠 Answer based on general knowledge (no relevant documents found)") # Add assistant response to chat history st.session_state.messages.append({ "role": "assistant", "content": response_text, "sources": sources, "mode": mode }) if __name__ == "__main__": main()