Update App_Function_Libraries/RAG/RAG_Libary_2.py

App_Function_Libraries/RAG/RAG_Libary_2.py

@@ -35,6 +35,63 @@ config.read('config.txt')
 
 
 
+# RAG Search with keyword filtering
+def enhanced_rag_pipeline(query: str, api_choice: str, keywords: str = None) -> Dict[str, Any]:
+    try:
+        # Load embedding provider from config, or fallback to 'openai'
+        embedding_provider = config.get('Embeddings', 'provider', fallback='openai')
+
+        # Log the provider used
+        logging.debug(f"Using embedding provider: {embedding_provider}")
+
+        # Process keywords if provided
+        keyword_list = [k.strip().lower() for k in keywords.split(',')] if keywords else []
+        logging.debug(f"enhanced_rag_pipeline - Keywords: {keyword_list}")
+
+        # Fetch relevant media IDs based on keywords if keywords are provided
+        relevant_media_ids = fetch_relevant_media_ids(keyword_list) if keyword_list else None
+        logging.debug(f"enhanced_rag_pipeline - relevant media IDs: {relevant_media_ids}")
+
+        # Perform vector search
+        vector_results = perform_vector_search(query, relevant_media_ids)
+        logging.debug(f"enhanced_rag_pipeline - Vector search results: {vector_results}")
+
+        # Perform full-text search
+        fts_results = perform_full_text_search(query, relevant_media_ids)
+        logging.debug(f"enhanced_rag_pipeline - Full-text search results: {fts_results}")
+
+        # Combine results
+        all_results = vector_results + fts_results
+        # FIXME
+        if not all_results:
+            logging.info(f"No results found. Query: {query}, Keywords: {keywords}")
+            return {
+                "answer": "I couldn't find any relevant information based on your query and keywords.",
+                "context": ""
+            }
+
+        # FIXME - Apply Re-Ranking of results here
+        apply_re_ranking = False
+        if apply_re_ranking:
+            # Implement re-ranking logic here
+            pass
+        # Extract content from results
+        context = "\n".join([result['content'] for result in all_results[:10]])  # Limit to top 10 results
+        logging.debug(f"Context length: {len(context)}")
+        logging.debug(f"Context: {context[:200]}")
+        # Generate answer using the selected API
+        answer = generate_answer(api_choice, context, query)
+
+        return {
+            "answer": answer,
+            "context": context
+        }
+    except Exception as e:
+        logging.error(f"Error in enhanced_rag_pipeline: {str(e)}")
+        return {
+            "answer": "An error occurred while processing your request.",
+            "context": ""
+        }