enhanced the RAG

f1_ai.py

@@ -178,17 +178,24 @@ class F1AI:
             # Format context from documents
             context = "\n\n".join([f"Document {i+1}: {doc.page_content}" for i, doc in enumerate(docs)])
             
-            # Create prompt for the LLM
+            # Create enhanced prompt for the LLM with better instructions
             prompt = f"""
-            Answer the question based on the provided context. Include relevant citations using [1], [2], etc.
-            If you're unsure or if the context doesn't contain the information, acknowledge the uncertainty.
-
+            You are an expert Formula 1 knowledge assistant. Using the provided context, answer the question comprehensively and naturally.
+            
+            Guidelines:
+            1. Provide detailed, well-structured responses that flow naturally
+            2. Use source citations [1], [2], etc. to support key facts
+            3. If information is uncertain or missing from context, acknowledge it
+            4. Organize complex answers with clear paragraphs
+            5. Add relevant examples or explanations when helpful
+            6. dont fill the ouput with citations only
+            
             Context:
             {context}
             
             Question: {question}
-
-            Answer with citations:
+            
+            Provide a comprehensive answer with appropriate citations:
             """
             
             # Get response from LLM
@@ -203,17 +210,27 @@ class F1AI:
             else:  # LangChain LLM
                 response_text = self.llm.invoke(prompt)
             
-            # Format sources
-            sources = [{
-                "url": doc.metadata["source"],
-                "chunk_index": doc.metadata.get("chunk_index", 0),
-                "timestamp": doc.metadata.get("timestamp", "")
-            } for doc in docs]
+            # Process and format sources with better attribution
+            sources = []
+            for i, doc in enumerate(docs, 1):
+                source = {
+                    "index": i,
+                    "url": doc.metadata["source"],
+                    "chunk_index": doc.metadata.get("chunk_index", 0),
+                    "timestamp": doc.metadata.get("timestamp", ""),
+                    "excerpt": doc.page_content[:200] + "..." if len(doc.page_content) > 200 else doc.page_content
+                }
+                sources.append(source)
             
-            # Format response
+            # Format response with enhanced structure
             formatted_response = {
                 "answer": response_text,
-                "sources": sources
+                "sources": sources,
+                "metadata": {
+                    "total_sources": len(sources),
+                    "query_timestamp": datetime.now().isoformat(),
+                    "response_format_version": "2.0"
+                }
             }
             
             return formatted_response
@@ -258,8 +275,16 @@ async def main():
         console.print(Markdown(response['answer']))
         
         console.print("\n[bold yellow]Sources:[/bold yellow]")
-        for i, source in enumerate(response['sources']):
-            console.print(f"[{i+1}] {source['url']}")
+        for source in response['sources']:
+            console.print(f"[{source['index']}] {source['url']}")
+            console.print(f"[dim]Excerpt:[/dim] {source['excerpt']}\n")
+        
+        # Print metadata
+        console.print("\n[bold blue]Query Info:[/bold blue]")
+        console.print(f"Total sources: {response['metadata']['total_sources']}")
+        console.print(f"Query time: {response['metadata']['query_timestamp']}")
+        console.print(f"Response version: {response['metadata']['response_format_version']}")
+
     else:
         parser.print_help()