Update app.py

app.py

@@ -143,22 +143,37 @@ def get_conversation_history(messages: List[ChatMessage]) -> List[Dict[str, str]
 @app.post("/v1/chat/completions")
 async def create_chat_completion(request: ChatCompletionRequest):
     try:
-        # Извлекаем system prompt
-        system_prompt = extract_system_prompt(request.messages)
+        # Получаем system prompt и последнее сообщение пользователя
+        system_prompt = ""
+        last_user_message = None
         
-        # Получаем историю диалога
-        conversation_history = get_conversation_history(request.messages)
+        for msg in request.messages:
+            if msg.role == "system":
+                system_prompt = msg.content
+            elif msg.role == "user":
+                last_user_message = msg
         
-        # Получаем последнее сообщение пользователя
-        last_user_message = next((msg for msg in reversed(request.messages) if msg.role == "user"), None)
         if not last_user_message:
             raise HTTPException(status_code=400, detail="No user message found")
         
+        # Объединяем system prompt с сообщением пользователя
+        combined_message = last_user_message.content
+        if system_prompt:
+            combined_message = f"{system_prompt}\n\n{combined_message}"
+        
+        # Формируем историю диалога для Flowise
+        history = []
+        for msg in request.messages:
+            if msg.role in ["user", "assistant"]:
+                history.append({
+                    "role": msg.role,
+                    "content": msg.content
+                })
+        
         # Формируем запрос к Flowise
         flowise_request = {
-            "question": last_user_message.content,
-            "system_prompt": system_prompt,
-            "conversation_history": conversation_history
+            "question": combined_message,
+            "history": history
         }
         
         # Засекаем время начала запроса
@@ -184,7 +199,7 @@ async def create_chat_completion(request: ChatCompletionRequest):
             )
         
         # Подсчитываем токены
-        prompt_tokens = count_tokens(last_user_message.content)
+        prompt_tokens = count_tokens(combined_message)
         completion_tokens = count_tokens(assistant_response)
         
         response = JSONResponse({