better ratelimit handle using tenacit

chatbot_simulator.py

@@ -1,10 +1,10 @@
 from openai import OpenAI
 import json_repair
 from transformers import AutoTokenizer
-from openai import RateLimitError
-import time
 from prompts import *
 import re
+from tenacity import retry, wait_fixed, stop_after_attempt, retry_if_exception_type
+from openai import RateLimitError
 
 
 class ChatbotSimulation:
@@ -50,8 +50,6 @@ class ChatbotSimulation:
 
     def _generate_system_prompt(self):
         """Create a dynamic system prompt based on the current state."""
-        #current_page = self.user_state['current_page']
-        #last_page = self.user_state['last_page']
         current_page = self.page_history[-1] if len(self.page_history) >= 1 else "Home"
         last_page = self.page_history[-2] if len(self.page_history) > 1 else "Home"
         page_info = self._get_page_details(current_page)
@@ -67,29 +65,20 @@ class ChatbotSimulation:
                                  page_info=page_info
                                  )
 
+    @retry(
+        retry=retry_if_exception_type(RateLimitError),
+        wait=wait_fixed(5),  # Waits for 5 seconds between retries
+        stop=stop_after_attempt(5)  # Stops after 5 failed attempts
+    )
     def _get_openai_response(self, prompt):
-        """Fetch response from OpenAI API."""
-        self._trim_conversation()
-        while True:
-            try:
-                response = self.client.chat.completions.create(
-                    model="gpt-4",
-                    messages=prompt,
-                    max_tokens=self.buffer_tokens,  # Adjusted max_tokens if needed
-                    temperature=0.7,
-                )
-                return response.choices[0].message.content
-            except RateLimitError as e:
-                # Parse the suggested retry time from the error message, default to 5s if not available
-                wait_time = 5
-                try:
-                    # Attempt to get the time from the error message
-                    wait_time = float(e.response['error']['message'].split("in ")[1].split("s")[0])
-                except (KeyError, IndexError, ValueError):
-                    print("Could not parse wait time from error message. Defaulting to 5 seconds.")
-
-                print(f"Rate limit reached. Retrying in {wait_time} seconds...")
-                time.sleep(wait_time)
+        """Fetch response from OpenAI API using tenacity for handling retries."""
+        response = self.client.chat.completions.create(
+            model="gpt-4",
+            messages=prompt,
+            max_tokens=self.buffer_tokens,  # Adjusted max_tokens if needed
+            temperature=0.7,
+        )
+        return response.choices[0].message.content
 
     def _calculate_token_count(self, conversation):
         """Accurately calculate the token count in the conversation using a tokenizer."""