added global rate limiting

agent.py

@@ -137,7 +137,7 @@ NEVER fake or simulate tool output yourself. If you are unable to make progreess
 Thought: I now know the final answer
 Final Answer: YOUR FINAL ANSWER should be a number OR as few words as possible OR a comma separated list of numbers and/or strings. If you are asked for a number, don't use comma to write your number neither use units such as $ or percent sign unless specified otherwise. If you are asked for a string, don't use articles, neither abbreviations (e.g. for cities), and write the digits in plain text unless specified otherwise. If you are asked for a comma separated list, apply the above rules depending of whether the element to be put in the list is a number or a string. For one word answers, start with the word with a capital letter.
 Make sure to follow any formatting instructions given by the user.
-Now begin! Reminder to ALWAYS use the exact characters `Final Answer:` when you provide a definitive answer."""
+Now begin! Reminder to ALWAYS use the exact characters `Final Answer:` when you provide a definitive answer. DO NOT USE TAVILY SEARCH FOR CLARIFICATION"""
 
 # Generate the chat interface, including the tools
 llm = ChatOpenAI(
@@ -1391,7 +1391,7 @@ class TurboNerd:
 
 # Example usage:
 if __name__ == "__main__":
-    agent = TurboNerd(max_iterations=35)
+    agent = TurboNerd(max_iterations=25)
     response = agent("""The attached Excel file contains the sales of menu items for a local fast-food chain. What were the total sales that the chain made from food (not including drinks)? Express your answer in USD with two decimal places. TEMPP\excel.xlsx """)
     print("\nFinal Response:")
     print(response)

app.py

@@ -157,6 +157,17 @@ def chat_with_agent(question: str, file_uploads, history: list) -> tuple:
         session_histories[session_id].append({"role": "assistant", "content": formatted_response})
         
         return history, "", f"Remaining queries this hour: {remaining_queries}/{query_limiter.max_queries}"
+    except RecursionError as e:
+        error_message = (
+            "I apologize, but I've reached my thinking limit while trying to answer your question. "
+            "This usually happens when the question requires too many steps to solve and therefore too much money"
+            "Could you please try breaking down your question into smaller, more specific parts? "
+            "For example, instead of asking about multiple things at once, try asking about one aspect at a time."
+        )
+        history.append({"role": "assistant", "content": error_message})
+        if session_id in session_histories:
+            session_histories[session_id].append({"role": "assistant", "content": error_message})
+        return history, "", "Remaining queries this hour: 5/5"
     except Exception as e:
         error_message = f"Error: {str(e)}"
         history.append({"role": "assistant", "content": error_message})

rate_limiter.py

@@ -5,22 +5,19 @@ import threading
 class QueryRateLimiter:
     def __init__(self, max_queries_per_hour: int = 10):
         """
-        Initialize rate limiter for queries per hour.
+        Initialize rate limiter for global queries per hour.
         
         Args:
-            max_queries_per_hour: Maximum number of queries allowed per hour
+            max_queries_per_hour: Maximum number of queries allowed per hour globally
         """
         self.max_queries = max_queries_per_hour
-        self.queries = defaultdict(list)  # ip_address -> list of timestamps
+        self.queries = []  # List of timestamps for all queries
         self.lock = threading.Lock()
     
-    def is_allowed(self, ip_address: str) -> bool:
+    def is_allowed(self, _: str = None) -> bool:
         """
-        Check if an IP address is allowed to make another query.
+        Check if another query is allowed within the hourly limit.
         
-        Args:
-            ip_address: IP address of the client
-            
         Returns:
             bool: True if query is allowed, False if rate limited
         """
@@ -29,22 +26,19 @@ class QueryRateLimiter:
         
         with self.lock:
             # Remove queries older than 1 hour
-            self.queries[ip_address] = [t for t in self.queries[ip_address] if t > hour_ago]
+            self.queries = [t for t in self.queries if t > hour_ago]
             
             # Check if under rate limit
-            if len(self.queries[ip_address]) < self.max_queries:
-                self.queries[ip_address].append(current_time)
+            if len(self.queries) < self.max_queries:
+                self.queries.append(current_time)
                 return True
             
             return False
     
-    def get_remaining_queries(self, ip_address: str) -> int:
+    def get_remaining_queries(self, _: str = None) -> int:
         """
-        Get number of remaining queries for an IP address in the current hour.
+        Get number of remaining queries in the current hour.
         
-        Args:
-            ip_address: IP address of the client
-            
         Returns:
             int: Number of remaining queries
         """
@@ -53,27 +47,24 @@ class QueryRateLimiter:
         
         with self.lock:
             # Remove queries older than 1 hour
-            self.queries[ip_address] = [t for t in self.queries[ip_address] if t > hour_ago]
+            self.queries = [t for t in self.queries if t > hour_ago]
             
-            return self.max_queries - len(self.queries[ip_address])
+            return self.max_queries - len(self.queries)
     
-    def get_time_until_reset(self, ip_address: str) -> float:
+    def get_time_until_reset(self, _: str = None) -> float:
         """
-        Get time in seconds until the rate limit resets for an IP address.
+        Get time in seconds until the rate limit resets.
         
-        Args:
-            ip_address: IP address of the client
-            
         Returns:
             float: Seconds until rate limit reset
         """
         current_time = time.time()
         
         with self.lock:
-            if not self.queries[ip_address]:
+            if not self.queries:
                 return 0.0
             
-            oldest_query = min(self.queries[ip_address])
+            oldest_query = min(self.queries)
             reset_time = oldest_query + 3600  # 1 hour in seconds
             
             return max(0.0, reset_time - current_time)