updated 4

app.py

@@ -3,15 +3,44 @@ import gradio as gr
 import requests
 import inspect
 import pandas as pd
+import re
+import time
+from datetime import datetime, timedelta
+from collections import deque
+import random
 from smolagents import CodeAgent, DuckDuckGoSearchTool, load_tool, tool
 from smolagents.models import Model, ChatMessage, MessageRole, Tool
 from tools import FinalAnswerTool
 import google.generativeai as genai
-import re
 
 # (Keep Constants as is)
 # --- Constants ---
 DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
+MAX_RETRIES = 3
+INITIAL_RETRY_DELAY = 1  # seconds
+MAX_RETRY_DELAY = 32  # seconds
+
+class RateLimiter:
+    def __init__(self, requests_per_minute):
+        self.requests_per_minute = requests_per_minute
+        self.window_size = 60  # 60 seconds = 1 minute
+        self.requests = deque()
+
+    def wait_if_needed(self):
+        now = datetime.now()
+        
+        # Remove requests older than our window
+        while self.requests and (now - self.requests[0]).total_seconds() > self.window_size:
+            self.requests.popleft()
+        
+        # If we've hit our limit, wait until the oldest request expires
+        if len(self.requests) >= self.requests_per_minute:
+            wait_time = self.window_size - (now - self.requests[0]).total_seconds()
+            if wait_time > 0:
+                time.sleep(wait_time + 0.1)  # Add a small buffer
+        
+        # Add the current request
+        self.requests.append(now)
 
 # --- Basic Agent Definition ---
 
@@ -23,6 +52,7 @@ class GeminiModel(Model):
         self.api_key = api_key
         genai.configure(api_key=api_key)
         self.model = genai.GenerativeModel('models/gemini-2.0-flash-lite')
+        self.rate_limiter = RateLimiter(requests_per_minute=25)  # Setting to 25 to be safe
         
         # System prompt for smolagents format
         self.system_prompt = """You are a highly focused AI assistant tasked with answering specific questions accurately using available tools. Your primary goal is to find and provide precise answers to questions using the tools provided.
@@ -103,79 +133,99 @@ Remember:
         tools_to_call_from: list[Tool] | None = None,
         **kwargs,
     ) -> ChatMessage:
-        try:
-            # Handle different prompt types
-            if isinstance(messages, list) and len(messages) > 0:
-                # Extract content from the last message (most recent)
-                last_message = messages[-1]
-                
-                if isinstance(last_message, dict) and 'content' in last_message:
-                    content = last_message['content']
-                elif isinstance(last_message, ChatMessage) and last_message.content:
-                    content = last_message.content
+        retry_count = 0
+        delay = INITIAL_RETRY_DELAY
+
+        while True:
+            try:
+                # Wait if we need to due to rate limiting
+                self.rate_limiter.wait_if_needed()
+
+                # Handle different prompt types
+                if isinstance(messages, list) and len(messages) > 0:
+                    last_message = messages[-1]
+                    
+                    if isinstance(last_message, dict) and 'content' in last_message:
+                        content = last_message['content']
+                    elif isinstance(last_message, ChatMessage) and last_message.content:
+                        content = last_message.content
+                    else:
+                        content = ""
+                        for msg in messages:
+                            if isinstance(msg, dict) and 'content' in msg:
+                                content += str(msg['content']) + "\n"
+                            elif isinstance(msg, ChatMessage) and msg.content:
+                                content += str(msg.content) + "\n"
+                            else:
+                                content += str(msg) + "\n"
                 else:
-                    # Fallback: combine all message content
-                    content = ""
-                    for msg in messages:
-                        if isinstance(msg, dict) and 'content' in msg:
-                            content += str(msg['content']) + "\n"
-                        elif isinstance(msg, ChatMessage) and msg.content:
-                            content += str(msg.content) + "\n"
-                        else:
-                            content += str(msg) + "\n"
-            else:
-                content = str(messages)
-
-            # Ensure content is a simple string for Gemini API
-            if isinstance(content, list):
-                # Handle list of content parts
-                text_parts = []
-                for part in content:
-                    if isinstance(part, dict):
-                        if 'text' in part:
-                            text_parts.append(part['text'])
-                        elif 'content' in part:
-                            text_parts.append(part['content'])
+                    content = str(messages)
+
+                # Ensure content is a simple string for Gemini API
+                if isinstance(content, list):
+                    text_parts = []
+                    for part in content:
+                        if isinstance(part, dict):
+                            if 'text' in part:
+                                text_parts.append(part['text'])
+                            elif 'content' in part:
+                                text_parts.append(part['content'])
+                            else:
+                                text_parts.append(str(part))
                         else:
                             text_parts.append(str(part))
+                    content = "\n".join(text_parts)
+                elif isinstance(content, dict):
+                    if 'text' in content:
+                        content = content['text']
+                    elif 'content' in content:
+                        content = content['content']
                     else:
-                        text_parts.append(str(part))
-                content = "\n".join(text_parts)
-            elif isinstance(content, dict):
-                # Handle dict content - extract text if possible
-                if 'text' in content:
-                    content = content['text']
-                elif 'content' in content:
-                    content = content['content']
-                else:
-                    content = str(content)
-            
-            # Combine system prompt with user content
-            full_prompt = f"{self.system_prompt}\n\nTask: {content}"
+                        content = str(content)
+                
+                # Combine system prompt with user content
+                full_prompt = f"{self.system_prompt}\n\nTask: {content}"
 
-            # Generate response
-            response = self.model.generate_content(full_prompt)
-            
-            # Extract text from response
-            if hasattr(response, 'text'):
-                response_text = response.text
-            elif isinstance(response, str):
-                response_text = response
-            elif hasattr(response, 'content'):
-                response_text = response.content
-            else:
-                response_text = str(response)
+                # Generate response
+                response = self.model.generate_content(full_prompt)
                 
-            # Return ChatMessage object as expected by smolagents
-            return ChatMessage(
-                role=MessageRole.ASSISTANT,
-                content=response_text,
-                raw=response
-            )
+                # Extract text from response
+                if hasattr(response, 'text'):
+                    response_text = response.text
+                elif isinstance(response, str):
+                    response_text = response
+                elif hasattr(response, 'content'):
+                    response_text = response.content
+                else:
+                    response_text = str(response)
+                    
+                # Return ChatMessage object as expected by smolagents
+                return ChatMessage(
+                    role=MessageRole.ASSISTANT,
+                    content=response_text,
+                    raw=response
+                )
+                    
+            except Exception as e:
+                error_str = str(e)
                 
-        except Exception as e:
-            print(f"Error in generate: {e}")
-            raise e
+                # Check if it's a rate limit error
+                if "429" in error_str and retry_count < MAX_RETRIES:
+                    retry_count += 1
+                    
+                    # Add some random jitter to prevent all retries happening at exactly the same time
+                    jitter = random.uniform(0, 0.1) * delay
+                    sleep_time = delay + jitter
+                    
+                    print(f"Rate limit hit. Retrying in {sleep_time:.2f} seconds (attempt {retry_count}/{MAX_RETRIES})")
+                    time.sleep(sleep_time)
+                    
+                    # Exponential backoff
+                    delay = min(delay * 2, MAX_RETRY_DELAY)
+                    continue
+                    
+                print(f"Error in generate: {e}")
+                raise e
 
 class MyAgent:
     def __init__(self):