app.py

# Hey, stranger! this code is for use of free rate of gemini llm 
# which is limited by RPM (15/30). Testing has shown that if I put 
# request delay 10 then search drops out timed out.
# Nevertheless, it scrored 35% which is good for me while two questions 
# were dropped due to exceeding RPM. So, it is still possible to improve,
# e.g. deploying gemini 2.0 flash lite which has double RPM limit.
# Try it out!

import os
import gradio as gr
import requests
import inspect
import pandas as pd
import aiohttp
import asyncio
import json
from agent import MagAgent  
from token_bucket import Limiter, MemoryStorage 

# (Keep Constants as is)
# --- Constants ---
DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"

# Rate limiting configuration
RATE_LIMIT = 10  # Requests per minute
TOKEN_BUCKET_CAPACITY = RATE_LIMIT
TOKEN_BUCKET_REFILL_RATE = RATE_LIMIT / 60.0  # Tokens per second

# Initialize global token bucket with MemoryStorage
storage = MemoryStorage()
token_bucket = Limiter(rate=TOKEN_BUCKET_REFILL_RATE, capacity=TOKEN_BUCKET_CAPACITY, storage=storage)

async def fetch_questions(session: aiohttp.ClientSession, questions_url: str) -> list:
    """Fetch questions asynchronously."""
    try:
        async with session.get(questions_url, timeout=15) as response:
            response.raise_for_status()
            questions_data = await response.json()
            if not questions_data:
                print("Fetched questions list is empty.")
                return []
            print(f"Fetched {len(questions_data)} questions.")
            return questions_data
    except aiohttp.ClientError as e:
        print(f"Error fetching questions: {e}")
        return None
    except Exception as e:
        print(f"An unexpected error occurred fetching questions: {e}")
        return None

async def submit_answers(session: aiohttp.ClientSession, submit_url: str,
 submission_data: dict) -> dict:
    """Submit answers asynchronously."""
    try:
        async with session.post(submit_url, json=submission_data, timeout=60) as response:
            response.raise_for_status()
            return await response.json()
    except aiohttp.ClientResponseError as e:
        print(f"Submission Failed: Server responded with status {e.status}. Detail: {e.message}")
        return None
    except aiohttp.ClientError as e:
        print(f"Submission Failed: Network error - {e}")
        return None
    except Exception as e:
        print(f"An unexpected error occurred during submission: {e}")
        return None

async def process_question(agent, question_text: str, task_id: str,
 results_log: list):
    """Process a single question with global rate limiting."""
    submitted_answer = None
    try:
        # Retry until a token is available
        while not token_bucket.consume(1):
            print(f"Rate limit reached for task {task_id}. Waiting to retry...")
            await asyncio.sleep(60 / RATE_LIMIT)
        submitted_answer = await agent(question_text)
        results_log.append({"Task ID": task_id, "Question": question_text, "Submitted Answer": submitted_answer})
        return {"task_id": task_id, "submitted_answer": submitted_answer}
    except aiohttp.ClientResponseError as e:
        if e.status == 429:
            print(f"Rate limit hit for task {task_id}. Retrying after delay...")
            await asyncio.sleep(60 / RATE_LIMIT)
            while not token_bucket.consume(1):
                await asyncio.sleep(60 / RATE_LIMIT)
            try:
                submitted_answer = await agent(question_text)
                results_log.append({"Task ID": task_id, "Question": question_text, "Submitted Answer": submitted_answer})
                return {"task_id": task_id, "submitted_answer": submitted_answer}
            except Exception as retry_e:
                submitted_answer = f"AGENT ERROR: {retry_e}"
                results_log.append({"Task ID": task_id, "Question": question_text, "Submitted Answer": submitted_answer})
                return None
        else:
            submitted_answer = f"AGENT ERROR: {e}"
            results_log.append({"Task ID": task_id, "Question": question_text, "Submitted Answer": submitted_answer})
            return None
    except Exception as e:
        submitted_answer = f"AGENT ERROR: {e}"
        results_log.append({"Task ID": task_id, "Question": question_text, "Submitted Answer": submitted_answer})
        return None

async def run_and_submit_all(profile: gr.OAuthProfile | None):
    """
    Fetches all questions asynchronously, runs the MagAgent on them, submits all answers,
    and displays the results.
    """
    # --- Determine HF Space Runtime URL and Repo URL ---
    space_id = os.getenv("SPACE_ID")

    if profile:
        username = f"{profile.username}"
        print(f"User logged in: {username}")
    else:
        print("User not logged in.")
        return "Please Login to Hugging Face with the button.", None

    api_url = DEFAULT_API_URL
    questions_url = f"{api_url}/questions"
    submit_url = f"{api_url}/submit"

    # 1. Instantiate Agent
    try:
        agent =MagAgent()
    except Exception as e:
        print(f"Error instantiating agent: {e}")
        return f"Error initializing agent: {e}", None

    agent_code = f"https://huggingface.co/spaces/{space_id}/tree/main"
    print(agent_code)

    # 2. Fetch Questions Asynchronously
    async with aiohttp.ClientSession() as session:
        questions_data = await fetch_questions(session, questions_url)
        if questions_data is None:
            return "Error fetching questions.", None
        if not questions_data:
            return "Fetched questions list is empty or invalid format.", None

        # 3. Run Agent on Questions
        # Process questions sequentially with rate limiting
        results_log = []
        answers_payload = []
        print(f"Running agent on {len(questions_data)} questions...")

        for item in questions_data:
            if item.get("task_id") and item.get("question"):
                result = await process_question(agent, item["question"], item["task_id"], results_log)
                if result:
                    answers_payload.append(result)

        if not answers_payload:
            print("Agent did not produce any answers to submit.")
            return "Agent did not produce any answers to submit.", pd.DataFrame(results_log)
            
        # 4. Prepare Submission
        submission_data = {"username": username.strip(), "agent_code": agent_code, "answers": answers_payload}
        status_update = f"Agent finished. Submitting {len(answers_payload)} answers for user '{username}'..."
        print(status_update)

        # 5. Submit Answers Asynchronously
        result_data = await submit_answers(session, submit_url, submission_data)
        if result_data is None:
            status_message = "Submission Failed."
            print(status_message)
            results_df = pd.DataFrame(results_log)
            return status_message, results_df

        final_status = (
            f"Submission Successful!\n"
            f"User: {result_data.get('username')}\n"
            f"Overall Score: {result_data.get('score', 'N/A')}% "
            f"({result_data.get('correct_count', '?')}/{result_data.get('total_attempted', '?')} correct)\n"
            f"Message: {result_data.get('message', 'No message received.')}"
        )
        print("Submission successful.")
        results_df = pd.DataFrame(results_log)
        return final_status, results_df

# --- Build Gradio Interface using Blocks ---
with gr.Blocks() as demo:
    gr.Markdown("# Magus Agent Evaluation Runner")
    gr.Markdown(
        """
        **Instructions:**
        1. Log in to your Hugging Face account using the button below.
        2. Click 'Run Evaluation & Submit All Answers' to fetch questions, run your agent, and submit answers.
        ---
        **Notes:**
        The agent uses asynchronous operations for efficiency. Answers are processed and submitted asynchronously.
        """
    )

    gr.LoginButton()

    run_button = gr.Button("Run Evaluation & Submit All Answers")

    status_output = gr.Textbox(label="Run Status / Submission Result", lines=5, interactive=False)
    # Removed max_rows=10 from DataFrame constructor
    results_table = gr.DataFrame(label="Questions and Agent Answers", wrap=True)

    run_button.click(
        fn=run_and_submit_all,
        outputs=[status_output, results_table]
    )

if __name__ == "__main__":
    print("\n" + "-"*30 + " App Starting " + "-"*30)
    # Check for SPACE_HOST and SPACE_ID at startup for information
    space_host_startup = os.getenv("SPACE_HOST")
    space_id_startup = os.getenv("SPACE_ID") # Get SPACE_ID at startup

    if space_host_startup:
        print(f"✅ SPACE_HOST found: {space_host_startup}")
        print(f"   Runtime URL should be: https://{space_host_startup}.hf.space")
    else:
        print("ℹ️  SPACE_HOST environment variable not found (running locally?).")

    if space_id_startup: # Print repo URLs if SPACE_ID is found
        print(f"✅ SPACE_ID found: {space_id_startup}")
        print(f"   Repo URL: https://huggingface.co/spaces/{space_id_startup}")
        print(f"   Repo Tree URL: https://huggingface.co/spaces/{space_id_startup}/tree/main")
    else:
        print("ℹ️  SPACE_ID environment variable not found (running locally?). Repo URL cannot be determined.")

    print("-"*(60 + len(" App Starting ")) + "\n")

    print("Launching Gradio Interface for Mag Agent Evaluation...")
    demo.launch(debug=True, share=False)