app.py

import os
import gradio as gr
import requests
import pandas as pd
import time

# Import your newly refactored smolagents GaiaAgent
from agent import GaiaAgent

# --- Constants ---
DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"

# Global placeholder layout to keep the UI dataframe stable during initialization steps
COLUMNS = ["Task ID", "Question", "Submitted Answer"]
INITIAL_DF = pd.DataFrame(columns=COLUMNS)

def run_and_submit_all(profile: gr.OAuthProfile | None):
    """
    Fetches all questions, runs the GaiaAgent on them, submits all answers,
    and streams the results back to the Gradio UI asynchronously.
    """
    space_id = os.getenv("SPACE_ID") 

    if profile:
        username = f"{profile.username}"
        print(f"User logged in: {username}")
    else:
        yield "Please Login to Hugging Face with the button.", INITIAL_DF
        return

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

    yield "Initializing Agent...", INITIAL_DF

    # 1. Instantiate Agent
    try:
        agent = GaiaAgent()
    except Exception as e:
        yield f"Error initializing agent: {e}", INITIAL_DF
        return

    agent_code = f"https://huggingface.co/spaces/{space_id}/tree/main" if space_id else "Local Run"

    # 2. Fetch Questions
    yield "Fetching questions from server...", INITIAL_DF
    try:
        response = requests.get(questions_url, timeout=15)
        response.raise_for_status()
        questions_data = response.json()
        
        if not questions_data:
            yield "Fetched questions list is empty or invalid format.", INITIAL_DF
            return
            
    except Exception as e:
        yield f"An unexpected error occurred fetching questions: {e}", INITIAL_DF
        return

    total_questions = len(questions_data)
    results_log = []
    answers_payload = []

    # 3. Run Agent (Streaming Updates via Yield)
    for i, item in enumerate(questions_data, 1):
        task_id = item.get("task_id")
        question_text = item.get("question")
        
        if not task_id or question_text is None:
            continue
            
        yield f"Processing question {i}/{total_questions} (Task ID: {task_id})...", pd.DataFrame(results_log, columns=COLUMNS)
        
        try:
            submitted_answer = agent(task_id, question_text)
            answers_payload.append({"task_id": task_id, "submitted_answer": submitted_answer})
            results_log.append({
                "Task ID": task_id, 
                "Question": question_text, 
                "Submitted Answer": submitted_answer
            })
        except Exception as e:
            results_log.append({
                "Task ID": task_id, 
                "Question": question_text, 
                "Submitted Answer": f"AGENT ERROR: {e}"
            })
            
        # Yield updated table immediately so the user sees real-time progress
        yield f"Completed {i}/{total_questions}. Moving to next...", pd.DataFrame(results_log, columns=COLUMNS)
        
        # Small buffer to prevent overwhelming the server between consecutive model inquiries
        time.sleep(2)

    if not answers_payload:
        yield "Agent did not produce any answers to submit.", pd.DataFrame(results_log, columns=COLUMNS)
        return

    # 4. Prepare & Submit
    submission_data = {
        "username": username.strip(), 
        "agent_code": agent_code, 
        "answers": answers_payload
    }
    
    yield f"Agent finished. Submitting {len(answers_payload)} answers...", pd.DataFrame(results_log, columns=COLUMNS)

    try:
        response = requests.post(submit_url, json=submission_data, timeout=60)
        response.raise_for_status()
        result_data = response.json()
        
        final_status = (
            f"🎉 Submission Successful!\n"
            f"User: {result_data.get('username')}\n"
            f"Overall Score: {result_data.get('score', 'N/A')}%\n"
            f"Correct: {result_data.get('correct_count', '?')}/{result_data.get('total_attempted', '?')}\n"
            f"Message: {result_data.get('message', 'No message received.')}"
        )
        yield final_status, pd.DataFrame(results_log, columns=COLUMNS)
        
    except Exception as e:
        yield f"Submission Failed: {e}", pd.DataFrame(results_log, columns=COLUMNS)


# --- Build Gradio Interface using Blocks ---
with gr.Blocks(theme=gr.themes.Soft()) as demo:
    gr.Markdown("# 🤖 GAIA Agent Evaluation Runner")
    gr.Markdown(
        """
        **Instructions:**
        1. Log in to your Hugging Face account using the button below. This uses your HF username for submission.
        2. Click 'Run Evaluation & Submit All Answers' to fetch questions and run your agent.
        3. Watch the table update in real-time as your agent processes each question!
        """
    )

    gr.LoginButton()

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

    status_output = gr.Textbox(label="Run Status / Submission Result", lines=6, interactive=False)
    results_table = gr.DataFrame(value=INITIAL_DF, label="Questions and Agent Answers", wrap=True)

    # Gradio handles the gr.OAuthProfile backend injection natively when clicked
    run_button.click(
        fn=run_and_submit_all,
        inputs=None,
        outputs=[status_output, results_table]
    )

if __name__ == "__main__":
    print("\n" + "-"*30 + " App Starting " + "-"*30)
    
    space_host_startup = os.getenv("SPACE_HOST")
    space_id_startup = os.getenv("SPACE_ID")

    if space_host_startup:
        print(f"✅ SPACE_HOST found: {space_host_startup}")
    else:
        print("ℹ️  SPACE_HOST environment variable not found (running locally?).")

    if space_id_startup:
        print(f"✅ SPACE_ID found: {space_id_startup}")
    else:
        print("ℹ️  SPACE_ID environment variable not found (running locally?).")

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

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