app.py

""" Basic Agent Evaluation Runner – invia sempre tutte le risposte """

import os
import requests
import gradio as gr
import pandas as pd
from langchain_core.messages import HumanMessage
from agent import build_graph


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


# --- Agent wrapper -------------------------------------------------------- #
class BasicAgent:
    """LangGraph agent ready for evaluation."""
    def __init__(self):
        print("BasicAgent initialized (provider=groq).")
        self.graph = build_graph(provider="groq")

    def __call__(self, question: str) -> str:
        print(f"Agent received question (first 50 chars): {question[:50]}...")
        msgs = [HumanMessage(content=question)]
        result = self.graph.invoke({"messages": msgs})
        answer = result["messages"][-1].content
        # rimuove la parte "FINAL ANSWER: "
        return answer[14:]


# --- Main evaluation logic ------------------------------------------------ #
def run_and_submit_all(profile: gr.OAuthProfile | None):
    # 0. Check login
    if not profile:
        return "Please Login to Hugging Face with the button.", None
    username = profile.username
    print(f"User logged in: {username}")

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

    # 2. Fetch questions
    try:
        resp = requests.get(f"{DEFAULT_API_URL}/questions", timeout=15)
        resp.raise_for_status()
        questions_data = resp.json()
        if not questions_data:
            return "Fetched questions list is empty.", None
    except Exception as e:
        return f"Error fetching questions: {e}", None

    # 3. Run agent and build payload
    answers_payload = []
    results_log = []

    for item in questions_data:
        task_id = item.get("task_id")
        q_text = item.get("question")

        submitted_answer = "errore"  # default in caso di failure
        try:
            submitted_answer = agent(q_text)
        except Exception as e:
            print(f"Error running agent on task {task_id}: {e}")

        # in ogni caso inseriamo la risposta (successo o errore)
        answers_payload.append(
            {"task_id": task_id, "submitted_answer": submitted_answer}
        )
        results_log.append(
            {
                "Task ID": task_id,
                "Question": q_text,
                "Submitted Answer": submitted_answer,
            }
        )

    # 4. Submit answers
    submission = {
        "username": username,
        "agent_code": f"https://huggingface.co/spaces/{os.getenv('SPACE_ID', '')}/tree/main",
        "answers": answers_payload,
    }

    try:
        resp = requests.post(f"{DEFAULT_API_URL}/submit", json=submission, timeout=60)
        resp.raise_for_status()
        data = resp.json()
        status_msg = (
            f"Submission Successful!\nUser: {data.get('username')}\n"
            f"Overall Score: {data.get('score', 'N/A')}% "
            f"({data.get('correct_count', '?')}/{data.get('total_attempted', '?')} correct)\n"
            f"Message: {data.get('message', 'No message received.')}"
        )
    except Exception as e:
        status_msg = f"Submission Failed: {e}"

    return status_msg, pd.DataFrame(results_log)


# --- Gradio UI ------------------------------------------------------------ #
with gr.Blocks() as demo:
    gr.Markdown("# Basic Agent Evaluation Runner (retry & error-safe)")
    gr.LoginButton()
    run_btn = gr.Button("Run Evaluation & Submit All Answers")
    status_box = gr.Textbox(lines=5, label="Run Status / Submission Result")
    results_tbl = gr.DataFrame(label="Questions and Agent Answers", wrap=True)
    run_btn.click(fn=run_and_submit_all, outputs=[status_box, results_tbl])

if __name__ == "__main__":
    demo.launch(debug=True, share=False)