import os
import gradio as gr
import requests
import pandas as pd
from transformers import pipeline, AutoTokenizer, AutoModelForCausalLM

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

# --- Smart Agent Definition Using Open LLM ---
class BasicAgent:
    def __init__(self):
        print("Loading open instruct model...")
        model_id = "meta-llama/Llama-3-8B-instruct"  # <— change to a model you have access to
        self.tokenizer = AutoTokenizer.from_pretrained(model_id)
        self.model = AutoModelForCausalLM.from_pretrained(model_id)
        self.pipeline = pipeline(
            "text-generation",
            model=self.model,
            tokenizer=self.tokenizer,
            max_new_tokens=150,
            temperature=0.3,
            do_sample=False,
        )

    def __call__(self, question: str) -> str:
        print(f"Agent received question: {question[:60]}")

        prompt = f"""You are a helpful assistant. Answer the question concisely and output only the final answer (no explanation).

Question: {question}

Answer:"""

        try:
            output = self.pipeline(prompt)[0]["generated_text"]
            # After generation, remove the prompt part
            answer = output[len(prompt):].strip()
            # Sometimes model may append new lines or extra text; trim to first line
            answer = answer.split("\n")[0].strip()
            return answer
        except Exception as e:
            print("Error during inference:", e)
            return "I don't know"

def run_and_submit_all(profile: gr.OAuthProfile | None):
    space_id = os.getenv("SPACE_ID")
    if profile:
        username = profile.username
    else:
        return "Please Login to Hugging Face with the button.", None

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

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

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

    try:
        response = requests.get(questions_url, timeout=15)
        response.raise_for_status()
        questions_data = response.json()
    except Exception as e:
        return f"Error fetching questions: {e}", None

    results_log = []
    answers_payload = []

    for item in questions_data:
        task_id = item.get("task_id")
        question_text = item.get("question")
        if not task_id or question_text is None:
            continue
        try:
            answer = agent(question_text)
            answers_payload.append({"task_id": task_id, "submitted_answer": answer})
            results_log.append({"Task ID": task_id, "Question": question_text, "Submitted Answer": answer})
        except Exception as e:
            results_log.append({"Task ID": task_id, "Question": question_text, "Submitted Answer": f"ERROR: {e}"})

    if not answers_payload:
        return "No answers to submit.", pd.DataFrame(results_log)

    submission_data = {
        "username": username.strip(),
        "agent_code": agent_code,
        "answers": answers_payload
    }

    try:
        resp = requests.post(submit_url, json=submission_data, timeout=60)
        resp.raise_for_status()
        result = resp.json()
        final_status = (
            f"Submission Successful!\n"
            f"User: {result.get('username')}\n"
            f"Score: {result.get('score', 'N/A')}% "
            f"({result.get('correct_count', '?')}/{result.get('total_attempted', '?')})\n"
            f"Message: {result.get('message', '')}"
        )
        return final_status, pd.DataFrame(results_log)
    except Exception as e:
        return f"Submission Failed: {e}", pd.DataFrame(results_log)


with gr.Blocks() as demo:
    gr.Markdown("# Basic Agent Evaluation Runner")
    gr.Markdown("This uses an open access instruct model (no gated repo).")

    gr.LoginButton()
    run_button = gr.Button("Run Evaluation & Submit All Answers")
    status_output = gr.Textbox(label="Run Status", lines=5, interactive=False)
    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__":
    demo.launch(debug=True, share=False)