app.py

import os
import gradio as gr
import requests
import pandas as pd
from smolagents import CodeAgent, DuckDuckGoSearchTool, HfApiModel
import logging

# Настройка логирования
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)

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

# --- Internet-Enabled Agent Definition ---
class InternetAgent:
    def __init__(self):
        print("🌐 InternetAgent initializing with web search capabilities...")
        
        try:
            # Используем модель от Hugging Face и инструмент поиска
            self.model = HfApiModel("Qwen/Qwen2.5-Coder-32B-Instruct")
            self.search_tool = DuckDuckGoSearchTool()
            
            # Создаем агента с доступом к поиску
            self.agent = CodeAgent(
                tools=[self.search_tool],
                model=self.model,
                max_steps=6,  # Ограничиваем шаги для скорости
                add_base_tools=False  # Используем только наши инструменты
            )
            print("✅ InternetAgent initialized successfully with web search")
            
        except Exception as e:
            print(f"❌ Error initializing InternetAgent: {e}")
            self.agent = None
            # Резервная база знаний на случай ошибки
            self.fallback_knowledge = {
                "capital of france": "Paris",
                "capital of germany": "Berlin", 
                "capital of uk": "London",
                "capital of usa": "Washington D.C.",
                "2+2": "4",
                "largest planet": "Jupiter",
            }

    def __call__(self, question: str) -> str:
        print(f"🤖 Processing: {question}")
        
        if not self.agent:
            # Используем резервную базу знаний если агент не инициализирован
            question_lower = question.lower()
            for key, answer in self.fallback_knowledge.items():
                if key in question_lower:
                    return answer
            return "I need internet access to answer this question properly."
        
        try:
            # Создаем оптимизированный промпт для лучших результатов
            optimized_prompt = f"""
            Please provide a clear, concise, and accurate answer to the following question.
            If you need to search for information, use the search tool.
            Keep your answer brief and to the point.
            
            Question: {question}
            
            Answer:
            """
            
            # Запускаем агента
            response = self.agent.run(optimized_prompt)
            
            # Очищаем ответ
            clean_response = self.clean_response(response)
            print(f"✅ Answer: {clean_response[:100]}...")
            
            return clean_response
            
        except Exception as e:
            print(f"❌ Error in agent execution: {e}")
            return f"I encountered an error while searching for the answer: {str(e)}"

    def clean_response(self, response: str) -> str:
        """Очищает ответ от лишней информации"""
        # Удаляем мета-комментарии агента
        lines = response.split('\n')
        clean_lines = []
        
        for line in lines:
            # Пропускаем строки с инструментами или процессами
            if any(term in line.lower() for term in ['tool:', 'searching', 'step', 'using tool']):
                continue
            # Пропускаем пустые строки в начале
            if not clean_lines and not line.strip():
                continue
            clean_lines.append(line)
        
        clean_response = '\n'.join(clean_lines).strip()
        
        # Если ответ слишком длинный, берем первую часть
        if len(clean_response) > 500:
            clean_response = clean_response[:497] + "..."
            
        return clean_response if clean_response else "I couldn't find a clear answer to that question."

# --- Упрощенная версия для тестирования ---
class LiteInternetAgent:
    def __init__(self):
        print("🌐 LiteInternetAgent initializing...")
        self.search_tool = DuckDuckGoSearchTool()
        
    def __call__(self, question: str) -> str:
        try:
            # Прямой поиск через инструмент
            result = self.search_tool(question)
            return f"According to web search: {result[:300]}..." if len(result) > 300 else result
        except Exception as e:
            return f"Search failed: {str(e)}"

# --- Основная функция ---
def run_and_submit_all(profile: gr.OAuthProfile | None):
    """
    Fetches all questions, runs the InternetAgent on them, submits all answers,
    and displays the results.
    """
    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 our Internet Agent
    try:
        # Пробуем полную версию, если не работает - упрощенную
        agent = InternetAgent()
        if agent.agent is None:
            agent = LiteInternetAgent()
        print("✅ InternetAgent created successfully")
    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(f"Agent code URL: {agent_code}")

    # 2. Fetch Questions
    print(f"Fetching questions from: {questions_url}")
    try:
        response = requests.get(questions_url, timeout=30)
        response.raise_for_status()
        questions_data = response.json()
        if not questions_data:
            print("Fetched questions list is empty.")
            return "Fetched questions list is empty or invalid format.", None
        print(f"✅ Fetched {len(questions_data)} questions.")
    except Exception as e:
        error_msg = f"❌ Error fetching questions: {e}"
        print(error_msg)
        
        # Демо-режим с разными типами вопросов
        demo_questions = [
            {"task_id": "demo1", "question": "What is the capital of France?"},
            {"task_id": "demo2", "question": "What is the current weather in Tokyo?"},
            {"task_id": "demo3", "question": "Who won the Nobel Prize in Physics in 2023?"},
            {"task_id": "demo4", "question": "What is the population of Brazil?"},
            {"task_id": "demo5", "question": "Explain quantum computing in simple terms"},
        ]
        questions_data = demo_questions
        print("🚨 Using demo questions since API is unavailable")

    # 3. Run your Agent
    results_log = []
    answers_payload = []
    print(f"Running agent on {len(questions_data)} questions...")
    
    for i, item in enumerate(questions_data):
        task_id = item.get("task_id")
        question_text = item.get("question")
        
        if not task_id or question_text is None:
            print(f"Skipping item with missing task_id or question: {item}")
            continue
            
        print(f"🔍 Processing question {i+1}/{len(questions_data)}: {question_text[:50]}...")
        
        try:
            submitted_answer = agent(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:
            print(f"❌ Error running agent on task {task_id}: {e}")
            results_log.append({"Task ID": task_id, "Question": question_text, "Submitted Answer": f"AGENT ERROR: {e}"})

    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. Processed {len(answers_payload)} answers for user '{username}'"
    print(status_update)

    # 5. Submit answers (только если вопросы реальные, не демо)
    if "demo" not in str(questions_data[0].get("task_id", "")):
        print(f"Submitting {len(answers_payload)} answers to: {submit_url}")
        try:
            response = requests.post(submit_url, json=submission_data, timeout=120)
            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')}% "
                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
            
        except Exception as e:
            error_message = f"❌ Submission Failed: {str(e)}"
            print(error_message)
            results_df = pd.DataFrame(results_log)
            return error_message, results_df
    else:
        # Демо-режим: показываем ответы но не отправляем
        demo_status = (
            f"🧪 DEMO MODE (API Unavailable)\n"
            f"👤 User: {username}\n"
            f"📊 Processed: {len(answers_payload)} demo questions\n"
            f"🌐 Agent used web search for answers\n"
            f"💬 Real submission disabled - API not accessible\n\n"
            f"Check the web-powered answers below!"
        )
        print("✅ Demo completed - showing results without submission")
        results_df = pd.DataFrame(results_log)
        return demo_status, results_df

# --- Gradio Interface ---
with gr.Blocks(title="Internet-Enabled AI Agent") as demo:
    gr.Markdown("""
    # 🌐 Internet-Enabled AI Agent
    **Powered by web search and large language models**
    
    ### 🔧 Capabilities:
    - **Web Search**: Real-time information from DuckDuckGo
    - **LLM Power**: Qwen2.5-32B model for understanding
    - **Multi-step Reasoning**: Complex question answering
    
    ### 📚 Example questions:
    - *"Current weather in any city"*
    - *"Latest news headlines"* 
    - *"Historical facts and data"*
    - *"Scientific explanations"*
    - *"Complex calculations"*
    """)
    
    gr.Markdown("""
    ### ⚠️ Important Notes:
    - This agent requires internet access
    - Responses may take longer due to web searches
    - Some questions might not have clear online answers
    """)
    
    with gr.Row():
        with gr.Column():
            gr.LoginButton()
            run_button = gr.Button("🚀 Run Evaluation", variant="primary")
    
    with gr.Row():
        with gr.Column():
            status_output = gr.Textbox(
                label="Status", 
                lines=4, 
                interactive=False
            )
        with gr.Column():
            results_table = gr.DataFrame(
                label="Questions & Answers", 
                wrap=True
            )

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

if __name__ == "__main__":
    print("🚀 Starting Internet-Enabled AI Agent...")
    demo.launch(debug=True, share=False)