app.py

import os
import gradio as gr
import requests
import pandas as pd
import time
import re
from smolagents import LiteLLMModel, CodeAgent, Tool

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

# --- Answer Extraction Function ---
def extract_answer(text: str, original_question: str) -> str:
    """Extract the answer from the LLM response, being robust to various formats."""
    if not text:
        return "- unknown"
    
    # Clean the text
    cleaned = text.strip()
    
    # If the response is the same as the question, it's not an answer
    if cleaned == original_question.strip():
        return "- unknown"
    
    # Remove common prefixes
    prefixes_to_remove = [
        '[ANSWER]:',
        '[ANSWER]',
        'Final answer:',
        'Final Answer:',
        'Answer:',
        'answer:',
        'The answer is',
        'The final answer is',
    ]
    
    for prefix in prefixes_to_remove:
        if cleaned.startswith(prefix):
            cleaned = cleaned[len(prefix):].strip()
    
    # If it's a "how many" question, try to extract just the number
    if 'how many' in original_question.lower():
        # Look for numbers in the response

        number_patterns = [
            r'(?:is|are)\s*(\d+)',
            r'^\s*(\d+)',
            r'\D(\d+)\D*$',
            r'(\d+)'
        ]
    
        for pattern in number_patterns:
            numbers = re.findall(pattern, cleaned)
            if numbers:
                return numbers[0]
            
        numbers = re.findall(r'\d+', cleaned)
        if numbers:
            return numbers[0]  # Return the first number found
    
    # If it's asking for a year, try to extract just the year
    if re.search(r'\b(19|20)\d{2}\b', original_question):
        years = re.findall(r'\b(19|20)\d{2}\b', cleaned)
        if years:
            return years[0]  # Return the first year found
    
    # If we still have the full question in the response, try to extract what comes after it
    if original_question.strip() in cleaned:
        # Split by the question and take what comes after
        parts = cleaned.split(original_question.strip())
        if len(parts) > 1 and parts[1].strip():
            cleaned = parts[1].strip()
        else:
            # Try to find numbers or short answers in the response
            # Look for a line that might contain the answer
            lines = cleaned.split('\n')
            for line in lines:
                line = line.strip()
                if line and line != original_question.strip():
                    # If it's a short line, it might be the answer
                    if len(line) < 100 or 'how many' in original_question.lower():
                        cleaned = line
                        break
    
    # If the cleaned answer is still very long and contains the question, 
    # try to extract just the essential part
    if len(cleaned) > 200 and original_question.strip() in cleaned:
        # Try to find a short line that might be the answer
        lines = cleaned.split('\n')
        for line in lines:
            line = line.strip()
            if line and len(line) < 100 and line != original_question.strip():
                # Check if it looks like an answer (short and possibly numeric)
                if re.match(r'^[\w\s\d\-\.,]+$', line):  # Simple alphanumeric answer
                    return line
    
    # If we still have a very long response, try to extract just the last line
    # which might be the answer
    if len(cleaned) > 200:
        lines = cleaned.split('\n')
        # Take the last non-empty line that isn't too long
        for line in reversed(lines):
            line = line.strip()
            if line and len(line) < 100:
                cleaned = line
                break
    
    # Final fallback - if the result is still the same as the question, return unknown
    if cleaned == original_question.strip():
        return "- unknown"
    
    return cleaned if cleaned else "- unknown"

# --- Agent Tools ---
class MathSolver(Tool):
    name = "math_solver"
    description = "Safely evaluate basic math expressions."
    inputs = {"input": {"type": "string", "description": "Math expression to evaluate."}}
    output_type = "string"

    def forward(self, input: str) -> str:
        try:
            # Safe evaluation of math expressions
            allowed_names = {
                k: v for k, v in __builtins__.items() if k in [
                    'abs', 'round', 'min', 'max', 'sum', 'pow'
                ]
            }
            allowed_names.update({
                'int': int, 'float': float, 'str': str,
                '__builtins__': {}
            })
            return str(eval(input, allowed_names))
        except Exception as e:
            return f"Math error: {e}"

class FileAttachmentQueryTool(Tool):
    name = "run_query_with_file"
    description = "Downloads a file mentioned in a user prompt, adds it to the context, and runs a query on it."
    inputs = {
        "task_id": {
            "type": "string",
            "description": "A unique identifier for the task related to this file, used to download it.",
            "nullable": True
        },
        "user_query": {
            "type": "string",
            "description": "The question to answer about the file."
        }
    }
    output_type = "string"

    def forward(self, task_id: str | None, user_query: str) -> str:
        if not task_id:
            return "No task_id provided for file download."
        
        file_url = f"https://agents-course-unit4-scoring.hf.space/files/{task_id}"
        try:
            file_response = requests.get(file_url)
            if file_response.status_code != 200:
                return f"Failed to download file: {file_response.status_code}"
            
            # For text-based files, return content directly
            file_content = file_response.text[:2000]  # Limit content size
            return f"Relevant information from file: {file_content}"
        except Exception as e:
            return f"File download error: {e}"

class WikipediaSearchTool(Tool):
    name = "wikipedia_search"
    description = "Search Wikipedia for information relevant to the user's query."
    inputs = {"query": {"type": "string", "description": "The search query."}}
    output_type = "string"

    def forward(self, query: str) -> str:
        try:
            search_url = f"https://en.wikipedia.org/api/rest_v1/page/summary/{requests.utils.quote(query)}"
            response = requests.get(search_url)
            if response.status_code == 200:
                data = response.json()
                return data.get("extract", "No summary available.")
            else:
                return f"Wikipedia search error: {response.status_code}"
        except Exception as e:
            return f"Wikipedia search exception: {e}"
            
# --- Agent Implementation ---
def select_model(provider="groq"):
    """Select and return a model based on the provider."""
    GROQ_MODEL_NAME = "groq/llama3-70b-8192"
    HF_MODEL_NAME = "huggingfaceh4/zephyr-7b-beta"
    
    if provider == "groq":
        api_key = os.getenv("GROQ_API_KEY")

        if api_key:
            return LiteLLMModel(model_id="groq/llama-3.1-8b-instant",
                                api_key=os.getenv("GROQ_API_KEY"))
        if not api_key:
            raise ValueError("GROQ_API_KEY environment variable is not set")
        
        return LiteLLMModel(model_id=GROQ_MODEL_NAME, api_key=api_key)
    elif provider == "hf":
        api_key = os.getenv("HF_TOKEN")
        if not api_key:
            raise ValueError("HF_TOKEN environment variable is not set")
        return LiteLLMModel(model_id=HF_MODEL_NAME, api_key=api_key)
    else:
        # Default to Groq if no valid provider specified
        api_key = os.getenv("GROQ_API_KEY")
        if not api_key:
            raise ValueError("GROQ_API_KEY environment variable is not set")
        return LiteLLMModel(model_id=GROQ_MODEL_NAME, api_key=api_key)

class BasicAgent:
    def __init__(self, provider="groq"):
        model = select_model(provider)
        tools = [
            MathSolver(),
            FileAttachmentQueryTool(),
            WikipediaSearchTool()
        ]
        self.agent = CodeAgent(
            model=model,
            tools=tools,
            add_base_tools=False,
            max_steps=15,
        )
        # System prompt to enforce exact answer format
        self.agent.prompt_templates["system_prompt"] = (
            "You are a GAIA benchmark AI assistant. Your sole purpose is to output the minimal, final answer. "
            "You must NEVER output explanations, intermediate steps, reasoning, or comments — only the answer. "
            "For numerical answers, use digits only, e.g., `4` not `four`. "
            "For string answers, omit articles ('a', 'the') and use full words. "
            "For lists, output in comma-separated format with no conjunctions. "
            "If the answer is not found, say `- unknown`."
            "IMPORTANT: Respond with ONLY the answer, nothing else. No prefixes, no explanations."
        )

    def __call__(self, question: str) -> str:
        max_retries = 3
        retry_delay = 10  # Start with 10 seconds
        
        for attempt in range(max_retries):
            try:
                result = self.agent.run(question)
                # Use our enhanced extraction function
                final_str = extract_answer(str(result), question)
                return final_str
            except Exception as e:
                # Check if it's a rate limit error
                if "RateLimitError" in str(e) or "rate_limit_exceeded" in str(e):
                    if attempt < max_retries - 1:  # Not the last attempt
                        print(f"Rate limit hit. Waiting {retry_delay} seconds before retry {attempt + 1}/{max_retries}")
                        time.sleep(retry_delay)
                        retry_delay *= 2  # Exponential backoff
                        continue
                    else:
                        return f"Rate limit error after {max_retries} attempts: {e}"
                else:
                    # Not a rate limit error, re-raise
                    raise e
        
        return f"Failed to get response after {max_retries} attempts"

# --- Main Application Functions ---
def run_and_submit_all(profile: gr.OAuthProfile | None):
    """
    Fetches all questions, runs the BasicAgent 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 = BasicAgent()
    except Exception as e:
        print(f"Error instantiating agent: {e}")
        return f"Error initializing agent: {e}", None
    
    # In the case of an app running as a hugging Face space, this link points toward your codebase
    agent_code = f"https://huggingface.co/spaces/{space_id}/tree/main"
    print(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 requests.exceptions.RequestException as e:
        print(f"Error fetching questions: {e}")
        return f"Error fetching questions: {e}", None
    except requests.exceptions.JSONDecodeError as e:
         print(f"Error decoding JSON response from questions endpoint: {e}")
         print(f"Response text: {response.text[:500]}")
         return f"Error decoding server response for questions: {e}", None
    except Exception as e:
        print(f"An unexpected error occurred fetching questions: {e}")
        return f"An unexpected error occurred fetching questions: {e}", None

    # 3. Run your Agent
    results_log = []
    answers_payload = []
    print(f"Running agent on {len(questions_data)} questions...")
    
    # Progress tracking
    progress_count = 0
    total_questions = len(questions_data)
    
    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:
            print(f"Skipping item with missing task_id or question: {item}")
            continue
            
        # Update progress
        progress_count += 1
        print(f"Processing question {progress_count}/{total_questions}")
        
        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})

            break
            
            # Add a small delay between questions to help with rate limiting
            if progress_count < total_questions:  # Don't delay after the last question
                time.sleep(2)
        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. Submitting {len(answers_payload)} answers for user '{username}'..."
    print(status_update)

    # 5. Submit
    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 requests.exceptions.HTTPError as e:
        error_detail = f"Server responded with status {e.response.status_code}."
        try:
            error_json = e.response.json()
            error_detail += f" Detail: {error_json.get('detail', e.response.text)}"
        except requests.exceptions.JSONDecodeError:
            error_detail += f" Response: {e.response.text[:500]}"
        status_message = f"❌ Submission Failed: {error_detail}"
        print(status_message)
        results_df = pd.DataFrame(results_log)
        return status_message, results_df
    except requests.exceptions.Timeout:
        status_message = "❌ Submission Failed: The request timed out. Please try again."
        print(status_message)
        results_df = pd.DataFrame(results_log)
        return status_message, results_df
    except requests.exceptions.RequestException as e:
        status_message = f"❌ Submission Failed: Network error - {e}"
        print(status_message)
        results_df = pd.DataFrame(results_log)
        return status_message, results_df
    except Exception as e:
        status_message = f"❌ An unexpected error occurred during submission: {e}"
        print(status_message)
        results_df = pd.DataFrame(results_log)
        return status_message, results_df

def test_agent(question: str, provider: str):
    """Test the agent with a single question."""
    try:
        agent = BasicAgent(provider=provider)
        answer = agent(question)
        return f"Question: {question}\nAnswer: {answer}"
    except Exception as e:
        return f"Error testing agent: {e}"

# --- Build Gradio Interface using Blocks ---
with gr.Blocks(title="GAIA Agent Evaluator") as demo:
    gr.Markdown("# 🤖 GAIA Agent Evaluator")
    gr.Markdown(
        """
        This interface allows you to evaluate your agent against the GAIA benchmark questions.
        
        **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
        3. View your results and score in the output panel
        
        **For Testing:**
        Use the test section below to verify your agent works correctly with sample questions.
        """
    )
    
    with gr.Tab("Evaluation"):
        gr.Markdown("## 🚀 Run Full Evaluation")
        gr.LoginButton()
        
        with gr.Row():
            run_button = gr.Button("Run Evaluation & Submit All Answers", variant="primary")
        
        status_output = gr.Textbox(label="📊 Status / Submission Result", lines=8, 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]
        )
    
    with gr.Tab("Testing"):
        gr.Markdown("## 🧪 Test Your Agent")
        with gr.Row():
            with gr.Column():
                test_question = gr.Textbox(
                    label="Question",
                    placeholder="Enter a test question...",
                    value="What is 2+2?"
                )
                provider_choice = gr.Radio(
                    choices=["groq", "hf"],
                    value="groq",
                    label="Provider"
                )
                test_button = gr.Button("Test Agent")
            with gr.Column():
                test_output = gr.Textbox(label="Agent Response", lines=10, interactive=False)
        
        test_button.click(
            fn=test_agent,
            inputs=[test_question, provider_choice],
            outputs=test_output
        )

if __name__ == "__main__":
    print("\n" + "="*50)
    print("🚀 GAIA Agent Evaluator Starting")
    print("="*50)
    
    # 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")
    
    if space_host_startup:
        print(f"✅ SPACE_HOST found: {space_host_startup}")
        print(f"   Runtime URL: https://{space_host_startup}.hf.space")
    else:
        print("ℹ️  Running locally (SPACE_HOST not found)")
    
    if space_id_startup:
        print(f"✅ SPACE_ID found: {space_id_startup}")
        print(f"   Repo URL: https://huggingface.co/spaces/{space_id_startup}")
    else:
        print("ℹ️  SPACE_ID not found (Repo URL cannot be determined)")
    
    print("="*50)
    print("Launching Gradio Interface...")
    demo.launch(debug=True, share=False)