app.py

import os
import gradio as gr
import requests
import inspect
import pandas as pd
import smolagents
import traceback
from smolagents import DuckDuckGoSearchTool, VisitWebpageTool
import time 
from functools import lru_cache  

import google.generativeai as genai
from google.generativeai.types import HarmCategory, HarmBlockThreshold

from youtube_transcript_api import YouTubeTranscriptApi
import re

GOOGLE_API_KEY = os.getenv("GOOGLE_API_KEY") 
genai.configure(api_key=GOOGLE_API_KEY)

class YouTubeVideoTool:
    def __init__(self):
        self.name = "youtube_video_tool"
        
    def __call__(self, query):
        """
        Extract information from a YouTube video.
        
        Args:
            query: Either a YouTube URL or video ID
        
        Returns:
            String with the transcript of the video
        """
        try:
            # Extract video ID from URL if needed
            video_id = self._extract_video_id(query)
            if not video_id:
                return "Could not extract a valid YouTube video ID"
                
            # Get the transcript
            transcript_list = YouTubeTranscriptApi.get_transcript(video_id)
            
            # Combine the transcript text
            transcript_text = " ".join([item['text'] for item in transcript_list])
            
            return f"Transcript from YouTube video {video_id}:\n{transcript_text}"
        except Exception as e:
            return f"Error processing YouTube video: {str(e)}"
    
    
    
    def _extract_video_id(self, url_or_id):
        """Extract YouTube video ID from various URL formats or return the ID if already provided."""
        # Handle direct video ID
        if len(url_or_id) == 11 and re.match(r'^[A-Za-z0-9_-]{11}$', url_or_id):
            return url_or_id
            
        # Common YouTube URL patterns
        patterns = [
            r'(?:youtube\.com\/watch\?v=|youtu\.be\/|youtube\.com\/embed\/|youtube\.com\/v\/)([A-Za-z0-9_-]{11})',
            r'youtube\.com\/watch\?.*v=([A-Za-z0-9_-]{11})',
            r'youtube\.com\/shorts\/([A-Za-z0-9_-]{11})'
        ]
        
        for pattern in patterns:
            match = re.search(pattern, url_or_id)
            if match:
                return match.group(1)
                
        return None        


# TOOLS
search_tool = DuckDuckGoSearchTool()
visit_webpage = VisitWebpageTool()
youtube_tool = YouTubeVideoTool()

# (Keep Constants as is)
# --- Constants ---
DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"

# Cache Wrapper
@lru_cache(maxsize=100)
def cached_search(query):
    try:
        print(f"Performing search for: {query[:50000]}...")  
        result = search_tool(query)
        print(f"Search successful, returned {len(result)} characters")  
        return result
    except Exception as e:
        print(f"Search error: {str(e)}")  
        return f"Search error: {str(e)}"  

# --- Basic Agent Definition ---
# ----- THIS IS WERE YOU CAN BUILD WHAT YOU WANT ------



        
class BasicAgent:
    
    
    def __init__(self, model=None, tools=None):
        self.model = model
        self.tools = tools if tools is not None else []
        self.history = []
        print(f"BasicAgent initialized with model: {model} and {len(self.tools)} tools.")
        

        if self.model and self.model.startswith('gemini'):
            try:
                self._init_gemini_model()
                print("Successfully initialized Gemini model")
            except Exception as e:
                print(f"Error initializing Gemini model: {e}")
                print("Will try again when needed")
                self.gemini_model = None
        else:
            self.gemini_model = None

    def _init_gemini_model(self):
        
        """Initialize the Gemini model with appropriate settings"""
        generation_config = {
            "temperature": 0.2,
            "top_p": 0.8,
            "top_k": 30,
            "max_output_tokens": 300000,
        }
        
        safety_settings = {
            HarmCategory.HARM_CATEGORY_HARASSMENT: HarmBlockThreshold.BLOCK_NONE,
            HarmCategory.HARM_CATEGORY_HATE_SPEECH: HarmBlockThreshold.BLOCK_NONE,
            HarmCategory.HARM_CATEGORY_SEXUALLY_EXPLICIT: HarmBlockThreshold.BLOCK_NONE,
            HarmCategory.HARM_CATEGORY_DANGEROUS_CONTENT: HarmBlockThreshold.BLOCK_NONE,
        }
        
        model_name = "gemini-pro"
        if "gemini-2.0" in self.model:
            model_name = "gemini-2.5-pro-exp-03-25"  
            
        self.gemini_model = genai.GenerativeModel(
            model_name=model_name,
            generation_config=generation_config,
            safety_settings=safety_settings
        )

    def __call__(self, question: str) -> str:
        print(f"Agent received question: {question[:500]}...")  
        try:
            final_answer = self.process_question(question)
            print(f"Agent returning answer: {final_answer[:500]}...")
            return final_answer
        except Exception as e:
            print(f"Agent error: {str(e)}")
            traceback.print_exc()  
            return f"I apologize, but I encountered an error while processing your question. Error: {str(e)}"

    def _process_image_query(self, question, image_url):
        try:
            response = requests.get(image_url)
            image_data = response.content
            multimodal_prompt = [{"text": question}, {"image": {"data": image_data}}]
            response = self.gemini_model.generate_content(multimodal_prompt)
            return response.text
        except Exception as e:
            print(f"Error processing image: {str(e)}")
            return f"I encountered an error analyzing the image: {str(e)}"

    def _calculate(self, expression):
        try:
            # Basic calculator using Python's eval with safety restrictions
            import math, re
            safe_dict = {k: v for k, v in math.__dict__.items() if not k.startswith('__')}
            # Only allow safe math operations
            if re.match(r'^[\d\s\+\-\*\/\(\)\.\,\^\%\w]+$', expression):
                result = eval(expression, {"__builtins__": {}}, safe_dict)
                return f"Calculation result: {result}"
            return "Invalid calculation expression"
        except Exception as e:
            return f"Calculation error: {str(e)}"

    def _read_csv(self, file_url):
        try:
            response = requests.get(file_url)
            import io, csv, pandas as pd
            df = pd.read_csv(io.StringIO(response.text))
            summary = f"CSV contains {len(df)} rows, {len(df.columns)} columns.\nColumns: {', '.join(df.columns)}\nSample data: {df.head(3).to_string()}"
            return summary
        except Exception as e:
            return f"CSV reading error: {str(e)}"
            
    def _read_excel(self, file_url):
        try:
            response = requests.get(file_url)
            import io, pandas as pd
            df = pd.read_excel(io.BytesIO(response.content))
            summary = f"Excel contains {len(df)} rows, {len(df.columns)} columns.\nColumns: {', '.join(df.columns)}\nSample data: {df.head(3).to_string()}"
            return summary
        except Exception as e:
            return f"Excel reading error: {str(e)}"


    def _read_code_file(self, file_url):
        try:
            response = requests.get(file_url)
            code_content = response.text
            
            # Determine file extension for language detection
            file_extension = file_url.split('.')[-1].lower()
            
            # Map common extensions to language names
            language_map = {
                'py': 'Python', 'js': 'JavaScript', 'html': 'HTML', 'css': 'CSS',
                'java': 'Java', 'c': 'C', 'cpp': 'C++', 'cs': 'C#',
                'php': 'PHP', 'rb': 'Ruby', 'go': 'Go', 'rs': 'Rust',
                'ts': 'TypeScript', 'sh': 'Shell', 'sql': 'SQL'
            }
            
            language = language_map.get(file_extension, 'Unknown')
            
            # Count lines of code
            line_count = code_content.count('\n') + 1
            
            # Create a summary of the code file
            summary = f"Code file ({language}) with {line_count} lines.\n"
            # Fixed line with proper quote handling
            summary += f"First 10 lines:\n" + "\n".join(code_content.split("\n")[:10])
            
            return summary
        except Exception as e:
            return f"Code file reading error: {str(e)}"

           
    def process_question(self, question: str) -> str:
        try:
            # Processing image
            image_url_match = re.search(r'https?://\S+\.(jpg|jpeg|png|gif|webp)', question, re.IGNORECASE)
            if image_url_match and self.gemini_model:
                image_url = image_url_match.group(0)
                return self._process_image_query(question, image_url)

            #Excel
            excel_url_match = re.search(r'https?://\S+\.(xlsx|xls|xlsm)', question, re.IGNORECASE)
            if excel_url_match:
                return self._read_excel(excel_url_match.group(0))

            # Code
            code_extensions = ['py', 'js', 'html', 'css', 'java', 'c', 'cpp', 'cs', 'php', 'rb', 'go', 'rs', 'ts', 'sh', 'sql']
            code_pattern = '|'.join(code_extensions)
            code_url_match = re.search(f'https?://\\S+\\.({code_pattern})', question, re.IGNORECASE)
            if code_url_match:
                return self._read_code_file(code_url_match.group(0))

            # Read csv
            if "calculate" in question.lower() and any(c in question for c in "+-*/"):
                calculation = re.search(r'calculate\s+([\d\s\+\-\*\/\(\)\.\,\^\%]+)', question, re.IGNORECASE)
                if calculation:
                    return self._calculate(calculation.group(1))
                        
            csv_url_match = re.search(r'https?://\S+\.csv', question, re.IGNORECASE)
            if csv_url_match:
                return self._read_csv(csv_url_match.group(0))
                
            # Check if this is a request about a YouTube video
            youtube_patterns = ["youtube.com", "youtu.be", "watch youtube", "youtube video"]
            use_youtube_tool = any(pattern in question.lower() for pattern in youtube_patterns)
            
            search_results = ""  
            youtube_info = ""    
            
            # Step 1: Gather information
            if use_youtube_tool and any(isinstance(tool, YouTubeVideoTool) for tool in self.tools):
                # Extract potential YouTube URL or ID
                url_match = re.search(r'(?:https?:\/\/)?(?:www\.)?(?:youtube\.com|youtu\.be)\/[^\s]+', question)
                youtube_url = url_match.group(0) if url_match else question
                
                print(f"Using YouTube tool with URL: {youtube_url}") 
                # Use YouTube tool
                youtube_tool_instance = next((tool for tool in self.tools if isinstance(tool, YouTubeVideoTool)), None)
                if youtube_tool_instance:
                    youtube_info = youtube_tool_instance(youtube_url)
                    print(f"YouTube info retrieved: {len(youtube_info)} characters")  
                
            #  Always search as backup or additional context
            if any(isinstance(tool, DuckDuckGoSearchTool) for tool in self.tools):
                search_results = cached_search(question)
                print(f"Search results: {len(search_results)} characters")  
            
            #  Determine what information to use
            if youtube_info and "Error processing YouTube video" not in youtube_info:
                primary_info = youtube_info
                print("Using YouTube info as primary source")  
            else:
                primary_info = search_results
                print("Using search results as primary source")  
            
            # Extract key information
            relevant_info = self._extract_key_info(primary_info, question)
            print(f"Extracted relevant info: {len(relevant_info)} characters")  
            
            # Formulate an answer
            return self._formulate_direct_answer(relevant_info, question)
            
        except Exception as e:
            print(f"Error in process_question: {str(e)}") 
            traceback.print_exc() 
            if "too many requests" in str(e).lower():
                time.sleep(2)
                try:
                    search_results = cached_search(question)
                    relevant_info = self._extract_key_info(search_results, question)
                    return self._formulate_direct_answer(relevant_info, question)
                except Exception as retry_error:
                    print(f"Error in retry: {str(retry_error)}") 
                    return self._get_fallback_answer(question)
            return self._get_fallback_answer(question)
    
    
    
    
    def _extract_key_info(self, search_results, question):
        # Basic check for empty results
        if not search_results or len(search_results) < 15:
            return "No relevant information found."
            
        # For YouTube transcripts, extract the most relevant portion
        if "Transcript from YouTube video" in search_results:
            # Split by sentences but keep limited context
            max_chars = 30000  # Keep a reasonable chunk size
            if len(search_results) > max_chars:
                # Take a portion from the middle of the transcript for better relevance
                start_idx = search_results.find("\n") + 1  # Skip the first line which is the header
                # Get content chunk
                return search_results[start_idx:start_idx+max_chars]
            return search_results
        
        # For search results
        # Split results into sentences and find most relevant
        sentences = search_results.split('. ')
        if len(sentences) <= 2000:
            return search_results[:50000]  
        
        # Try to find sentences with keywords from question
        keywords = [w for w in question.lower().split() if len(w) > 2]
        relevant_sentences = []  # NEW LINE
        
        for sentence in sentences:
            sentence_lower = sentence.lower()
            if any(keyword in sentence_lower for keyword in keywords):
                relevant_sentences.append(sentence)  
                if len(relevant_sentences) >= 10000:  
                    break  
        
        # If we found relevant sentences, use them
        if relevant_sentences:
            return '. '.join(relevant_sentences)
        
        # Fallback to first few sentences
        return '. '.join(sentences[:10000])  
    
    def _formulate_direct_answer(self, relevant_info, question):
  
        if not self.model:
            return f"{relevant_info}"
            
        if self.model.startswith('gemini'):
            try:
              
                if not hasattr(self, 'gemini_model') or self.gemini_model is None:
                    self._init_gemini_model()
                
            
                prompt = f"""
                   
                 
                You: You are a general AI assistant. I will ask you a question. YOUR FINAL ANSWER should be a number OR as few words as possible OR a comma separated list of numbers and/or strings. If you are asked for a number, don't use comma to write your number neither use units such as $ or percent sign unless specified otherwise. If you are asked for a string, don't use articles, neither abbreviations (e.g. for cities), and write the digits in plain text unless specified otherwise. If you are asked for a comma separated list, apply the above rules depending of whether the element to be put in the list is a number or a string.            
                Instructions:
                1. Read the question and think about what elements you need in order to answer it. If the question prompt does not contain all the elements, check the relevant information below. Pay close attention to the question instructions. Do not use the relevant information unless you need additional information from the question.  
                2. If the question is not comprehensible, try reading each letter backwards, from the last character in the last word, to the first letter of the first word. Read carefully, all the way to the very beginning of the question.  The text may be an instruction. Think about the instruction and follow it before providing the final answer. Don't provide comments. 
                3. If the question is still not comprehensible, try seeing if it is in another language. 
                4. Think about whether you need to elaborate on the information. For example, if you know that John and Jane are kids of Joan, you know Joan has at least two kids. In other words, if you don't have a number that is asked of you, see if you can count to produce an answer. Once you have counted, just answer the number. Be succinct, coesive, I would even say tight in your answers. If the question asks "how many?", just reply back the number that answers. In the example I just gave, you would answer: 2. Without quotes, of course.
                5. Provide a direct answer.
                6. If the information doesn't contain the answer, say so honestly.
                7. Do not invent anything. You can apply method to elaborate, but based on facts. Do not provide comments. Just the raw answer. Try hard to always answer it.  
                8. Format your response as a direct answer. For example, if you are asked the year in which World War II began, just reply: 1939
                9. Think thoroughly, but do not include your thoughts in your response. Only the final answer can be in your response. Your thoughts are important so that you can process the question. But do not share your thoughts.
                
                Question: {question}
    
                Relevant information: {relevant_info}
                """
                
                response = self.gemini_model.generate_content(prompt)
                if response and hasattr(response, 'text'):
                    return response.text
                else:
                    print("Gemini response was empty or invalid")  
                    return f"Based on the information: {relevant_info[:200]}..."
                    
            except Exception as e:
                print(f"Error using Gemini model: {e}") 
                traceback.print_exc()  
                return f"Based on the search: {relevant_info[:200]}..."
                
        return f"Based on the information: {relevant_info[:200]}..." 
  
    
    
    def _get_fallback_answer(self, question):
        return f"I cannot provide a specific answer to your question about {question.split()[0:3]}..."



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") # Get the SPACE_ID for sending link to the code

    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 ( modify this part to create your agent)
    try:
        agent = BasicAgent(model= 'gemini/gemini-2.0-flash-exp', tools=[search_tool, visit_webpage, youtube_tool])
    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 ( usefull for others so please keep it public)
    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=15)
        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...")
    for idx, 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
        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})
            if idx % 1 == 0 and idx > 0:  
                time.sleep(14)  
        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=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')}% "
            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."
        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


    
# --- Build Gradio Interface using Blocks ---
with gr.Blocks() as demo:
    gr.Markdown("# Basic Agent Evaluation Runner")
    gr.Markdown(
        """
        **Instructions:**

        1.  Please clone this space, then modify the code to define your agent's logic, the tools, the necessary packages, etc ...
        2.  Log in to your Hugging Face account using the button below. This uses your HF username for submission.
        3.  Click 'Run Evaluation & Submit All Answers' to fetch questions, run your agent, submit answers, and see the score.

        ---
        **Disclaimers:**
        Once clicking on the "submit button, it can take quite some time ( this is the time for the agent to go through all the questions).
        This space provides a basic setup and is intentionally sub-optimal to encourage you to develop your own, more robust solution. For instance for the delay process of the submit button, a solution could be to cache the answers and submit in a seperate action or even to answer the questions in async.
        """
    )

    gr.LoginButton()

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

    status_output = gr.Textbox(label="Run Status / Submission Result", lines=5, interactive=False)
    # Removed max_rows=10 from DataFrame constructor
    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__":
    print("\n" + "-"*30 + " App Starting " + "-"*30)
    # 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") # Get SPACE_ID at startup

    if space_host_startup:
        print(f"✅ SPACE_HOST found: {space_host_startup}")
        print(f"   Runtime URL should be: https://{space_host_startup}.hf.space")
    else:
        print("ℹ️  SPACE_HOST environment variable not found (running locally?).")

    if space_id_startup: # Print repo URLs if SPACE_ID is found
        print(f"✅ SPACE_ID found: {space_id_startup}")
        print(f"   Repo URL: https://huggingface.co/spaces/{space_id_startup}")
        print(f"   Repo Tree URL: https://huggingface.co/spaces/{space_id_startup}/tree/main")
    else:
        print("ℹ️  SPACE_ID environment variable not found (running locally?). Repo URL cannot be determined.")

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

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