scamrakshak_agents.py

import gradio as gr
from groq import Groq
import json
import requests
from datetime import datetime
import logging
import os
from typing import Dict, List, Optional
import time
from googlesearch import search
import threading
import queue
import colorama
from colorama import Fore, Style
import random
import pandas as pd
import csv
from PIL import Image
from io import BytesIO
from selenium import webdriver
from selenium.webdriver.common.by import By
import pytesseract

# Initialize colorama for colored console output
colorama.init()

# Configure logging
logging.basicConfig(
    level=logging.INFO,
    format='%(asctime)s | %(levelname)s | %(message)s',
    handlers=[
        logging.StreamHandler(),
        logging.FileHandler('agent_chat.log')
    ]
)

# Initialize Groq client
GROQ_API_KEY = "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"
client = Groq(api_key=GROQ_API_KEY)

def google_search(query: str, num_results: int = 5) -> List[str]:
    """Perform a Google search and return results"""
    try:
        search_results = []
        for result in search(query, stop=num_results):
            search_results.append(result)
        return search_results
    except Exception as e:
        logging.error(f"Google search error: {str(e)}")
        return []

class ConversationManager:
    def __init__(self):
        self.markdown_file = "conversation_history.md"
        self.text_file = "conversation_history.txt"
        self.current_session = datetime.now().strftime("%Y-%m-%d %H:%M:%S")
        
    def log_conversation(self, message: str, agent: str, is_task: bool = False):
        """Log conversation to both markdown and text files"""
        # Log to markdown file
        with open(self.markdown_file, "a", encoding="utf-8") as f:
            if not os.path.getsize(self.markdown_file):
                f.write(f"# Scamrakshak Team Conversations\n\n")
            
            if is_task:
                f.write(f"\n### Task Assignment ({self.current_session})\n")
                f.write(f"**From CEO to {agent}**:\n")
                f.write(f"```\n{message}\n```\n")
            else:
                f.write(f"\n### {agent} Response ({self.current_session})\n")
                f.write(f"{message}\n")
            
            f.write("\n---\n")
        
        # Log to text file
        with open(self.text_file, "a", encoding="utf-8") as f:
            if not os.path.getsize(self.text_file):
                f.write("=== SCAMRAKSHAK TEAM CONVERSATIONS ===\n\n")
            
            timestamp = datetime.now().strftime("%Y-%m-%d %H:%M:%S")
            
            if is_task:
                f.write(f"\n[{timestamp}] TASK ASSIGNMENT\n")
                f.write(f"From: CEO\n")
                f.write(f"To: {agent}\n")
                f.write(f"Task: {message}\n")
            else:
                f.write(f"\n[{timestamp}] {agent} RESPONSE\n")
                f.write(f"{message}\n")
            
            f.write("\n" + "="*50 + "\n")

class Agent:
    def __init__(self, name: str, role: str, system_prompt: str, conversation_manager: ConversationManager):
        self.name = name
        self.role = role
        self.system_prompt = system_prompt
        self.conversation_manager = conversation_manager
        self.conversation_history: List[Dict] = []
        self.task_queue = queue.Queue()
        self.research_results = {}
        self.detection_running = False
        self.stop_requested = False
        
    def get_response(self, user_input: str, from_agent: str = None) -> str:
        # First check for scam detection commands
        scam_detection_response = self.handle_scam_detection(user_input)
        if scam_detection_response:
            return scam_detection_response
        
        # Continue with normal response processing
        try:
            # Add context about who is sending the message
            sender_context = f"Message from {from_agent}: " if from_agent else ""
            
            # Perform research if needed
            research_results = []
            if "research" in user_input.lower() or "search" in user_input.lower():
                research_results = google_search(user_input)
                research_context = "\n\nResearch results:\n" + "\n".join(research_results)
            else:
                research_context = ""
            
            # Prepare messages including conversation history
            messages = [{"role": "system", "content": self.system_prompt}]
            messages.extend(self.conversation_history)
            messages.append({
                "role": "user", 
                "content": f"{sender_context}{user_input}{research_context}"
            })
            
            # Get response from Groq
            chat_completion = client.chat.completions.create(
                messages=messages,
                model="llama-3.2-90b-text-preview",
                temperature=0.7,
                max_tokens=1000
            )
            
            response = chat_completion.choices[0].message.content
            
            # Log the response
            self.conversation_manager.log_conversation(
                response,
                self.name,
                is_task=False
            )
            
            # Update conversation history
            self.conversation_history.append({"role": "user", "content": user_input})
            self.conversation_history.append({"role": "assistant", "content": response})
            
            # Keep only last 10 messages to prevent context length issues
            if len(self.conversation_history) > 10:
                self.conversation_history = self.conversation_history[-10:]
                
            return f"{self.name}: {response}"
            
        except Exception as e:
            logging.error(f"Error getting response from {self.name}: {str(e)}")
            return f"Error: Could not get response from {self.name}. Please try again."
    
    def assign_task(self, task: str, from_agent: str):
        """Add a task to the agent's queue"""
        self.task_queue.put((task, from_agent))
        self.conversation_manager.log_conversation(
            task,
            self.name,
            is_task=True
        )
    
    def process_task(self) -> Optional[str]:
        """Process the next task in the queue"""
        if not self.task_queue.empty():
            task, from_agent = self.task_queue.get()
            response = self.get_response(task, from_agent)
            return response
        return None

    def log_communication(self, message: str, from_agent: str = None, to_agent: str = None):
        """Log communication between agents"""
        timestamp = datetime.now().strftime("%Y-%m-%d %H:%M:%S")
        if from_agent and to_agent:
            print(f"{Fore.YELLOW}[{timestamp}] {Fore.GREEN}{from_agent} → {to_agent}{Fore.WHITE}: {message}{Style.RESET_ALL}")
        elif from_agent:
            print(f"{Fore.YELLOW}[{timestamp}] {Fore.BLUE}{from_agent}{Fore.WHITE}: {message}{Style.RESET_ALL}")
        else:
            print(f"{Fore.YELLOW}[{timestamp}]{Fore.WHITE}: {message}{Style.RESET_ALL}")

    def handle_scam_detection(self, message: str) -> str:
        """Handle scam detection commands"""
        if message.lower() == "scam_status":
            try:
                cumulative_path = os.path.join('data', 'reports', 'cumulative_analysis.txt')
                if not os.path.exists(cumulative_path):
                    return f"{self.name}: No analysis data available yet. Start detection with 'scam_detect'."
                
                with open(cumulative_path, 'r', encoding='utf-8') as f:
                    analysis = f.read()
                
                return f"{self.name}: Current Analysis Report:\n\n{analysis}"
                
            except Exception as e:
                return f"{self.name}: Error reading analysis data: {str(e)}"
        
        if "scam_detect" in message.lower():
            if self.detection_running:
                return f"{self.name}: Scam detection is already running. Use 'stop_detect' to stop it."
            
            try:
                self.detection_running = True
                self.stop_requested = False
                self.log_communication("Initializing scam detection process...", self.name)
                
                # Create necessary directories
                directories = ['data/images', 'data/texts', 'data/reports']
                for directory in directories:
                    os.makedirs(directory, exist_ok=True)
                
                # Start detection in background
                def run_detection():
                    try:
                        self.log_communication("Starting image scraping...", self.name)
                        image_urls = self.scrape_scam_images()
                        
                        if image_urls and not self.stop_requested:
                            self.log_communication(f"Found {len(image_urls)} images. Processing...", self.name)
                            self.process_scam_images(image_urls)
                        
                        # Clean up images
                        images_dir = os.path.join('data', 'images')
                        if os.path.exists(images_dir):
                            import shutil
                            shutil.rmtree(images_dir)
                            os.makedirs(images_dir)
                        
                        self.detection_running = False
                        if self.stop_requested:
                            self.log_communication("Scam detection stopped by user.", self.name)
                        else:
                            self.log_communication("Scam detection completed and images cleaned up.", self.name)
                        
                    except Exception as e:
                        self.detection_running = False
                        self.log_communication(f"Error in scam detection: {str(e)}", self.name)
                
                # Start detection in background thread
                import threading
                detection_thread = threading.Thread(target=run_detection)
                detection_thread.start()
                
                return f"{self.name}: I've initiated the scam detection process. Use 'stop_detect' to stop or 'scam_detect status' to check status."
                
            except Exception as e:
                self.detection_running = False
                return f"{self.name}: Error starting scam detection: {str(e)}"
        
        elif message.lower() == "stop_detect":
            if not self.detection_running:
                return f"{self.name}: No scam detection process is currently running."
            
            self.stop_requested = True
            return f"{self.name}: Stopping scam detection process... This may take a moment to clean up."
        
        elif "scam_detect status" in message.lower():
            return self.get_scam_detection_status()
        
        return None

    def scrape_scam_images(self):
        """Scrape images from Bing"""
        chrome_options = webdriver.ChromeOptions()
        chrome_options.add_argument('--headless')
        chrome_options.add_argument('--no-sandbox')
        chrome_options.add_argument('--disable-dev-shm-usage')

        driver = webdriver.Chrome(options=chrome_options)
        image_urls = []

        try:
            search_query = "indian scam sms"
            encoded_query = search_query.replace(' ', '+')
            driver.get(f"https://www.bing.com/images/search?q={encoded_query}")

            self.log_communication("Loading images...", self.name)
            time.sleep(3)

            for i in range(5):
                driver.execute_script("window.scrollTo(0, document.body.scrollHeight);")
                time.sleep(2)
                self.log_communication(f"Scroll {i+1}/5 completed", self.name)

            selectors = [".mimg", ".iusc"]
            for selector in selectors:
                elements = driver.find_elements(By.CSS_SELECTOR, selector)
                for element in elements:
                    try:
                        if selector == ".mimg":
                            url = element.get_attribute('src')
                        else:
                            m = element.get_attribute('m')
                            if m:
                                m_json = json.loads(m)
                                url = m_json.get('murl')
                            else:
                                continue

                        if url and url.startswith('http') and url not in image_urls:
                            image_urls.append(url)
                    except Exception as e:
                        self.log_communication(f"Error getting URL: {str(e)}", self.name)

            return image_urls

        finally:
            driver.quit()

    def process_scam_images(self, image_urls):
        """Process scraped images with OCR, Groq formatting, prediction, and storage"""
        timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
        last_report_time = datetime.now()
        stats = {
            'total_processed': 0,
            'scam_count': 0,
            'ham_count': 0,
            'failed_count': 0
        }

        try:
            self.log_communication(f"Starting to process {len(image_urls)} images...", self.name)
            
            for i, url in enumerate(image_urls, 1):
                if self.stop_requested:
                    self.log_communication("Stopping image processing as requested...", self.name)
                    break
                    
                try:
                    self.log_communication(f"Processing image {i}/{len(image_urls)}", self.name)
                    
                    # Download and process image
                    response = requests.get(url, timeout=10)
                    img = Image.open(BytesIO(response.content))
                    
                    # Save image temporarily with proper path tracking
                    img_filename = f"image_{timestamp}_{i}.png"
                    img_path = os.path.join('data', 'images', img_filename)
                    img.save(img_path)
                    
                    # Extract text using OCR
                    text = pytesseract.image_to_string(img)
                    if text.strip():
                        # Format text using Groq
                        try:
                            prompt = f"""
                            Format the following extracted text from an SMS image.
                            Keep the original content intact but improve the formatting and remove any OCR artifacts:

                            {text.strip()}
                            """
                            
                            completion = client.chat.completions.create(
                                messages=[{"role": "user", "content": prompt}],
                                model="llama-3.2-90b-text-preview",
                                temperature=0.3,
                                max_tokens=1024
                            )
                            
                            formatted_text = completion.choices[0].message.content.strip()
                            
                            # Send formatted text to prediction API
                            if formatted_text:
                                try:
                                    prediction_response = requests.post(
                                        "https://varun324242-sssssss.hf.space/predict",
                                        json={"message": formatted_text},
                                        timeout=30
                                    )
                                    prediction_response.raise_for_status()
                                    prediction = prediction_response.json().get("predicted_result", "unknown")
                                    
                                    # Update stats
                                    stats['total_processed'] += 1
                                    if prediction == "scam":
                                        stats['scam_count'] += 1
                                    elif prediction == "ham":
                                        stats['ham_count'] += 1
                                    
                                    # Store message with prediction
                                    message_data = [{
                                        'message': formatted_text,
                                        'prediction': prediction,
                                        'timestamp': datetime.now().strftime('%Y-%m-%d %H:%M:%S')
                                    }]
                                    self.update_scam_csv(message_data)
                                    self.log_communication(f"Message from image {i} processed and stored (Prediction: {prediction})", self.name)
                                    
                                except Exception as e:
                                    self.log_communication(f"Prediction API error for image {i}: {str(e)}", self.name)
                                    stats['failed_count'] += 1
                        
                        except Exception as e:
                            self.log_communication(f"Error formatting text with Groq for image {i}: {str(e)}", self.name)
                            stats['failed_count'] += 1
                    
                    # Generate analysis report every 30 seconds
                    if (datetime.now() - last_report_time).total_seconds() >= 30:
                        self.generate_analysis_report(stats, is_final=False)
                        last_report_time = datetime.now()
                    
                except Exception as e:
                    stats['failed_count'] += 1
                    self.log_communication(f"Error processing image {i}: {str(e)}", self.name)
                    continue
                
                finally:
                    # Delete processed image
                    try:
                        if os.path.exists(img_path):
                            os.remove(img_path)
                            self.log_communication(f"Deleted image: {img_filename}", self.name)
                    except Exception as e:
                        self.log_communication(f"Error deleting image {img_filename}: {str(e)}", self.name)

            # Generate final analysis report
            if not self.stop_requested:
                self.generate_analysis_report(stats, is_final=True)

        except Exception as e:
            self.log_communication(f"Critical error in image processing: {str(e)}", self.name)
        
        finally:
            # Clean up images directory
            images_dir = os.path.join('data', 'images')
            try:
                if os.path.exists(images_dir):
                    import shutil
                    shutil.rmtree(images_dir)
                    os.makedirs(images_dir)
                    self.log_communication("Images directory cleaned successfully", self.name)
            except Exception as e:
                self.log_communication(f"Error cleaning images directory: {str(e)}", self.name)

    def update_scam_csv(self, new_data):
        """Update scam123.csv immediately with new messages"""
        csv_path = os.path.join('data', 'scam123.csv')
        
        try:
            # Read existing messages
            existing_messages = set()
            if os.path.exists(csv_path):
                with open(csv_path, 'r', encoding='utf-8') as f:
                    reader = csv.DictReader(f)
                    existing_messages = {row['message'] for row in reader}

            # Add new messages
            messages_added = 0
            for item in new_data:
                message = item.get('message', '').strip()
                if message and message not in existing_messages:
                    existing_messages.add(message)
                    messages_added += 1

            # Write all messages to CSV
            with open(csv_path, 'w', encoding='utf-8', newline='') as f:
                writer = csv.DictWriter(f, fieldnames=['message'])
                writer.writeheader()
                for message in existing_messages:
                    writer.writerow({'message': message})

            # Create backup
            backup_path = os.path.join('data', 'backups', f'scam123_backup_{datetime.now().strftime("%Y%m%d_%H%M%S")}.csv')
            os.makedirs(os.path.join('data', 'backups'), exist_ok=True)
            with open(backup_path, 'w', encoding='utf-8', newline='') as f:
                writer = csv.DictWriter(f, fieldnames=['message'])
                writer.writeheader()
                for message in existing_messages:
                    writer.writerow({'message': message})

            if messages_added > 0:
                self.log_communication(f"Added {messages_added} new messages to scam123.csv", self.name)

        except Exception as e:
            self.log_communication(f"Error updating CSV: {str(e)}", self.name)

    def generate_analysis_report(self, stats, is_final=False):
        """Generate cumulative analysis report"""
        try:
            timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
            report_path = os.path.join('data', 'reports', f'analysis_report_{timestamp}.txt')
            cumulative_path = os.path.join('data', 'reports', 'cumulative_analysis.txt')
            
            # Read previous cumulative stats if exists
            cumulative_stats = {
                'total_processed': 0,
                'scam_count': 0,
                'ham_count': 0,
                'failed_count': 0,
                'last_update': None
            }
            
            if os.path.exists(cumulative_path):
                with open(cumulative_path, 'r', encoding='utf-8') as f:
                    for line in f:
                        if 'Total Messages Processed:' in line:
                            cumulative_stats['total_processed'] = int(line.split(':')[1].strip())
                        elif 'Scam Messages Detected:' in line:
                            cumulative_stats['scam_count'] = int(line.split(':')[1].strip())
                        elif 'Ham Messages Detected:' in line:
                            cumulative_stats['ham_count'] = int(line.split(':')[1].strip())
                        elif 'Failed Processing:' in line:
                            cumulative_stats['failed_count'] = int(line.split(':')[1].strip())
            
            # Update cumulative stats
            cumulative_stats['total_processed'] += stats['total_processed']
            cumulative_stats['scam_count'] += stats['scam_count']
            cumulative_stats['ham_count'] += stats['ham_count']
            cumulative_stats['failed_count'] += stats['failed_count']
            cumulative_stats['last_update'] = datetime.now()
            
            # Write current analysis report
            with open(report_path, 'w', encoding='utf-8') as f:
                f.write(f"Scam Detection Analysis Report\n")
                f.write(f"Generated at: {datetime.now().strftime('%Y-%m-%d %H:%M:%S')}\n")
                f.write(f"{'=' * 50}\n\n")
                
                f.write("Current Session Statistics:\n")
                f.write(f"Total Messages Processed: {stats['total_processed']}\n")
                f.write(f"Scam Messages Detected: {stats['scam_count']}\n")
                f.write(f"Ham Messages Detected: {stats['ham_count']}\n")
                f.write(f"Failed Processing: {stats['failed_count']}\n\n")
                
                f.write("Cumulative Statistics:\n")
                f.write(f"Total Messages Processed: {cumulative_stats['total_processed']}\n")
                f.write(f"Scam Messages Detected: {cumulative_stats['scam_count']}\n")
                f.write(f"Ham Messages Detected: {cumulative_stats['ham_count']}\n")
                f.write(f"Failed Processing: {cumulative_stats['failed_count']}\n\n")
                
                if cumulative_stats['total_processed'] > 0:
                    scam_percentage = (cumulative_stats['scam_count'] / cumulative_stats['total_processed']) * 100
                    ham_percentage = (cumulative_stats['ham_count'] / cumulative_stats['total_processed']) * 100
                    f.write("Analysis:\n")
                    f.write(f"Scam Percentage: {scam_percentage:.2f}%\n")
                    f.write(f"Ham Percentage: {ham_percentage:.2f}%\n\n")
                
                if is_final:
                    f.write("\nFinal Status:\n")
                    f.write("Processing completed successfully\n")
            
            # Update cumulative analysis file
            with open(cumulative_path, 'w', encoding='utf-8') as f:
                f.write(f"Cumulative Scam Detection Analysis\n")
                f.write(f"Last Updated: {cumulative_stats['last_update'].strftime('%Y-%m-%d %H:%M:%S')}\n")
                f.write(f"{'=' * 50}\n\n")
                f.write(f"Total Messages Processed: {cumulative_stats['total_processed']}\n")
                f.write(f"Scam Messages Detected: {cumulative_stats['scam_count']}\n")
                f.write(f"Ham Messages Detected: {cumulative_stats['ham_count']}\n")
                f.write(f"Failed Processing: {cumulative_stats['failed_count']}\n\n")
                
                if cumulative_stats['total_processed'] > 0:
                    scam_percentage = (cumulative_stats['scam_count'] / cumulative_stats['total_processed']) * 100
                    ham_percentage = (cumulative_stats['ham_count'] / cumulative_stats['total_processed']) * 100
                    f.write("Overall Analysis:\n")
                    f.write(f"Scam Percentage: {scam_percentage:.2f}%\n")
                    f.write(f"Ham Percentage: {ham_percentage:.2f}%\n")

            self.log_communication(
                f"Analysis reports generated:\n"
                f"Current: {report_path}\n"
                f"Cumulative: {cumulative_path}", 
                self.name
            )

        except Exception as e:
            self.log_communication(f"Error generating analysis report: {str(e)}", self.name)

    def get_scam_detection_status(self) -> str:
        """Get current status of scam detection"""
        try:
            # Check scam123.csv
            csv_file = os.path.join('data', 'scam123.csv')
            
            if not os.path.exists(csv_file):
                return f"{self.name}: No scam detection data available yet. Process hasn't started or no messages detected."
            
            try:
                # Read CSV and get statistics
                with open(csv_file, 'r', encoding='utf-8') as f:
                    reader = csv.DictReader(f)
                    messages = list(reader)
                    total_messages = len(messages)
                    
                    status = (
                        f"{self.name}: Current Scam Detection Status:\n"
                        f"Total Messages Collected: {total_messages}\n"
                    )
                    
                    # Check if process is currently running
                    images_dir = os.path.join('data', 'images')
                    if os.path.exists(images_dir) and len(os.listdir(images_dir)) > 0:
                        status += "\nStatus: RUNNING - Currently processing images..."
                    else:
                        status += "\nStatus: IDLE - Waiting for new detection run"
                    
                    # Add last update time if file exists
                    if total_messages > 0:
                        file_modified_time = datetime.fromtimestamp(os.path.getmtime(csv_file))
                        status += f"\nLast Updated: {file_modified_time.strftime('%Y-%m-%d %H:%M:%S')}"
                    
                    return status
                    
            except Exception as e:
                return f"{self.name}: Error reading scam detection data: {str(e)}"
                
        except Exception as e:
            return f"{self.name}: Error checking status: {str(e)}"

# Initialize agents with their roles and prompts
ceo_agent = Agent(
    name="CEO",
    role="Executive",
    system_prompt="""You are the CEO of Scamrakshak, a company dedicated to protecting users from digital scams. 
    You can delegate tasks to the Tech Support and Research teams.
    When given a task about technical implementation or research:
    1. Break it down into specific sub-tasks
    2. Assign appropriate tasks to Tech Support and Research teams
    3. Synthesize their responses into a comprehensive plan
    4. Provide strategic oversight and direction
    
    Format task assignments as: "TASK FOR [AGENT]: [specific task description]"
    """,
    conversation_manager=ConversationManager()
)

tech_support_agent = Agent(
    name="Tech Support",
    role="Support",
    system_prompt="""You are Scamrakshak's Technical Support specialist.
    When assigned tasks by the CEO:
    1. Analyze technical requirements
    2. Provide detailed implementation steps
    3. Consider security implications
    4. Suggest best practices and potential challenges
    5. Research technical solutions using available resources
    
    Focus on practical, secure, and efficient solutions.
    Always consider Android best practices and security guidelines.
    """,
    conversation_manager=ConversationManager()
)

researcher_agent = Agent(
    name="Researcher",
    role="Analyst",
    system_prompt="""You are Scamrakshak's Research Analyst specializing in scam trends and prevention.
    When assigned tasks by the CEO:
    1. Research current trends and solutions
    2. Analyze market data and competitor approaches
    3. Provide data-backed recommendations
    4. Consider regulatory and compliance aspects
    5. Identify potential risks and opportunities
    
    Use research results to provide comprehensive analysis.
    Focus on actionable insights and industry best practices.
    """,
    conversation_manager=ConversationManager()
)

class AgentSystem:
    def __init__(self):
        self.agents = {
            "CEO": ceo_agent,
            "Tech Support": tech_support_agent,
            "Researcher": researcher_agent
        }
        self.current_agent = "CEO"
        self.conversation_log = []
        
    def switch_agent(self, agent_name: str) -> str:
        if agent_name in self.agents:
            self.current_agent = agent_name
            return f"Switched to {agent_name}"
        return f"Invalid agent name. Available agents: {', '.join(self.agents.keys())}"
    
    def process_task_chain(self, initial_input: str) -> List[str]:
        """Process a task through multiple agents"""
        responses = []
        
        # Log initial request
        print(f"\n{Fore.CYAN}=== Starting New Task Chain ==={Style.RESET_ALL}")
        self.agents["CEO"].log_communication(initial_input, "User")
        
        # CEO processes initial request
        print(f"\n{Fore.CYAN}=== CEO Analyzing Request ==={Style.RESET_ALL}")
        ceo_response = self.agents["CEO"].get_response(initial_input)
        responses.append(ceo_response)
        
        # Extract and process tasks immediately
        tasks_found = False
        for line in ceo_response.split('\n'):
            if "TASK FOR" in line:
                tasks_found = True
                target_agent = line.split("TASK FOR")[1].split(":")[0].strip()
                task = line.split(":", 1)[1].strip()
                
                if target_agent.upper() == "RESEARCH TEAM":
                    target_agent = "Researcher"  # Map to correct agent name
                elif target_agent.upper() == "TECH SUPPORT TEAM":
                    target_agent = "Tech Support"  # Map to correct agent name
                
                if target_agent in self.agents:
                    print(f"\n{Fore.CYAN}=== {target_agent} Processing Task ==={Style.RESET_ALL}")
                    # Assign and process task immediately
                    self.agents[target_agent].assign_task(task, "CEO")
                    response = self.agents[target_agent].process_task()
                    if response:
                        responses.append(response)
                        print(f"\n{Fore.GREEN}=== {target_agent} Task Complete ==={Style.RESET_ALL}")
        
        if tasks_found:
            # CEO synthesizes all responses
            print(f"\n{Fore.CYAN}=== CEO Synthesizing All Responses ==={Style.RESET_ALL}")
            synthesis_prompt = (
                "Based on the research team and tech support findings above, "
                "provide a comprehensive summary and strategic recommendations. "
                "Include specific action items and next steps."
            )
            final_response = self.agents["CEO"].get_response(synthesis_prompt)
            responses.append(final_response)
        else:
            print(f"{Fore.RED}No tasks were delegated in the CEO's response{Style.RESET_ALL}")
        
        print(f"\n{Fore.CYAN}=== Task Chain Complete ==={Style.RESET_ALL}\n")
        return responses
    
    def get_response(self, user_input: str) -> str:
        if "implement" in user_input.lower() or "research" in user_input.lower():
            # Process as a task chain
            responses = self.process_task_chain(user_input)
            return "\n\n".join(responses)
        else:
            # Normal single-agent response
            self.agents[self.current_agent].log_communication(user_input, "User")
            response = self.agents[self.current_agent].get_response(user_input)
            return response

# Initialize agent system
agent_system = AgentSystem()

def chat_interface(message: str, history: List[List[str]]) -> str:
    """Handle chat interactions and agent responses"""
    
    print(f"\n{Fore.CYAN}=== New User Message ==={Style.RESET_ALL}")
    
    # Check for agent switch command
    if message.startswith("/switch"):
        try:
            _, agent_name = message.split(" ", 1)
            response = agent_system.switch_agent(agent_name)
            print(f"{Fore.YELLOW}[SYSTEM] {response}{Style.RESET_ALL}")
            return response
        except ValueError:
            error_msg = "Invalid switch command. Use: /switch [CEO|Tech Support|Researcher]"
            print(f"{Fore.RED}[ERROR] {error_msg}{Style.RESET_ALL}")
            return error_msg
    else:
        # Get response from current agent
        return agent_system.get_response(message)

# Add this new class for team chat
class TeamChat:
    def __init__(self, agents: Dict[str, Agent]):
        self.agents = agents
        self.is_active = False
        self.conversation_manager = ConversationManager()
        
    def process_team_message(self, message: str, from_role: str = "Founder") -> List[str]:
        """Process a message in team chat mode"""
        responses = []
        
        if from_role == "Founder":
            # CEO responds to founder's task
            ceo_prompt = f"As CEO, respond briefly to the founder's request: {message}. Keep it under 50 words and professional."
            ceo_response = self.agents["CEO"].get_response(ceo_prompt)
            responses.append(ceo_response)
            
            # CEO delegates if needed
            if "implement" in message.lower() or "research" in message.lower():
                delegation_prompt = f"Delegate this task briefly to team members: {message}. Keep each delegation under 30 words."
                delegation = self.agents["CEO"].get_response(delegation_prompt)
                responses.append(delegation)
                
                # Team members acknowledge
                for agent_name in ["Tech Support", "Researcher"]:
                    ack_prompt = f"Acknowledge the task briefly and professionally. Keep it under 20 words."
                    ack = self.agents[agent_name].get_response(ack_prompt)
                    responses.append(ack)
        else:
            # Normal team member response
            response_prompt = f"Respond briefly to the team chat message: {message}. Keep it under 30 words and professional."
            response = self.agents[from_role].get_response(response_prompt)
            responses.append(response)
            
        return responses

def create_interface():
    """Create and configure the Gradio interface"""
    
    with gr.Blocks(
        title="Scamrakshak AI Assistant",
        theme=gr.themes.Soft(),
        css="""
            .gradio-container { 
                font-family: 'Arial', sans-serif;
                max-width: 1000px;
                margin: auto;
            }
            .agent-status {
                padding: 1rem;
                margin: 1rem 0;
                border-radius: 0.5rem;
                background-color: #f8f9fa;
                border: 1px solid #dee2e6;
            }
            .agent-indicator {
                display: inline-block;
                padding: 0.25rem 0.5rem;
                border-radius: 0.25rem;
                margin-right: 0.5rem;
                font-weight: bold;
            }
            .ceo-color { background-color: #e3f2fd; color: #1565c0; }
            .tech-color { background-color: #f3e5f5; color: #7b1fa2; }
            .research-color { background-color: #e8f5e9; color: #2e7d32; }
            .chat-message { 
                padding: 1rem;
                margin: 0.5rem;
                border-radius: 0.5rem;
                border-left: 4px solid;
            }
            .ceo-message { border-left-color: #1565c0; }
            .tech-message { border-left-color: #7b1fa2; }
            .research-message { border-left-color: #2e7d32; }
            .user-message { border-left-color: #ff9800; }
            .task-delegation {
                background-color: #fff3e0;
                border: 1px solid #ffe0b2;
                padding: 0.5rem;
                margin: 0.5rem 0;
                border-radius: 0.25rem;
            }
        """
    ) as interface:
        
        with gr.Row():
            gr.Markdown("""
            # 🤖 Scamrakshak AI Assistant
            
            An advanced AI system with three specialized agents working together to protect you from scams.
            """)
        
        # Agent Status Panel
        with gr.Row() as agent_status:
            with gr.Column(scale=1):
                gr.Markdown("""
                ### Active Agents
                """)
                
                with gr.Group(elem_classes="agent-status"):
                    current_agent = gr.Textbox(
                        label="Current Active Agent",
                        value="CEO",
                        interactive=False,
                        elem_classes="agent-indicator ceo-color"
                    )
                    
                    gr.Markdown("""
                    #### Available Agents:
                    - 👔 **CEO** - Strategic oversight and task delegation
                    - 🛠️ **Tech Support** - Technical implementation and security
                    - 🔍 **Researcher** - Trend analysis and market research
                    
                    Use `/switch [agent]` to change agents
                    """)
        
        # Main Chat Interface
        with gr.Row():
            with gr.Column(scale=3):
                chatbot = gr.Chatbot(
                    label="Conversation",
                    height=600,
                    container=True,
                    show_label=True,
                    elem_id="chatbot"
                )
                
                with gr.Row():
                    with gr.Column(scale=4):
                        msg = gr.Textbox(
                            label="Your message",
                            placeholder="Ask a question or use /team to start team chat...",
                            lines=2,
                            show_label=True,
                            container=True
                        )
                    with gr.Column(scale=1):
                        with gr.Row():
                            send = gr.Button("Send", variant="primary")
                            clear = gr.Button("Clear", variant="stop")
            
                # Add team chat controls
                with gr.Row():
                    team_chat_active = gr.Checkbox(
                        label="Team Chat Mode",
                        value=False,
                        interactive=True
                    )
                    current_role = gr.Dropdown(
                        choices=["Founder", "CEO", "Tech Support", "Researcher"],
                        value="Founder",
                        label="Speaking As",
                        interactive=True
                    )

        # Message handling functions remain the same
        def user_message(message: str, history: List[List[str]], is_team_chat: bool, role: str) -> tuple[List[List[str]], str]:
            if message.strip() == "":
                return history, ""
            
            if message.startswith("/team"):
                is_team_chat = True
                return history, ""
            
            if is_team_chat:
                # Process team chat message
                team_chat = TeamChat(agent_system.agents)
                responses = team_chat.process_team_message(message, role)
                
                # Format team chat messages
                history.append([
                    f'<div class="team-chat-message {role.lower()}-message">{role}: {message}</div>',
                    ""
                ])
                
                for response in responses:
                    agent = response.split(":")[0]
                    content = response.split(":", 1)[1]
                    history.append([
                        "",
                        f'<div class="team-chat-message {agent.lower()}-message">{response}</div>'
                    ])
            else:
                # Normal chat processing
                response = chat_interface(message, history)
                history.append([
                    f'<div class="user-message">{message}</div>',
                    response
                ])
            
            return history, ""

        # Connect interface elements
        msg.submit(
            user_message,
            [msg, chatbot, team_chat_active, current_role],
            [chatbot, msg]
        )
        send.click(
            user_message,
            [msg, chatbot, team_chat_active, current_role],
            [chatbot, msg]
        )
        clear.click(lambda: ([], ""), None, [chatbot, msg])
        
        # Update current agent display
        def update_current_agent(message: str) -> str:
            if message.startswith("/switch"):
                try:
                    _, agent_name = message.split(" ", 1)
                    if agent_name in ["CEO", "Tech Support", "Researcher"]:
                        return agent_name
                except:
                    pass
            return current_agent.value
        
        msg.submit(update_current_agent, [msg], [current_agent])
        send.click(update_current_agent, [msg], [current_agent])

        # Add Team Chat section
        with gr.Tab("Team Chat"):
            with gr.Column():
                gr.Markdown("""
                # 👥 Team Chat Room
                Watch the Scamrakshak team have spontaneous work discussions!
                """)
                
                team_chat_box = gr.Chatbot(
                    label="Team Discussion",
                    height=400
                )
                
                start_discussion = gr.Button("Start New Team Discussion", variant="primary")
                
                def trigger_team_discussion() -> List[List[str]]:
                    team_chat = TeamChat(agent_system.agents)
                    discussion = team_chat.start_team_discussion()
                    formatted_discussion = []
                    for msg in discussion:
                        agent = msg.split(":")[0]
                        content = msg.split(":", 1)[1]
                        formatted_discussion.append([
                            "",
                            f'<div class="{agent.lower()}-message">{msg}</div>'
                        ])
                    return formatted_discussion
                
                start_discussion.click(
                    trigger_team_discussion,
                    outputs=[team_chat_box]
                )
                
                gr.Markdown("""
                ### About Team Chat
                - Team members spontaneously discuss work-related topics
                - Discussions are focused on improving Scamrakshak's services
                - Watch how different team members contribute their expertise
                - Topics include security, features, market trends, and more
                """)

    return interface

if __name__ == "__main__":
    # Create and launch the interface
    demo = create_interface()
    demo.queue()  # Enable queuing for better handling of multiple requests
    demo.launch(
        server_name="0.0.0.0",
        server_port=7860,
        share=True,
        show_error=True,
        show_api=False
    )