from uniface import RetinaFace, ArcFace, compute_similarity
import gradio as gr
import cv2
import numpy as np
import pandas as pd
from datetime import datetime
import os
import pickle
from pathlib import Path
import warnings
warnings.filterwarnings('ignore')

# Initialize face detection and recognition models
detector = RetinaFace()
recognizer = ArcFace()

# Create necessary directories
DATA_DIR = Path("face_data")
ATTENDANCE_DIR = Path("attendance_records")
DATA_DIR.mkdir(exist_ok=True)
ATTENDANCE_DIR.mkdir(exist_ok=True)

# Database files
FACE_DB_FILE = DATA_DIR / "face_database.pkl"
USER_DB_FILE = DATA_DIR / "user_database.pkl"
ATTENDANCE_FILE = ATTENDANCE_DIR / f"attendance_{datetime.now().strftime('%Y_%m_%d')}.csv"

# Initialize databases
if FACE_DB_FILE.exists():
    with open(FACE_DB_FILE, 'rb') as f:
        face_database = pickle.load(f)
else:
    face_database = {}

if USER_DB_FILE.exists():
    with open(USER_DB_FILE, 'rb') as f:
        user_database = pickle.load(f)
else:
    user_database = {}

# Initialize attendance file
if not ATTENDANCE_FILE.exists():
    df_attendance = pd.DataFrame(columns=['Timestamp', 'User_ID', 'Name', 'Roll_No', 'Status'])
    df_attendance.to_csv(ATTENDANCE_FILE, index=False)

class FaceAttendanceSystem:
    def __init__(self):
        self.camera = None
        self.recognition_threshold = 0.4
        self.current_frame = None
        
    def start_camera(self):
        if self.camera is None or not self.camera.isOpened():
            self.camera = cv2.VideoCapture(0)
            if not self.camera.isOpened():
                return False, "Cannot open camera"
        return True, "Camera started successfully"
    
    def stop_camera(self):
        if self.camera is not None:
            self.camera.release()
            self.camera = None
        return "Camera stopped"
    
    def capture_frame(self):
        if self.camera is None or not self.camera.isOpened():
            return None
        ret, frame = self.camera.read()
        if ret:
            frame_rgb = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
            self.current_frame = frame_rgb
            return frame_rgb
        return None
    
    def register_face(self, name, roll_no, image):
        if image is None:
            return "No image provided", None
        
        if isinstance(image, np.ndarray):
            frame = image
        else:
            frame = np.array(image)
        
        try:
            faces = detector.detect(frame)
            
            if len(faces) == 0:
                return "No face detected", None
            elif len(faces) > 1:
                return "Multiple faces detected", None
            
            face = faces[0]
            landmarks = face['landmarks']
            embedding = recognizer.get_normalized_embedding(frame, landmarks)
            
            if embedding is None or len(embedding) == 0:
                return "Could not extract face features", None
            
            # Check if this face is already registered
            if face_database:
                for existing_user_id, db_embedding in face_database.items():
                    similarity = compute_similarity(embedding, db_embedding)
                    if similarity > self.recognition_threshold:
                        # Face already exists in database
                        existing_user = user_database.get(existing_user_id, {})
                        existing_name = existing_user.get('name', 'Unknown')
                        existing_roll = existing_user.get('roll_no', 'Unknown')
                        return f"⚠️ Already registered! This face matches:\n• User ID: {existing_user_id}\n• Name: {existing_name}\n• Roll No: {existing_roll}\n• Similarity: {similarity:.2%}", None
            
            user_id = f"USER_{len(user_database) + 1:03d}"
            
            face_database[user_id] = embedding
            user_database[user_id] = {
                'name': name,
                'roll_no': roll_no,
                'registration_date': datetime.now().strftime("%Y-%m-%d %H:%M:%S")
            }
            
            with open(FACE_DB_FILE, 'wb') as f:
                pickle.dump(face_database, f)
            with open(USER_DB_FILE, 'wb') as f:
                pickle.dump(user_database, f)
            
            return f"✅ Successfully registered {name} (Roll No: {roll_no}) as {user_id}", user_id
            
        except Exception as e:
            return f"Error: {str(e)}", None
    
    def recognize_face(self, frame):
        if not face_database:
            return frame, "No registered users", []
        
        try:
            faces = detector.detect(frame)
            
            if len(faces) == 0:
                return frame, "No face detected", []
            
            frame_with_boxes = frame.copy()
            recognized_users = []
            
            for face in faces:
                bbox = face['bbox']
                landmarks = face['landmarks']
                x1, y1, x2, y2 = bbox
                
                cv2.rectangle(frame_with_boxes, (int(x1), int(y1)), (int(x2), int(y2)), (0, 255, 0), 2)
                
                embedding = recognizer.get_normalized_embedding(frame, landmarks)
                
                if embedding is not None:
                    best_match = None
                    best_score = 0
                    
                    for user_id, db_embedding in face_database.items():
                        similarity = compute_similarity(embedding, db_embedding)
                        
                        if similarity > best_score and similarity > self.recognition_threshold:
                            best_score = similarity
                            best_match = user_id
                    
                    if best_match:
                        user_info = user_database[best_match]
                        attendance_status = self.mark_attendance(best_match)
                        
                        recognized_users.append({
                            'user_id': best_match,
                            'name': user_info['name'],
                            'roll_no': user_info['roll_no'],
                            'confidence': best_score,
                            'attendance_status': attendance_status
                        })
                        
                        cv2.putText(frame_with_boxes, 
                                   f"{user_info['name']}", 
                                   (int(x1), int(y1)-10), 
                                   cv2.FONT_HERSHEY_SIMPLEX, 
                                   0.5, (0, 255, 0), 2)
                        cv2.putText(frame_with_boxes, 
                                   f"{best_score:.2f}", 
                                   (int(x1), int(y2)+20), 
                                   cv2.FONT_HERSHEY_SIMPLEX, 
                                   0.4, (0, 255, 0), 1)
            
            if recognized_users:
                status = f"Recognized {len(recognized_users)} user(s)"
            else:
                status = "Unknown user"
            
            return frame_with_boxes, status, recognized_users
            
        except Exception as e:
            return frame, f"Error: {str(e)}", []
    
    def mark_attendance(self, user_id):
        """Mark attendance and return status: 'already_present', 'marked', or 'error'"""
        current_time = datetime.now().strftime("%Y-%m-%d %H:%M:%S")
        user_info = user_database.get(user_id)
        
        if not user_info:
            return 'error'
        
        df = pd.read_csv(ATTENDANCE_FILE)
        today = datetime.now().strftime("%Y-%m-%d")
        roll_no = user_info['roll_no']
        
        # Check if this user_id OR this roll_no already has attendance today
        # This prevents duplicate attendance even if same person registered multiple times
        already_marked = df[
            ((df['User_ID'] == user_id) | (df['Roll_No'] == roll_no)) & 
            (df['Timestamp'].str.startswith(today))
        ].shape[0] > 0
        
        if already_marked:
            return 'already_present'
        
        new_entry = {
            'Timestamp': current_time,
            'User_ID': user_id,
            'Name': user_info['name'],
            'Roll_No': user_info['roll_no'],
            'Status': 'Present'
        }
        
        df = pd.concat([df, pd.DataFrame([new_entry])], ignore_index=True)
        df.to_csv(ATTENDANCE_FILE, index=False)
        
        user_database[user_id]['last_attendance'] = current_time
        with open(USER_DB_FILE, 'wb') as f:
            pickle.dump(user_database, f)
        
        return 'marked'
    
    def get_attendance_report(self):
        if ATTENDANCE_FILE.exists():
            df = pd.read_csv(ATTENDANCE_FILE)
            return df
        return pd.DataFrame()
    
    def get_registered_users(self):
        users_list = []
        for user_id, info in user_database.items():
            users_list.append({
                'User ID': user_id,
                'Name': info['name'],
                'Roll No': info['roll_no'],
                'Registered On': info.get('registration_date', 'N/A'),
                'Last Attendance': info.get('last_attendance', 'Never')
            })
        return pd.DataFrame(users_list)

system = FaceAttendanceSystem()

def start_camera_interface():
    success, message = system.start_camera()
    return message

def capture_and_register(name, roll_no):
    if not name or not roll_no:
        return "Please provide both name and roll number", None
    
    frame = system.capture_frame()
    if frame is None:
        return "Camera not started", None
    
    message, user_id = system.register_face(name, roll_no, frame)
    
    try:
        faces = detector.detect(frame)
        if faces:
            bbox = faces[0]['bbox']
            x1, y1, x2, y2 = bbox
            frame_with_box = frame.copy()
            cv2.rectangle(frame_with_box, (int(x1), int(y1)), (int(x2), int(y2)), (0, 255, 0), 2)
            return message, frame_with_box
    except:
        pass
    
    return message, frame

def detect_and_recognize():
    import random
    
    # Funny quotes for unknown/absent users
    absent_quotes = [
        "🤔 Who dis? Register first, ghostie! 👻",
        "🕵️ Face not found in our database... Are you a spy? 🔍",
        "❓ Unknown face detected! Are you lost, traveler? 🧭",
        "🎭 Nice try, mystery person! Register to join the club! 🎪",
        "👽 Alien detected! Please register on Earth first! 🌍",
        "🦸 Even superheroes need to register their secret identity! 🦹",
        "🤖 Face not recognized. Beep boop, please register! 🔌",
    ]
    
    frame = system.capture_frame()
    if frame is None:
        return None, "Camera not available", ""
    
    frame_with_boxes, status, recognized_users = system.recognize_face(frame)
    
    user_messages = []
    for user in recognized_users:
        name = user['name']
        roll_no = user['roll_no']
        confidence = user['confidence']
        attendance_status = user.get('attendance_status', 'error')
        
        if attendance_status == 'already_present':
            user_messages.append(f"😊 Welcome back, {name}!\n   You're already marked PRESENT today. Chill! 🎉")
        elif attendance_status == 'marked':
            user_messages.append(f"✅ {name} ({roll_no})\n   Attendance marked! Have a great day! 🌟")
        else:
            user_messages.append(f"⚠️ {name} - Error marking attendance")
    
    if user_messages:
        return frame_with_boxes, status, "\n\n".join(user_messages)
    else:
        # Unknown user - show funny quote
        funny_quote = random.choice(absent_quotes)
        return frame_with_boxes, "Unknown Face 👀", funny_quote

def get_todays_attendance():
    df = system.get_attendance_report()
    if df.empty:
        return pd.DataFrame({'Message': ['No attendance records for today']})
    return df

def get_all_users():
    df = system.get_registered_users()
    if df.empty:
        return pd.DataFrame({'Message': ['No users registered yet']})
    return df

def get_all_attendance_files():
    """Get all attendance CSV files"""
    all_data = []
    for file in ATTENDANCE_DIR.glob("attendance_*.csv"):
        try:
            df = pd.read_csv(file)
            all_data.append(df)
        except:
            pass
    if all_data:
        return pd.concat(all_data, ignore_index=True)
    return pd.DataFrame()

def filter_attendance_by_date(filter_type, start_date, end_date):
    """Filter attendance records by date range"""
    df = get_all_attendance_files()
    
    if df.empty:
        return pd.DataFrame({'Message': ['No attendance records found']}), "No records found"
    
    # Parse timestamps
    df['Date'] = pd.to_datetime(df['Timestamp']).dt.date
    today = datetime.now().date()
    
    if filter_type == "Today":
        filtered = df[df['Date'] == today]
        date_info = f"📅 Today: {today.strftime('%d %b %Y')}"
    elif filter_type == "This Week":
        week_start = today - pd.Timedelta(days=today.weekday())
        filtered = df[df['Date'] >= week_start]
        date_info = f"📅 This Week: {week_start.strftime('%d %b')} - {today.strftime('%d %b %Y')}"
    elif filter_type == "This Month":
        month_start = today.replace(day=1)
        filtered = df[df['Date'] >= month_start]
        date_info = f"📅 This Month: {month_start.strftime('%d %b')} - {today.strftime('%d %b %Y')}"
    elif filter_type == "Custom Range":
        if start_date and end_date:
            start = pd.to_datetime(start_date).date()
            end = pd.to_datetime(end_date).date()
            filtered = df[(df['Date'] >= start) & (df['Date'] <= end)]
            date_info = f"📅 Custom: {start.strftime('%d %b %Y')} - {end.strftime('%d %b %Y')}"
        else:
            return pd.DataFrame({'Message': ['Please select both start and end dates']}), "⚠️ Select date range"
    else:  # All Time
        filtered = df
        if not df.empty:
            min_date = df['Date'].min()
            max_date = df['Date'].max()
            date_info = f"📅 All Time: {min_date.strftime('%d %b %Y')} - {max_date.strftime('%d %b %Y')}"
        else:
            date_info = "📅 All Time"
    
    if filtered.empty:
        return pd.DataFrame({'Message': ['No records found for selected period']}), f"{date_info} | 0 records"
    
    # Drop the temporary Date column and return
    result = filtered.drop(columns=['Date'])
    record_count = len(result)
    unique_users = result['User_ID'].nunique() if 'User_ID' in result.columns else 0
    
    return result, f"{date_info} | {record_count} records | {unique_users} unique users"

def download_report(filter_type, start_date, end_date):
    """Generate downloadable CSV report"""
    df, _ = filter_attendance_by_date(filter_type, start_date, end_date)
    
    if 'Message' in df.columns:
        return None
    
    # Create download file
    timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
    filename = f"attendance_report_{filter_type.lower().replace(' ', '_')}_{timestamp}.csv"
    filepath = ATTENDANCE_DIR / filename
    df.to_csv(filepath, index=False)
    
    return str(filepath)

def get_attendance_summary(filter_type, start_date, end_date):
    """Get summary statistics"""
    df, info = filter_attendance_by_date(filter_type, start_date, end_date)
    
    if 'Message' in df.columns:
        return df, info, pd.DataFrame({'Summary': ['No data available']})
    
    # Create summary
    summary_data = {
        'Metric': ['Total Records', 'Unique Users', 'Present Count'],
        'Value': [
            len(df),
            df['User_ID'].nunique() if 'User_ID' in df.columns else 0,
            len(df[df['Status'] == 'Present']) if 'Status' in df.columns else len(df)
        ]
    }
    summary_df = pd.DataFrame(summary_data)
    
    return df, info, summary_df

# Comprehensive Responsive CSS
custom_css = """
/* Base styles */
.gradio-container {
    max-width: 1400px !important;
    margin: 0 auto !important;
    padding: 20px !important;
    font-family: 'Segoe UI', Tahoma, Geneva, Verdana, sans-serif !important;
}

/* Header styling */
.main-header {
    background: linear-gradient(135deg, #667eea 0%, #764ba2 100%);
    -webkit-background-clip: text;
    -webkit-text-fill-color: transparent;
    background-clip: text;
    font-size: 2.5rem !important;
    font-weight: 700 !important;
    text-align: center;
    margin-bottom: 10px !important;
}

.sub-header {
    text-align: center;
    color: #6c757d;
    font-size: 1.1rem !important;
    margin-bottom: 20px !important;
}

/* Tab styling */
.tabs {
    border-radius: 15px !important;
    overflow: hidden !important;
}

/* Button styling */
button.primary {
    background: linear-gradient(135deg, #667eea 0%, #764ba2 100%) !important;
    border: none !important;
    border-radius: 12px !important;
    padding: 12px 24px !important;
    font-weight: 600 !important;
    transition: all 0.3s ease !important;
    box-shadow: 0 4px 15px rgba(102, 126, 234, 0.4) !important;
}

button.primary:hover {
    transform: translateY(-2px) !important;
    box-shadow: 0 6px 20px rgba(102, 126, 234, 0.6) !important;
}

button.secondary {
    background: linear-gradient(135deg, #f093fb 0%, #f5576c 100%) !important;
    border: none !important;
    border-radius: 12px !important;
    padding: 12px 24px !important;
    font-weight: 600 !important;
    transition: all 0.3s ease !important;
}

/* Card-like sections */
.gr-form, .gr-box {
    background: #ffffff !important;
    border-radius: 16px !important;
    box-shadow: 0 4px 20px rgba(0, 0, 0, 0.08) !important;
    padding: 20px !important;
    margin-bottom: 15px !important;
}

/* Input styling */
input, textarea {
    border-radius: 10px !important;
    border: 2px solid #e9ecef !important;
    padding: 12px !important;
    transition: border-color 0.3s ease !important;
}

input:focus, textarea:focus {
    border-color: #667eea !important;
    box-shadow: 0 0 0 3px rgba(102, 126, 234, 0.1) !important;
}

/* Image container */
.image-container img {
    border-radius: 12px !important;
    box-shadow: 0 4px 15px rgba(0, 0, 0, 0.1) !important;
}

/* Dataframe styling */
.dataframe {
    border-radius: 12px !important;
    overflow: hidden !important;
}

/* Status box styling */
.status-success {
    background: linear-gradient(135deg, #11998e 0%, #38ef7d 100%) !important;
    color: white !important;
    padding: 15px !important;
    border-radius: 12px !important;
}

/* Mobile First - Base styles (320px and up) */
@media (max-width: 480px) {
    .gradio-container {
        padding: 10px !important;
    }
    
    .main-header {
        font-size: 1.5rem !important;
    }
    
    .sub-header {
        font-size: 0.9rem !important;
    }
    
    button {
        width: 100% !important;
        margin-bottom: 10px !important;
        padding: 14px 20px !important;
        font-size: 16px !important;
        min-height: 50px !important;
    }
    
    .gr-form, .gr-box {
        padding: 12px !important;
        margin: 8px 0 !important;
    }
    
    input, textarea {
        font-size: 16px !important;
        padding: 14px !important;
    }
    
    .gr-row {
        flex-direction: column !important;
    }
    
    .gr-column {
        width: 100% !important;
        min-width: 100% !important;
    }
}

/* Tablet styles (481px to 768px) */
@media (min-width: 481px) and (max-width: 768px) {
    .gradio-container {
        padding: 15px !important;
    }
    
    .main-header {
        font-size: 2rem !important;
    }
    
    button {
        padding: 12px 20px !important;
        min-height: 48px !important;
    }
    
    .gr-row {
        flex-wrap: wrap !important;
    }
    
    .gr-column {
        min-width: 45% !important;
    }
}

/* Small desktop (769px to 1024px) */
@media (min-width: 769px) and (max-width: 1024px) {
    .gradio-container {
        padding: 20px !important;
        max-width: 95% !important;
    }
}

/* Large desktop (1025px and up) */
@media (min-width: 1025px) {
    .gradio-container {
        padding: 30px !important;
    }
    
    .main-header {
        font-size: 2.8rem !important;
    }
}

/* Animation for buttons */
@keyframes pulse {
    0% { transform: scale(1); }
    50% { transform: scale(1.02); }
    100% { transform: scale(1); }
}

button.primary:active {
    animation: pulse 0.3s ease;
}

/* Dark mode support */
@media (prefers-color-scheme: dark) {
    .gr-form, .gr-box {
        background: #1a1a2e !important;
    }
}
"""

# Create Gradio interface - compatible with different versions
demo = None
try:
    # Try newest Gradio with all features
    demo = gr.Blocks(title="Face Attendance System", theme=gr.themes.Soft(), css=custom_css)
except (TypeError, AttributeError):
    try:
        # Try without theme
        demo = gr.Blocks(title="Face Attendance System", css=custom_css)
    except TypeError:
        try:
            # Try without css
            demo = gr.Blocks(title="Face Attendance System")
        except TypeError:
            # Oldest version - no parameters
            demo = gr.Blocks()

with demo:
    # Header
    gr.Markdown("# 🎯 Face Recognition Attendance System")
    gr.Markdown("### Powered by UniFace (RetinaFace + ArcFace)")
    
    # Camera Tab
    with gr.Tab("📷 Camera"):
        gr.Markdown("## 📷 Camera Control")
        with gr.Row():
            start_btn = gr.Button("▶️ Start Camera", variant="primary")
            stop_btn = gr.Button("⏹️ Stop Camera", variant="secondary")
            refresh_btn = gr.Button("🔄 Refresh Preview", variant="secondary")
        camera_status = gr.Textbox(
            label="📡 Status", 
            interactive=False, 
            value="📷 Camera auto-started! Click 'Refresh Preview' to see the feed."
        )
        preview = gr.Image(label="📺 Live Preview", interactive=False)
    
    # Register Tab
    with gr.Tab("📝 Register"):
        gr.Markdown("## 📝 Register New User")
        gr.Markdown("*Fill in details and capture your face to register*")
        
        with gr.Row():
            with gr.Column():
                gr.Markdown("### 📋 User Details")
                name_input = gr.Textbox(
                    label="👤 Full Name", 
                    placeholder="Enter your full name..."
                )
                roll_no_input = gr.Textbox(
                    label="🔢 Roll Number / ID", 
                    placeholder="Enter roll number or ID..."
                )
                register_btn = gr.Button("📸 Capture & Register", variant="primary")
            
            with gr.Column():
                gr.Markdown("### 📊 Registration Status")
                registration_output = gr.Textbox(
                    label="ℹ️ Status", 
                    interactive=False, 
                    lines=5
                )
                registration_image = gr.Image(
                    label="📸 Captured Face", 
                    interactive=False
                )
    
    # Attendance Tab
    with gr.Tab("✅ Attendance"):
        gr.Markdown("## ✅ Mark Your Attendance")
        gr.Markdown("*Click the button below to detect your face and mark attendance*")
        
        detect_btn = gr.Button("🔍 Detect & Mark Attendance", variant="primary")
        
        with gr.Row():
            with gr.Column():
                detection_image = gr.Image(
                    label="📸 Detection Result", 
                    interactive=False
                )
            with gr.Column():
                detection_status = gr.Textbox(
                    label="📡 Status", 
                    interactive=False
                )
                recognized_users = gr.Textbox(
                    label="🎯 Result", 
                    lines=8, 
                    interactive=False
                )
    
    # Reports Tab
    with gr.Tab("📊 Reports"):
        gr.Markdown("## 📊 Attendance Reports & Analytics")
        
        # Filter Section
        with gr.Row():
            with gr.Column():
                gr.Markdown("### 📅 Select Date Range")
                filter_type = gr.Radio(
                    choices=["Today", "This Week", "This Month", "Custom Range", "All Time"],
                    value="Today",
                    label="Filter Period"
                )
            with gr.Column():
                gr.Markdown("### 📆 Custom Date Range")
                start_date = gr.Textbox(
                    label="Start Date", 
                    placeholder="YYYY-MM-DD (e.g., 2024-01-01)"
                )
                end_date = gr.Textbox(
                    label="End Date", 
                    placeholder="YYYY-MM-DD (e.g., 2024-12-31)"
                )
        
        # Action Buttons
        gr.Markdown("### 🎛️ Actions")
        with gr.Row():
            filter_btn = gr.Button("🔍 Apply Filter", variant="primary")
            users_btn = gr.Button("👥 All Users", variant="secondary")
            download_btn = gr.Button("📥 Download CSV", variant="secondary")
        
        # Report Info
        report_info = gr.Textbox(
            label="📊 Report Summary", 
            interactive=False, 
            value="👆 Select a filter and click 'Apply Filter' to view records"
        )
        
        # Summary Stats
        gr.Markdown("### 📈 Quick Stats")
        summary_output = gr.Dataframe(
            label="Statistics", 
            interactive=False
        )
        
        # Main Report Table
        gr.Markdown("### 📋 Attendance Records")
        report_output = gr.Dataframe(
            label="Records", 
            interactive=False
        )
        
        # Download Section
        download_file = gr.File(label="📥 Downloaded Report")
    
    # Event Handlers
    start_btn.click(start_camera_interface, inputs=None, outputs=camera_status)
    stop_btn.click(system.stop_camera, inputs=None, outputs=camera_status)
    
    # Camera preview refresh function
    def refresh_preview():
        frame = system.capture_frame()
        if frame is not None:
            return frame
        return None
    
    refresh_btn.click(refresh_preview, inputs=None, outputs=preview)
    
    register_btn.click(
        capture_and_register, 
        inputs=[name_input, roll_no_input], 
        outputs=[registration_output, registration_image]
    )
    detect_btn.click(
        detect_and_recognize, 
        inputs=None, 
        outputs=[detection_image, detection_status, recognized_users]
    )
    
    # Report filter handler
    def apply_filter(filter_type, start_date, end_date):
        df, info, summary = get_attendance_summary(filter_type, start_date, end_date)
        return df, info, summary
    
    filter_btn.click(
        apply_filter, 
        inputs=[filter_type, start_date, end_date], 
        outputs=[report_output, report_info, summary_output]
    )
    users_btn.click(get_all_users, inputs=None, outputs=report_output)
    download_btn.click(
        download_report, 
        inputs=[filter_type, start_date, end_date], 
        outputs=download_file
    )

if __name__ == "__main__":
    print("=" * 50)
    print("🎯 Face Recognition Attendance System")
    print("=" * 50)
    print("📦 Models: RetinaFace (detection) + ArcFace (recognition)")
    print("📷 Camera: Auto-starting...")
    system.start_camera()
    print("✅ Camera ready!")
    print("🌐 Launching web interface...")
    print("=" * 50)
    demo.launch(share=False, server_name="0.0.0.0", server_port=7860)