app.py

import serial
import serial.tools.list_ports
import time
import streamlit as st
import threading
import pickle
from sklearn.tree import DecisionTreeClassifier
from sklearn.preprocessing import LabelEncoder
import pandas as pd

# Global variables to store bus information and hardware statuses
bus_info = {
    'bus_name': 'N/A',
    'seat_status': 'N/A',
    'distance': 'N/A',
    'led_status': 'OFF',
    'buzzer_status': 'OFF',
}

# Function to collect available serial ports data
def collect_serial_ports():
    ports = serial.tools.list_ports.comports()
    port_data = []
    for port in ports:
        data = {
            'port': port.device,
            'description': port.description,
            'manufacturer': port.manufacturer if port.manufacturer else 'Unknown'
        }
        port_data.append(data)
    return port_data

# Function to preprocess and label the data for model training
def preprocess_data(port_data):
    df = pd.DataFrame(port_data)

    # Encode the 'description' and 'manufacturer' fields to numeric
    le = LabelEncoder()
    df['description'] = le.fit_transform(df['description'])
    df['manufacturer'] = le.fit_transform(df['manufacturer'])

    return df, le

# Train a simple decision tree classifier (this is just an example)
def train_model():
    # Example: manually labeled data where '1' represents Arduino and '0' is other devices
    port_data = collect_serial_ports()

    df, le = preprocess_data(port_data)
    df['label'] = [1, 0]  # 1 = Arduino, 0 = Other device (Replace with actual labeled data)

    # Train the classifier
    X = df.drop(columns=['port', 'label'])
    y = df['label']
    clf = DecisionTreeClassifier()
    clf.fit(X, y)

    # Save the model and label encoder
    with open('serial_port_model.pkl', 'wb') as model_file:
        pickle.dump(clf, model_file)
    with open('label_encoder.pkl', 'wb') as le_file:
        pickle.dump(le, le_file)

# Function to predict the serial port based on the AI model
def predict_serial_port(port_data):
    # Load the trained model and label encoder
    with open('serial_port_model.pkl', 'rb') as model_file:
        clf = pickle.load(model_file)
    with open('label_encoder.pkl', 'rb') as le_file:
        le = pickle.load(le_file)

    # Preprocess the data
    df, _ = preprocess_data(port_data)

    # Predict the label for the serial ports (Arduino or not)
    prediction = clf.predict(df.drop(columns=['port', 'label']))

    # Return the port if it's predicted to be an Arduino (label = 1)
    for i, pred in enumerate(prediction):
        if pred == 1:  # If prediction is 1, it's the Arduino
            return port_data[i]['port']

    return None  # No Arduino found

# Function to initialize the serial connection with the correct port
def initialize_serial_connection():
    port_data = collect_serial_ports()  # Get available ports data
    com_port = predict_serial_port(port_data)  # Get the correct port from the AI model

    if com_port:
        try:
            ser = serial.Serial(com_port, 9600)  # Open serial port
            time.sleep(2)  # Wait for the connection to initialize
            print(f"Connected to {com_port}")
            return ser
        except serial.SerialException as e:
            st.error(f"Error opening serial port: {e}")
            return None
    else:
        st.error("No Arduino device found.")
        return None

# Initialize serial connection
ser = initialize_serial_connection()

# Function to update bus information based on received serial data
def display_bus_info():
    global bus_info
    try:
        while True:
            if ser.in_waiting > 0:
                # Read data from serial port
                data = ser.readline().decode('utf-8', errors='ignore').strip()
                
                # Debug print to see the raw serial data
                print(f"Raw Data: {data}")

                if "Bus Name:" in data:
                    bus_info['bus_name'] = data.split("Bus Name:")[1].strip()

                elif "Seat Status:" in data:
                    bus_info['seat_status'] = data.split("Seat Status:")[1].strip()

                elif "Distance:" in data:
                    bus_info['distance'] = data.split("Distance:")[1].strip()

                elif "LED:" in data:
                    bus_info['led_status'] = data.split("LED:")[1].strip()

                elif "Buzzer:" in data:
                    bus_info['buzzer_status'] = data.split("Buzzer:")[1].strip()

            time.sleep(1)  # Small delay to prevent flooding the terminal

    except KeyboardInterrupt:
        print("Exiting...")

# Start the bus information display in a separate thread
bus_info_thread = threading.Thread(target=display_bus_info)
bus_info_thread.daemon = True
bus_info_thread.start()

# Streamlit Web Interface
st.title('Bus System Status')

# Display the bus information
col1, col2 = st.columns(2)

with col1:
    st.subheader('Bus Name')
    bus_name = st.empty()

with col2:
    st.subheader('Seat Status')
    seat_status = st.empty()

st.subheader('Distance to Bus')
distance_display = st.empty()

# Display LED and Buzzer status
st.subheader('LED Status')
led_status_display = st.empty()

st.subheader('Buzzer Status')
buzzer_status_display = st.empty()

# Update the information on the web interface every second
while True:
    bus_name.text(bus_info['bus_name'])
    seat_status.text(bus_info['seat_status'])
    distance_display.text(f"Distance: {bus_info['distance']} cm")
    led_status_display.text(f"LED: {bus_info['led_status']}")
    buzzer_status_display.text(f"Buzzer: {bus_info['buzzer_status']}")
    time.sleep(1)