import streamlit as st
from streamlit_drawable_canvas import st_canvas
import cv2
from tensorflow.keras.models import load_model
import numpy as np
from PIL import Image

# Define the list of Arabic characters
arabic_chars = ['alef', 'beh', 'teh', 'theh', 'jeem', 'hah', 'khah', 'dal', 'thal', 'reh', 'zain', 'seen', 'sheen',
                'sad', 'dad', 'tah', 'zah', 'ain', 'ghain', 'feh', 'qaf', 'kaf', 'lam', 'meem', 'noon', 'heh', 'waw', 'yeh']

# Load the model once at the beginning
model_path = "saved_model.h5"  # Update with your model path
model = load_model(model_path)

# Define the prediction function
def predict_image(image):
    # Convert to grayscale
    img = cv2.cvtColor(np.array(image), cv2.COLOR_BGR2GRAY)
    
    # Invert the image colors (black background with white letter)
    img = cv2.bitwise_not(img)
    
    # Resize the image to the size expected by the model
    img = cv2.resize(img, (32, 32))
    
    # Reshape and normalize the image
    img = img.reshape(1, 32, 32, 1)
    img = img.astype('float32') / 255.0

    # Predict the character
    pred = model.predict(img)
    predicted_label = arabic_chars[np.argmax(pred)]

    return predicted_label

# Streamlit app
st.title("Arabic Character Recognition App")

canvas_result = st_canvas(
    fill_color="rgba(255, 165, 0, 0.3)",  # Filled color
    stroke_width=12,  # Stroke width
    stroke_color="#FFFFFF",  # Stroke color (white)
    background_color="#000000",  # Canvas background color (black)
    update_streamlit=True,
    height=400,
    width=400,
    drawing_mode="freedraw",
    key="canvas",
)

if canvas_result.image_data is not None:
    # Display the drawn image
    st.image(canvas_result.image_data)

    # Convert the canvas image data to a PIL image
    image = Image.fromarray(canvas_result.image_data.astype('uint8'), 'RGBA').convert('RGB')
    #st.image(image)

    # Predict the character
    predicted_char = predict_image(image)

    # Display the predicted character
    st.subheader(f"Predicted Character: {predicted_char}")