import streamlit as st
import joblib
import numpy as np
import pandas as pd

# 1. Page Config
st.set_page_config(page_title="Digit Recognizer", page_icon="🔢")

st.title("🔢 Digit Recognizer")
st.write("This app uses a Random Forest model to predict hand-drawn digits.")

# 2. Load the Model
@st.cache_resource
def load_model():
    # Ensure digit_recognizer_rf.joblib is in the same folder
    return joblib.load("digit_recognizer_rf.joblib")

model = load_model()

# 3. Sidebar for Data Upload (Optional)
st.sidebar.header("Upload Data")
uploaded_file = st.sidebar.file_exists = st.sidebar.file_uploader("Upload test.csv", type="csv")

if uploaded_file is not None:
    test_data = pd.read_csv(uploaded_file)
    
    # Select a row to predict
    row_index = st.number_input("Select an image index from CSV", min_value=0, max_value=len(test_data)-1, value=0)
    
    if st.button("Predict Digit"):
        # Get pixels and normalize (if you did this during training)
        pixels = test_data.iloc[row_index].values.reshape(1, -1)
        
        # Display Image
        image_reshaped = pixels.reshape(28, 28)
        st.image(image_reshaped, caption=f"Image at index {row_index}", width=150)
        
        # Prediction
        prediction = model.predict(pixels)
        st.success(f"The model predicts this digit is: **{prediction[0]}**")
else:
    st.info("Please upload the 'test.csv' file from Kaggle in the sidebar to test the model.")

st.divider()
st.caption("Developed by Basak Tamer | Part of 20-Kaggle Competition Challenge")