import streamlit as st
import numpy as np
import tensorflow as tf
import pandas as pd
from sklearn.preprocessing import MinMaxScaler

# Load the saved model
model = tf.keras.models.load_model('models/emotion_model')

# Function to preprocess user inputs and perform prediction
def predict_emotion(spO2, heart_rate, body_temp):
    scaler = MinMaxScaler()  # Initialize scaler
    scaler.fit(pd.DataFrame(columns=['spO2', 'heart-rate', 'body-temperature'], data=[[70, 50, 95.0], [100, 120, 105.0]]))  # Fit scaler to specified range
    input_data = np.array([[spO2, heart_rate, body_temp]])
    input_data_scaled = scaler.transform(input_data)
    predicted_emotions = model.predict(input_data_scaled)
    return predicted_emotions[0]

def main():
    st.title('Emotion Prediction App')
    st.sidebar.title('Options')

    # User input fields
    st.sidebar.header('User Inputs')
    spO2 = st.sidebar.slider('Select spO2 level', min_value=70, max_value=100, value=98)
    heart_rate = st.sidebar.slider('Select heart rate', min_value=50, max_value=120, value=80)
    body_temp = st.sidebar.slider('Select body temperature', min_value=95.0, max_value=105.0, value=98.6)

    # Button to trigger emotion prediction
    if st.sidebar.button('Predict Emotion'):
        # Perform prediction using the loaded model
        predicted_emotions = predict_emotion(spO2, heart_rate, body_temp)
        emotions = ['Anger', 'Fear', 'Sadness', 'Disgust', 'Surprise', 'Anticipation', 'Trust', 'Joy']
        
        # Display predicted emotions
        st.subheader('Predicted Emotions')
        for emotion, score in zip(emotions, predicted_emotions):
            st.write(f'{emotion}: {score:.2f}')

        # Determine likely emotions and display them
        likely_emotions = [emotions[i] for i, score in enumerate(predicted_emotions) if score > 0.5]
        st.success(f'Most likely emotion(s): {", ".join(likely_emotions)}')

if __name__ == '__main__':
    main()