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

# Custom CSS for styling
st.markdown("""
    <style>
    /* General styling */
    body {
        background-color: #f0f2f6;
        font-family: 'Arial', sans-serif;
    }
    .stApp {
        max-width: 1200px;
        margin: 0 auto;
    }
    
    /* Title */
    .title {
        color: #2c3e50;
        font-size: 2.5em;
        text-align: center;
        margin-bottom: 0.5em;
    }
    
    /* Subheader */
    .subheader {
        color: #3498db;
        font-size: 1.2em;
        text-align: center;
        margin-bottom: 2em;
    }
    
    /* Input containers */
    .input-container {
        background-color: white;
        padding: 20px;
        border-radius: 10px;
        box-shadow: 0 4px 6px rgba(0, 0, 0, 0.1);
        margin-bottom: 20px;
    }
    
    /* Button styling */
    .stButton>button {
        background-color: #1a10e3;
        color: white;
        border: none;
        padding: 10px 20px;
        border-radius: 5px;
        font-weight: bold;
        transition: all 0.3s ease;
    }
    .stButton>button:hover {
        background-color: #c0392b;
        transform: scale(1.05);
    }
    
    /* Success message */
    .stSuccess {
        background-color: #2ecc71 !important;
        color: white !important;
        padding: 15px;
        border-radius: 5px;
        text-align: center;
        font-size: 1.2em;
    }
    
    /* Dataframe styling */
    .dataframe {
        border: 2px solid #3498db;
        border-radius: 5px;
        padding: 10px;
    }
    
    /* Footer */
    .footer {
        text-align: center;
        color: #7f8c8d;
        margin-top: 30px;
        font-size: 0.9em;
    }
    .footer b {
        color: #e74c3c;
    }
    
    /* Sidebar */
    .sidebar .sidebar-content {
        background-color: #34495e;
        color: white;
        padding: 20px;
    }
    </style>
""", unsafe_allow_html=True)

# Load the saved model and scaler
model = joblib.load('srn_rvp_model_version_2.pkl')
scaler = joblib.load('srn_rvp_scaler_version_2.pkl')

# Define feature names and default values
features = ['C_101_Top Temp', 'Stabiliser_feed', 'Kero_DOT ', 'Stab_Tray_3_temp ', 
            'Kero _reboiler_inlet_temp', 'Stab_top_pr', 'LGO_DOT', 'mp_stm_HGO_strp']
default_values = [130.0, 180.0, 200.0, 130.0, 250.0, 8.0, 270.0, 140.0]

# Sidebar for additional info
with st.sidebar:
    st.markdown("<h2 style='color: #ecf0f1;'>About</h2>", unsafe_allow_html=True)
    st.write("""
        This app predicts the **SRN RVP (Reid Vapor Pressure)** lab value for a Crude Distillation Unit (CDU) using a pre-trained machine learning model.
        
        **Features Used:**
        - Temperature measurements
        - Pressure readings
        - Flow rates
        
        
    """)
    st.image("distillation.jpg", caption="Refinery Process Predictive Modeling")

# Main app content
st.markdown("<h1 class='title'>🔬 CDU SRN 'RVP' Prediction Tool</h1>", unsafe_allow_html=True)
st.markdown("<p class='subheader'>Enter process parameters to predict the lab RVP value</p>", unsafe_allow_html=True)

# Input form in columns for better layout
st.markdown("<div class='input-container'>", unsafe_allow_html=True)
st.write("### Input Process Parameters")
col1, col2 = st.columns(2)

input_data = {}
for i, (feature, default) in enumerate(zip(features, default_values)):
    with col1 if i % 2 == 0 else col2:
        input_data[feature] = st.number_input(
            feature,
            min_value=0.0,
            max_value=1000.0,
            value=float(default),
            step=1.0,
            format="%.1f",
            key=feature
        )
st.markdown("</div>", unsafe_allow_html=True)

# Convert inputs to DataFrame
input_df = pd.DataFrame([input_data], columns=features)

# Predict button
if st.button("🔍 Predict Lab Value"):
    # Scale the input data
    input_scaled = scaler.transform(input_df)
    
    # Make prediction
    prediction = model.predict(input_scaled)[0]
    
    # Display result with animation
    st.markdown(f"""
        <div class='stSuccess'>
            Predicted RVP Lab Value: <b>{prediction:.4f} psi</b>
        </div>
    """, unsafe_allow_html=True)

# Display input values with corrected precision formatting
st.write("### Your Input Values")
# Use format() to set precision to 2 decimal places
styled_df = input_df.style.highlight_max(axis=0).format("{:.2f}")
st.dataframe(styled_df, use_container_width=True)

# Instructions expander
with st.expander("ℹ️ How to Use", expanded=False):
    st.markdown("""
        1. **Enter Values**: Adjust the input fields for each parameter.
        2. **Predict**: Click the "Predict Lab Value" button.
        3. **Review**: Check the predicted RVP and input values below.
        
        *Note*: This ML model is trained on refinery-specific data and uses scaled features for predictions.
    """)

# Footer
st.markdown("""
    <div class='footer'>
        Developed by <b>SKB</b> | © 2025 All Rights Reserved
    </div>
""", unsafe_allow_html=True)