#!/usr/bin/env python3
"""
Streamlit App for Tourism Package Prediction
"""
%%writefile tourism_project/deployment/app.py
import streamlit as st
import pandas as pd
import joblib
from huggingface_hub import hf_hub_download
import os

# Page configuration
st.set_page_config(
    page_title="Tourism Package Prediction",
    page_icon="✈️",
    layout="wide",
    initial_sidebar_state="expanded"
)

# Custom CSS for better styling
st.markdown("""
    <style>
    .main-header {
        font-size: 2.5rem;
        color: #1f77b4;
        text-align: center;
        margin-bottom: 2rem;
    }
    .sub-header {
        font-size: 1.5rem;
        color: #2c3e50;
        margin-top: 1rem;
        margin-bottom: 1rem;
    }
    .prediction-box {
        padding: 2rem;
        border-radius: 10px;
        margin-top: 2rem;
    }
    .success-box {
        background-color: #d4edda;
        border: 2px solid #28a745;
    }
    .danger-box {
        background-color: #f8d7da;
        border: 2px solid #dc3545;
    }
    </style>
""", unsafe_allow_html=True)

# Cache the model loading to avoid reloading on every interaction
@st.cache_resource
def load_model_and_preprocessor():
    """Load model and preprocessor from HuggingFace"""
    try:
        with st.spinner("Loading model and preprocessor from HuggingFace..."):
            model_path = hf_hub_download(
                repo_id="dararaje/Tourism_Package_Prediction",
                filename="tourism_model.pkl",
                repo_type="model"
            )
            preprocessor_path = hf_hub_download(
                repo_id="dararaje/Tourism_Package_Prediction",
                filename="tourism_preprocessor.pkl",
                repo_type="model"
            )

            model = joblib.load(model_path)
            preprocessor = joblib.load(preprocessor_path)

            return model, preprocessor
    except Exception as e:
        st.error(f"Error loading model: {str(e)}")
        st.error("Make sure you have run upload_model.py to upload the model files to HuggingFace")
        st.stop()

# Load model and preprocessor
model, preprocessor = load_model_and_preprocessor()

# Main title
st.markdown('<h1 class="main-header">🌍 Tourism Package Prediction</h1>', unsafe_allow_html=True)
st.markdown("### Predict if a customer will purchase the Tourism Package")
st.markdown("---")

# Create sidebar for information
with st.sidebar:
    st.image("https://img.icons8.com/fluency/96/000000/around-the-world.png", width=100)
    st.markdown("## About This App")
    st.info(
        """
        This application uses machine learning to predict whether a customer
        will purchase the Tourism Package based on:
        - Customer demographics
        - Travel preferences
        - Sales interaction quality
        """
    )

    st.markdown("## Model Information")
    st.markdown("""
    - **Algorithm**: Machine Learning Classifier
    - **Features**: 18 customer attributes
    - **Output**: Purchase probability
    """)

    st.markdown("## How to Use")
    st.markdown("""
    1. Fill in customer details in the form
    2. Click 'Predict Purchase Likelihood'
    3. View prediction results with confidence scores
    """)

# Create three columns for organized input
col1, col2, col3 = st.columns(3)

with col1:
    st.markdown('<p class="sub-header">👤 Customer Demographics</p>', unsafe_allow_html=True)
    age = st.number_input("Age", min_value=18, max_value=100, value=30, step=1)

    gender = st.selectbox("Gender", options=["Male", "Female"])

    marital_status = st.selectbox(
        "Marital Status",
        options=["Single", "Married", "Divorced", "Unmarried"]
    )

    city_tier = st.selectbox(
        "City Tier",
        options=[1, 2, 3],
        format_func=lambda x: f"Tier {x} {'(Metro)' if x==1 else '(Tier-2)' if x==2 else '(Tier-3)'}"
    )

    occupation = st.selectbox(
        "Occupation",
        options=["Salaried", "Small Business", "Large Business", "Free Lancer"]
    )

    designation = st.selectbox(
        "Designation",
        options=["Executive", "Manager", "Senior Manager", "AVP", "VP"]
    )

    monthly_income = st.number_input(
        "Monthly Income (₹)",
        min_value=0,
        max_value=1000000,
        value=20000,
        step=1000
    )

with col2:
    st.markdown('<p class="sub-header">🧳 Travel Preferences</p>', unsafe_allow_html=True)

    num_person_visiting = st.number_input(
        "Number of Persons Visiting",
        min_value=1,
        max_value=10,
        value=2,
        step=1
    )

    num_children_visiting = st.number_input(
        "Number of Children Visiting (below 5 years)",
        min_value=0,
        max_value=5,
        value=0,
        step=1
    )

    preferred_property_star = st.selectbox(
        "Preferred Property Star Rating",
        options=[3.0, 4.0, 5.0],
        format_func=lambda x: f"{int(x)} Star"
    )

    num_trips = st.number_input(
        "Number of Trips per Year",
        min_value=0,
        max_value=20,
        value=2,
        step=1
    )

    passport = st.selectbox(
        "Has Passport?",
        options=[1, 0],
        format_func=lambda x: "Yes" if x == 1 else "No"
    )

    own_car = st.selectbox(
        "Owns Car?",
        options=[1, 0],
        format_func=lambda x: "Yes" if x == 1 else "No"
    )

with col3:
    st.markdown('<p class="sub-header">💼 Sales Interaction Details</p>', unsafe_allow_html=True)

    type_of_contact = st.selectbox(
        "Type of Contact",
        options=["Company Invited", "Self Inquiry"]
    )

    product_pitched = st.selectbox(
        "Product Pitched",
        options=["Basic", "Standard", "Deluxe", "Super Deluxe", "King"]
    )

    pitch_satisfaction_score = st.slider(
        "Pitch Satisfaction Score",
        min_value=1,
        max_value=5,
        value=3,
        step=1
    )

    num_followups = st.number_input(
        "Number of Follow-ups",
        min_value=0,
        max_value=10,
        value=3,
        step=1
    )

    duration_of_pitch = st.number_input(
        "Duration of Pitch (minutes)",
        min_value=0,
        max_value=120,
        value=15,
        step=1
    )

# Center the predict button
st.markdown("---")
col_button1, col_button2, col_button3 = st.columns([1, 1, 1])
with col_button2:
    predict_button = st.button("🔮 Predict Purchase Likelihood", type="primary", use_container_width=True)

# Prediction logic
if predict_button:
    try:
        # Create input dataframe
        input_data = pd.DataFrame({
            'Age': [age],
            'TypeofContact': [type_of_contact],
            'CityTier': [city_tier],
            'Occupation': [occupation],
            'Gender': [gender],
            'NumberOfPersonVisiting': [num_person_visiting],
            'PreferredPropertyStar': [preferred_property_star],
            'MaritalStatus': [marital_status],
            'NumberOfTrips': [num_trips],
            'Passport': [passport],
            'OwnCar': [own_car],
            'NumberOfChildrenVisiting': [num_children_visiting],
            'Designation': [designation],
            'MonthlyIncome': [monthly_income],
            'PitchSatisfactionScore': [pitch_satisfaction_score],
            'ProductPitched': [product_pitched],
            'NumberOfFollowups': [num_followups],
            'DurationOfPitch': [duration_of_pitch]
        })

        # Preprocess and predict
        processed_data = preprocessor.transform(input_data)
        prediction = model.predict(processed_data)
        probability = model.predict_proba(processed_data)

        # Display results
        st.markdown("---")

        if prediction[0] == 1:
            st.markdown('<div class="prediction-box success-box">', unsafe_allow_html=True)
            st.markdown("### ✅ Prediction: WILL PURCHASE")
            st.markdown(f"#### Confidence: {probability[0][1] * 100:.2f}%")
            st.markdown('</div>', unsafe_allow_html=True)
        else:
            st.markdown('<div class="prediction-box danger-box">', unsafe_allow_html=True)
            st.markdown("### ❌ Prediction: WILL NOT PURCHASE")
            st.markdown(f"#### Confidence: {probability[0][0] * 100:.2f}%")
            st.markdown('</div>', unsafe_allow_html=True)

        # Probability breakdown
        st.markdown("### 📊 Probability Breakdown")

        col_prob1, col_prob2 = st.columns(2)
        with col_prob1:
            st.metric(
                label="Purchase Probability",
                value=f"{probability[0][1] * 100:.2f}%",
                delta=f"{(probability[0][1] - 0.5) * 100:.2f}% from baseline"
            )
        with col_prob2:
            st.metric(
                label="No Purchase Probability",
                value=f"{probability[0][0] * 100:.2f}%",
                delta=f"{(probability[0][0] - 0.5) * 100:.2f}% from baseline"
            )

        # Progress bars for visualization
        st.markdown("#### Visual Probability Distribution")
        st.progress(probability[0][1], text=f"Purchase: {probability[0][1] * 100:.1f}%")
        st.progress(probability[0][0], text=f"No Purchase: {probability[0][0] * 100:.1f}%")

        # Recommendation
        st.markdown("### 💡 Recommendation")
        if prediction[0] == 1:
            if probability[0][1] > 0.75:
                st.success("**High Priority Lead**: Strong likelihood of purchase. Recommend immediate follow-up with premium package details.")
            else:
                st.info("**Moderate Priority Lead**: Good potential for purchase. Consider personalized follow-up with package benefits.")
        else:
            if probability[0][0] > 0.75:
                st.warning("**Low Priority Lead**: Low likelihood of purchase. May require additional nurturing or different product offering.")
            else:
                st.info("**Uncertain Lead**: Probability is borderline. Additional information or engagement might help.")

    except Exception as e:
        st.error(f"Error making prediction: {str(e)}")
        st.exception(e)

# Footer
st.markdown("---")
st.markdown("""
    <div style='text-align: center; color: #7f8c8d;'>
        <p>Built with ❤️ using Streamlit and HuggingFace | Tourism Package Prediction MLOps Project</p>
    </div>
""", unsafe_allow_html=True)