# Streamlit frontend for the Airbnb Rental Price Prediction service.
# This UI collects property details from the user, sends them to the
# deployed Flask backend, and displays the predicted rental price.

import streamlit as st
import pandas as pd
import requests

# -----------------------------------------------------------------------------
# Configure the backend API endpoints
# -----------------------------------------------------------------------------
# Replace the base URL below with the URL of your deployed backend Space.
MODEL_ROOT_URL = "https://uohna-RentalPricePredictionBackend.hf.space"
ONLINE_URL = f"{MODEL_ROOT_URL}/v1/rental"
BATCH_URL = f"{MODEL_ROOT_URL}/v1/rentalbatch"

# -----------------------------------------------------------------------------
# Page configuration
# -----------------------------------------------------------------------------
st.set_page_config(
    page_title="Airbnb Rental Price Predictor",
    page_icon="🏡",
    layout="centered",
)

st.title("🏡 Airbnb Rental Price Predictor")
st.write(
    "Predict the nightly rental price for a property using our deployed "
    "machine learning model."
)

# Two tabs: one for single-property prediction, one for batch uploads
online_tab, batch_tab = st.tabs(["Single Prediction", "Batch Prediction"])

# -----------------------------------------------------------------------------
# Online (single property) inference
# -----------------------------------------------------------------------------
with online_tab:
    st.subheader("Enter Property Details")

    col1, col2 = st.columns(2)

    with col1:
        room_type = st.selectbox(
            "Room Type",
            ["Entire home/apt", "Private room", "Shared room"],
        )
        accommodates = st.number_input(
            "Accommodates", min_value=1, max_value=20, value=2, step=1
        )
        bathrooms = st.number_input(
            "Bathrooms", min_value=0.0, max_value=10.0, value=1.0, step=0.5
        )
        bedrooms = st.number_input(
            "Bedrooms", min_value=0, max_value=15, value=1, step=1
        )
        beds = st.number_input(
            "Beds", min_value=0, max_value=20, value=1, step=1
        )

    with col2:
        cancellation_policy = st.selectbox(
            "Cancellation Policy",
            ["flexible", "moderate", "strict", "super_strict_30", "super_strict_60"],
        )
        cleaning_fee = st.selectbox("Cleaning Fee Charged?", [True, False])
        instant_bookable = st.selectbox("Instant Bookable?", ["t", "f"])
        review_scores_rating = st.slider(
            "Review Scores Rating", min_value=0, max_value=100, value=90, step=1
        )

    if st.button("Predict Price", type="primary"):
        payload = {
            "room_type": room_type,
            "accommodates": int(accommodates),
            "bathrooms": float(bathrooms),
            "cancellation_policy": cancellation_policy,
            "cleaning_fee": bool(cleaning_fee),
            "instant_bookable": instant_bookable,
            "review_scores_rating": int(review_scores_rating),
            "bedrooms": int(bedrooms),
            "beds": int(beds),
        }

        try:
            response = requests.post(ONLINE_URL, json=payload, timeout=30)
            if response.status_code == 200:
                result = response.json()
                price = result.get("Predicted Price (in dollars)")
                st.success(f"💰 Predicted Nightly Price: **${price}**")
            else:
                st.error(f"Request failed ({response.status_code}): {response.text}")
        except requests.exceptions.RequestException as err:
            st.error(f"Could not reach the backend API: {err}")

# -----------------------------------------------------------------------------
# Batch inference
# -----------------------------------------------------------------------------
with batch_tab:
    st.subheader("Upload a CSV for Batch Predictions")
    st.caption(
        "Your CSV should contain the same feature columns the model was trained on "
        "(e.g., room_type, accommodates, bathrooms, cancellation_policy, cleaning_fee, "
        "instant_bookable, review_scores_rating, bedrooms, beds, and an 'id' column)."
    )

    uploaded_file = st.file_uploader("Choose a CSV file", type=["csv"])

    if uploaded_file is not None:
        df_preview = pd.read_csv(uploaded_file)
        st.write("Preview of uploaded data:")
        st.dataframe(df_preview.head())

        if st.button("Run Batch Prediction"):
            # Rewind so we can send the raw CSV bytes
            uploaded_file.seek(0)
            try:
                response = requests.post(
                    BATCH_URL,
                    files={"file": uploaded_file.getvalue()},
                    timeout=120,
                )
                if response.status_code == 200:
                    predictions = response.json()
                    st.success(f"✅ Received {len(predictions)} predictions.")
                    results_df = pd.DataFrame(
                        list(predictions.items()),
                        columns=["Property ID", "Predicted Price (in dollars)"],
                    )
                    st.dataframe(results_df)
                    st.download_button(
                        "Download Predictions as CSV",
                        data=results_df.to_csv(index=False).encode("utf-8"),
                        file_name="rental_price_predictions.csv",
                        mime="text/csv",
                    )
                else:
                    st.error(
                        f"Request failed ({response.status_code}): {response.text}"
                    )
            except requests.exceptions.RequestException as err:
                st.error(f"Could not reach the backend API: {err}")