import streamlit as st
import joblib
import numpy as np

# Load model pipeline
pipeline = joblib.load("src/logistic_model_pipeline")
model = pipeline["model"]
num_scaler = pipeline["num_scaler"]
cat_encoder = pipeline["cat_encoder"]
y_encoder = pipeline["y_encoder"]
num_columns = pipeline["num_columns"]
cat_columns = pipeline["cat_columns"]

# Streamlit page setup
st.set_page_config(page_title="🐧 Penguin Classifier", layout="centered")
st.title("🐧 Penguin Species Prediction")
st.subheader("📥 Enter the Penguin's Features")

# Numeric input using sliders with actual data ranges
st.markdown("### 🔢 Numeric Features")
num_inputs = []

slider_ranges = {
    "bill_length_mm": (32.1, 59.6),
    "bill_depth_mm": (13.1, 21.5),
    "flipper_length_mm": (172.0, 231.0),
    "body_mass_g": (2700.0, 6300.0)
}

for col in num_columns:
    min_val, max_val = slider_ranges.get(col, (0.0, 100.0))  # fallback in case
    val = st.slider(
        f"{col.replace('_', ' ').title()}",
        min_value=min_val,
        max_value=max_val,
        step=0.1
    )
    num_inputs.append(val)

# Categorical input
st.markdown("### 🧬 Categorical Features")
cat_inputs = []

for col in cat_columns:
    if "island" in col.lower():
        val = st.selectbox(f"{col.replace('_', ' ').title()}:", ["Dream", "Torgersen", "Biscoe"])
    else:
        val = st.text_input(f"{col.replace('_', ' ').title()}:")
    cat_inputs.append(val)

# Prediction button
if st.button("🚀 Predict"):
    try:
        # Transform inputs
        scaled_nums = num_scaler.transform([num_inputs])
        encoded_cats = cat_encoder.transform([cat_inputs])
        combined = np.hstack([scaled_nums, encoded_cats])

        # Predict
        result = model.predict(combined)
        final_pred = y_encoder.inverse_transform(result)

        st.success(f"✅ The predicted species is: **{final_pred[0]}**")
        st.balloons()

    except Exception as e:
        st.error(f"Oops! Something went wrong: {e}")