import streamlit as st
import joblib
import numpy as np
import cv2
from PIL import Image
import os
from skimage.feature import local_binary_pattern, graycomatrix, graycoprops

# --- KONFIGURASI HALAMAN ---
st.set_page_config(page_title="Weather Classifier", layout="wide")

# --- FUNGSI LOAD MODEL ---
@st.cache_resource
def load_models(algorithm):
    try:
        # Sesuaikan path ini dengan struktur foldermu
        base_path = "model" 
        
        if algorithm == 'SVM':
            # Pastikan nama file di dalam folder model/svm/ sesuai dengan yang ada
            model_path = os.path.join(base_path, "svm/svm_model_optimal_80_20.joblib")
            scaler_path = os.path.join(base_path, "svm/scaler_svm.joblib")
            le_path = os.path.join(base_path, "svm/label_encoder.joblib")
            
            model = joblib.load(model_path)
            scaler = joblib.load(scaler_path)
            le = joblib.load(le_path)
            
        elif algorithm == 'XGBoost':
            # Pastikan nama file di dalam folder model/xgboost/ sesuai
            model_path = os.path.join(base_path, "xgboost/xgb_model_optimal.joblib")
            le_path = os.path.join(base_path, "xgboost/label_encoder_xgb.joblib")
            
            model = joblib.load(model_path)
            scaler = None 
            le = joblib.load(le_path)
        
        return model, scaler, le
    except Exception as e:
        st.error(f"Error loading model files: {e}")
        return None, None, None

# --- FUNGSI EKSTRAKSI FITUR (GLCM, LBP, HSV) ---
def extract_features_final(image_pil):
    img_np = np.array(image_pil.convert('RGB'))
    
    # Resize (Samakan dengan training, misal 128x128 atau 224x224)
    img_np = cv2.resize(img_np, (224, 224)) 

    # Preprocessing
    hsv_image = cv2.cvtColor(img_np, cv2.COLOR_RGB2HSV)
    h = hsv_image[:, :, 0]
    s = hsv_image[:, :, 1]
    
    gray = cv2.cvtColor(img_np, cv2.COLOR_RGB2GRAY)
    clahe_builder = cv2.createCLAHE(clipLimit=2.0, tileGridSize=(8, 8))
    gray_clahe = clahe_builder.apply(gray)
    
    # Ekstraksi Fitur
    hist_h = cv2.calcHist([h], [0], None, [8], [0, 256]).flatten()
    hist_s = cv2.calcHist([s], [0], None, [8], [0, 256]).flatten()

    lbp = local_binary_pattern(gray_clahe, P=8, R=1, method='uniform')
    lbp_hist, _ = np.histogram(lbp.ravel(), bins=np.arange(0, 10), range=(0, 9))
    lbp_hist = lbp_hist.astype('float')
    lbp_hist /= (lbp_hist.sum() + 1e-6)

    glcm = graycomatrix(gray_clahe, distances=[1], angles=[0], symmetric=True, normed=True)
    contrast = graycoprops(glcm, 'contrast')[0, 0]
    energy = graycoprops(glcm, 'energy')[0, 0]
    homogeneity = graycoprops(glcm, 'homogeneity')[0, 0]

    features = np.hstack([hist_h, hist_s, lbp_hist, contrast, energy, homogeneity])
    return features.reshape(1, -1)

# --- UI UTAMA (WEATHER THEME) ---
st.title("☁️ Weather Classification App 🌦️")
st.write("Klasifikasi Cuaca (Cloudy, Rainy, Shine, Sunrise) menggunakan SVM & XGBoost")

algo_choice = st.sidebar.selectbox("Pilih Algoritma:", ["SVM", "XGBoost"])
uploaded_file = st.file_uploader("Upload foto cuaca...", type=["jpg", "png", "jpeg"])

if uploaded_file is not None:
    image = Image.open(uploaded_file)
    col1, col2 = st.columns(2)
    
    with col1:
        st.image(image, caption='Gambar Diupload', use_column_width=True)
    
    if st.button("Tebak Cuaca"):
        with st.spinner("Sedang menganalisis tekstur & warna langit..."):
            model, scaler, le = load_models(algo_choice)
            
            if model:
                try:
                    features = extract_features_final(image)
                    if scaler:
                        features = scaler.transform(features)
                    
                    prediction_index = model.predict(features)[0]
                    # Label otomatis menyesuaikan (Cloudy/Rainy/dll) dari file label_encoder
                    label = le.inverse_transform([prediction_index])[0]
                    
                    with col2:
                        st.success(f"### Prediksi: {label}")
                        st.info(f"Model: {algo_choice}")
                        
                except Exception as e:
                    st.error(f"Error: {e}")