src/streamlit_app.py

import streamlit as st
import tensorflow as tf
from PIL import Image
import numpy as np
import os
import json

# ==========================================
# 1. KONFIGURASI HALAMAN
# ==========================================
st.set_page_config(page_title="Plant Disease Expert System", layout="wide", page_icon="🌿")

# ==========================================
# 2. PATH DEFINISI (MENGGUNAKAN PATH RELATIF)
# ==========================================
# Di Hugging Face, gunakan path relatif terhadap root aplikasi
BASE_MODEL_DIR = "model" 

MODEL_PATHS = {
    "MobileNetV2 (Best Fine-Tuned)": os.path.join(BASE_MODEL_DIR, "model_skripsi_mobilenet_final_FT.h5"),
    "InceptionV3 (Fine-Tuned)": os.path.join(BASE_MODEL_DIR, "model_skripsi_inception_finetuned.h5"),
    "Deep Custom CNN": os.path.join(BASE_MODEL_DIR, "model_skripsi_cnn_deep.h5"),
    "MobileNetV2 (Standard)": os.path.join(BASE_MODEL_DIR, "model_skripsi_mobilenet.h5"),
    "InceptionV3 (Standard)": os.path.join(BASE_MODEL_DIR, "model_skripsi_inception.h5")
}

# Path untuk Knowledge Base / Label
PATH_JSON = os.path.join(BASE_MODEL_DIR, "knowledge_base.json")

# ==========================================
# 3. FUNGSI LOAD DATA & MODEL (CACHED)
# ==========================================

@st.cache_data
def get_labels_from_json(path):
    """Mengambil label dari knowledge_base.json agar tidak bergantung pada folder dataset"""
    try:
        if not os.path.exists(path):
            st.error(f"File {path} tidak ditemukan!")
            return {}
        with open(path, 'r') as f:
            return json.load(f)
    except Exception as e:
        st.error(f"Gagal membaca label JSON: {e}")
        return {}

@st.cache_resource
def load_my_model(path):
    """Memuat model h5"""
    return tf.keras.models.load_model(path)

# ==========================================
# 4. SIDEBAR: KONTROL & VALIDASI
# ==========================================
st.sidebar.header("⚙️ Kontrol Sistem")
model_choice = st.sidebar.selectbox("Pilih Arsitektur Model:", list(MODEL_PATHS.keys()))

# Ambil Knowledge Base (JSON)
KB_DATA = get_labels_from_json(PATH_JSON)
# Ambil list nama penyakit untuk mapping index
CLASS_NAMES = [KB_DATA[str(i)]['jenis'] if str(i) in KB_DATA else f"Class {i}" for i in range(len(KB_DATA))]

# Loading Model
current_model_path = MODEL_PATHS[model_choice]

if os.path.exists(current_model_path):
    model = load_my_model(current_model_path)
    
    # Deteksi info model otomatis
    num_classes_model = model.output_shape[-1]
    model_input_h = model.input_shape[1]
    model_input_w = model.input_shape[2]
    
    st.sidebar.divider()
    st.sidebar.success("✅ Model Loaded")
    st.sidebar.write(f"📊 Output Model: **{num_classes_model} Kelas**")
    st.sidebar.info(f"Dimensi Input: {model_input_h}x{model_input_w}")
else:
    st.sidebar.error(f"❌ File Model Tidak Ditemukan di: {current_model_path}")
    st.stop() # Hentikan eksekusi jika model tidak ada

# ==========================================
# 5. HALAMAN UTAMA: PREDIKSI
# ==========================================
st.title("🌿 Identifikasi Penyakit Tanaman")
st.write(f"Model Aktif: **{model_choice}**")

uploaded_file = st.file_uploader("Unggah foto daun...", type=["jpg", "jpeg", "png"])

if uploaded_file:
    col1, col2 = st.columns([1, 1])
    
    img_display = Image.open(uploaded_file).convert("RGB")
    
    with col1:
        st.image(img_display, caption="Preview Foto", use_container_width=True)
    
    with col2:
        if st.button("🚀 Jalankan Diagnosa"):
            with st.spinner('Menganalisis pola penyakit...'):
                # Preprocessing sesuai input shape model
                target_size = (model_input_h, model_input_w)
                img = img_display.resize(target_size)
                img_array = np.array(img) / 255.0
                img_array = np.expand_dims(img_array, axis=0)

                # Prediksi
                preds = model.predict(img_array)
                idx = np.argmax(preds)
                confidence = np.max(preds) * 100

                # Tampilan Hasil Diagnosa
                st.markdown("### 📊 Hasil Analisis")
                
                res_data = KB_DATA.get(str(idx))
                if res_data:
                    st.success(f"**Terdeteksi:** \n\n ### {res_data['jenis']}")
                    st.metric("Tingkat Keyakinan", f"{confidence:.2f}%")
                    
                    with st.expander("Lihat Detail Kondisi & Saran"):
                        st.write(f"**Kondisi:** {res_data.get('kondisi', 'N/A')}")
                        st.write(f"**Deskripsi:** {res_data.get('deskripsi', 'N/A')}")
                else:
                    st.error(f"Index {idx} tidak ditemukan di Knowledge Base.")

# ==========================================
# 6. FOOTER
# ==========================================
st.markdown("---")
st.caption("Aplikasi Prediksi Penyakit Tanaman - Penelitian Skripsi 2026")