import gradio as gr
from keras.models import load_model
from PIL import Image, ImageOps
import numpy as np
import os

# --- 1. KONFIGURASI ---
MODEL_PATH = "keras_model.h5"
LABELS_PATH = "labels.txt"

# MAPPING PASTI: Label di txt -> File MP3
# Saya sudah sesuaikan dengan daftar label & file yang kamu kasih
DAFTAR_SUARA = {
    "tempe": "tempe.mp3",
    "hidup": "hidup.mp3",
    "hidup_jokowi": "hidup_jokowi.mp3",
    "dsh": "dsh.mp3",
    "bakso": "bakso.mp3"
}

# Suara default (kalau ada error/label ga dikenal)
SUARA_DEFAULT = "dsh.mp3" 

# --- 2. LOAD MODEL ---
# Load model sekali saja di awal
model = load_model(MODEL_PATH, compile=False)
class_names = [line.strip() for line in open(LABELS_PATH, "r").readlines()]

# --- 3. FUNGSI DETEKSI (PASTI/MUTLAK) ---
def deteksi_pasti(image):
    if image is None:
        return "⚠️ WADUH! Gambarnya mana?", None

    # Persiapan Gambar (Standar Keras/Teachable Machine)
    data = np.ndarray(shape=(1, 224, 224, 3), dtype=np.float32)
    image = Image.fromarray(image).convert("RGB")
    size = (224, 224)
    image = ImageOps.fit(image, size, Image.Resampling.LANCZOS)
    image_array = np.asarray(image)
    normalized_image_array = (image_array.astype(np.float32) / 127.5) - 1
    data[0] = normalized_image_array

    # Prediksi
    prediction = model.predict(data)
    
    # AMBIL YANG PALING TINGGI (Index Max)
    index = np.argmax(prediction)
    class_name = class_names[index]
    
    # Ambil nama label bersih.
    # Contoh: "0 tempe" -> kita buang "0 " (2 karakter awal) -> jadi "tempe"
    label_bersih = class_name[2:].strip()

    # Cari File Suaranya
    if label_bersih in DAFTAR_SUARA:
        file_suara = DAFTAR_SUARA[label_bersih]
    else:
        # Jaga-jaga kalau ada label baru
        file_suara = SUARA_DEFAULT

    # Buat Pesan Teks
    teks_hasil = f"📢 HASIL: {label_bersih.upper()} !!"

    return teks_hasil, file_suara

# --- 4. TAMPILAN ANTARMUKA (UI) ---
css_custom = """
.container {text-align: center;}
#judul-utama {color: #FF5733; font-size: 3em; font-weight: 900; text-align: center; margin-bottom: 0;}
#sub-judul {text-align: center; font-style: italic; margin-bottom: 20px;}
#kotak-hasil {
    font-size: 32px; 
    font-weight: bold; 
    text-align: center; 
    color: white; 
    background-color: #28B463; 
    padding: 10px; 
    border-radius: 10px;
}
"""

with gr.Blocks(theme=gr.themes.Soft(), css=css_custom) as demo:
    gr.Markdown("# ⚡ MAGIC SOUND DETECTOR ⚡", elem_id="judul-utama")
    gr.Markdown("Arahkan kamera ke Bakso, Tempe, atau Foto Pak Jokowi!", elem_id="sub-judul")
    
    with gr.Row():
        with gr.Column(scale=1):
            input_img = gr.Image(
                sources=["webcam", "upload"], 
                type="numpy", 
                label="📸 Input Kamera/Foto",
                height=400
            )
            btn = gr.Button("🔥 CEK SEKARANG 🔥", variant="primary", size="lg")
        
        with gr.Column(scale=1):
            # Output Teks
            txt_output = gr.Textbox(label="Hasil Deteksi", elem_id="kotak-hasil", interactive=False)
            # Output Audio
            audio_output = gr.Audio(label="🔊 Suara Efek", autoplay=True)

    # Logika Tombol
    btn.click(fn=deteksi_pasti, inputs=input_img, outputs=[txt_output, audio_output])

# Jalankan
demo.launch()