import streamlit as st
import numpy as np
from PIL import Image
import base64
import io
from tensorflow.keras.models import load_model

st.set_page_config(page_title="Hurma Sınıflandırıcı", layout="centered")

st.title("📷 Hurma Resmi Sınıflandırma")
st.write("Lütfen bir hurma resmi yükleyin ve hangi tür olduğunu tahmin edelim.")

# --- MODEL ---
try:
    model = load_model("src/dates_classifier_model.h5")
except Exception as e:
    st.error(f"Model yüklenemedi: {e}")
    st.stop()

class_names = [
    'Rutab', 'Meneifi', 'Sokari', 'Galaxy', 'Shaishe',
    'Medjool', 'Ajwa', 'Nabtat Ali', 'Sugaey'
]

# --- IMAGE SESSION ---
def image_to_base64(image_bytes):
    return base64.b64encode(image_bytes).decode("utf-8")

def base64_to_image(base64_str):
    return Image.open(io.BytesIO(base64.b64decode(base64_str))).convert("RGB")

def process_image(img):
    img = img.resize((224, 224))
    img = np.array(img) / 255.0
    img = np.expand_dims(img, axis=0)
    return img

if "image_data" not in st.session_state:
    st.session_state.image_data = None

uploaded_file = st.file_uploader("Resim Seç (.jpg, .jpeg, .png)", type=["jpg", "jpeg", "png"])

if uploaded_file is not None:
    st.session_state.image_data = image_to_base64(uploaded_file.read())

if st.session_state.image_data:
    try:
        img = base64_to_image(st.session_state.image_data)
        st.image(img, caption="Yüklenen Resim", use_column_width=True)

        processed_img = process_image(img)
        prediction = model.predict(processed_img)
        predicted_class = np.argmax(prediction)

        st.success(f"Tahmin: **{class_names[predicted_class]}**")
    except Exception as e:
        st.error(f"Fotoğraf işlenemedi: {e}")