import streamlit as st
import os
import zipfile
import numpy as np
import pickle
import tensorflow as tf
from tensorflow.keras.applications.resnet50 import ResNet50, preprocess_input
from tensorflow.keras.preprocessing import image
from tensorflow.keras.layers import GlobalMaxPool2D
from sklearn.neighbors import NearestNeighbors
from numpy.linalg import norm
from PIL import Image

# --------------------------------
# 1. ZIP DOSYASINI AÇMA (En Önemli Kısım)
# --------------------------------
# Eğer images klasörü boşsa veya yoksa zip'i açar
if not os.path.exists("images") or len(os.listdir("images")) < 10:
    if os.path.exists("images.zip"):
        with st.spinner("Resimler paketten çıkarılıyor (296 MB)..."):
            with zipfile.ZipFile("images.zip", "r") as zip_ref:
                zip_ref.extractall(".")
    else:
        st.error("Hata: images.zip dosyası bulunamadı!")

# --------------------------------
# SAYFA AYARI
# --------------------------------
st.set_page_config(page_title="Moda Öneri Sistemi", layout="centered")
st.title("🛍️ Moda Öneri Sistemi")

# --------------------------------
# MODEL YÜKLE
# --------------------------------
@st.cache_resource
def load_model():
    base_model = ResNet50(weights="imagenet", include_top=False, input_shape=(224,224,3))
    base_model.trainable = False
    model = tf.keras.models.Sequential([base_model, GlobalMaxPool2D()])
    return model

model = load_model()

# --------------------------------
# DATA YÜKLE
# --------------------------------
@st.cache_resource
def load_data():
    features = np.array(pickle.load(open("Images_features.pkl","rb")))
    filenames = pickle.load(open("filenames.pkl","rb"))
    return features, filenames

feature_list, filenames = load_data()

# --------------------------------
# FEATURE ÇIKARMA
# --------------------------------
def extract_features(img_path, model):
    img = image.load_img(img_path, target_size=(224,224))
    img_array = image.img_to_array(img)
    expanded_img = np.expand_dims(img_array, axis=0)
    preprocessed = preprocess_input(expanded_img)
    result = model.predict(preprocessed).flatten()
    normalized = result / norm(result)
    return normalized

# --------------------------------
# ARAYÜZ VE RESİM YÜKLEME
# --------------------------------
uploaded_file = st.file_uploader("Bir kıyafet resmi yükleyin", type=["jpg","png","jpeg"])

if uploaded_file is not None:
    img = Image.open(uploaded_file)
    st.image(img, width=300, caption="Seçtiğiniz ürün")

    with open("temp.jpg","wb") as f:
        f.write(uploaded_file.getbuffer())

    with st.spinner("Benzer ürünler bulunuyor..."):
        input_feature = extract_features("temp.jpg", model)
        neighbors = NearestNeighbors(n_neighbors=6, algorithm="brute", metric="euclidean")
        neighbors.fit(feature_list)
        distances, indices = neighbors.kneighbors([input_feature])

    st.subheader("✨ Benzer Ürünler")
    cols = st.columns(5)

    # Akıllı Resim Arama (Hangi klasörde olursa olsun bulur)
    image_db = {}
    for root, dirs, files in os.walk('images'):
        for f in files:
            image_db[f.lower()] = os.path.join(root, f)

    for i in range(1, 6):
        with cols[i-1]:
            raw_path = filenames[indices[0][i]].replace("\\", "/")
            file_name = os.path.basename(raw_path).lower()

            if file_name in image_db:
                st.image(image_db[file_name], use_container_width=True)
            else:
                st.write("Eksik:", file_name)

# Hata Ayıklama Paneli (Yan tarafta)
if os.path.exists("images"):
    file_count = sum([len(files) for r, d, files in os.walk("images")])
    st.sidebar.write(f"📂 images klasöründeki resim sayısı: {file_count}")