app.py

import joblib
import gradio as gr
import numpy as np

# =========================
# تحميل البيانات
# =========================
books = joblib.load("books.pkl")
theses = joblib.load("Theses.pkl")
books_emb = joblib.load("books_embeddings.pkl")
theses_emb = joblib.load("theses_embeddings.pkl")

# =========================
# مثال دالة بحث بسيطة
# =========================
from sklearn.metrics.pairwise import cosine_similarity

def search_library(query_embedding, top_k=5):
    # دمج كل Embeddings
    all_emb = np.vstack([books_emb, theses_emb])
    similarity = cosine_similarity([query_embedding], all_emb)[0]
    top_indices = similarity.argsort()[-top_k:][::-1]
    
    results = []
    for idx in top_indices:
        if idx < len(books):
            results.append(books[idx])
        else:
            results.append(theses[idx - len(books)])
    return results

# =========================
# واجهة Gradio
# =========================
def search_interface(query_embedding):
    return search_library(query_embedding)

demo = gr.Interface(
    fn=search_interface,
    inputs=gr.inputs.Dataframe(),  # هنا ممكن تغيّري حسب نوع المدخل
    outputs="text",
    title="Library Search",
    description="بحث في مكتبة الكتب والرسائل"
)

demo.launch()