app.py

import pandas as pd
from sentence_transformers import SentenceTransformer, util
import gradio as gr

# ─── Load catalog & compute embeddings ─────────────────────────
PRODUCTS_CSV = "products.csv"
df = pd.read_csv(PRODUCTS_CSV)
descriptions = df["description"].tolist()

model = SentenceTransformer("sentence-transformers/all-MiniLM-L6-v2")
embeddings = model.encode(descriptions, convert_to_tensor=True, normalize_embeddings=True)

# ─── Semantic-search function ─────────────────────────────────
def search_products(query: str, top_k: int = 5):
    if not query.strip():
        return pd.DataFrame(columns=df.columns.tolist() + ["score"])
    q_emb = model.encode(query, convert_to_tensor=True, normalize_embeddings=True)
    hits = util.cos_sim(q_emb, embeddings)[0].topk(k=top_k)
    indices = hits.indices.cpu().tolist()
    scores = [round(float(s), 3) for s in hits.values.cpu().tolist()]

    results = df.iloc[indices].copy()
    results["score"] = scores
    return results.reset_index(drop=True)

# ─── Gradio UI ────────────────────────────────────────────────
with gr.Blocks(title="🛍️ Salon Catalog Semantic Search") as demo:
    gr.Markdown("""
    # 🛍️ Salon Product Search  
    **Natural-language queries** → **top matching products** via MiniLM embeddings  
    (Runs entirely on free CPU; no paid APIs)
    """)

    with gr.Row():
        query_in = gr.Textbox(placeholder="e.g. sulfate-free shampoo under $15", label="Search query")
        topk     = gr.Slider(1, 10, value=5, step=1, label="Number of results")
    search_btn = gr.Button("Search 🔍", variant="primary")

    table = gr.Dataframe(
        headers=list(df.columns) + ["score"],
        datatype=["number","str","str","str","number","number"],
        interactive=False,
        row_count= topk.value,
        label="Results"
    )

    search_btn.click(search_products, [query_in, topk], table)

if __name__ == "__main__":
    demo.launch(server_name="0.0.0.0")