app.py

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

# ---------- Load data & model (all CPU-friendly) ----------
faq_df = pd.read_csv("faqs.csv")
questions = faq_df["question"].tolist()
answers   = faq_df["answer"].tolist()

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

# ---------- Search function ----------
def semantic_search(user_query, top_k=3):
    query_embedding = model.encode(user_query, convert_to_tensor=True, normalize_embeddings=True)
    scores = util.cos_sim(query_embedding, question_embeddings)[0]
    top_k_idx = scores.topk(k=top_k).indices.cpu().numpy()
    
    results = []
    for idx in top_k_idx:
        results.append(
            {
                "FAQ Question": questions[idx],
                "FAQ Answer"  : answers[idx],
                "Similarity"  : f"{scores[idx]:.3f}"
            }
        )
    return results

# ---------- Gradio UI ----------
with gr.Blocks(title="MiniLM Semantic FAQ Search") as demo:
    gr.Markdown(
        """
        # 🔍 Semantic FAQ Search  
        Enter a salon-related question. The model finds the closest FAQs and displays their answers.
        """)
    
    with gr.Row():
        query_box = gr.Textbox(
            label="Ask a question",
            placeholder="e.g. Which spray protects hair from heat?"
        )
        topk_slider = gr.Slider(
            1, 5, value=3, step=1, label="Number of results"
        )
    search_btn = gr.Button("Search")
    out = gr.Dataframe(headers=["FAQ Question", "FAQ Answer", "Similarity"], visible=True, wrap=True)
    
    search_btn.click(semantic_search, [query_box, topk_slider], out)

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