app.py

import streamlit as st
from pymongo import MongoClient
import pandas as pd
from sentence_transformers import SentenceTransformer
from sklearn.metrics.pairwise import cosine_similarity

# Load embedding model
@st.cache_resource
def load_model():
    return SentenceTransformer("all-MiniLM-L6-v2")

model = load_model()

# MongoDB connection
MONGO_URI = "mongodb+srv://muthurajlingam788:VHassOJn4N4niYqg@cluster0.vgobnxf.mongodb.net/mydatabase?retryWrites=true&w=majority"
client = MongoClient(MONGO_URI)
db = client["shopping_mall"]
collection = db["customers"]

# Streamlit UI
st.title("🛍️ Shopping Mall Customer Management")

menu = st.sidebar.radio("Menu", ["Add Customer", "View Customers", "Ask AI"])

# Insert new customer
if menu == "Add Customer":
    st.header("Add Customer Details")
    name = st.text_input("Name")
    age = st.number_input("Age", min_value=0)
    gender = st.selectbox("Gender", ["Male", "Female", "Other"])
    items = st.text_input("Items Purchased (comma-separated)")
    amount = st.number_input("Total Amount Spent (₹)", min_value=0.0)

    if st.button("Submit"):
        customer = {
            "name": name,
            "age": age,
            "gender": gender,
            "items": [item.strip() for item in items.split(",")],
            "amount": amount
        }
        collection.insert_one(customer)
        st.success("Customer added successfully!")

# View customers
elif menu == "View Customers":
    st.header("Customer List")
    data = list(collection.find({}, {"_id": 0}))  # Exclude _id field

    if data:
        df = pd.DataFrame(data)
        search = st.text_input("Search by Name")
        if search:
            df = df[df["name"].str.contains(search, case=False)]
        st.dataframe(df)
    else:
        st.info("No customer records found.")

# AI Query Assistant
elif menu == "Ask AI":
    st.header("🧠 Ask Questions About Customers")
    question = st.text_input("Enter your question")

    # Load and format data
    data = list(collection.find({}, {"_id": 0}))
    if not data:
        st.warning("No data available to query.")
    else:
        context_texts = []
        for c in data:
            context_texts.append(
                f"{c['name']} is a {c['age']}-year-old {c['gender']} who bought {', '.join(c['items'])} and spent ₹{c['amount']}."
            )

        # Encode and search best match
        question_embedding = model.encode([question])
        context_embeddings = model.encode(context_texts)

        scores = cosine_similarity(question_embedding, context_embeddings)[0]
        best_idx = scores.argmax()
        response = context_texts[best_idx]

        if question:
            st.write("📌 Most Relevant Info:")
            st.success(response)