app.py

import streamlit as st
import pickle
import nltk
from nltk.stem.porter import PorterStemmer

ps = PorterStemmer()

def transform_text(text):
    text = text.lower()
    words = nltk.word_tokenize(text)
    words = [ps.stem(i) for i in words if i.isalnum()]

    return " ".join(words)

tfidf = pickle.load(open('vectorizer.pkl', 'rb'))
model = pickle.load(open('model.pkl', 'rb'))

st.title("SMS Spam Classifier")

input_sms = st.text_input("Enter the message")

if st.button('Predict'):
    # 1. preprocess
    transformed_sms = transform_text(input_sms)
    # 2. vectorize
    vector_input = tfidf.transform([transformed_sms]).toarray()
    # 3. predict
    result = model.predict(vector_input)[0]
    # 4. Display
    if result == 1:
        st.header("Spam")
    else:
        st.header("Not Spam")