app.py

import streamlit as st
import joblib
import string
import re
import spacy
import time

st.set_page_config(page_title="Spam Email Detector",page_icon="✉️")

@st.cache_resource(ttl=3600)
def load_model():
    learn_inf = joblib.load('checkpoints/spam_detection_model.pkl')
    vectorizer = joblib.load('checkpoints/count_vectorizer.pkl')
    return learn_inf,vectorizer


def clean_text(s): 
    for cs in s:
        if  not cs in string.ascii_letters:
            s = s.replace(cs, ' ')
    return s.rstrip('\r\n')

def remove_little(s): 
    wordsList = s.split()
    k_length=2
    resultList = [element for element in wordsList if len(element) > k_length]
    resultString = ' '.join(resultList)
    return resultString

def lemmatize_text(text):
    nlp = spacy.load("en_core_web_sm")
    doc = nlp(text)
    lemmatized_text = " ".join([token.lemma_ for token in doc])
    return lemmatized_text

def preprocess(text):
    return lemmatize_text(remove_little(clean_text(text)))

def classify_email(model,vectorizer,email):
    prediction = model.predict(vectorizer.transform([email]))
    return prediction

def main():
    st.title("Spam Email Detector")
    output = st.empty()
    status_bar = st.empty()

    with st.spinner('Loading the webpage...'):
        user_input = st.text_area('Enter the email text:', '',placeholder='Congratulations!! You have won Rs. 100000. \nClick here to Redeem!!')
    
    if st.button("Check for Spam"):
        output.empty()
        status_bar.empty()
        if user_input:
            with status_bar.status("Loading the model.....", expanded=True) as status:
                model,vectorizer = load_model()
                time.sleep(2)

                # st.write("Analyzing the email.....")
                status.update(label="Analyzing the email.....!", state="running", expanded=True)
                user_input = preprocess(user_input)
                time.sleep(2)

                # st.write("Checking for Spam.....")
                status.update(label="Checking for Spam.....", state="running", expanded=True)
                prediction = classify_email(model,vectorizer,user_input)
                time.sleep(2)

                status.update(label="Detection Completed!", state="complete", expanded=False)


            status_bar.empty()
            if prediction == 1:
                output.error('Spam Detected!')
            else:
                output.success('Not Spam')
        else:
            output.warning("Kindly enter the text to detect !!")

if __name__ == "__main__":
    main()