app.py

import streamlit as st
import requests
import pandas as pd
import socket
import whois
from urllib.parse import urlparse
from bs4 import BeautifulSoup
from datetime import datetime
import pickle


def extract_features(url):
    try:
        socket.inet_aton(urlparse(url).netloc)
        having_IP_Address = 1
    except:
        having_IP_Address = 0

    URL_Length = 1 if len(url) >= 54 else 0

    try:
        response = requests.get(url, timeout=5)
        soup = BeautifulSoup(response.content, "html.parser")
        anchors = soup.find_all("a", href=True)
        if len(anchors) == 0:
            URL_of_Anchor = 1
        else:
            unsafe = [a for a in anchors if not a['href'].startswith(url)]
            URL_of_Anchor = 1 if len(unsafe) / len(anchors) > 0.5 else 0
    except:
        URL_of_Anchor = 1

    try:
        domain_info = whois.whois(urlparse(url).netloc)
        if isinstance(domain_info.creation_date, list):
            creation_date = domain_info.creation_date[0]
        else:
            creation_date = domain_info.creation_date
        age_of_domain = 1 if (datetime.now() - creation_date).days > 180 else 0
    except:
        age_of_domain = 0

    SSLfinal_State = 1 if url.startswith("https") else 0

    try:
        request_response = requests.get(url, timeout=5)
        if request_response.url == url:
            Request_URL = 0
        else:
            Request_URL = 1
    except:
        Request_URL = 1

    try:
        forms = soup.find_all("form", action=True)
        if len(forms) == 0:
            SFH = 1
        else:
            for form in forms:
                if form['action'] == "about:blank" or not form['action'].startswith("http"):
                    SFH = 1
                    break
            else:
                SFH = 0
    except:
        SFH = 1

    try:
        if "window.open" in response.text:
            popUpWidnow = 1
        else:
            popUpWidnow = 0
    except:
        popUpWidnow = 0

    return [SFH, popUpWidnow, SSLfinal_State, Request_URL, URL_of_Anchor, URL_Length, age_of_domain, having_IP_Address]


def predict_url(url, model):
    features = extract_features(url)
    X_columns = ['SFH', 'popUpWidnow', 'SSLfinal_State', 'Request_URL', 'URL_of_Anchor', 'URL_Length', 'age_of_domain', 'having_IP_Address']
    features_df = pd.DataFrame([features], columns=X_columns)
    prediction = model.predict(features_df)
    if prediction[0] == 1:
        return "Phishing"
    elif prediction[0] == 0:
        return "Legitimate"
    else:
        return "Unknown"


# Streamlit app configuration
st.set_page_config(page_title='Phishing URL Detection', layout='centered')

# App Header
st.markdown("""
    <style>
    body { background-color: #f0f2f6; }
    .main { background-color: white; padding: 2rem; border-radius: 12px; box-shadow: 0 0 10px rgba(0,0,0,0.1); }
    </style>
""", unsafe_allow_html=True)

st.title('🔍 Phishing URL Detection App')
st.write('Enter a URL to check if it is Phishing or Legitimate.')


# Load the trained model
with open('phishing_model.pkl', 'rb') as f:
    model = pickle.load(f)
    

# Input URL
url_input = st.text_input('Enter URL:', '')

if st.button('Check URL'):
    if url_input:
        try:
            # Make prediction
            result = predict_url(url_input, model)

            if result == 'Phishing':
                st.error('🚨 This URL is likely a **Phishing Site**. Be careful!')
            elif result == 'Legitimate':
                st.success('✅ This URL is likely **Legitimate**.')
            else:
                st.warning('⚠️ Unable to determine. Try again later.')

        except Exception as e:
            st.error(f'Error: {e}')
    else:
        st.warning('Please enter a valid URL.')