app.py

import streamlit as st
import pandas as pd
import joblib
import pickle
import numpy as np

# Load model and preprocessing artifacts
model = joblib.load("ensemble_model.pkl")
with open("features_to_drop.pkl", "rb") as f:
    features_to_drop = pickle.load(f)

# Column names from the raw 49-column dataset (before feature engineering)
raw_columns = [
    'srcip', 'sport', 'dstip', 'dsport', 'proto', 'state', 'dur', 'sbytes', 'dbytes',
    'sttl', 'dttl', 'sloss', 'dloss', 'service', 'Sload', 'Dload', 'Spkts', 'Dpkts',
    'swin', 'dwin', 'stcpb', 'dtcpb', 'smeansz', 'dmeansz', 'trans_depth', 'res_bdy_len',
    'Sjit', 'Djit', 'Stime', 'Ltime', 'Sintpkt', 'Dintpkt', 'tcprtt', 'synack', 'ackdat',
    'is_sm_ips_ports', 'ct_state_ttl', 'ct_flw_http_mthd', 'is_ftp_login', 'ct_ftp_cmd',
    'ct_srv_src', 'ct_srv_dst', 'ct_dst_ltm', 'ct_src_ ltm', 'ct_src_dport_ltm',
    'ct_dst_sport_ltm', 'ct_dst_src_ltm', 'attack_cat', 'Label'
]

# Function to preprocess a single input row
def preprocess_input(row_values):
    if len(row_values) != 49:
        raise ValueError(f"❌ Expected 49 values, but got {len(row_values)}.")
    
    # Create DataFrame from input
    input_df = pd.DataFrame([row_values], columns=raw_columns)

    # Convert all columns to numeric
    input_df = input_df.apply(pd.to_numeric, errors='coerce')

    # Feature engineering
    input_df['duration'] = input_df['Ltime'] - input_df['Stime']
    input_df['byte_ratio'] = input_df['sbytes'] / (input_df['dbytes'] + 1)
    input_df['pkt_ratio'] = input_df['Spkts'] / (input_df['Dpkts'] + 1)

    # ✅ Fix: convert features_to_drop to list before adding with another list
    input_df = input_df.drop(columns=list(features_to_drop) + ['attack_cat', 'Label'], errors='ignore')

    return input_df

# Streamlit UI
st.title("🌐 ML-Driven Anomaly Detector")
st.markdown("Paste a **single row** of raw features from the dataset (49 values, tab-separated):")

user_input = st.text_area("Input Row", height=150)

if st.button("Predict"):
    try:
        # Parse the input
        values = user_input.strip().split("\t")

        # Preprocess the input row
        processed_df = preprocess_input(values)

        # Predict using the preprocessed data
        prediction = model.predict(processed_df)[0]
        st.success(f"✅ Predicted Attack Category: **{prediction}**")

    except Exception as e:
        st.error(f"❌ Error processing input: {e}")