src/streamlit_app.py

import streamlit as st
import pandas as pd
import numpy as np
import joblib
import os
import ipaddress
import tensorflow as tf
from tensorflow.keras.models import load_model
from tensorflow.keras.optimizers.legacy import SGD

# ==== File Paths ====
MODEL_FILE = "src/model1.h5"
WEIGHTS_FILE = "src/weights.h5"
SCALER_FILE = "src/standard_scaler1.pkl"
LABEL_ENCODER_FILE = "src/label_encoder1.pkl"

ENCODER_PATHS = {
    "proto": "src/categorical_label_encoder_proto1.pkl",
    "conn_state": "src/categorical_label_encoder_conn_state1.pkl",
    "history": "src/categorical_label_encoder_history1.pkl"
}

SAMPLE_FILE = "src/sample_input.csv"

# ==== Class for Malware Prediction ====
class MalwareClassifier:
    def __init__(self):
        for path in [MODEL_FILE, WEIGHTS_FILE, SCALER_FILE, LABEL_ENCODER_FILE] + list(ENCODER_PATHS.values()):
            if not os.path.exists(path):
                raise FileNotFoundError(f"Missing file: {path}")

        self.model = load_model(MODEL_FILE, compile=False)
        self.model.load_weights(WEIGHTS_FILE)
        self.model.compile(optimizer=SGD(), loss="categorical_crossentropy", metrics=["accuracy"])

        self.scaler = joblib.load(SCALER_FILE)
        self.label_encoder = joblib.load(LABEL_ENCODER_FILE)
        self.encoders = {k: joblib.load(v) for k, v in ENCODER_PATHS.items()}

    def _validate_numeric_column(self, col_name, values):
        if not values.astype(str).str.isdigit().all():
            raise ValueError(f"Non-integer value found in column {col_name}")
        if (values < 0).any():
            raise ValueError(f"Negative value found in column {col_name}")

    def _validate_ip_address(self, col_name, values):
        for value in values:
            try:
                ipaddress.ip_address(value)
            except ValueError:
                raise ValueError(f"Invalid IP address in column {col_name}: {value}")

    def _validate_input_data(self, data):
        required_columns = {
            'id.orig_h', 'id.orig_p', 'id.resp_h', 'id.resp_p',
            'proto', 'conn_state', 'history',
            'orig_pkts', 'orig_ip_bytes', 'resp_pkts', 'resp_ip_bytes'
        }

        if data.empty:
            raise ValueError("CSV is empty.")

        missing = required_columns - set(data.columns)
        if missing:
            raise ValueError(f"Missing required columns: {missing}")

        for col in data.columns:
            if col in {'id.orig_p', 'id.resp_p', 'orig_pkts', 'orig_ip_bytes', 'resp_pkts', 'resp_ip_bytes'}:
                self._validate_numeric_column(col, data[col])
            elif col in {'id.orig_h', 'id.resp_h'}:
                self._validate_ip_address(col, data[col])

    def _encode_data(self, df):
        for col in ['proto', 'conn_state', 'history']:
            df[col] = self.encoders[col].transform(df[col])
        df['id.orig_h'] = df['id.orig_h'].apply(lambda x: int(ipaddress.ip_address(x)))
        df['id.resp_h'] = df['id.resp_h'].apply(lambda x: int(ipaddress.ip_address(x)))
        return df

    def _preprocess_data(self, df):
        self._validate_input_data(df)
        df = self._encode_data(df)
        model_columns = [
            'id.orig_h', 'id.orig_p', 'id.resp_h', 'id.resp_p',
            'proto', 'conn_state', 'history',
            'orig_pkts', 'orig_ip_bytes', 'resp_pkts', 'resp_ip_bytes'
        ]
        return self.scaler.transform(df[model_columns])

    def predict(self, df):
        preprocessed = self._preprocess_data(df)
        preds = self.model.predict(preprocessed)

        results = []
        for pred in preds:
            label = self.label_encoder.inverse_transform([np.argmax(pred)])[0]
            scores = {label: f"{score:.6f}" for label, score in zip(self.label_encoder.classes_, pred)}
            results.append({"result": label, "scores": scores})
        return results

# ==== Streamlit UI ====
def main():
    st.set_page_config(page_title="Malware Detection System", page_icon="🛡️")
    st.title("🛡️ Malware Detection System")
    st.markdown("Upload a CSV file with network traffic logs to detect malware.")

    st.markdown("📄 Need a sample file to test? Download below:")

    if os.path.exists(SAMPLE_FILE):
        with open(SAMPLE_FILE, "rb") as f:
            st.download_button(
                label="📥 Download Sample CSV",
                data=f,
                file_name="sample_input.csv",  # ✅ Do NOT include `src/` in file_name
                mime="text/csv"
            )
    else:
        st.warning("⚠️ 'sample_input.csv' not found in the 'src' directory.")

    try:
        classifier = MalwareClassifier()
    except Exception as e:
        st.error(f"❌ Model loading failed: {e}")
        return

    uploaded_file = st.file_uploader("📂 Upload your network CSV", type=["csv"])

    if uploaded_file is not None:
        try:
            df = pd.read_csv(uploaded_file)
            df.columns = df.columns.str.strip()

            required_prediction_columns = [
                'id.orig_h', 'id.orig_p', 'id.resp_h', 'id.resp_p',
                'proto', 'conn_state', 'history',
                'orig_pkts', 'orig_ip_bytes', 'resp_pkts', 'resp_ip_bytes'
            ]

            missing_columns = set(required_prediction_columns) - set(df.columns)
            if missing_columns:
                st.error(f"❌ Missing required columns for prediction: {missing_columns}")
                st.write("📋 Detected columns:", df.columns.tolist())
                return

            prediction_input = df[required_prediction_columns].copy()
            predictions = classifier.predict(prediction_input)
            st.success("✅ Prediction complete!")

            for i, result in enumerate(predictions):
                st.subheader(f"Prediction {i + 1}")
                st.write(f"**Predicted Label:** {result['result']}")
                st.json(result['scores'])

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

if __name__ == "__main__":
    main()