app.py

import gradio as gr
import torch
import torch.nn as nn
import pickle
import numpy as np
import os

# ---------------------------------------------------------
# 1. Define the Missing Model Architecture
# ---------------------------------------------------------
# Your .pt file only has weights. We need this class to hold them.
class PhishingNet(nn.Module):
    def __init__(self, input_size=5, hidden_size=10, output_size=2):
        super(PhishingNet, self).__init__()
        # Standard architecture for these types of models
        self.fc1 = nn.Linear(input_size, hidden_size)
        self.relu = nn.ReLU()
        self.fc2 = nn.Linear(hidden_size, output_size)
        
    def forward(self, x):
        out = self.fc1(x)
        out = self.relu(out)
        out = self.fc2(out)
        return out

# ---------------------------------------------------------
# 2. Define a "Dummy" Scaler (Backup Plan)
# ---------------------------------------------------------
# If your scaler.pkl is broken, we use this so the code doesn't crash.
class DummyScaler:
    def transform(self, x):
        return x # Do nothing, just pass the data through

# ---------------------------------------------------------
# 3. Load Resources (Safely)
# ---------------------------------------------------------
MODEL_PATH = "models/phishing_rf_model.pt"
SCALER_PATH = "models/scaler.pkl"

model = None
scaler = None
status_log = "System Startup Log:\n"

# --- Try Loading Scaler ---
if os.path.exists(SCALER_PATH):
    try:
        with open(SCALER_PATH, "rb") as f:
            scaler = pickle.load(f)
        status_log += "✅ Scaler loaded successfully.\n"
    except Exception as e:
        status_log += f"⚠️ Scaler is corrupt ({str(e)}). Using raw data instead.\n"
        scaler = DummyScaler()
else:
    status_log += "⚠️ Scaler file not found. Using raw data.\n"
    scaler = DummyScaler()

# --- Try Loading Model ---
if os.path.exists(MODEL_PATH):
    try:
        # Load the weights (OrderedDict)
        state_dict = torch.load(MODEL_PATH, map_location=torch.device('cpu'))
        
        # Initialize the architecture
        # We try to guess the shape from the weights if possible
        input_shape = 5 # Default
        if 'fc1.weight' in state_dict:
            input_shape = state_dict['fc1.weight'].shape[1]
            
        model = PhishingNet(input_size=input_shape)
        model.load_state_dict(state_dict)
        model.eval()
        status_log += f"✅ Model weights loaded (Input size: {input_shape}).\n"
    except Exception as e:
        status_log += f"❌ MODEL CRASH: {str(e)}\n"
        model = None
else:
    status_log += "❌ Model file not found.\n"

# ---------------------------------------------------------
# 4. Feature Extraction & Prediction
# ---------------------------------------------------------
def extract_features(url: str) -> np.ndarray:
    length = len(url)
    dots = url.count('.')
    hyphens = url.count('-')
    digits = sum(c.isdigit() for c in url)
    at_sign = url.count('@')
    return np.array([[length, dots, hyphens, digits, at_sign]], dtype=float)

def predict_phishing(url):
    if model is None:
        return {"Error": 0}, "Model failed to load. Check status."
    
    if not url:
        return None, "Please enter a URL."

    try:
        # 1. Extract Features
        features = extract_features(url)
        
        # 2. Scale (Real or Dummy)
        features_scaled = scaler.transform(features)
        
        # 3. Convert to Tensor
        features_tensor = torch.tensor(features_scaled, dtype=torch.float32)
        
        # 4. Predict
        with torch.no_grad():
            logits = model(features_tensor)
            probs = torch.nn.functional.softmax(logits, dim=1)
            
            # Map probabilities (Index 0 = Safe, Index 1 = Phishing)
            safe_conf = float(probs[0][0])
            phish_conf = float(probs[0][1])

        return {"✅ Safe": safe_conf, "🚨 Phishing": phish_conf}, "Success"

    except Exception as e:
        return {"Error": 0}, f"Prediction Error: {str(e)}"

# ---------------------------------------------------------
# 5. UI Setup
# ---------------------------------------------------------
with gr.Blocks(theme=gr.themes.Soft()) as iface:
    gr.Markdown("# 🛡️ PhishScope Final")
    
    with gr.Row():
        url_input = gr.Textbox(label="URL", placeholder="https://example.com")
        btn = gr.Button("Scan", variant="primary")
        
    with gr.Row():
        label_out = gr.Label(label="Result")
        debug_out = gr.Textbox(label="System Status (Debug)", value=status_log, lines=5)

    btn.click(fn=predict_phishing, inputs=url_input, outputs=[label_out, debug_out])

iface.launch()