Update app.py

app.py

@@ -2,37 +2,55 @@ import gradio as gr
 import numpy as np
 import matplotlib.pyplot as plt
 import matplotlib
-from transformers import pipeline
-import asyncio
-
 matplotlib.use('Agg')
+from transformers import pipeline
 
-# Load the phishing detection model asynchronously
+# Load the phishing detection model
 model = pipeline('text-classification', model="Ajay1311/phish")
 
-# Function to create a simplified gauge chart
-def create_simple_gauge_chart(confidence, is_phishing):
+# Function to create a gauge chart for confidence visualization
+def create_gauge_chart(confidence, is_phishing):
+    # Create figure and axis
     fig, ax = plt.subplots(figsize=(6, 3), subplot_kw={'projection': 'polar'})
     
+    # Set gauge properties
     gauge_min, gauge_max = 0, 100
+    gauge_range = gauge_max - gauge_min
+    
+    # Convert confidence to percentage
     confidence_pct = confidence * 100
-    color = 'red' if is_phishing else 'green'
-    label = f"PHISHING: {confidence_pct:.1f}%" if is_phishing else f"BENIGN: {confidence_pct:.1f}%"
     
-    # Plotting the filled gauge chart
-    ax.plot([0, np.pi * (confidence_pct / 100)], [0, 1], color=color, linewidth=10)
+    # Set colors based on phishing status
+    if is_phishing:
+        color = 'red'
+        label = f"PHISHING DETECTED: {confidence_pct:.1f}% Confidence"
+    else:
+        color = 'green'
+        label = f"BENIGN: {confidence_pct:.1f}% Confidence"
+    
+    # Plot gauge
+    theta = np.linspace(np.pi/2, -np.pi/2, 100)
+    ax.plot(theta, [1]*100, color='lightgray', linewidth=10, alpha=0.5)
     
+    # Plot filled portion
+    filled_theta = np.linspace(np.pi/2, np.pi/2 - (confidence_pct/100) * np.pi, 100)
+    ax.plot(filled_theta, [1]*len(filled_theta), color=color, linewidth=10)
+    
+    # Customize gauge appearance
     ax.set_rticks([])  # No radial ticks
     ax.set_xticks([])  # No angular ticks
-    ax.spines['polar'].set_visible(False)
+    ax.spines['polar'].set_visible(False)  # Hide the circular spine
     
+    # Add text
     ax.text(0, 0, label, ha='center', va='center', fontsize=12, fontweight='bold')
+    
+    # Set limits
     ax.set_ylim(0, 1.5)
     
     return fig
 
-# Function to analyze phishing asynchronously
-async def analyze_phishing(text):
+# Function to analyze and classify text
+def analyze_phishing(text):
     if not text.strip():
         return {
             "result": "Please enter a URL or email text to analyze",
@@ -40,23 +58,25 @@ async def analyze_phishing(text):
             "analysis": "No input provided"
         }
     
-    # Get model prediction
-    result = await asyncio.to_thread(model, text)
+    # Get model prediction - this returns the format you showed
+    result = model(text)
     
+    # Extract label and score from your specific output format
     label = result[0]['label']
     score = result[0]['score']
     
+    # Determine if phishing
     is_phishing = label.lower() == 'phishing'
     
-    # Create simplified chart
-    chart = create_simple_gauge_chart(score, is_phishing)
+    # Create visualization
+    chart = create_gauge_chart(score, is_phishing)
     
     # Format the raw JSON for display
     raw_json = f"""```json
 {str(result)}
 ```"""
     
-    # Generate analysis based on phishing or benign classification
+    # Generate detailed analysis
     if is_phishing:
         analysis = f"""
         ⚠️ **POTENTIAL PHISHING DETECTED**
@@ -65,6 +85,14 @@ async def analyze_phishing(text):
         
         **Raw Model Output:**
         {raw_json}
+        
+        **Common phishing indicators:**
+        - Suspicious URLs or email domains
+        - Urgent requests for personal information
+        - Grammatical errors or unusual formatting
+        - Requests to click on suspicious links
+        
+        **Recommendation:** Exercise caution and verify the source before proceeding.
         """
     else:
         analysis = f"""
@@ -74,6 +102,13 @@ async def analyze_phishing(text):
         
         **Raw Model Output:**
         {raw_json}
+        
+        **Always exercise caution when:**
+        - Sharing personal information online
+        - Clicking on links from unknown sources
+        - Responding to unexpected requests
+        
+        **Recommendation:** Continue with normal caution.
         """
     
     return {
@@ -82,6 +117,15 @@ async def analyze_phishing(text):
         "analysis": analysis
     }
 
+# Example inputs for demonstration
+examples = [
+    ["Please verify your account by clicking on the link: http://amaz0n-security-alert.com/verify"],
+    ["Your Netflix subscription is about to expire. Update payment at netfl1x-accounts.com/renew"],
+    ["Meeting scheduled for tomorrow at 2 PM in the conference room. Agenda attached."],
+    ["Your Amazon order #12345 has been shipped and will arrive on Friday."],
+    ["URGENT: Your account has been compromised! Click here to reset: bit.ly/2xCvZ9"]
+]
+
 # Create Gradio interface with custom theme
 theme = gr.themes.Soft(
     primary_hue="blue",
@@ -94,7 +138,6 @@ theme = gr.themes.Soft(
     block_title_text_color="*primary_500",
 )
 
-# Gradio interface
 with gr.Blocks(theme=theme, css=""" 
     .footer {text-align: center; margin-top: 20px; color: #666;}
     .container {max-width: 800px; margin: 0 auto;}
@@ -106,19 +149,29 @@ with gr.Blocks(theme=theme, css="""
     .safe {color: #2ca02c; background: rgba(44, 160, 44, 0.1); padding: 10px; border-radius: 5px;}
 """) as demo:
     with gr.Column(elem_classes="container"):
-        gr.HTML("<div class='header'><h1>🛡️ PhishGuard AI</h1><p>Advanced phishing detection powered by machine learning</p></div>")
+        # Header
+        gr.HTML("""
+        <div class="header">
+            <h1>🛡️ PhishGuard AI</h1>
+            <p>Advanced phishing detection powered by machine learning</p>
+        </div>
+        """)
         
-        # Input
+        # Input section
         with gr.Group():
             gr.Markdown("### Enter URL or Email Text to Analyze")
-            input_text = gr.Textbox(placeholder="Paste a suspicious URL, email, or message here...", lines=5, label="Input Text")
+            input_text = gr.Textbox(
+                placeholder="Paste a suspicious URL, email, or message here...",
+                lines=5,
+                label="Input Text"
+            )
         
-        # Buttons
+        # Action buttons
         with gr.Row():
             analyze_btn = gr.Button("🔍 Analyze", variant="primary")
             clear_btn = gr.Button("🗑️ Clear")
         
-        # Output
+        # Output section
         with gr.Group():
             gr.Markdown("### Analysis Results")
             with gr.Row():
@@ -129,13 +182,8 @@ with gr.Blocks(theme=theme, css="""
             
             analysis_md = gr.Markdown(label="Detailed Analysis")
         
-        # Set up Examples (for demonstration)
-        examples = [
-            ["Please verify your account by clicking on the link: http://amaz0n-security-alert.com/verify"],
-            ["Your Netflix subscription is about to expire. Update payment at netfl1x-accounts.com/renew"],
-            ["Meeting scheduled for tomorrow at 2 PM in the conference room. Agenda attached."]
-        ]
-        
+        # Examples section
+        gr.Markdown("### Example Inputs")
         gr.Examples(
             examples=examples,
             inputs=input_text,
@@ -145,7 +193,11 @@ with gr.Blocks(theme=theme, css="""
         )
         
         # Footer
-        gr.HTML("<div class='footer'><p>Created for Cybersecurity Hackathon 2025 | PhishGuard AI Team</p></div>")
+        gr.HTML("""
+        <div class="footer">
+            <p>Created for Cybersecurity Hackathon 2025 | PhishGuard AI Team</p>
+        </div>
+        """)
         
         # Set up event handlers
         analyze_btn.click(
@@ -159,6 +211,21 @@ with gr.Blocks(theme=theme, css="""
             inputs=None,
             outputs=[input_text, result_plot, analysis_md]
         )
+        
+        # Update CSS class based on result
+        def update_result_class(result):
+            if "PHISHING DETECTED" in result:
+                return "phishing"  # Set class to "phishing" for phishing results
+            elif "BENIGN" in result:
+                return "safe"  # Set class to "safe" for benign results
+            return ""  # Default to no class if no match
+
+        result_text.change(
+            update_result_class,
+            inputs=result_text,
+            outputs=result_text  # Use the component itself as output
+        )
 
-# Increase the timeout to allow model inference to finish
-demo.launch(server_name="0.0.0.0", server_port=7860, share=True, timeout=120)
+# Launch the app
+if __name__ == "__main__":
+    demo.launch()