feat: Add domain analysis to scam detection AI

Author: google-labs-jules[bot]

collections/scam-detection-ai/email_analyzer.py

@@ -1,5 +1,17 @@
 import re
 
+import requests
+
+def get_domain_reputation(domain):
+    """
+    Checks the reputation of a domain using a mock API.
+    In a real application, this would integrate with a service like VirusTotal.
+    """
+    # This is a mock response.
+    if "fake" in domain or "suspicious" in domain:
+        return "Malicious"
+    return "Clean"
+
 def analyze_email(email_text):
     """
     Analyzes an email for potential scam indicators.
@@ -16,7 +28,8 @@ def analyze_email(email_text):
         "suspicious_links": 0,
         "unusual_sender": 0,
         "payment_requests": 0,
-        "attachments": 0
+        "attachments": 0,
+        "domain_analysis": {}
     }
 
     # Urgency keywords
@@ -54,6 +67,14 @@ def analyze_email(email_text):
     if "Content-Disposition: attachment" in email_text:
         scam_indicators["attachments"] = 1
 
+    # Domain analysis
+    urls = re.findall(r'http[s]?://(?:[a-zA-Z]|[0-9]|[$-_@.&+]|[!*\\(\\),]|(?:%[0-9a-fA-F][0-9a-fA-F]))+', email_text)
+    domains = [re.search(r'//(.*?)/', url).group(1) for url in urls if re.search(r'//(.*?)/', url)]
+
+    for domain in domains:
+        reputation = get_domain_reputation(domain)
+        scam_indicators["domain_analysis"][domain] = reputation
+
     return scam_indicators
 
 if __name__ == '__main__':

collections/scam-detection-ai/templates/result.html

@@ -17,9 +17,19 @@ <h1>Analysis Result</h1>
         <h2>{{ 'This email is likely a scam!' if is_scam else 'This email seems legitimate.' }}</h2>
         <ul>
             {% for indicator, value in analysis.items() %}
-                <li>{{ indicator.replace('_', ' ')|title }}: {{ 'Detected' if value else 'Not Detected' }}</li>
+                {% if indicator != 'domain_analysis' %}
+                    <li>{{ indicator.replace('_', ' ')|title }}: {{ 'Detected' if value else 'Not Detected' }}</li>
+                {% endif %}
             {% endfor %}
         </ul>
+        {% if analysis.domain_analysis %}
+            <h3>Domain Analysis</h3>
+            <ul>
+                {% for domain, status in analysis.domain_analysis.items() %}
+                    <li>{{ domain }}: {{ status }}</li>
+                {% endfor %}
+            </ul>
+        {% endif %}
     </div>
     <br>
     <a href="/">Analyze another email</a>

collections/scam-detection-ai/templates/results_list.html

@@ -21,9 +21,19 @@ <h3>Subject: {{ email.subject }}</h3>
                 <h4>{{ 'This email is likely a scam!' if email.is_scam else 'This email seems legitimate.' }}</h4>
                 <ul>
                     {% for indicator, value in email.analysis.items() %}
-                        <li>{{ indicator.replace('_', ' ')|title }}: {{ 'Detected' if value else 'Not Detected' }}</li>
+                        {% if indicator != 'domain_analysis' %}
+                            <li>{{ indicator.replace('_', ' ')|title }}: {{ 'Detected' if value else 'Not Detected' }}</li>
+                        {% endif %}
                     {% endfor %}
                 </ul>
+                {% if email.analysis.domain_analysis %}
+                    <h4>Domain Analysis</h4>
+                    <ul>
+                        {% for domain, status in email.analysis.domain_analysis.items() %}
+                            <li>{{ domain }}: {{ status }}</li>
+                        {% endfor %}
+                    </ul>
+                {% endif %}
             </div>
         {% endfor %}
     {% else %}

collections/scam-detection-ai/test_email_analyzer.py

@@ -20,6 +20,7 @@ def test_scam_email(self):
         self.assertEqual(analysis['generic_greeting'], 1)
         self.assertEqual(analysis['suspicious_links'], 1)
         self.assertEqual(analysis['payment_requests'], 1)
+        self.assertEqual(analysis['domain_analysis']['fake-prize.com'], 'Malicious')
 
     def test_legitimate_email(self):
         legitimate_email = """