Update app.py

app.py

@@ -119,77 +119,90 @@ class ImageScraper:
 
 
     def extract_key_topics(self, script: str) -> List[str]:
-        """Extract key topics from a long text prompt with improved accuracy"""
+        """Extract key topics from script with improved VaultGenix-specific processing"""
         try:
             # Define relevant categories for VaultGenix
             categories = {
-                'security': ['security', 'encryption', 'protection', 'privacy', 'safe', 'secure'],
-                'digital': ['digital', 'online', 'virtual', 'cyber', 'electronic'],
-                'legacy': ['legacy', 'inheritance', 'heir', 'posthumous', 'estate'],
-                'management': ['management', 'planning', 'organization', 'control', 'administration'],
-                'technology': ['AI', 'artificial intelligence', 'technology', 'platform', 'system'],
-                'family': ['family', 'heir', 'custodian', 'relative', 'loved ones']
+                'legacy': [
+                    'digital legacy', 'legacy management', 'digital estate', 
+                    'posthumous', 'inheritance', 'heir', 'custodian'
+                ],
+                'security': [
+                    'encryption', 'security', 'protection', 'privacy', 'AES-256',
+                    'data security', 'secure', 'authentication'
+                ],
+                'technology': [
+                    'AI', 'artificial intelligence', 'platform', 'digital', 
+                    'automation', 'analytics'
+                ],
+                'management': [
+                    'asset management', 'directive', 'planning', 'preservation',
+                    'customization', 'optimization'
+                ],
+                'identity': [
+                    'digital identity', 'presence', 'account', 'profile',
+                    'digital footprint'
+                ]
             }
             
             # Process text
             text = script.lower()
             found_topics = set()
             
-            # Extract single-word matches
-            words = text.split()
+            # Extract category-based matches
             for category, terms in categories.items():
                 for term in terms:
                     if term in text:
+                        # Add both the term and its category combination
                         found_topics.add(term)
-                        found_topics.add(category)
+                        if category in ['legacy', 'security', 'technology']:
+                            found_topics.add(f"digital {term}")
+                            found_topics.add(f"{category} management")
             
-            # Extract meaningful phrases
+            # Extract key compound phrases
             important_phrases = [
-                'digital legacy',
-                'legacy management',
-                'digital security',
-                'data protection',
-                'artificial intelligence',
-                'digital estate',
-                'digital identity',
-                'secure platform',
-                'family protection',
-                'digital inheritance'
+                'digital legacy management',
+                'AI-driven platform',
+                'digital estate planning',
+                'legacy preservation',
+                'secure inheritance',
+                'digital asset protection',
+                'intelligent legacy system',
+                'automated legacy management',
+                'digital identity preservation',
+                'secure legacy platform'
             ]
             
             for phrase in important_phrases:
-                if phrase in text:
+                if phrase.lower() in text:
                     found_topics.add(phrase)
             
-            # Combine related topics
-            combined_topics = []
-            for topic in found_topics:
-                # Create meaningful combinations
-                if topic in ['digital', 'secure', 'smart', 'AI']:
-                    related = ['legacy', 'security', 'protection', 'management']
-                    for rel in related:
-                        if rel in found_topics:
-                            combined_topics.append(f"{topic} {rel}")
-            
-            # Add combined topics to results
-            found_topics.update(combined_topics)
-            
-            # Prioritize topics
-            priority_topics = [
-                topic for topic in found_topics 
-                if any(key in topic for key in ['digital', 'security', 'legacy', 'AI'])
-            ]
-            
-            # Ensure we have enough topics
-            if len(priority_topics) < 3:
-                priority_topics.extend(['digital security', 'legacy management', 'data protection'][:3 - len(priority_topics)])
+            # Prioritize topics based on VaultGenix focus
+            priority_topics = sorted(
+                found_topics,
+                key=lambda x: (
+                    'digital legacy' in x,
+                    'security' in x or 'secure' in x,
+                    'AI' in x.lower() or 'intelligence' in x.lower(),
+                    'management' in x,
+                    len(x.split())  # Prefer compound terms
+                ),
+                reverse=True
+            )
             
-            return list(set(priority_topics))[:5]  # Return top 5 unique topics
+            # Return top unique topics
+            return list(dict.fromkeys(priority_topics))[:8]
             
         except Exception as e:
-            print(f"Topic extraction error: {e}")
-            return ['digital security', 'legacy management', 'data protection']
-
+            self.logger.error(f"Topic extraction error: {e}")
+            return [
+                'digital legacy management',
+                'secure inheritance',
+                'AI-driven platform',
+                'digital asset protection',
+                'legacy preservation'
+            ]
+    
     def get_images_for_keyword(self, keyword: str) -> List[Dict[str, str]]:
         """Get images for a specific keyword with improved relevance"""
         try: