Update app.py

app.py

@@ -50,8 +50,8 @@ def get_language_specific_data(proposal_data, field, language='en'):
     
     # If it's the new multi-language format, return language-specific data
     if isinstance(proposal_data.get(field), dict):
-        return proposal_data.get(field, {}).get(language, 
-                                               proposal_data.get(field, {}).get('en', ''))
+        # Only return data for the requested language, no fallback
+        return proposal_data.get(field, {}).get(language, '')
     
     return ''
 
@@ -152,22 +152,27 @@ def semantic_search(query: str, top_k=1, category_filter=None, language='en'):
                     category = get_language_specific_data(proposal_data, "category", language)
                     thumb_url = metadata.get("thumbUrl", "")
                     
-                    result = {
-                        "title": title,
-                        "summary": summary,
-                        "costLKR": costLKR,
-                        "category": category,
-                        "pdfUrl": f"assets/pdfs/{file_path}" if file_path else "",
-                        "thumbUrl": f"assets/thumbs/{thumb_url}" if thumb_url else "",
-                        "score": score,
-                        "relevance_percentage": int(score * 100),
-                        "file_path": file_path,
-                        "id": match["id"],
-                        "content": metadata.get("content", "")  # Add the actual content
-                    }
-                    
-                    results.append(result)
-                    doc_count += 1
+                    # Only include documents that have meaningful content in the requested language
+                    # Skip documents where title and summary are empty or "Unknown"/"No summary available"
+                    if (title and title.strip() and title not in ["Unknown", "Unknown Title", ""] and
+                        summary and summary.strip() and summary not in ["No summary available", ""]):
+                        
+                        result = {
+                            "title": title,
+                            "summary": summary,
+                            "costLKR": costLKR,
+                            "category": category,
+                            "pdfUrl": f"assets/pdfs/{file_path}" if file_path else "",
+                            "thumbUrl": f"assets/thumbs/{thumb_url}" if thumb_url else "",
+                            "score": score,
+                            "relevance_percentage": int(score * 100),
+                            "file_path": file_path,
+                            "id": match["id"],
+                            "content": metadata.get("content", "")  # Add the actual content
+                        }
+                        
+                        results.append(result)
+                        doc_count += 1
                     break
         
         return results
@@ -233,8 +238,9 @@ def get_all_proposals(category_filter=None, language='en'):
             thumb_url = metadata.get("thumbUrl", "")
             
             # Only include documents that have meaningful content in the requested language
-            # Skip documents where title is empty or "Unknown", but allow "No summary available"
-            if (title and title.strip() and title not in ["Unknown", "Unknown Title"]):
+            # Skip documents where title and summary are empty or "Unknown"/"No summary available"
+            if (title and title.strip() and title not in ["Unknown", "Unknown Title", ""] and
+                summary and summary.strip() and summary not in ["No summary available", ""]):
                 
                 result = {
                     "title": title,
@@ -340,7 +346,12 @@ def get_categories():
         for file_path, metadata in DYNAMIC_METADATA.items():
             category = metadata.get("category")
             if category:
-                categories.add(category)
+                # Handle both string and dict formats
+                if isinstance(category, dict):
+                    # Extract English category from dict
+                    category = category.get("en", "")
+                if category:
+                    categories.add(category)
         
         # If no categories from metadata, fallback to Pinecone
         if not categories: