Update app.py

app.py

@@ -278,6 +278,13 @@ def semantic_search(query: str, top_k=1, category_filter=None, language='en'):
             threshold = max_score * 0.5  # Show documents within 50% of best score
             max_docs = 5
         
+        # Create a lookup dictionary for efficient metadata retrieval
+        metadata_lookup = {}
+        for match in res["matches"]:
+            file_path_key = match["metadata"].get("file_path", "")
+            if file_path_key not in metadata_lookup:
+                metadata_lookup[file_path_key] = match
+        
         results = []
         doc_count = 0
         
@@ -285,47 +292,47 @@ def semantic_search(query: str, top_k=1, category_filter=None, language='en'):
             if doc_count >= max_docs or score < threshold:
                 break
             
-            # Get the metadata for this document
-            for match in res["matches"]:
+            # Get the metadata for this document using the lookup
+            if file_path in metadata_lookup:
+                match = metadata_lookup[file_path]
                 metadata = match["metadata"]
-                if metadata.get("file_path", "") == file_path:
-                    # Use the DYNAMIC_METADATA mapping if available, otherwise use metadata
-                    proposal_data = DYNAMIC_METADATA.get(file_path, {
-                        "title": metadata.get("title", "Unknown Title"),
-                        "summary": metadata.get("summary", ""),
-                        "category": metadata.get("category", "Budget Proposal"),
-                        "costLKR": metadata.get("costLKR", "No Costing Available")
-                    })
+                
+                # Use the DYNAMIC_METADATA mapping if available, otherwise use metadata
+                proposal_data = DYNAMIC_METADATA.get(file_path, {
+                    "title": metadata.get("title", "Unknown Title"),
+                    "summary": metadata.get("summary", ""),
+                    "category": metadata.get("category", "Budget Proposal"),
+                    "costLKR": metadata.get("costLKR", "No Costing Available")
+                })
+                
+                # Get language-specific data
+                title = get_language_specific_data(proposal_data, "title", language)
+                summary = get_language_specific_data(proposal_data, "summary", language)
+                costLKR = get_language_specific_data(proposal_data, "costLKR", language)
+                category = get_language_specific_data(proposal_data, "category", language)
+                thumb_url = metadata.get("thumbUrl", "")
+                
+                # 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", ""]):
                     
-                    # Get language-specific data
-                    title = get_language_specific_data(proposal_data, "title", language)
-                    summary = get_language_specific_data(proposal_data, "summary", language)
-                    costLKR = get_language_specific_data(proposal_data, "costLKR", language)
-                    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
+                    }
                     
-                    # 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
+                    results.append(result)
+                    doc_count += 1
         
         return results
     except Exception as e: