Method create_single_index

demo_batch_indexing.py:139–195

Create a single index from configuration.

(self, index_config: Dict[str, Any])

        return created_indexes
    
    def create_single_index(self, index_config: Dict[str, Any]) -> Optional[str]:
        """Create a single index from configuration."""
        try:
            # Extract index metadata
            index_name = index_config.get("name", "Unnamed Index")
            index_description = index_config.get("description", "")
            documents = index_config.get("documents", [])
            
            if not documents:
                print(f"⚠️  No documents specified for index '{index_name}', skipping...")
                return None
            
            # Validate documents
            valid_documents = self.validate_documents(documents)
            if not valid_documents:
                print(f"❌ No valid documents found for index '{index_name}'")
                return None
            
            print(f"\n🚀 Creating index: {index_name}")
            print(f"📄 Processing {len(valid_documents)} documents")
            
            # Create index record in database
            index_metadata = {
                "created_by": "demo_batch_indexing.py",
                "created_at": datetime.now().isoformat(),
                "document_count": len(valid_documents),
                "config_used": index_config.get("processing_options", {})
            }
            
            index_id = self.db.create_index(
                name=index_name,
                description=index_description,
                metadata=index_metadata
            )
            
            # Add documents to index
            for doc_path in valid_documents:
                filename = os.path.basename(doc_path)
                self.db.add_document_to_index(index_id, filename, doc_path)
            
            # Process documents through pipeline
            start_time = time.time()
            self.pipeline.process_documents(valid_documents)
            processing_time = time.time() - start_time
            
            print(f"✅ Index '{index_name}' created successfully!")
            print(f"   Index ID: {index_id}")
            print(f"   Processing time: {processing_time:.2f} seconds")
            print(f"   Documents processed: {len(valid_documents)}")
            
            return index_id
            
        except Exception as e:
            print(f"❌ Error creating index '{index_name}': {e}")
            import traceback
            traceback.print_exc()
            return None