""" Production Setup Example - Enterprise-grade configuration. This example shows how to set up the SDK for production use with: - Multiple vector store options (Pinecone, Qdrant, Weaviate, pgVector) - Error handling and monitoring - Performance optimization - Auto-monitoring for new products """ import sys from pathlib import Path import os # Add SDK to path sdk_path = Path(__file__).parent.parent sys.path.insert(0, str(sdk_path)) from image_search_sdk import ImageSearchClient, SearchOptions def setup_with_pinecone(): """Production setup with Pinecone (Managed Vector Database)""" print("Setting up with Pinecone (Managed)") client = ImageSearchClient( data_dir="./production_data", model_name="clip-ViT-L-14", vector_store_type="pinecone", vector_store_config={ "api_key": os.getenv("PINECONE_API_KEY"), "environment": os.getenv("PINECONE_ENVIRONMENT", "us-west1-gcp"), "index_name": "product-search" } ) return client def setup_with_qdrant(): """Production setup with Qdrant (Self-hosted or Cloud)""" print("Setting up with Qdrant") client = ImageSearchClient( data_dir="./production_data", model_name="clip-ViT-L-14", vector_store_type="qdrant", vector_store_config={ "url": os.getenv("QDRANT_URL", "http://localhost:6333"), "api_key": os.getenv("QDRANT_API_KEY"), # Optional for cloud "collection_name": "products" } ) return client def setup_with_weaviate(): """Production setup with Weaviate""" print("Setting up with Weaviate") client = ImageSearchClient( data_dir="./production_data", model_name="clip-ViT-L-14", vector_store_type="weaviate", vector_store_config={ "url": os.getenv("WEAVIATE_URL", "http://localhost:8080"), "api_key": os.getenv("WEAVIATE_API_KEY"), # Optional } ) return client def setup_with_pgvector(): """Production setup with pgVector (PostgreSQL extension)""" print("Setting up with pgVector (PostgreSQL)") client = ImageSearchClient( data_dir="./production_data", model_name="clip-ViT-L-14", vector_store_type="pgvector", vector_store_config={ "host": os.getenv("PG_HOST", "localhost"), "port": int(os.getenv("PG_PORT", "5432")), "database": os.getenv("PG_DATABASE", "products"), "user": os.getenv("PG_USER", "postgres"), "password": os.getenv("PG_PASSWORD"), "table_name": "product_vectors" } ) return client def setup_with_chromadb(): """Development/Small-scale setup with ChromaDB""" print("Setting up with ChromaDB (Local/Development)") client = ImageSearchClient( data_dir="./production_data", model_name="clip-ViT-L-14", vector_store_type="chromadb", vector_store_config={ "mode": "persistent", "path": "./production_data/chroma_vectors" } ) return client def main(): print("Production Setup Example") print("=" * 70) # Choose your vector store (set via environment variable) vector_store = os.getenv("VECTOR_STORE", "chromadb").lower() print(f"\nUsing vector store: {vector_store}") print("=" * 70) # Initialize based on choice if vector_store == "pinecone": client = setup_with_pinecone() elif vector_store == "qdrant": client = setup_with_qdrant() elif vector_store == "weaviate": client = setup_with_weaviate() elif vector_store == "pgvector": client = setup_with_pgvector() else: client = setup_with_chromadb() print("\n✓ SDK initialized") # Connect to your production database print("\nConnecting to production database...") try: client.connect_database( connection_string=os.getenv("DATABASE_URL"), table_name=os.getenv("PRODUCTS_TABLE", "products"), id_column="id", title_column="title", image_url_column="image_url", category_column="category", brand_column="brand", price_column="price", rating_column="rating", description_column="description" ) print("✓ Database connected") except Exception as e: print(f"✗ Database connection failed: {e}") return # Initialize embeddings (with error handling) print("\nInitializing embeddings...") try: stats = client.initialize(batch_size=100) print(f"✓ Initialization complete") print(f" - Indexed: {stats['indexed_products']:,} products") print(f" - Failed: {stats['failed_products']:,} products") print(f" - Rate: {stats['products_per_second']:.1f} products/sec") if stats['errors']: print(f"\n⚠ Errors encountered: {len(stats['errors'])}") # Log errors to file or monitoring system with open("indexing_errors.log", "w") as f: for error in stats['errors']: f.write(f"{error}\n") except Exception as e: print(f"✗ Initialization failed: {e}") return # Enable auto-monitoring for production print("\nEnabling auto-monitoring...") try: monitor_interval = int(os.getenv("MONITOR_INTERVAL", "60")) client.enable_auto_monitor( poll_interval=monitor_interval, enable=True ) print(f"✓ Auto-monitoring enabled (interval: {monitor_interval}s)") except Exception as e: print(f"⚠ Auto-monitoring failed: {e}") # Test search functionality print("\nTesting search...") try: test_results = client.search_by_text( "test query", options=SearchOptions(top_k=1) ) print(f"✓ Search working ({len(test_results)} result)") except Exception as e: print(f"✗ Search test failed: {e}") # Display statistics print("\n" + "=" * 70) print("PRODUCTION STATISTICS") print("=" * 70) stats = client.get_stats() print(f"Vector Store: {stats.get('store_type', 'N/A')}") print(f"Total Vectors: {stats.get('total_vectors', 0):,}") print(f"Status: {'Operational' if stats.get('is_initialized') else 'Not Ready'}") print("\n" + "=" * 70) print("Production setup complete!") print("The system is now running and monitoring for new products.") print("=" * 70) # Keep running (for monitoring) print("\nPress Ctrl+C to shutdown...") try: import time while True: time.sleep(60) # Optionally log health metrics here current_stats = client.get_stats() print(f"[Health Check] Vectors: {current_stats.get('total_vectors', 0):,}") except KeyboardInterrupt: print("\n\nShutting down...") client.close() print("✓ Shutdown complete") if __name__ == "__main__": main()