ocr speed improved

LightRAG-main/lightrag/api/lightrag_server.py

@@ -244,6 +244,23 @@ def create_app(args):
                 task.add_done_callback(app.state.background_tasks.discard)
                 logger.info(f"Process {os.getpid()} auto scan task started at startup for workspace '{args.workspace}'.")
 
+            # Warm up OCR processor in background to avoid cold‑start delay on first upload
+            async def warm_up_ocr_processor():
+                try:
+                    logger.info("Starting OCR processor warm‑up...")
+                    # Import inside function to avoid unnecessary dependency if OCR not used
+                    from lightrag.document_processor import get_document_processor
+                    # This will initialize OptimizedOCRProcessor (≈9 seconds)
+                    processor = get_document_processor()
+                    logger.info("OCR processor warmed up successfully")
+                except Exception as e:
+                    logger.warning(f"OCR warm‑up failed (non‑critical): {e}")
+
+            # Schedule warm‑up as a background task (non‑blocking)
+            warm_up_task = asyncio.create_task(warm_up_ocr_processor())
+            app.state.background_tasks.add(warm_up_task)
+            warm_up_task.add_done_callback(app.state.background_tasks.discard)
+
             ASCIIColors.green("\nServer is ready to accept connections! 🚀\n")
 
             yield