| 
									
										
										
										
											2024-04-24 15:02:29 +08:00
										 |  |  | import datetime | 
					
						
							|  |  |  | import logging | 
					
						
							|  |  |  | import time | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | import click | 
					
						
							| 
									
										
										
										
											2024-12-24 18:38:51 +08:00
										 |  |  | from celery import shared_task  # type: ignore | 
					
						
							| 
									
										
										
										
											2024-04-24 15:02:29 +08:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2024-07-12 12:25:38 +08:00
										 |  |  | from configs import dify_config | 
					
						
							| 
									
										
										
										
											2024-09-11 16:40:52 +08:00
										 |  |  | from core.indexing_runner import DocumentIsPausedError, IndexingRunner | 
					
						
							| 
									
										
										
										
											2024-04-24 15:02:29 +08:00
										 |  |  | from core.rag.index_processor.index_processor_factory import IndexProcessorFactory | 
					
						
							|  |  |  | from extensions.ext_database import db | 
					
						
							|  |  |  | from models.dataset import Dataset, Document, DocumentSegment | 
					
						
							|  |  |  | from services.feature_service import FeatureService | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2024-08-26 13:38:37 +08:00
										 |  |  | @shared_task(queue="dataset") | 
					
						
							| 
									
										
										
										
											2024-04-24 15:02:29 +08:00
										 |  |  | def duplicate_document_indexing_task(dataset_id: str, document_ids: list): | 
					
						
							|  |  |  |     """
 | 
					
						
							|  |  |  |     Async process document | 
					
						
							|  |  |  |     :param dataset_id: | 
					
						
							|  |  |  |     :param document_ids: | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2025-03-16 11:16:28 +08:00
										 |  |  |     Usage: duplicate_document_indexing_task.delay(dataset_id, document_ids) | 
					
						
							| 
									
										
										
										
											2024-04-24 15:02:29 +08:00
										 |  |  |     """
 | 
					
						
							|  |  |  |     documents = [] | 
					
						
							|  |  |  |     start_at = time.perf_counter() | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |     dataset = db.session.query(Dataset).filter(Dataset.id == dataset_id).first() | 
					
						
							| 
									
										
										
										
											2024-12-24 18:38:51 +08:00
										 |  |  |     if dataset is None: | 
					
						
							| 
									
										
										
										
											2025-04-07 20:31:26 +08:00
										 |  |  |         logging.info(click.style("Dataset not found: {}".format(dataset_id), fg="red")) | 
					
						
							|  |  |  |         db.session.close() | 
					
						
							|  |  |  |         return | 
					
						
							| 
									
										
										
										
											2024-04-24 15:02:29 +08:00
										 |  |  | 
 | 
					
						
							|  |  |  |     # check document limit | 
					
						
							|  |  |  |     features = FeatureService.get_features(dataset.tenant_id) | 
					
						
							|  |  |  |     try: | 
					
						
							|  |  |  |         if features.billing.enabled: | 
					
						
							|  |  |  |             vector_space = features.vector_space | 
					
						
							|  |  |  |             count = len(document_ids) | 
					
						
							| 
									
										
										
										
											2025-04-01 16:45:31 +08:00
										 |  |  |             if features.billing.subscription.plan == "sandbox" and count > 1: | 
					
						
							|  |  |  |                 raise ValueError("Your current plan does not support batch upload, please upgrade your plan.") | 
					
						
							| 
									
										
										
										
											2024-07-12 12:25:38 +08:00
										 |  |  |             batch_upload_limit = int(dify_config.BATCH_UPLOAD_LIMIT) | 
					
						
							| 
									
										
										
										
											2024-04-24 15:02:29 +08:00
										 |  |  |             if count > batch_upload_limit: | 
					
						
							|  |  |  |                 raise ValueError(f"You have reached the batch upload limit of {batch_upload_limit}.") | 
					
						
							|  |  |  |             if 0 < vector_space.limit <= vector_space.size: | 
					
						
							| 
									
										
										
										
											2024-08-26 13:38:37 +08:00
										 |  |  |                 raise ValueError( | 
					
						
							|  |  |  |                     "Your total number of documents plus the number of uploads have over the limit of " | 
					
						
							|  |  |  |                     "your subscription." | 
					
						
							|  |  |  |                 ) | 
					
						
							| 
									
										
										
										
											2024-04-24 15:02:29 +08:00
										 |  |  |     except Exception as e: | 
					
						
							|  |  |  |         for document_id in document_ids: | 
					
						
							| 
									
										
										
										
											2024-08-26 13:38:37 +08:00
										 |  |  |             document = ( | 
					
						
							|  |  |  |                 db.session.query(Document).filter(Document.id == document_id, Document.dataset_id == dataset_id).first() | 
					
						
							|  |  |  |             ) | 
					
						
							| 
									
										
										
										
											2024-04-24 15:02:29 +08:00
										 |  |  |             if document: | 
					
						
							| 
									
										
										
										
											2024-08-26 13:38:37 +08:00
										 |  |  |                 document.indexing_status = "error" | 
					
						
							| 
									
										
										
										
											2024-04-24 15:02:29 +08:00
										 |  |  |                 document.error = str(e) | 
					
						
							| 
									
										
										
										
											2025-03-17 16:13:11 +08:00
										 |  |  |                 document.stopped_at = datetime.datetime.now(datetime.UTC).replace(tzinfo=None) | 
					
						
							| 
									
										
										
										
											2024-04-24 15:02:29 +08:00
										 |  |  |                 db.session.add(document) | 
					
						
							|  |  |  |         db.session.commit() | 
					
						
							|  |  |  |         return | 
					
						
							| 
									
										
										
										
											2025-04-07 20:31:26 +08:00
										 |  |  |     finally: | 
					
						
							|  |  |  |         db.session.close() | 
					
						
							| 
									
										
										
										
											2024-04-24 15:02:29 +08:00
										 |  |  | 
 | 
					
						
							|  |  |  |     for document_id in document_ids: | 
					
						
							| 
									
										
										
										
											2024-08-26 13:38:37 +08:00
										 |  |  |         logging.info(click.style("Start process document: {}".format(document_id), fg="green")) | 
					
						
							| 
									
										
										
										
											2024-04-24 15:02:29 +08:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2024-08-26 13:38:37 +08:00
										 |  |  |         document = ( | 
					
						
							|  |  |  |             db.session.query(Document).filter(Document.id == document_id, Document.dataset_id == dataset_id).first() | 
					
						
							|  |  |  |         ) | 
					
						
							| 
									
										
										
										
											2024-04-24 15:02:29 +08:00
										 |  |  | 
 | 
					
						
							|  |  |  |         if document: | 
					
						
							|  |  |  |             # clean old data | 
					
						
							|  |  |  |             index_type = document.doc_form | 
					
						
							|  |  |  |             index_processor = IndexProcessorFactory(index_type).init_index_processor() | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |             segments = db.session.query(DocumentSegment).filter(DocumentSegment.document_id == document_id).all() | 
					
						
							|  |  |  |             if segments: | 
					
						
							|  |  |  |                 index_node_ids = [segment.index_node_id for segment in segments] | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |                 # delete from vector index | 
					
						
							| 
									
										
										
										
											2024-12-25 19:49:07 +08:00
										 |  |  |                 index_processor.clean(dataset, index_node_ids, with_keywords=True, delete_child_chunks=True) | 
					
						
							| 
									
										
										
										
											2024-04-24 15:02:29 +08:00
										 |  |  | 
 | 
					
						
							|  |  |  |                 for segment in segments: | 
					
						
							|  |  |  |                     db.session.delete(segment) | 
					
						
							|  |  |  |                 db.session.commit() | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2024-08-26 13:38:37 +08:00
										 |  |  |             document.indexing_status = "parsing" | 
					
						
							| 
									
										
										
										
											2025-03-17 16:13:11 +08:00
										 |  |  |             document.processing_started_at = datetime.datetime.now(datetime.UTC).replace(tzinfo=None) | 
					
						
							| 
									
										
										
										
											2024-04-24 15:02:29 +08:00
										 |  |  |             documents.append(document) | 
					
						
							|  |  |  |             db.session.add(document) | 
					
						
							|  |  |  |     db.session.commit() | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |     try: | 
					
						
							|  |  |  |         indexing_runner = IndexingRunner() | 
					
						
							|  |  |  |         indexing_runner.run(documents) | 
					
						
							|  |  |  |         end_at = time.perf_counter() | 
					
						
							| 
									
										
										
										
											2024-08-26 13:38:37 +08:00
										 |  |  |         logging.info(click.style("Processed dataset: {} latency: {}".format(dataset_id, end_at - start_at), fg="green")) | 
					
						
							| 
									
										
										
										
											2024-09-11 16:40:52 +08:00
										 |  |  |     except DocumentIsPausedError as ex: | 
					
						
							| 
									
										
										
										
											2024-08-26 13:38:37 +08:00
										 |  |  |         logging.info(click.style(str(ex), fg="yellow")) | 
					
						
							| 
									
										
										
										
											2024-04-24 15:02:29 +08:00
										 |  |  |     except Exception: | 
					
						
							|  |  |  |         pass | 
					
						
							| 
									
										
										
										
											2025-04-07 20:31:26 +08:00
										 |  |  |     finally: | 
					
						
							|  |  |  |         db.session.close() |