Underground-Digital/Workflow-Engine
0
1import datetime2import logging3import time4 5import click6from celery import shared_task7from werkzeug.exceptions import NotFound8 9from core.rag.index_processor.index_processor_factory import IndexProcessorFactory10from core.rag.models.document import Document11from extensions.ext_database import db12from extensions.ext_redis import redis_client13from models.dataset import DocumentSegment14 15 16@shared_task(queue="dataset")17def enable_segment_to_index_task(segment_id: str):18 """19 Async enable segment to index20 :param segment_id:21 22 Usage: enable_segment_to_index_task.delay(segment_id)23 """24 logging.info(click.style("Start enable segment to index: {}".format(segment_id), fg="green"))25 start_at = time.perf_counter()26 27 segment = db.session.query(DocumentSegment).filter(DocumentSegment.id == segment_id).first()28 if not segment:29 raise NotFound("Segment not found")30 31 if segment.status != "completed":32 raise NotFound("Segment is not completed, enable action is not allowed.")33 34 indexing_cache_key = "segment_{}_indexing".format(segment.id)35 36 try:37 document = Document(38 page_content=segment.content,39 metadata={40 "doc_id": segment.index_node_id,41 "doc_hash": segment.index_node_hash,42 "document_id": segment.document_id,43 "dataset_id": segment.dataset_id,44 },45 )46 47 dataset = segment.dataset48 49 if not dataset:50 logging.info(click.style("Segment {} has no dataset, pass.".format(segment.id), fg="cyan"))51 return52 53 dataset_document = segment.document54 55 if not dataset_document:56 logging.info(click.style("Segment {} has no document, pass.".format(segment.id), fg="cyan"))57 return58 59 if not dataset_document.enabled or dataset_document.archived or dataset_document.indexing_status != "completed":60 logging.info(click.style("Segment {} document status is invalid, pass.".format(segment.id), fg="cyan"))61 return62 63 index_processor = IndexProcessorFactory(dataset_document.doc_form).init_index_processor()64 # save vector index65 index_processor.load(dataset, [document])66 67 end_at = time.perf_counter()68 logging.info(69 click.style("Segment enabled to index: {} latency: {}".format(segment.id, end_at - start_at), fg="green")70 )71 except Exception as e:72 logging.exception("enable segment to index failed")73 segment.enabled = False74 segment.disabled_at = datetime.datetime.now(datetime.timezone.utc).replace(tzinfo=None)75 segment.status = "error"76 segment.error = str(e)77 db.session.commit()78 finally:79 redis_client.delete(indexing_cache_key)80 