remove_document_from_index_task.py 2.6 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273
  1. import datetime
  2. import logging
  3. import time
  4. import click
  5. from celery import shared_task # type: ignore
  6. from werkzeug.exceptions import NotFound
  7. from core.rag.index_processor.index_processor_factory import IndexProcessorFactory
  8. from extensions.ext_database import db
  9. from extensions.ext_redis import redis_client
  10. from models.dataset import Document, DocumentSegment
  11. @shared_task(queue="dataset")
  12. def remove_document_from_index_task(document_id: str):
  13. """
  14. Async Remove document from index
  15. :param document_id: document id
  16. Usage: remove_document_from_index.delay(document_id)
  17. """
  18. logging.info(click.style("Start remove document segments from index: {}".format(document_id), fg="green"))
  19. start_at = time.perf_counter()
  20. document = db.session.query(Document).filter(Document.id == document_id).first()
  21. if not document:
  22. raise NotFound("Document not found")
  23. if document.indexing_status != "completed":
  24. return
  25. indexing_cache_key = "document_{}_indexing".format(document.id)
  26. try:
  27. dataset = document.dataset
  28. if not dataset:
  29. raise Exception("Document has no dataset")
  30. index_processor = IndexProcessorFactory(document.doc_form).init_index_processor()
  31. segments = db.session.query(DocumentSegment).filter(DocumentSegment.document_id == document.id).all()
  32. index_node_ids = [segment.index_node_id for segment in segments]
  33. if index_node_ids:
  34. try:
  35. index_processor.clean(dataset, index_node_ids, with_keywords=True, delete_child_chunks=False)
  36. except Exception:
  37. logging.exception(f"clean dataset {dataset.id} from index failed")
  38. # update segment to disable
  39. db.session.query(DocumentSegment).filter(DocumentSegment.document_id == document.id).update(
  40. {
  41. DocumentSegment.enabled: False,
  42. DocumentSegment.disabled_at: datetime.datetime.now(datetime.UTC).replace(tzinfo=None),
  43. DocumentSegment.disabled_by: document.disabled_by,
  44. DocumentSegment.updated_at: datetime.datetime.now(datetime.UTC).replace(tzinfo=None),
  45. }
  46. )
  47. db.session.commit()
  48. end_at = time.perf_counter()
  49. logging.info(
  50. click.style(
  51. "Document removed from index: {} latency: {}".format(document.id, end_at - start_at), fg="green"
  52. )
  53. )
  54. except Exception:
  55. logging.exception("remove document from index failed")
  56. if not document.archived:
  57. document.enabled = True
  58. db.session.commit()
  59. finally:
  60. redis_client.delete(indexing_cache_key)