clean_notion_document_task.py 2.1 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758
  1. import logging
  2. import time
  3. from typing import List
  4. import click
  5. from celery import shared_task
  6. from core.index.keyword_table_index import KeywordTableIndex
  7. from core.index.vector_index import VectorIndex
  8. from extensions.ext_database import db
  9. from models.dataset import DocumentSegment, Dataset, Document
  10. @shared_task
  11. def clean_notion_document_task(document_ids: List[str], dataset_id: str):
  12. """
  13. Clean document when document deleted.
  14. :param document_ids: document ids
  15. :param dataset_id: dataset id
  16. Usage: clean_notion_document_task.delay(document_ids, dataset_id)
  17. """
  18. logging.info(click.style('Start clean document when import form notion document deleted: {}'.format(dataset_id), fg='green'))
  19. start_at = time.perf_counter()
  20. try:
  21. dataset = db.session.query(Dataset).filter(Dataset.id == dataset_id).first()
  22. if not dataset:
  23. raise Exception('Document has no dataset')
  24. vector_index = VectorIndex(dataset=dataset)
  25. keyword_table_index = KeywordTableIndex(dataset=dataset)
  26. for document_id in document_ids:
  27. document = db.session.query(Document).filter(
  28. Document.id == document_id
  29. ).first()
  30. db.session.delete(document)
  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. # delete from vector index
  34. vector_index.del_nodes(index_node_ids)
  35. # delete from keyword index
  36. if index_node_ids:
  37. keyword_table_index.del_nodes(index_node_ids)
  38. for segment in segments:
  39. db.session.delete(segment)
  40. db.session.commit()
  41. end_at = time.perf_counter()
  42. logging.info(
  43. click.style('Clean document when import form notion document deleted end :: {} latency: {}'.format(
  44. dataset_id, end_at - start_at),
  45. fg='green'))
  46. except Exception:
  47. logging.exception("Cleaned document when import form notion document deleted failed")