add_annotation_to_index_task.py 1.9 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859
  1. import logging
  2. import time
  3. import click
  4. from celery import shared_task
  5. from langchain.schema import Document
  6. from core.index.index import IndexBuilder
  7. from models.dataset import Dataset
  8. from services.dataset_service import DatasetCollectionBindingService
  9. @shared_task(queue='dataset')
  10. def add_annotation_to_index_task(annotation_id: str, question: str, tenant_id: str, app_id: str,
  11. collection_binding_id: str):
  12. """
  13. Add annotation to index.
  14. :param annotation_id: annotation id
  15. :param question: question
  16. :param tenant_id: tenant id
  17. :param app_id: app id
  18. :param collection_binding_id: embedding binding id
  19. Usage: clean_dataset_task.delay(dataset_id, tenant_id, indexing_technique, index_struct)
  20. """
  21. logging.info(click.style('Start build index for annotation: {}'.format(annotation_id), fg='green'))
  22. start_at = time.perf_counter()
  23. try:
  24. dataset_collection_binding = DatasetCollectionBindingService.get_dataset_collection_binding_by_id_and_type(
  25. collection_binding_id,
  26. 'annotation'
  27. )
  28. dataset = Dataset(
  29. id=app_id,
  30. tenant_id=tenant_id,
  31. indexing_technique='high_quality',
  32. collection_binding_id=dataset_collection_binding.id
  33. )
  34. document = Document(
  35. page_content=question,
  36. metadata={
  37. "annotation_id": annotation_id,
  38. "app_id": app_id,
  39. "doc_id": annotation_id
  40. }
  41. )
  42. index = IndexBuilder.get_index(dataset, 'high_quality')
  43. if index:
  44. index.add_texts([document])
  45. end_at = time.perf_counter()
  46. logging.info(
  47. click.style(
  48. 'Build index successful for annotation: {} latency: {}'.format(annotation_id, end_at - start_at),
  49. fg='green'))
  50. except Exception:
  51. logging.exception("Build index for annotation failed")