update_annotation_to_index_task.py 2.2 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061
  1. import logging
  2. import time
  3. import click
  4. from celery import shared_task
  5. from core.rag.datasource.vdb.vector_factory import Vector
  6. from core.rag.models.document import Document
  7. from models.dataset import Dataset
  8. from services.dataset_service import DatasetCollectionBindingService
  9. @shared_task(queue='dataset')
  10. def update_annotation_to_index_task(annotation_id: str, question: str, tenant_id: str, app_id: str,
  11. collection_binding_id: str):
  12. """
  13. Update 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 update 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. embedding_model_provider=dataset_collection_binding.provider_name,
  33. embedding_model=dataset_collection_binding.model_name,
  34. collection_binding_id=dataset_collection_binding.id
  35. )
  36. document = Document(
  37. page_content=question,
  38. metadata={
  39. "annotation_id": annotation_id,
  40. "app_id": app_id,
  41. "doc_id": annotation_id
  42. }
  43. )
  44. vector = Vector(dataset, attributes=['doc_id', 'annotation_id', 'app_id'])
  45. vector.delete_by_metadata_field('annotation_id', annotation_id)
  46. vector.add_texts([document])
  47. end_at = time.perf_counter()
  48. logging.info(
  49. click.style(
  50. 'Build index successful for annotation: {} latency: {}'.format(annotation_id, end_at - start_at),
  51. fg='green'))
  52. except Exception:
  53. logging.exception("Build index for annotation failed")