hit_testing.py 3.8 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112
  1. import logging
  2. from flask_login import login_required, current_user
  3. from flask_restful import Resource, reqparse, marshal, fields
  4. from werkzeug.exceptions import InternalServerError, NotFound, Forbidden
  5. import services
  6. from controllers.console import api
  7. from controllers.console.app.error import ProviderNotInitializeError, ProviderQuotaExceededError, \
  8. ProviderModelCurrentlyNotSupportError
  9. from controllers.console.datasets.error import HighQualityDatasetOnlyError, DatasetNotInitializedError
  10. from controllers.console.setup import setup_required
  11. from controllers.console.wraps import account_initialization_required
  12. from core.model_providers.error import ProviderTokenNotInitError, QuotaExceededError, ModelCurrentlyNotSupportError
  13. from libs.helper import TimestampField
  14. from services.dataset_service import DatasetService
  15. from services.hit_testing_service import HitTestingService
  16. document_fields = {
  17. 'id': fields.String,
  18. 'data_source_type': fields.String,
  19. 'name': fields.String,
  20. 'doc_type': fields.String,
  21. }
  22. segment_fields = {
  23. 'id': fields.String,
  24. 'position': fields.Integer,
  25. 'document_id': fields.String,
  26. 'content': fields.String,
  27. 'answer': fields.String,
  28. 'word_count': fields.Integer,
  29. 'tokens': fields.Integer,
  30. 'keywords': fields.List(fields.String),
  31. 'index_node_id': fields.String,
  32. 'index_node_hash': fields.String,
  33. 'hit_count': fields.Integer,
  34. 'enabled': fields.Boolean,
  35. 'disabled_at': TimestampField,
  36. 'disabled_by': fields.String,
  37. 'status': fields.String,
  38. 'created_by': fields.String,
  39. 'created_at': TimestampField,
  40. 'indexing_at': TimestampField,
  41. 'completed_at': TimestampField,
  42. 'error': fields.String,
  43. 'stopped_at': TimestampField,
  44. 'document': fields.Nested(document_fields),
  45. }
  46. hit_testing_record_fields = {
  47. 'segment': fields.Nested(segment_fields),
  48. 'score': fields.Float,
  49. 'tsne_position': fields.Raw
  50. }
  51. class HitTestingApi(Resource):
  52. @setup_required
  53. @login_required
  54. @account_initialization_required
  55. def post(self, dataset_id):
  56. dataset_id_str = str(dataset_id)
  57. dataset = DatasetService.get_dataset(dataset_id_str)
  58. if dataset is None:
  59. raise NotFound("Dataset not found.")
  60. try:
  61. DatasetService.check_dataset_permission(dataset, current_user)
  62. except services.errors.account.NoPermissionError as e:
  63. raise Forbidden(str(e))
  64. # only high quality dataset can be used for hit testing
  65. if dataset.indexing_technique != 'high_quality':
  66. raise HighQualityDatasetOnlyError()
  67. parser = reqparse.RequestParser()
  68. parser.add_argument('query', type=str, location='json')
  69. args = parser.parse_args()
  70. query = args['query']
  71. if not query or len(query) > 250:
  72. raise ValueError('Query is required and cannot exceed 250 characters')
  73. try:
  74. response = HitTestingService.retrieve(
  75. dataset=dataset,
  76. query=query,
  77. account=current_user,
  78. limit=10,
  79. )
  80. return {"query": response['query'], 'records': marshal(response['records'], hit_testing_record_fields)}
  81. except services.errors.index.IndexNotInitializedError:
  82. raise DatasetNotInitializedError()
  83. except ProviderTokenNotInitError as ex:
  84. raise ProviderNotInitializeError(ex.description)
  85. except QuotaExceededError:
  86. raise ProviderQuotaExceededError()
  87. except ModelCurrentlyNotSupportError:
  88. raise ProviderModelCurrentlyNotSupportError()
  89. except ValueError as e:
  90. raise ValueError(str(e))
  91. except Exception as e:
  92. logging.exception("Hit testing failed.")
  93. raise InternalServerError(str(e))
  94. api.add_resource(HitTestingApi, '/datasets/<uuid:dataset_id>/hit-testing')