airbyte-cdk 6.6.6__py3-none-any.whl → 6.6.8__py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (106) hide show
  1. airbyte_cdk/cli/source_declarative_manifest/_run.py +2 -1
  2. airbyte_cdk/config_observation.py +2 -1
  3. airbyte_cdk/connector.py +1 -0
  4. airbyte_cdk/connector_builder/connector_builder_handler.py +1 -1
  5. airbyte_cdk/connector_builder/main.py +2 -1
  6. airbyte_cdk/destinations/destination.py +2 -1
  7. airbyte_cdk/destinations/vector_db_based/config.py +2 -1
  8. airbyte_cdk/destinations/vector_db_based/document_processor.py +4 -3
  9. airbyte_cdk/destinations/vector_db_based/embedder.py +5 -4
  10. airbyte_cdk/entrypoint.py +3 -2
  11. airbyte_cdk/logger.py +2 -1
  12. airbyte_cdk/models/airbyte_protocol.py +2 -1
  13. airbyte_cdk/sources/config.py +2 -1
  14. airbyte_cdk/sources/declarative/auth/jwt.py +1 -0
  15. airbyte_cdk/sources/declarative/auth/oauth.py +1 -0
  16. airbyte_cdk/sources/declarative/auth/selective_authenticator.py +1 -0
  17. airbyte_cdk/sources/declarative/auth/token.py +2 -1
  18. airbyte_cdk/sources/declarative/auth/token_provider.py +3 -2
  19. airbyte_cdk/sources/declarative/concurrent_declarative_source.py +6 -4
  20. airbyte_cdk/sources/declarative/decoders/json_decoder.py +3 -2
  21. airbyte_cdk/sources/declarative/decoders/noop_decoder.py +1 -0
  22. airbyte_cdk/sources/declarative/decoders/pagination_decoder_decorator.py +1 -0
  23. airbyte_cdk/sources/declarative/decoders/xml_decoder.py +1 -0
  24. airbyte_cdk/sources/declarative/extractors/dpath_extractor.py +1 -0
  25. airbyte_cdk/sources/declarative/extractors/http_selector.py +1 -0
  26. airbyte_cdk/sources/declarative/extractors/record_selector.py +1 -0
  27. airbyte_cdk/sources/declarative/extractors/response_to_file_extractor.py +2 -1
  28. airbyte_cdk/sources/declarative/incremental/datetime_based_cursor.py +2 -1
  29. airbyte_cdk/sources/declarative/interpolation/jinja.py +5 -4
  30. airbyte_cdk/sources/declarative/manifest_declarative_source.py +4 -3
  31. airbyte_cdk/sources/declarative/migrations/legacy_to_per_partition_state_migration.py +1 -1
  32. airbyte_cdk/sources/declarative/parsers/model_to_component_factory.py +3 -2
  33. airbyte_cdk/sources/declarative/partition_routers/substream_partition_router.py +1 -0
  34. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/constant_backoff_strategy.py +1 -0
  35. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/exponential_backoff_strategy.py +1 -0
  36. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/wait_time_from_header_backoff_strategy.py +1 -0
  37. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/wait_until_time_from_header_backoff_strategy.py +1 -0
  38. airbyte_cdk/sources/declarative/requesters/error_handlers/composite_error_handler.py +1 -0
  39. airbyte_cdk/sources/declarative/requesters/error_handlers/default_error_handler.py +1 -0
  40. airbyte_cdk/sources/declarative/requesters/error_handlers/default_http_response_filter.py +1 -0
  41. airbyte_cdk/sources/declarative/requesters/error_handlers/http_response_filter.py +1 -0
  42. airbyte_cdk/sources/declarative/requesters/http_job_repository.py +2 -1
  43. airbyte_cdk/sources/declarative/requesters/http_requester.py +1 -0
  44. airbyte_cdk/sources/declarative/requesters/paginators/default_paginator.py +1 -0
  45. airbyte_cdk/sources/declarative/requesters/paginators/no_pagination.py +1 -0
  46. airbyte_cdk/sources/declarative/requesters/paginators/paginator.py +1 -0
  47. airbyte_cdk/sources/declarative/requesters/paginators/strategies/cursor_pagination_strategy.py +1 -0
  48. airbyte_cdk/sources/declarative/requesters/paginators/strategies/offset_increment.py +1 -0
  49. airbyte_cdk/sources/declarative/requesters/paginators/strategies/page_increment.py +1 -0
  50. airbyte_cdk/sources/declarative/requesters/paginators/strategies/pagination_strategy.py +1 -0
  51. airbyte_cdk/sources/declarative/requesters/paginators/strategies/stop_condition.py +1 -0
  52. airbyte_cdk/sources/declarative/requesters/request_options/interpolated_request_options_provider.py +2 -1
  53. airbyte_cdk/sources/declarative/requesters/requester.py +1 -0
  54. airbyte_cdk/sources/declarative/retrievers/async_retriever.py +2 -1
  55. airbyte_cdk/sources/declarative/retrievers/simple_retriever.py +1 -0
  56. airbyte_cdk/sources/declarative/stream_slicers/declarative_partition_generator.py +1 -1
  57. airbyte_cdk/sources/declarative/transformations/add_fields.py +1 -0
  58. airbyte_cdk/sources/declarative/transformations/remove_fields.py +1 -0
  59. airbyte_cdk/sources/declarative/yaml_declarative_source.py +1 -0
  60. airbyte_cdk/sources/embedded/tools.py +1 -0
  61. airbyte_cdk/sources/file_based/config/abstract_file_based_spec.py +2 -1
  62. airbyte_cdk/sources/file_based/config/avro_format.py +2 -1
  63. airbyte_cdk/sources/file_based/config/csv_format.py +2 -1
  64. airbyte_cdk/sources/file_based/config/excel_format.py +2 -1
  65. airbyte_cdk/sources/file_based/config/file_based_stream_config.py +2 -1
  66. airbyte_cdk/sources/file_based/config/jsonl_format.py +2 -1
  67. airbyte_cdk/sources/file_based/config/parquet_format.py +2 -1
  68. airbyte_cdk/sources/file_based/config/unstructured_format.py +2 -1
  69. airbyte_cdk/sources/file_based/file_based_source.py +2 -1
  70. airbyte_cdk/sources/file_based/file_based_stream_reader.py +2 -1
  71. airbyte_cdk/sources/file_based/file_types/avro_parser.py +1 -0
  72. airbyte_cdk/sources/file_based/file_types/csv_parser.py +2 -1
  73. airbyte_cdk/sources/file_based/file_types/excel_parser.py +5 -5
  74. airbyte_cdk/sources/file_based/file_types/jsonl_parser.py +2 -1
  75. airbyte_cdk/sources/file_based/file_types/parquet_parser.py +2 -1
  76. airbyte_cdk/sources/file_based/file_types/unstructured_parser.py +9 -8
  77. airbyte_cdk/sources/file_based/stream/abstract_file_based_stream.py +2 -1
  78. airbyte_cdk/sources/file_based/stream/concurrent/adapters.py +2 -1
  79. airbyte_cdk/sources/http_logger.py +1 -0
  80. airbyte_cdk/sources/streams/call_rate.py +1 -2
  81. airbyte_cdk/sources/streams/concurrent/abstract_stream.py +2 -1
  82. airbyte_cdk/sources/streams/concurrent/adapters.py +2 -2
  83. airbyte_cdk/sources/streams/concurrent/availability_strategy.py +2 -1
  84. airbyte_cdk/sources/streams/concurrent/state_converters/datetime_stream_state_converter.py +1 -1
  85. airbyte_cdk/sources/streams/core.py +2 -1
  86. airbyte_cdk/sources/streams/http/error_handlers/default_error_mapping.py +2 -1
  87. airbyte_cdk/sources/streams/http/error_handlers/http_status_error_handler.py +1 -0
  88. airbyte_cdk/sources/streams/http/error_handlers/json_error_message_parser.py +1 -0
  89. airbyte_cdk/sources/streams/http/error_handlers/response_models.py +2 -1
  90. airbyte_cdk/sources/streams/http/http.py +3 -2
  91. airbyte_cdk/sources/streams/http/http_client.py +34 -2
  92. airbyte_cdk/sources/streams/http/requests_native_auth/abstract_oauth.py +2 -1
  93. airbyte_cdk/sources/streams/http/requests_native_auth/oauth.py +1 -0
  94. airbyte_cdk/sources/utils/schema_helpers.py +3 -2
  95. airbyte_cdk/sql/secrets.py +2 -1
  96. airbyte_cdk/sql/shared/sql_processor.py +8 -6
  97. airbyte_cdk/test/entrypoint_wrapper.py +4 -3
  98. airbyte_cdk/test/mock_http/mocker.py +1 -0
  99. airbyte_cdk/utils/schema_inferrer.py +2 -1
  100. airbyte_cdk/utils/slice_hasher.py +1 -1
  101. airbyte_cdk/utils/traced_exception.py +2 -1
  102. {airbyte_cdk-6.6.6.dist-info → airbyte_cdk-6.6.8.dist-info}/METADATA +8 -1
  103. {airbyte_cdk-6.6.6.dist-info → airbyte_cdk-6.6.8.dist-info}/RECORD +106 -106
  104. {airbyte_cdk-6.6.6.dist-info → airbyte_cdk-6.6.8.dist-info}/LICENSE.txt +0 -0
  105. {airbyte_cdk-6.6.6.dist-info → airbyte_cdk-6.6.8.dist-info}/WHEEL +0 -0
  106. {airbyte_cdk-6.6.6.dist-info → airbyte_cdk-6.6.8.dist-info}/entry_points.txt +0 -0
@@ -11,6 +11,8 @@ from typing import Any, Callable, Dict, List, Mapping, Optional, Tuple, Union
11
11
  import orjson
12
12
  import requests
13
13
  import requests_cache
14
+ from requests.auth import AuthBase
15
+
14
16
  from airbyte_cdk.models import (
15
17
  AirbyteMessageSerializer,
16
18
  AirbyteStreamStatus,
@@ -43,12 +45,13 @@ from airbyte_cdk.sources.streams.http.rate_limiting import (
43
45
  rate_limit_default_backoff_handler,
44
46
  user_defined_backoff_handler,
45
47
  )
48
+ from airbyte_cdk.sources.utils.types import JsonType
49
+ from airbyte_cdk.utils.airbyte_secrets_utils import filter_secrets
46
50
  from airbyte_cdk.utils.constants import ENV_REQUEST_CACHE_PATH
47
51
  from airbyte_cdk.utils.stream_status_utils import (
48
52
  as_airbyte_message as stream_status_as_airbyte_message,
49
53
  )
50
54
  from airbyte_cdk.utils.traced_exception import AirbyteTracedException
51
- from requests.auth import AuthBase
52
55
 
53
56
  BODY_REQUEST_METHODS = ("GET", "POST", "PUT", "PATCH")
54
57
 
@@ -333,6 +336,29 @@ class HttpClient:
333
336
 
334
337
  return response # type: ignore # will either return a valid response of type requests.Response or raise an exception
335
338
 
339
+ def _get_response_body(self, response: requests.Response) -> Optional[JsonType]:
340
+ """
341
+ Extracts and returns the body of an HTTP response.
342
+
343
+ This method attempts to parse the response body as JSON. If the response
344
+ body is not valid JSON, it falls back to decoding the response content
345
+ as a UTF-8 string. If both attempts fail, it returns None.
346
+
347
+ Args:
348
+ response (requests.Response): The HTTP response object.
349
+
350
+ Returns:
351
+ Optional[JsonType]: The parsed JSON object as a string, the decoded
352
+ response content as a string, or None if both parsing attempts fail.
353
+ """
354
+ try:
355
+ return str(response.json())
356
+ except requests.exceptions.JSONDecodeError:
357
+ try:
358
+ return response.content.decode("utf-8")
359
+ except Exception:
360
+ return "The Content of the Response couldn't be decoded."
361
+
336
362
  def _handle_error_resolution(
337
363
  self,
338
364
  response: Optional[requests.Response],
@@ -361,12 +387,18 @@ class HttpClient:
361
387
 
362
388
  if error_resolution.response_action == ResponseAction.FAIL:
363
389
  if response is not None:
364
- error_message = f"'{request.method}' request to '{request.url}' failed with status code '{response.status_code}' and error message '{self._error_message_parser.parse_response_error_message(response)}'"
390
+ filtered_response_message = filter_secrets(
391
+ f"Request (body): '{str(request.body)}'. Response (body): '{self._get_response_body(response)}'. Response (headers): '{response.headers}'."
392
+ )
393
+ error_message = f"'{request.method}' request to '{request.url}' failed with status code '{response.status_code}' and error message: '{self._error_message_parser.parse_response_error_message(response)}'. {filtered_response_message}"
365
394
  else:
366
395
  error_message = (
367
396
  f"'{request.method}' request to '{request.url}' failed with exception: '{exc}'"
368
397
  )
369
398
 
399
+ # ensure the exception message is emitted before raised
400
+ self._logger.error(error_message)
401
+
370
402
  raise MessageRepresentationAirbyteTracedErrors(
371
403
  internal_message=error_message,
372
404
  message=error_resolution.error_message or error_message,
@@ -10,12 +10,13 @@ from typing import Any, List, Mapping, MutableMapping, Optional, Tuple, Union
10
10
  import backoff
11
11
  import pendulum
12
12
  import requests
13
+ from requests.auth import AuthBase
14
+
13
15
  from airbyte_cdk.models import FailureType, Level
14
16
  from airbyte_cdk.sources.http_logger import format_http_message
15
17
  from airbyte_cdk.sources.message import MessageRepository, NoopMessageRepository
16
18
  from airbyte_cdk.utils import AirbyteTracedException
17
19
  from airbyte_cdk.utils.airbyte_secrets_utils import add_to_secrets
18
- from requests.auth import AuthBase
19
20
 
20
21
  from ..exceptions import DefaultBackoffException
21
22
 
@@ -6,6 +6,7 @@ from typing import Any, List, Mapping, Optional, Sequence, Tuple, Union
6
6
 
7
7
  import dpath
8
8
  import pendulum
9
+
9
10
  from airbyte_cdk.config_observation import (
10
11
  create_connector_config_control_message,
11
12
  emit_configuration_as_airbyte_control_message,
@@ -10,12 +10,13 @@ import pkgutil
10
10
  from typing import Any, ClassVar, Dict, List, Mapping, MutableMapping, Optional, Tuple
11
11
 
12
12
  import jsonref
13
- from airbyte_cdk.models import ConnectorSpecification, FailureType
14
- from airbyte_cdk.utils.traced_exception import AirbyteTracedException
15
13
  from jsonschema import RefResolver, validate
16
14
  from jsonschema.exceptions import ValidationError
17
15
  from pydantic.v1 import BaseModel, Field
18
16
 
17
+ from airbyte_cdk.models import ConnectorSpecification, FailureType
18
+ from airbyte_cdk.utils.traced_exception import AirbyteTracedException
19
+
19
20
 
20
21
  class JsonFileLoader:
21
22
  """
@@ -6,9 +6,10 @@ from __future__ import annotations
6
6
  import json
7
7
  from typing import TYPE_CHECKING, Any
8
8
 
9
- from airbyte_cdk.sql import exceptions as exc
10
9
  from pydantic_core import CoreSchema, core_schema
11
10
 
11
+ from airbyte_cdk.sql import exceptions as exc
12
+
12
13
  if TYPE_CHECKING:
13
14
  from pydantic import GetCoreSchemaHandler, GetJsonSchemaHandler, ValidationInfo
14
15
  from pydantic.json_schema import JsonSchemaValue
@@ -13,6 +13,12 @@ from typing import TYPE_CHECKING, Any, final
13
13
  import pandas as pd
14
14
  import sqlalchemy
15
15
  import ulid
16
+ from airbyte_protocol_dataclasses.models import AirbyteStateMessage
17
+ from pandas import Index
18
+ from pydantic import BaseModel, Field
19
+ from sqlalchemy import Column, Table, and_, create_engine, insert, null, select, text, update
20
+ from sqlalchemy.exc import ProgrammingError, SQLAlchemyError
21
+
16
22
  from airbyte_cdk.sql import exceptions as exc
17
23
  from airbyte_cdk.sql._util.hashing import one_way_hash
18
24
  from airbyte_cdk.sql._util.name_normalizers import LowerCaseNormalizer
@@ -24,16 +30,10 @@ from airbyte_cdk.sql.constants import (
24
30
  )
25
31
  from airbyte_cdk.sql.secrets import SecretString
26
32
  from airbyte_cdk.sql.types import SQLTypeConverter
27
- from airbyte_protocol_dataclasses.models import AirbyteStateMessage
28
- from pandas import Index
29
- from pydantic import BaseModel, Field
30
- from sqlalchemy import Column, Table, and_, create_engine, insert, null, select, text, update
31
- from sqlalchemy.exc import ProgrammingError, SQLAlchemyError
32
33
 
33
34
  if TYPE_CHECKING:
34
35
  from collections.abc import Generator
35
36
 
36
- from airbyte_cdk.sql.shared.catalog_providers import CatalogProvider
37
37
  from sqlalchemy.engine import Connection, Engine
38
38
  from sqlalchemy.engine.cursor import CursorResult
39
39
  from sqlalchemy.engine.reflection import Inspector
@@ -41,6 +41,8 @@ if TYPE_CHECKING:
41
41
  from sqlalchemy.sql.elements import TextClause
42
42
  from sqlalchemy.sql.type_api import TypeEngine
43
43
 
44
+ from airbyte_cdk.sql.shared.catalog_providers import CatalogProvider
45
+
44
46
 
45
47
  class SQLRuntimeError(Exception):
46
48
  """Raised when an SQL operation fails."""
@@ -23,6 +23,10 @@ from io import StringIO
23
23
  from pathlib import Path
24
24
  from typing import Any, List, Mapping, Optional, Union
25
25
 
26
+ from orjson import orjson
27
+ from pydantic import ValidationError as V2ValidationError
28
+ from serpyco_rs import SchemaValidationError
29
+
26
30
  from airbyte_cdk.entrypoint import AirbyteEntrypoint
27
31
  from airbyte_cdk.exception_handler import assemble_uncaught_exception
28
32
  from airbyte_cdk.logger import AirbyteLogFormatter
@@ -40,9 +44,6 @@ from airbyte_cdk.models import (
40
44
  Type,
41
45
  )
42
46
  from airbyte_cdk.sources import Source
43
- from orjson import orjson
44
- from pydantic import ValidationError as V2ValidationError
45
- from serpyco_rs import SchemaValidationError
46
47
 
47
48
 
48
49
  class EntrypointOutput:
@@ -7,6 +7,7 @@ from types import TracebackType
7
7
  from typing import Callable, List, Optional, Union
8
8
 
9
9
  import requests_mock
10
+
10
11
  from airbyte_cdk.test.mock_http import HttpRequest, HttpRequestMatcher, HttpResponse
11
12
 
12
13
 
@@ -5,11 +5,12 @@
5
5
  from collections import defaultdict
6
6
  from typing import Any, Dict, List, Mapping, Optional
7
7
 
8
- from airbyte_cdk.models import AirbyteRecordMessage
9
8
  from genson import SchemaBuilder, SchemaNode
10
9
  from genson.schema.strategies.object import Object
11
10
  from genson.schema.strategies.scalar import Number
12
11
 
12
+ from airbyte_cdk.models import AirbyteRecordMessage
13
+
13
14
  # schema keywords
14
15
  _TYPE = "type"
15
16
  _NULL_TYPE = "null"
@@ -1,6 +1,6 @@
1
1
  import hashlib
2
2
  import json
3
- from typing import Any, Mapping, Optional, Final
3
+ from typing import Any, Final, Mapping, Optional
4
4
 
5
5
 
6
6
  class SliceEncoder(json.JSONEncoder):
@@ -5,6 +5,8 @@ import time
5
5
  import traceback
6
6
  from typing import Optional
7
7
 
8
+ from orjson import orjson
9
+
8
10
  from airbyte_cdk.models import (
9
11
  AirbyteConnectionStatus,
10
12
  AirbyteErrorTraceMessage,
@@ -18,7 +20,6 @@ from airbyte_cdk.models import (
18
20
  )
19
21
  from airbyte_cdk.models import Type as MessageType
20
22
  from airbyte_cdk.utils.airbyte_secrets_utils import filter_secrets
21
- from orjson import orjson
22
23
 
23
24
 
24
25
  class AirbyteTracedException(Exception):
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.1
2
2
  Name: airbyte-cdk
3
- Version: 6.6.6
3
+ Version: 6.6.8
4
4
  Summary: A framework for writing Airbyte Connectors.
5
5
  Home-page: https://airbyte.com
6
6
  License: MIT
@@ -170,6 +170,13 @@ Installing all extras is required to run the full suite of unit tests.
170
170
 
171
171
  To see all available scripts, run `poetry run poe`.
172
172
 
173
+ #### Formatting the code
174
+
175
+ - Iterate on the CDK code locally
176
+ - Run `poetry run ruff format` to format your changes.
177
+
178
+ To see all available `ruff` options, run `poetry run ruff`.
179
+
173
180
  ##### Autogenerated files
174
181
 
175
182
  Low-code CDK models are generated from `sources/declarative/declarative_component_schema.yaml`. If