airbyte-cdk 6.7.0rc3__py3-none-any.whl → 6.7.1__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 (123) 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/__init__.py +2 -0
  13. airbyte_cdk/models/airbyte_protocol.py +2 -1
  14. airbyte_cdk/sources/concurrent_source/concurrent_read_processor.py +3 -3
  15. airbyte_cdk/sources/concurrent_source/concurrent_source.py +1 -1
  16. airbyte_cdk/sources/config.py +2 -1
  17. airbyte_cdk/sources/declarative/auth/jwt.py +1 -0
  18. airbyte_cdk/sources/declarative/auth/oauth.py +1 -0
  19. airbyte_cdk/sources/declarative/auth/selective_authenticator.py +1 -0
  20. airbyte_cdk/sources/declarative/auth/token.py +2 -1
  21. airbyte_cdk/sources/declarative/auth/token_provider.py +3 -2
  22. airbyte_cdk/sources/declarative/concurrent_declarative_source.py +24 -54
  23. airbyte_cdk/sources/declarative/declarative_component_schema.yaml +167 -0
  24. airbyte_cdk/sources/declarative/decoders/json_decoder.py +3 -2
  25. airbyte_cdk/sources/declarative/decoders/noop_decoder.py +1 -0
  26. airbyte_cdk/sources/declarative/decoders/pagination_decoder_decorator.py +1 -0
  27. airbyte_cdk/sources/declarative/decoders/xml_decoder.py +1 -0
  28. airbyte_cdk/sources/declarative/extractors/dpath_extractor.py +1 -0
  29. airbyte_cdk/sources/declarative/extractors/http_selector.py +1 -0
  30. airbyte_cdk/sources/declarative/extractors/record_filter.py +6 -48
  31. airbyte_cdk/sources/declarative/extractors/record_selector.py +32 -4
  32. airbyte_cdk/sources/declarative/extractors/response_to_file_extractor.py +2 -1
  33. airbyte_cdk/sources/declarative/incremental/datetime_based_cursor.py +2 -1
  34. airbyte_cdk/sources/declarative/incremental/global_substream_cursor.py +5 -2
  35. airbyte_cdk/sources/declarative/incremental/per_partition_cursor.py +5 -2
  36. airbyte_cdk/sources/declarative/incremental/per_partition_with_global.py +1 -3
  37. airbyte_cdk/sources/declarative/interpolation/jinja.py +5 -4
  38. airbyte_cdk/sources/declarative/manifest_declarative_source.py +4 -3
  39. airbyte_cdk/sources/declarative/migrations/legacy_to_per_partition_state_migration.py +1 -1
  40. airbyte_cdk/sources/declarative/models/declarative_component_schema.py +122 -0
  41. airbyte_cdk/sources/declarative/parsers/model_to_component_factory.py +26 -17
  42. airbyte_cdk/sources/declarative/partition_routers/substream_partition_router.py +1 -0
  43. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/constant_backoff_strategy.py +1 -0
  44. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/exponential_backoff_strategy.py +1 -0
  45. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/wait_time_from_header_backoff_strategy.py +1 -0
  46. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/wait_until_time_from_header_backoff_strategy.py +1 -0
  47. airbyte_cdk/sources/declarative/requesters/error_handlers/composite_error_handler.py +1 -0
  48. airbyte_cdk/sources/declarative/requesters/error_handlers/default_error_handler.py +1 -0
  49. airbyte_cdk/sources/declarative/requesters/error_handlers/default_http_response_filter.py +1 -0
  50. airbyte_cdk/sources/declarative/requesters/error_handlers/http_response_filter.py +1 -0
  51. airbyte_cdk/sources/declarative/requesters/http_job_repository.py +2 -1
  52. airbyte_cdk/sources/declarative/requesters/http_requester.py +1 -0
  53. airbyte_cdk/sources/declarative/requesters/paginators/default_paginator.py +1 -0
  54. airbyte_cdk/sources/declarative/requesters/paginators/no_pagination.py +1 -0
  55. airbyte_cdk/sources/declarative/requesters/paginators/paginator.py +1 -0
  56. airbyte_cdk/sources/declarative/requesters/paginators/strategies/cursor_pagination_strategy.py +1 -0
  57. airbyte_cdk/sources/declarative/requesters/paginators/strategies/offset_increment.py +1 -0
  58. airbyte_cdk/sources/declarative/requesters/paginators/strategies/page_increment.py +1 -0
  59. airbyte_cdk/sources/declarative/requesters/paginators/strategies/pagination_strategy.py +1 -0
  60. airbyte_cdk/sources/declarative/requesters/paginators/strategies/stop_condition.py +9 -3
  61. airbyte_cdk/sources/declarative/requesters/request_options/interpolated_request_options_provider.py +2 -1
  62. airbyte_cdk/sources/declarative/requesters/requester.py +1 -0
  63. airbyte_cdk/sources/declarative/retrievers/async_retriever.py +2 -1
  64. airbyte_cdk/sources/declarative/retrievers/simple_retriever.py +12 -4
  65. airbyte_cdk/sources/declarative/stream_slicers/declarative_partition_generator.py +7 -4
  66. airbyte_cdk/sources/declarative/transformations/add_fields.py +1 -0
  67. airbyte_cdk/sources/declarative/transformations/remove_fields.py +1 -0
  68. airbyte_cdk/sources/declarative/yaml_declarative_source.py +1 -0
  69. airbyte_cdk/sources/embedded/tools.py +1 -0
  70. airbyte_cdk/sources/file_based/config/abstract_file_based_spec.py +2 -1
  71. airbyte_cdk/sources/file_based/config/avro_format.py +2 -1
  72. airbyte_cdk/sources/file_based/config/csv_format.py +2 -1
  73. airbyte_cdk/sources/file_based/config/excel_format.py +2 -1
  74. airbyte_cdk/sources/file_based/config/file_based_stream_config.py +2 -1
  75. airbyte_cdk/sources/file_based/config/jsonl_format.py +2 -1
  76. airbyte_cdk/sources/file_based/config/parquet_format.py +2 -1
  77. airbyte_cdk/sources/file_based/config/unstructured_format.py +2 -1
  78. airbyte_cdk/sources/file_based/file_based_source.py +2 -1
  79. airbyte_cdk/sources/file_based/file_based_stream_reader.py +2 -1
  80. airbyte_cdk/sources/file_based/file_types/avro_parser.py +1 -0
  81. airbyte_cdk/sources/file_based/file_types/csv_parser.py +2 -1
  82. airbyte_cdk/sources/file_based/file_types/excel_parser.py +5 -5
  83. airbyte_cdk/sources/file_based/file_types/jsonl_parser.py +2 -1
  84. airbyte_cdk/sources/file_based/file_types/parquet_parser.py +2 -1
  85. airbyte_cdk/sources/file_based/file_types/unstructured_parser.py +9 -8
  86. airbyte_cdk/sources/file_based/stream/abstract_file_based_stream.py +2 -1
  87. airbyte_cdk/sources/file_based/stream/concurrent/adapters.py +5 -4
  88. airbyte_cdk/sources/file_based/stream/concurrent/cursor/abstract_concurrent_file_based_cursor.py +1 -1
  89. airbyte_cdk/sources/file_based/stream/concurrent/cursor/file_based_concurrent_cursor.py +1 -1
  90. airbyte_cdk/sources/file_based/stream/concurrent/cursor/file_based_final_state_cursor.py +1 -1
  91. airbyte_cdk/sources/http_logger.py +1 -0
  92. airbyte_cdk/sources/streams/call_rate.py +1 -2
  93. airbyte_cdk/sources/streams/concurrent/abstract_stream.py +2 -1
  94. airbyte_cdk/sources/streams/concurrent/adapters.py +8 -4
  95. airbyte_cdk/sources/streams/concurrent/availability_strategy.py +2 -1
  96. airbyte_cdk/sources/streams/concurrent/cursor.py +30 -6
  97. airbyte_cdk/sources/streams/concurrent/partitions/partition.py +1 -1
  98. airbyte_cdk/sources/streams/concurrent/partitions/types.py +1 -1
  99. airbyte_cdk/sources/streams/concurrent/state_converters/datetime_stream_state_converter.py +1 -1
  100. airbyte_cdk/sources/streams/core.py +2 -1
  101. airbyte_cdk/sources/streams/http/error_handlers/default_error_mapping.py +2 -1
  102. airbyte_cdk/sources/streams/http/error_handlers/http_status_error_handler.py +1 -0
  103. airbyte_cdk/sources/streams/http/error_handlers/json_error_message_parser.py +1 -0
  104. airbyte_cdk/sources/streams/http/error_handlers/response_models.py +2 -1
  105. airbyte_cdk/sources/streams/http/http.py +3 -2
  106. airbyte_cdk/sources/streams/http/http_client.py +49 -2
  107. airbyte_cdk/sources/streams/http/requests_native_auth/abstract_oauth.py +2 -1
  108. airbyte_cdk/sources/streams/http/requests_native_auth/oauth.py +1 -0
  109. airbyte_cdk/sources/types.py +14 -1
  110. airbyte_cdk/sources/utils/schema_helpers.py +3 -2
  111. airbyte_cdk/sql/secrets.py +2 -1
  112. airbyte_cdk/sql/shared/sql_processor.py +8 -6
  113. airbyte_cdk/test/entrypoint_wrapper.py +4 -3
  114. airbyte_cdk/test/mock_http/mocker.py +1 -0
  115. airbyte_cdk/utils/schema_inferrer.py +2 -1
  116. airbyte_cdk/utils/slice_hasher.py +1 -1
  117. airbyte_cdk/utils/traced_exception.py +2 -1
  118. {airbyte_cdk-6.7.0rc3.dist-info → airbyte_cdk-6.7.1.dist-info}/METADATA +9 -2
  119. {airbyte_cdk-6.7.0rc3.dist-info → airbyte_cdk-6.7.1.dist-info}/RECORD +122 -123
  120. airbyte_cdk/sources/streams/concurrent/partitions/record.py +0 -35
  121. {airbyte_cdk-6.7.0rc3.dist-info → airbyte_cdk-6.7.1.dist-info}/LICENSE.txt +0 -0
  122. {airbyte_cdk-6.7.0rc3.dist-info → airbyte_cdk-6.7.1.dist-info}/WHEEL +0 -0
  123. {airbyte_cdk-6.7.0rc3.dist-info → airbyte_cdk-6.7.1.dist-info}/entry_points.txt +0 -0
@@ -14,9 +14,8 @@ from urllib import parse
14
14
 
15
15
  import requests
16
16
  import requests_cache
17
- from pyrate_limiter import InMemoryBucket, Limiter
17
+ from pyrate_limiter import InMemoryBucket, Limiter, RateItem, TimeClock
18
18
  from pyrate_limiter import Rate as PyRateRate
19
- from pyrate_limiter import RateItem, TimeClock
20
19
  from pyrate_limiter.exceptions import BucketFullException
21
20
 
22
21
  # prevents mypy from complaining about missing session attributes in LimiterMixin
@@ -5,12 +5,13 @@
5
5
  from abc import ABC, abstractmethod
6
6
  from typing import Any, Iterable, Mapping, Optional
7
7
 
8
+ from deprecated.classic import deprecated
9
+
8
10
  from airbyte_cdk.models import AirbyteStream
9
11
  from airbyte_cdk.sources.source import ExperimentalClassWarning
10
12
  from airbyte_cdk.sources.streams.concurrent.availability_strategy import StreamAvailability
11
13
  from airbyte_cdk.sources.streams.concurrent.cursor import Cursor
12
14
  from airbyte_cdk.sources.streams.concurrent.partitions.partition import Partition
13
- from deprecated.classic import deprecated
14
15
 
15
16
 
16
17
  @deprecated("This class is experimental. Use at your own risk.", category=ExperimentalClassWarning)
@@ -8,6 +8,8 @@ import logging
8
8
  from functools import lru_cache
9
9
  from typing import Any, Iterable, List, Mapping, MutableMapping, Optional, Tuple, Union
10
10
 
11
+ from deprecated.classic import deprecated
12
+
11
13
  from airbyte_cdk.models import (
12
14
  AirbyteLogMessage,
13
15
  AirbyteMessage,
@@ -37,12 +39,10 @@ from airbyte_cdk.sources.streams.concurrent.helpers import (
37
39
  )
38
40
  from airbyte_cdk.sources.streams.concurrent.partitions.partition import Partition
39
41
  from airbyte_cdk.sources.streams.concurrent.partitions.partition_generator import PartitionGenerator
40
- from airbyte_cdk.sources.streams.concurrent.partitions.record import Record
41
42
  from airbyte_cdk.sources.streams.core import StreamData
43
+ from airbyte_cdk.sources.types import Record
42
44
  from airbyte_cdk.sources.utils.schema_helpers import InternalConfig
43
45
  from airbyte_cdk.sources.utils.slice_logger import SliceLogger
44
- from deprecated.classic import deprecated
45
-
46
46
  from airbyte_cdk.utils.slice_hasher import SliceHasher
47
47
 
48
48
  """
@@ -294,7 +294,11 @@ class StreamPartition(Partition):
294
294
  self._stream.transformer.transform(
295
295
  data_to_return, self._stream.get_json_schema()
296
296
  )
297
- yield Record(data_to_return, self)
297
+ yield Record(
298
+ data=data_to_return,
299
+ stream_name=self.stream_name(),
300
+ associated_slice=self._slice,
301
+ )
298
302
  else:
299
303
  self._message_repository.emit_message(record_data)
300
304
  except Exception as e:
@@ -6,9 +6,10 @@ import logging
6
6
  from abc import ABC, abstractmethod
7
7
  from typing import Optional
8
8
 
9
- from airbyte_cdk.sources.source import ExperimentalClassWarning
10
9
  from deprecated.classic import deprecated
11
10
 
11
+ from airbyte_cdk.sources.source import ExperimentalClassWarning
12
+
12
13
 
13
14
  class StreamAvailability(ABC):
14
15
  @abstractmethod
@@ -3,6 +3,7 @@
3
3
  #
4
4
 
5
5
  import functools
6
+ import logging
6
7
  from abc import ABC, abstractmethod
7
8
  from typing import Any, Callable, Iterable, List, Mapping, MutableMapping, Optional, Protocol, Tuple
8
9
 
@@ -10,12 +11,13 @@ from airbyte_cdk.sources.connector_state_manager import ConnectorStateManager
10
11
  from airbyte_cdk.sources.message import MessageRepository
11
12
  from airbyte_cdk.sources.streams import NO_CURSOR_STATE_KEY
12
13
  from airbyte_cdk.sources.streams.concurrent.partitions.partition import Partition
13
- from airbyte_cdk.sources.streams.concurrent.partitions.record import Record
14
14
  from airbyte_cdk.sources.streams.concurrent.partitions.stream_slicer import StreamSlicer
15
15
  from airbyte_cdk.sources.streams.concurrent.state_converters.abstract_stream_state_converter import (
16
16
  AbstractStreamStateConverter,
17
17
  )
18
- from airbyte_cdk.sources.types import StreamSlice
18
+ from airbyte_cdk.sources.types import Record, StreamSlice
19
+
20
+ LOGGER = logging.getLogger("airbyte")
19
21
 
20
22
 
21
23
  def _extract_value(mapping: Mapping[str, Any], path: List[str]) -> Any:
@@ -173,9 +175,11 @@ class ConcurrentCursor(Cursor):
173
175
  self.start, self._concurrent_state = self._get_concurrent_state(stream_state)
174
176
  self._lookback_window = lookback_window
175
177
  self._slice_range = slice_range
176
- self._most_recent_cursor_value_per_partition: MutableMapping[Partition, Any] = {}
178
+ self._most_recent_cursor_value_per_partition: MutableMapping[StreamSlice, Any] = {}
177
179
  self._has_closed_at_least_one_slice = False
178
180
  self._cursor_granularity = cursor_granularity
181
+ # Flag to track if the logger has been triggered (per stream)
182
+ self._should_be_synced_logger_triggered = False
179
183
 
180
184
  @property
181
185
  def state(self) -> MutableMapping[str, Any]:
@@ -210,12 +214,12 @@ class ConcurrentCursor(Cursor):
210
214
 
211
215
  def observe(self, record: Record) -> None:
212
216
  most_recent_cursor_value = self._most_recent_cursor_value_per_partition.get(
213
- record.partition
217
+ record.associated_slice
214
218
  )
215
219
  cursor_value = self._extract_cursor_value(record)
216
220
 
217
221
  if most_recent_cursor_value is None or most_recent_cursor_value < cursor_value:
218
- self._most_recent_cursor_value_per_partition[record.partition] = cursor_value
222
+ self._most_recent_cursor_value_per_partition[record.associated_slice] = cursor_value
219
223
 
220
224
  def _extract_cursor_value(self, record: Record) -> Any:
221
225
  return self._connector_state_converter.parse_value(self._cursor_field.extract_value(record))
@@ -231,7 +235,9 @@ class ConcurrentCursor(Cursor):
231
235
  self._has_closed_at_least_one_slice = True
232
236
 
233
237
  def _add_slice_to_state(self, partition: Partition) -> None:
234
- most_recent_cursor_value = self._most_recent_cursor_value_per_partition.get(partition)
238
+ most_recent_cursor_value = self._most_recent_cursor_value_per_partition.get(
239
+ partition.to_slice()
240
+ )
235
241
 
236
242
  if self._slice_boundary_fields:
237
243
  if "slices" not in self.state:
@@ -442,3 +448,21 @@ class ConcurrentCursor(Cursor):
442
448
  return lower + step
443
449
  except OverflowError:
444
450
  return self._end_provider()
451
+
452
+ def should_be_synced(self, record: Record) -> bool:
453
+ """
454
+ Determines if a record should be synced based on its cursor value.
455
+ :param record: The record to evaluate
456
+
457
+ :return: True if the record's cursor value falls within the sync boundaries
458
+ """
459
+ try:
460
+ record_cursor_value: CursorValueType = self._extract_cursor_value(record) # type: ignore # cursor_field is converted to an InterpolatedString in __post_init__
461
+ except ValueError:
462
+ if not self._should_be_synced_logger_triggered:
463
+ LOGGER.warning(
464
+ f"Could not find cursor field `{self.cursor_field.cursor_field_key}` in record. The incremental sync will assume it needs to be synced"
465
+ )
466
+ self._should_be_synced_logger_triggered = True
467
+ return True
468
+ return self.start <= record_cursor_value <= self._end_provider()
@@ -5,7 +5,7 @@
5
5
  from abc import ABC, abstractmethod
6
6
  from typing import Any, Iterable, Mapping, Optional
7
7
 
8
- from airbyte_cdk.sources.streams.concurrent.partitions.record import Record
8
+ from airbyte_cdk.sources.types import Record
9
9
 
10
10
 
11
11
  class Partition(ABC):
@@ -8,7 +8,7 @@ from airbyte_cdk.sources.concurrent_source.partition_generation_completed_sentin
8
8
  PartitionGenerationCompletedSentinel,
9
9
  )
10
10
  from airbyte_cdk.sources.streams.concurrent.partitions.partition import Partition
11
- from airbyte_cdk.sources.streams.concurrent.partitions.record import Record
11
+ from airbyte_cdk.sources.types import Record
12
12
 
13
13
 
14
14
  class PartitionCompleteSentinel:
@@ -7,6 +7,7 @@ from datetime import datetime, timedelta, timezone
7
7
  from typing import Any, Callable, List, MutableMapping, Optional, Tuple
8
8
 
9
9
  import pendulum
10
+ from pendulum.datetime import DateTime
10
11
 
11
12
  # FIXME We would eventually like the Concurrent package do be agnostic of the declarative package. However, this is a breaking change and
12
13
  # the goal in the short term is only to fix the issue we are seeing for source-declarative-manifest.
@@ -16,7 +17,6 @@ from airbyte_cdk.sources.streams.concurrent.state_converters.abstract_stream_sta
16
17
  AbstractStreamStateConverter,
17
18
  ConcurrencyCompatibleStateType,
18
19
  )
19
- from pendulum.datetime import DateTime
20
20
 
21
21
 
22
22
  class DateTimeStreamStateConverter(AbstractStreamStateConverter):
@@ -10,6 +10,8 @@ from dataclasses import dataclass
10
10
  from functools import cached_property, lru_cache
11
11
  from typing import Any, Dict, Iterable, Iterator, List, Mapping, MutableMapping, Optional, Union
12
12
 
13
+ from deprecated import deprecated
14
+
13
15
  import airbyte_cdk.sources.utils.casing as casing
14
16
  from airbyte_cdk.models import (
15
17
  AirbyteMessage,
@@ -35,7 +37,6 @@ from airbyte_cdk.sources.types import StreamSlice
35
37
  from airbyte_cdk.sources.utils.schema_helpers import InternalConfig, ResourceSchemaLoader
36
38
  from airbyte_cdk.sources.utils.slice_logger import DebugSliceLogger, SliceLogger
37
39
  from airbyte_cdk.sources.utils.transform import TransformConfig, TypeTransformer
38
- from deprecated import deprecated
39
40
 
40
41
  # A stream's read method can return one of the following types:
41
42
  # Mapping[str, Any]: The content of an AirbyteRecordMessage
@@ -4,12 +4,13 @@
4
4
 
5
5
  from typing import Mapping, Type, Union
6
6
 
7
+ from requests.exceptions import InvalidSchema, InvalidURL, RequestException
8
+
7
9
  from airbyte_cdk.models import FailureType
8
10
  from airbyte_cdk.sources.streams.http.error_handlers.response_models import (
9
11
  ErrorResolution,
10
12
  ResponseAction,
11
13
  )
12
- from requests.exceptions import InvalidSchema, InvalidURL, RequestException
13
14
 
14
15
  DEFAULT_ERROR_MAPPING: Mapping[Union[int, str, Type[Exception]], ErrorResolution] = {
15
16
  InvalidSchema: ErrorResolution(
@@ -7,6 +7,7 @@ from datetime import timedelta
7
7
  from typing import Mapping, Optional, Union
8
8
 
9
9
  import requests
10
+
10
11
  from airbyte_cdk.models import FailureType
11
12
  from airbyte_cdk.sources.streams.http.error_handlers.default_error_mapping import (
12
13
  DEFAULT_ERROR_MAPPING,
@@ -5,6 +5,7 @@
5
5
  from typing import Optional
6
6
 
7
7
  import requests
8
+
8
9
  from airbyte_cdk.sources.streams.http.error_handlers import ErrorMessageParser
9
10
  from airbyte_cdk.sources.utils.types import JsonType
10
11
 
@@ -5,9 +5,10 @@ from enum import Enum
5
5
  from typing import Optional, Union
6
6
 
7
7
  import requests
8
+ from requests import HTTPError
9
+
8
10
  from airbyte_cdk.models import FailureType
9
11
  from airbyte_cdk.utils.airbyte_secrets_utils import filter_secrets
10
- from requests import HTTPError
11
12
 
12
13
 
13
14
  class ResponseAction(Enum):
@@ -9,6 +9,9 @@ from typing import Any, Callable, Iterable, List, Mapping, MutableMapping, Optio
9
9
  from urllib.parse import urljoin
10
10
 
11
11
  import requests
12
+ from deprecated import deprecated
13
+ from requests.auth import AuthBase
14
+
12
15
  from airbyte_cdk.models import AirbyteMessage, FailureType, SyncMode
13
16
  from airbyte_cdk.models import Type as MessageType
14
17
  from airbyte_cdk.sources.message.repository import InMemoryMessageRepository
@@ -33,8 +36,6 @@ from airbyte_cdk.sources.streams.http.error_handlers.response_models import (
33
36
  from airbyte_cdk.sources.streams.http.http_client import HttpClient
34
37
  from airbyte_cdk.sources.types import Record, StreamSlice
35
38
  from airbyte_cdk.sources.utils.types import JsonType
36
- from deprecated import deprecated
37
- from requests.auth import AuthBase
38
39
 
39
40
  # list of all possible HTTP methods which can be used for sending of request bodies
40
41
  BODY_REQUEST_METHODS = ("GET", "POST", "PUT", "PATCH")
@@ -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
 
@@ -73,6 +76,7 @@ class MessageRepresentationAirbyteTracedErrors(AirbyteTracedException):
73
76
  class HttpClient:
74
77
  _DEFAULT_MAX_RETRY: int = 5
75
78
  _DEFAULT_MAX_TIME: int = 60 * 10
79
+ _ACTIONS_TO_RETRY_ON = {ResponseAction.RETRY, ResponseAction.RATE_LIMITED}
76
80
 
77
81
  def __init__(
78
82
  self,
@@ -333,6 +337,40 @@ class HttpClient:
333
337
 
334
338
  return response # type: ignore # will either return a valid response of type requests.Response or raise an exception
335
339
 
340
+ def _get_response_body(self, response: requests.Response) -> Optional[JsonType]:
341
+ """
342
+ Extracts and returns the body of an HTTP response.
343
+
344
+ This method attempts to parse the response body as JSON. If the response
345
+ body is not valid JSON, it falls back to decoding the response content
346
+ as a UTF-8 string. If both attempts fail, it returns None.
347
+
348
+ Args:
349
+ response (requests.Response): The HTTP response object.
350
+
351
+ Returns:
352
+ Optional[JsonType]: The parsed JSON object as a string, the decoded
353
+ response content as a string, or None if both parsing attempts fail.
354
+ """
355
+ try:
356
+ return str(response.json())
357
+ except requests.exceptions.JSONDecodeError:
358
+ try:
359
+ return response.content.decode("utf-8")
360
+ except Exception:
361
+ return "The Content of the Response couldn't be decoded."
362
+
363
+ def _evict_key(self, prepared_request: requests.PreparedRequest) -> None:
364
+ """
365
+ Addresses high memory consumption when enabling concurrency in https://github.com/airbytehq/oncall/issues/6821.
366
+
367
+ The `_request_attempt_count` attribute keeps growing as multiple requests are made using the same `http_client`.
368
+ To mitigate this issue, we evict keys for completed requests once we confirm that no further retries are needed.
369
+ This helps manage memory usage more efficiently while maintaining the necessary logic for retry attempts.
370
+ """
371
+ if prepared_request in self._request_attempt_count:
372
+ del self._request_attempt_count[prepared_request]
373
+
336
374
  def _handle_error_resolution(
337
375
  self,
338
376
  response: Optional[requests.Response],
@@ -341,6 +379,9 @@ class HttpClient:
341
379
  error_resolution: ErrorResolution,
342
380
  exit_on_rate_limit: Optional[bool] = False,
343
381
  ) -> None:
382
+ if error_resolution.response_action not in self._ACTIONS_TO_RETRY_ON:
383
+ self._evict_key(request)
384
+
344
385
  # Emit stream status RUNNING with the reason RATE_LIMITED to log that the rate limit has been reached
345
386
  if error_resolution.response_action == ResponseAction.RATE_LIMITED:
346
387
  # TODO: Update to handle with message repository when concurrent message repository is ready
@@ -361,12 +402,18 @@ class HttpClient:
361
402
 
362
403
  if error_resolution.response_action == ResponseAction.FAIL:
363
404
  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)}'"
405
+ filtered_response_message = filter_secrets(
406
+ f"Request (body): '{str(request.body)}'. Response (body): '{self._get_response_body(response)}'. Response (headers): '{response.headers}'."
407
+ )
408
+ 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
409
  else:
366
410
  error_message = (
367
411
  f"'{request.method}' request to '{request.url}' failed with exception: '{exc}'"
368
412
  )
369
413
 
414
+ # ensure the exception message is emitted before raised
415
+ self._logger.error(error_message)
416
+
370
417
  raise MessageRepresentationAirbyteTracedErrors(
371
418
  internal_message=error_message,
372
419
  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,
@@ -6,6 +6,8 @@ from __future__ import annotations
6
6
 
7
7
  from typing import Any, ItemsView, Iterator, KeysView, List, Mapping, Optional, ValuesView
8
8
 
9
+ import orjson
10
+
9
11
  # A FieldPointer designates a path to a field inside a mapping. For example, retrieving ["k1", "k1.2"] in the object {"k1" :{"k1.2":
10
12
  # "hello"}] returns "hello"
11
13
  FieldPointer = List[str]
@@ -15,9 +17,17 @@ StreamState = Mapping[str, Any]
15
17
 
16
18
 
17
19
  class Record(Mapping[str, Any]):
18
- def __init__(self, data: Mapping[str, Any], associated_slice: Optional[StreamSlice]):
20
+ def __init__(
21
+ self,
22
+ data: Mapping[str, Any],
23
+ stream_name: str,
24
+ associated_slice: Optional[StreamSlice] = None,
25
+ is_file_transfer_message: bool = False,
26
+ ):
19
27
  self._data = data
20
28
  self._associated_slice = associated_slice
29
+ self.stream_name = stream_name
30
+ self.is_file_transfer_message = is_file_transfer_message
21
31
 
22
32
  @property
23
33
  def data(self) -> Mapping[str, Any]:
@@ -139,3 +149,6 @@ class StreamSlice(Mapping[str, Any]):
139
149
 
140
150
  def __json_serializable__(self) -> Any:
141
151
  return self._stream_slice
152
+
153
+ def __hash__(self) -> int:
154
+ return hash(orjson.dumps(self._stream_slice, option=orjson.OPT_SORT_KEYS))
@@ -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: 6.7.0rc3
3
+ Version: 6.7.1
4
4
  Summary: A framework for writing Airbyte Connectors.
5
5
  Home-page: https://airbyte.com
6
6
  License: MIT
@@ -25,7 +25,7 @@ Requires-Dist: Deprecated (>=1.2,<1.3)
25
25
  Requires-Dist: Jinja2 (>=3.1.2,<3.2.0)
26
26
  Requires-Dist: PyYAML (>=6.0.1,<7.0.0)
27
27
  Requires-Dist: Sphinx (>=4.2,<4.3) ; extra == "sphinx-docs"
28
- Requires-Dist: airbyte-protocol-models-dataclasses (>=0.13,<0.14)
28
+ Requires-Dist: airbyte-protocol-models-dataclasses (>=0.14,<0.15)
29
29
  Requires-Dist: avro (>=1.11.2,<1.12.0) ; extra == "file-based"
30
30
  Requires-Dist: backoff
31
31
  Requires-Dist: cachetools
@@ -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