airbyte-cdk 6.6.8rc1__tar.gz → 6.6.8rc12__tar.gz

{airbyte_cdk-6.6.8rc1 → airbyte_cdk-6.6.8rc12}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: airbyte-cdk
-Version: 6.6.8rc1
+Version: 6.6.8rc12
 Summary: A framework for writing Airbyte Connectors.
 Home-page: https://airbyte.com
 License: MIT

{airbyte_cdk-6.6.8rc1 → airbyte_cdk-6.6.8rc12}/airbyte_cdk/sources/declarative/concurrent_declarative_source.py

@@ -24,6 +24,8 @@ from airbyte_cdk.sources.declarative.interpolation import InterpolatedString
 from airbyte_cdk.sources.declarative.manifest_declarative_source import ManifestDeclarativeSource
 from airbyte_cdk.sources.declarative.models.declarative_component_schema import (
     ConcurrencyLevel as ConcurrencyLevelModel,
+)
+from airbyte_cdk.sources.declarative.models.declarative_component_schema import (
     DatetimeBasedCursor as DatetimeBasedCursorModel,
     DeclarativeStream as DeclarativeStreamModel,
 )
@@ -46,7 +48,6 @@ from airbyte_cdk.sources.streams.concurrent.abstract_stream import AbstractStrea
 from airbyte_cdk.sources.streams.concurrent.availability_strategy import (
     AlwaysAvailableAvailabilityStrategy,
 )
-from airbyte_cdk.sources.streams.concurrent.cursor import FinalStateCursor
 from airbyte_cdk.sources.streams.concurrent.default_stream import DefaultStream
 from airbyte_cdk.sources.streams.concurrent.helpers import get_primary_key_from_stream
 
@@ -66,15 +67,6 @@ class ConcurrentDeclarativeSource(ManifestDeclarativeSource, Generic[TState]):
         component_factory: Optional[ModelToComponentFactory] = None,
         **kwargs: Any,
     ) -> None:
-        # To reduce the complexity of the concurrent framework, we are not enabling RFR with synthetic
-        # cursors. We do this by no longer automatically instantiating RFR cursors when converting
-        # the declarative models into runtime components. Concurrent sources will continue to checkpoint
-        # incremental streams running in full refresh.
-        component_factory = component_factory or ModelToComponentFactory(
-            emit_connector_builder_messages=emit_connector_builder_messages,
-            disable_resumable_full_refresh=True,
-        )
-
         super().__init__(
             source_config=source_config,
             debug=debug,
@@ -201,17 +193,6 @@ class ConcurrentDeclarativeSource(ManifestDeclarativeSource, Generic[TState]):
                     declarative_stream.name
                 ].get("incremental_sync")
 
-                partition_router_component_definition = (
-                    name_to_stream_mapping[declarative_stream.name]
-                    .get("retriever")
-                    .get("partition_router")
-                )
-
-                is_substream_without_incremental = (
-                    partition_router_component_definition
-                    and not datetime_based_cursor_component_definition
-                )
-
                 if (
                     datetime_based_cursor_component_definition
                     and datetime_based_cursor_component_definition.get("type", "")
@@ -264,41 +245,6 @@ class ConcurrentDeclarativeSource(ManifestDeclarativeSource, Generic[TState]):
                             cursor=cursor,
                         )
                     )
-                elif is_substream_without_incremental and hasattr(
-                    declarative_stream.retriever, "stream_slicer"
-                ):
-                    partition_generator = StreamSlicerPartitionGenerator(
-                        DeclarativePartitionFactory(
-                            declarative_stream.name,
-                            declarative_stream.get_json_schema(),
-                            self._retriever_factory(
-                                name_to_stream_mapping[declarative_stream.name],
-                                config,
-                                {},
-                            ),
-                            self.message_repository,
-                        ),
-                        declarative_stream.retriever.stream_slicer,
-                    )
-
-                    cursor = FinalStateCursor(
-                        stream_name=declarative_stream.name,
-                        stream_namespace=declarative_stream.namespace,
-                        message_repository=self.message_repository,
-                    )
-
-                    concurrent_streams.append(
-                        DefaultStream(
-                            partition_generator=partition_generator,
-                            name=declarative_stream.name,
-                            json_schema=declarative_stream.get_json_schema(),
-                            availability_strategy=AlwaysAvailableAvailabilityStrategy(),
-                            primary_key=get_primary_key_from_stream(declarative_stream.primary_key),
-                            cursor_field=None,
-                            logger=self.logger,
-                            cursor=cursor,
-                        )
-                    )
                 else:
                     synchronous_streams.append(declarative_stream)
             else:

{airbyte_cdk-6.6.8rc1 → airbyte_cdk-6.6.8rc12}/airbyte_cdk/sources/declarative/parsers/model_to_component_factory.py

@@ -383,7 +383,6 @@ class ModelToComponentFactory:
         emit_connector_builder_messages: bool = False,
         disable_retries: bool = False,
         disable_cache: bool = False,
-        disable_resumable_full_refresh: bool = False,
         message_repository: Optional[MessageRepository] = None,
     ):
         self._init_mappings()
@@ -392,7 +391,6 @@ class ModelToComponentFactory:
         self._emit_connector_builder_messages = emit_connector_builder_messages
         self._disable_retries = disable_retries
         self._disable_cache = disable_cache
-        self._disable_resumable_full_refresh = disable_resumable_full_refresh
         self._message_repository = message_repository or InMemoryMessageRepository(  # type: ignore
             self._evaluate_log_level(emit_connector_builder_messages)
         )
@@ -1336,8 +1334,6 @@ class ModelToComponentFactory:
                 if model.incremental_sync
                 else None
             )
-        elif self._disable_resumable_full_refresh:
-            return stream_slicer
         elif stream_slicer:
             # For the Full-Refresh sub-streams, we use the nested `ChildPartitionResumableFullRefreshCursor`
             return PerPartitionCursor(

{airbyte_cdk-6.6.8rc1 → airbyte_cdk-6.6.8rc12}/airbyte_cdk/sources/streams/http/http_client.py

@@ -1,7 +1,6 @@
 #
 # Copyright (c) 2023 Airbyte, Inc., all rights reserved.
 #
-
 import logging
 import os
 import urllib
@@ -43,6 +42,7 @@ from airbyte_cdk.sources.streams.http.rate_limiting import (
     rate_limit_default_backoff_handler,
     user_defined_backoff_handler,
 )
+from airbyte_cdk.utils.airbyte_secrets_utils import filter_secrets
 from airbyte_cdk.utils.constants import ENV_REQUEST_CACHE_PATH
 from airbyte_cdk.utils.stream_status_utils import (
     as_airbyte_message as stream_status_as_airbyte_message,
@@ -73,6 +73,7 @@ class MessageRepresentationAirbyteTracedErrors(AirbyteTracedException):
 class HttpClient:
     _DEFAULT_MAX_RETRY: int = 5
     _DEFAULT_MAX_TIME: int = 60 * 10
+    _ACTIONS_TO_RETRY_ON = {ResponseAction.RETRY, ResponseAction.RATE_LIMITED}
 
     def __init__(
         self,
@@ -333,6 +334,17 @@ class HttpClient:
 
         return response  # type: ignore # will either return a valid response of type requests.Response or raise an exception
 
+    def _evict_key(self, prepared_request: requests.PreparedRequest) -> None:
+        """
+        Addresses high memory consumption when enabling concurrency in https://github.com/airbytehq/oncall/issues/6821.
+
+        The `_request_attempt_count` attribute keeps growing as multiple requests are made using the same `http_client`.
+        To mitigate this issue, we evict keys for completed requests once we confirm that no further retries are needed.
+        This helps manage memory usage more efficiently while maintaining the necessary logic for retry attempts.
+        """
+        if prepared_request in self._request_attempt_count:
+            del self._request_attempt_count[prepared_request]
+
     def _handle_error_resolution(
         self,
         response: Optional[requests.Response],
@@ -341,6 +353,9 @@ class HttpClient:
         error_resolution: ErrorResolution,
         exit_on_rate_limit: Optional[bool] = False,
     ) -> None:
+        if error_resolution.response_action not in self._ACTIONS_TO_RETRY_ON:
+            self._evict_key(request)
+
         # Emit stream status RUNNING with the reason RATE_LIMITED to log that the rate limit has been reached
         if error_resolution.response_action == ResponseAction.RATE_LIMITED:
             # TODO: Update to handle with message repository when concurrent message repository is ready
@@ -361,12 +376,13 @@ class HttpClient:
 
         if error_resolution.response_action == ResponseAction.FAIL:
             if response is not None:
-                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)}'"
+                error_message = f"'{request.method}' request to '{request.url}' failed with status code '{response.status_code}' and error message '{response.content}'"
             else:
                 error_message = (
                     f"'{request.method}' request to '{request.url}' failed with exception: '{exc}'"
                 )
 
+            self._logger.warning(filter_secrets(error_message))
             raise MessageRepresentationAirbyteTracedErrors(
                 internal_message=error_message,
                 message=error_resolution.error_message or error_message,

{airbyte_cdk-6.6.8rc1 → airbyte_cdk-6.6.8rc12}/pyproject.toml

@@ -23,7 +23,7 @@ classifiers = [
 keywords = ["airbyte", "connector-development-kit", "cdk"]
 
 # Python CDK uses dynamic versioning: https://github.com/mtkennerly/poetry-dynamic-versioning
-version = "6.6.8rc1"  # Version will be calculated dynamically.
+version = "6.6.8rc12"  # Version will be calculated dynamically.
 
 [tool.poetry-dynamic-versioning]
 enable = false
@@ -94,6 +94,7 @@ poethepoet = "^0.24.2"
 pyproject-flake8 = "^6.1.0"
 pytest = "^7"
 pytest-memray = "^1.6.0"
+pympler = "*"
 pytest-cov = "*"
 pytest-httpserver = "*"
 pytest-mock = "*"