airbyte-cdk 7.0.1__tar.gz → 7.0.3__tar.gz

{airbyte_cdk-7.0.1 → airbyte_cdk-7.0.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: airbyte-cdk
-Version: 7.0.1
+Version: 7.0.3
 Summary: A framework for writing Airbyte Connectors.
 Home-page: https://airbyte.com
 License: MIT

{airbyte_cdk-7.0.1 → airbyte_cdk-7.0.3}/airbyte_cdk/sources/declarative/incremental/concurrent_partition_cursor.py

@@ -189,6 +189,7 @@ class ConcurrentPerPartitionCursor(Cursor):
         # FIXME this is a temporary field the time of the migration from declarative cursors to concurrent ones
         self._attempt_to_create_cursor_if_not_provided = attempt_to_create_cursor_if_not_provided
         self._synced_some_data = False
+        self._logged_regarding_datetime_format_error = False
 
     @property
     def cursor_field(self) -> CursorField:
@@ -518,10 +519,23 @@ class ConcurrentPerPartitionCursor(Cursor):
         except ValueError:
             return
 
+        try:
+            record_cursor = self._connector_state_converter.output_format(
+                self._connector_state_converter.parse_value(record_cursor_value)
+            )
+        except ValueError as exception:
+            if not self._logged_regarding_datetime_format_error:
+                logger.warning(
+                    "Skipping cursor update for stream '%s': failed to parse cursor field '%s' value %r: %s",
+                    self._stream_name,
+                    self._cursor_field.cursor_field_key,
+                    record_cursor_value,
+                    exception,
+                )
+                self._logged_regarding_datetime_format_error = True
+            return
+
         self._synced_some_data = True
-        record_cursor = self._connector_state_converter.output_format(
-            self._connector_state_converter.parse_value(record_cursor_value)
-        )
         self._update_global_cursor(record_cursor)
         if not self._use_global_cursor:
             self._cursor_per_partition[

{airbyte_cdk-7.0.1 → airbyte_cdk-7.0.3}/airbyte_cdk/sources/declarative/parsers/model_to_component_factory.py

@@ -1976,7 +1976,10 @@ class ModelToComponentFactory:
         primary_key = model.primary_key.__root__ if model.primary_key else None
 
         partition_router = self._build_stream_slicer_from_partition_router(
-            model.retriever, config, stream_name=model.name
+            model.retriever,
+            config,
+            stream_name=model.name,
+            **kwargs,
         )
         concurrent_cursor = self._build_concurrent_cursor(model, partition_router, config)
         if model.incremental_sync and isinstance(model.incremental_sync, DatetimeBasedCursorModel):
@@ -2155,10 +2158,11 @@ class ModelToComponentFactory:
         ],
         config: Config,
         stream_name: Optional[str] = None,
+        **kwargs: Any,
     ) -> PartitionRouter:
         if (
             hasattr(model, "partition_router")
-            and isinstance(model, SimpleRetrieverModel | AsyncRetrieverModel)
+            and isinstance(model, (SimpleRetrieverModel, AsyncRetrieverModel, CustomRetrieverModel))
             and model.partition_router
         ):
             stream_slicer_model = model.partition_router
@@ -2172,6 +2176,23 @@ class ModelToComponentFactory:
                     ],
                     parameters={},
                 )
+            elif isinstance(stream_slicer_model, dict):
+                # partition router comes from CustomRetrieverModel therefore has not been parsed as a model
+                params = stream_slicer_model.get("$parameters")
+                if not isinstance(params, dict):
+                    params = {}
+                    stream_slicer_model["$parameters"] = params
+
+                if stream_name is not None:
+                    params["stream_name"] = stream_name
+
+                return self._create_nested_component(  # type: ignore[no-any-return] # There is no guarantee that this will return a stream slicer. If not, we expect an AttributeError during the call to `stream_slices`
+                    model,
+                    "partition_router",
+                    stream_slicer_model,
+                    config,
+                    **kwargs,
+                )
             else:
                 return self._create_component_from_model(  # type: ignore[no-any-return] # Will be created PartitionRouter as stream_slicer_model is model.partition_router
                     model=stream_slicer_model, config=config, stream_name=stream_name or ""
@@ -2886,7 +2907,7 @@ class ModelToComponentFactory:
         )
 
     def create_parent_stream_config(
-        self, model: ParentStreamConfigModel, config: Config, stream_name: str, **kwargs: Any
+        self, model: ParentStreamConfigModel, config: Config, *, stream_name: str, **kwargs: Any
     ) -> ParentStreamConfig:
         declarative_stream = self._create_component_from_model(
             model.stream,
@@ -3693,14 +3714,19 @@ class ModelToComponentFactory:
         )
 
     def create_substream_partition_router(
-        self, model: SubstreamPartitionRouterModel, config: Config, **kwargs: Any
+        self,
+        model: SubstreamPartitionRouterModel,
+        config: Config,
+        *,
+        stream_name: str,
+        **kwargs: Any,
     ) -> SubstreamPartitionRouter:
         parent_stream_configs = []
         if model.parent_stream_configs:
             parent_stream_configs.extend(
                 [
                     self.create_parent_stream_config_with_substream_wrapper(
-                        model=parent_stream_config, config=config, **kwargs
+                        model=parent_stream_config, config=config, stream_name=stream_name, **kwargs
                     )
                     for parent_stream_config in model.parent_stream_configs
                 ]
@@ -3720,7 +3746,7 @@ class ModelToComponentFactory:
 
         # This flag will be used exclusively for StateDelegatingStream when a parent stream is created
         has_parent_state = bool(
-            self._connector_state_manager.get_stream_state(kwargs.get("stream_name", ""), None)
+            self._connector_state_manager.get_stream_state(stream_name, None)
             if model.incremental_dependency
             else False
         )
@@ -4113,11 +4139,17 @@ class ModelToComponentFactory:
         )
 
     def create_grouping_partition_router(
-        self, model: GroupingPartitionRouterModel, config: Config, **kwargs: Any
+        self,
+        model: GroupingPartitionRouterModel,
+        config: Config,
+        *,
+        stream_name: str,
+        **kwargs: Any,
     ) -> GroupingPartitionRouter:
         underlying_router = self._create_component_from_model(
             model=model.underlying_partition_router,
             config=config,
+            stream_name=stream_name,
             **kwargs,
         )
         if model.group_size < 1:

{airbyte_cdk-7.0.1 → airbyte_cdk-7.0.3}/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 = "7.0.1"  # Version will be calculated dynamically.
+version = "7.0.3"  # Version will be calculated dynamically.
 
 [tool.poetry-dynamic-versioning]
 enable = false
@@ -120,6 +120,7 @@ types-python-dateutil = "^2.9.0.20241003"
 types-pyyaml = "^6.0.12.20240917"
 types-cachetools = "^5.5.0.20240820"
 deptry = "^0.23.0"
+dagger-io = "0.18.6"
 
 [tool.poetry.extras]
 file-based = ["avro", "fastavro", "pyarrow", "unstructured", "pdf2image", "pdfminer.six", "unstructured.pytesseract", "pytesseract", "markdown", "python-calamine", "python-snappy"]
@@ -147,7 +148,7 @@ pre-commit = {cmd = "poetry run pre-commit run --all-files", help = "Run all pre
 # Build tasks
 assemble = {cmd = "bin/generate-component-manifest-dagger.sh", help = "Generate component manifest files."}
 build-package = {cmd = "poetry build", help = "Build the python package: source and wheels archives."}
-build = {sequence = ["assemble", "openapi-generate", "build-package"], help = "Run all tasks to build the package."}
+build = {sequence = ["assemble", "openapi-generate", "build-package", "ruff-fix"], help = "Run all tasks to build the package."}
 
 # Format check tasks
 format-check = {sequence = ["_format-check-ruff", "_format-check-prettier"], help = "Check formatting for all file types via Ruff and Prettier.", ignore_fail = "return_non_zero"}