airbyte-cdk 6.28.0__tar.gz → 6.28.0.dev0__tar.gz

{airbyte_cdk-6.28.0 → airbyte_cdk-6.28.0.dev0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: airbyte-cdk
-Version: 6.28.0
+Version: 6.28.0.dev0
 Summary: A framework for writing Airbyte Connectors.
 Home-page: https://airbyte.com
 License: MIT

{airbyte_cdk-6.28.0 → airbyte_cdk-6.28.0.dev0}/airbyte_cdk/sources/declarative/incremental/concurrent_partition_cursor.py

@@ -22,6 +22,9 @@ from airbyte_cdk.sources.streams.checkpoint.per_partition_key_serializer import
 )
 from airbyte_cdk.sources.streams.concurrent.cursor import ConcurrentCursor, Cursor, CursorField
 from airbyte_cdk.sources.streams.concurrent.partitions.partition import Partition
+from airbyte_cdk.sources.streams.concurrent.state_converters.abstract_stream_state_converter import (
+    AbstractStreamStateConverter,
+)
 from airbyte_cdk.sources.types import Record, StreamSlice, StreamState
 
 logger = logging.getLogger("airbyte")
@@ -72,6 +75,7 @@ class ConcurrentPerPartitionCursor(Cursor):
         stream_state: Any,
         message_repository: MessageRepository,
         connector_state_manager: ConnectorStateManager,
+        connector_state_converter: AbstractStreamStateConverter,
         cursor_field: CursorField,
     ) -> None:
         self._global_cursor: Optional[StreamState] = {}
@@ -79,6 +83,7 @@ class ConcurrentPerPartitionCursor(Cursor):
         self._stream_namespace = stream_namespace
         self._message_repository = message_repository
         self._connector_state_manager = connector_state_manager
+        self._connector_state_converter = connector_state_converter
         self._cursor_field = cursor_field
 
         self._cursor_factory = cursor_factory
@@ -301,8 +306,7 @@ class ConcurrentPerPartitionCursor(Cursor):
         ):
             # We assume that `stream_state` is in a global format that can be applied to all partitions.
             # Example: {"global_state_format_key": "global_state_format_value"}
-            self._global_cursor = deepcopy(stream_state)
-            self._new_global_cursor = deepcopy(stream_state)
+            self._set_global_state(stream_state)
 
         else:
             self._use_global_cursor = stream_state.get("use_global_cursor", False)
@@ -319,8 +323,7 @@ class ConcurrentPerPartitionCursor(Cursor):
 
             # set default state for missing partitions if it is per partition with fallback to global
             if self._GLOBAL_STATE_KEY in stream_state:
-                self._global_cursor = deepcopy(stream_state[self._GLOBAL_STATE_KEY])
-                self._new_global_cursor = deepcopy(stream_state[self._GLOBAL_STATE_KEY])
+                self._set_global_state(stream_state[self._GLOBAL_STATE_KEY])
 
         # Set initial parent state
         if stream_state.get("parent_state"):
@@ -329,6 +332,27 @@ class ConcurrentPerPartitionCursor(Cursor):
         # Set parent state for partition routers based on parent streams
         self._partition_router.set_initial_state(stream_state)
 
+    def _set_global_state(self, stream_state: Mapping[str, Any]) -> None:
+        """
+        Initializes the global cursor state from the provided stream state.
+
+        If the cursor field key is present in the stream state, its value is parsed,
+        formatted, and stored as the global cursor. This ensures consistency in state
+        representation across partitions.
+        """
+        if self.cursor_field.cursor_field_key in stream_state:
+            global_state_value = stream_state[self.cursor_field.cursor_field_key]
+            final_format_global_state_value = self._connector_state_converter.output_format(
+                self._connector_state_converter.parse_value(global_state_value)
+            )
+
+            fixed_global_state = {
+                self.cursor_field.cursor_field_key: final_format_global_state_value
+            }
+
+            self._global_cursor = deepcopy(fixed_global_state)
+            self._new_global_cursor = deepcopy(fixed_global_state)
+
     def observe(self, record: Record) -> None:
         if not self._use_global_cursor and self.limit_reached():
             self._use_global_cursor = True
@@ -372,5 +396,10 @@ class ConcurrentPerPartitionCursor(Cursor):
         cursor = self._cursor_per_partition[partition_key]
         return cursor
 
+    def _extract_cursor_value_from_state(self, state: StreamState) -> Any:
+        return self._connector_state_converter.parse_value(
+            state[self.cursor_field.cursor_field_key]
+        )
+
     def limit_reached(self) -> bool:
         return self._over_limit > self.DEFAULT_MAX_PARTITIONS_NUMBER

{airbyte_cdk-6.28.0 → airbyte_cdk-6.28.0.dev0}/airbyte_cdk/sources/declarative/parsers/model_to_component_factory.py

@@ -1202,6 +1202,22 @@ class ModelToComponentFactory:
         )
         cursor_field = CursorField(interpolated_cursor_field.eval(config=config))
 
+        datetime_format = datetime_based_cursor_model.datetime_format
+
+        cursor_granularity = (
+            parse_duration(datetime_based_cursor_model.cursor_granularity)
+            if datetime_based_cursor_model.cursor_granularity
+            else None
+        )
+
+        connector_state_converter: DateTimeStreamStateConverter
+        connector_state_converter = CustomFormatConcurrentStreamStateConverter(
+            datetime_format=datetime_format,
+            input_datetime_formats=datetime_based_cursor_model.cursor_datetime_formats,
+            is_sequential_state=True,  # ConcurrentPerPartitionCursor only works with sequential state
+            cursor_granularity=cursor_granularity,
+        )
+
         # Create the cursor factory
         cursor_factory = ConcurrentCursorFactory(
             partial(
@@ -1225,6 +1241,7 @@ class ModelToComponentFactory:
             stream_state=stream_state,
             message_repository=self._message_repository,  # type: ignore
             connector_state_manager=state_manager,
+            connector_state_converter=connector_state_converter,
             cursor_field=cursor_field,
         )
 

{airbyte_cdk-6.28.0 → airbyte_cdk-6.28.0.dev0}/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 = "6.28.0"  # Version will be calculated dynamically.
+version = "6.28.0dev0"  # Version will be calculated dynamically.
 
 [tool.poetry-dynamic-versioning]
 enable = false