airbyte-cdk 6.60.5__tar.gz → 6.60.7__tar.gz

{airbyte_cdk-6.60.5 → airbyte_cdk-6.60.7}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: airbyte-cdk
-Version: 6.60.5
+Version: 6.60.7
 Summary: A framework for writing Airbyte Connectors.
 Home-page: https://airbyte.com
 License: MIT

{airbyte_cdk-6.60.5 → airbyte_cdk-6.60.7}/airbyte_cdk/connector_builder/connector_builder_handler.py

@@ -3,8 +3,8 @@
 #
 
 
-from dataclasses import asdict, dataclass, field
-from typing import Any, ClassVar, Dict, List, Mapping
+from dataclasses import asdict
+from typing import Any, Dict, List, Mapping, Optional
 
 from airbyte_cdk.connector_builder.test_reader import TestReader
 from airbyte_cdk.models import (
@@ -15,45 +15,32 @@ from airbyte_cdk.models import (
     Type,
 )
 from airbyte_cdk.models import Type as MessageType
+from airbyte_cdk.sources.declarative.concurrent_declarative_source import (
+    ConcurrentDeclarativeSource,
+    TestLimits,
+)
 from airbyte_cdk.sources.declarative.declarative_source import DeclarativeSource
 from airbyte_cdk.sources.declarative.manifest_declarative_source import ManifestDeclarativeSource
-from airbyte_cdk.sources.declarative.parsers.model_to_component_factory import (
-    ModelToComponentFactory,
-)
 from airbyte_cdk.utils.airbyte_secrets_utils import filter_secrets
 from airbyte_cdk.utils.datetime_helpers import ab_datetime_now
 from airbyte_cdk.utils.traced_exception import AirbyteTracedException
 
-DEFAULT_MAXIMUM_NUMBER_OF_PAGES_PER_SLICE = 5
-DEFAULT_MAXIMUM_NUMBER_OF_SLICES = 5
-DEFAULT_MAXIMUM_RECORDS = 100
-DEFAULT_MAXIMUM_STREAMS = 100
-
 MAX_PAGES_PER_SLICE_KEY = "max_pages_per_slice"
 MAX_SLICES_KEY = "max_slices"
 MAX_RECORDS_KEY = "max_records"
 MAX_STREAMS_KEY = "max_streams"
 
 
-@dataclass
-class TestLimits:
-    __test__: ClassVar[bool] = False  # Tell Pytest this is not a Pytest class, despite its name
-
-    max_records: int = field(default=DEFAULT_MAXIMUM_RECORDS)
-    max_pages_per_slice: int = field(default=DEFAULT_MAXIMUM_NUMBER_OF_PAGES_PER_SLICE)
-    max_slices: int = field(default=DEFAULT_MAXIMUM_NUMBER_OF_SLICES)
-    max_streams: int = field(default=DEFAULT_MAXIMUM_STREAMS)
-
-
 def get_limits(config: Mapping[str, Any]) -> TestLimits:
     command_config = config.get("__test_read_config", {})
-    max_pages_per_slice = (
-        command_config.get(MAX_PAGES_PER_SLICE_KEY) or DEFAULT_MAXIMUM_NUMBER_OF_PAGES_PER_SLICE
+    return TestLimits(
+        max_records=command_config.get(MAX_RECORDS_KEY, TestLimits.DEFAULT_MAX_RECORDS),
+        max_pages_per_slice=command_config.get(
+            MAX_PAGES_PER_SLICE_KEY, TestLimits.DEFAULT_MAX_PAGES_PER_SLICE
+        ),
+        max_slices=command_config.get(MAX_SLICES_KEY, TestLimits.DEFAULT_MAX_SLICES),
+        max_streams=command_config.get(MAX_STREAMS_KEY, TestLimits.DEFAULT_MAX_STREAMS),
     )
-    max_slices = command_config.get(MAX_SLICES_KEY) or DEFAULT_MAXIMUM_NUMBER_OF_SLICES
-    max_records = command_config.get(MAX_RECORDS_KEY) or DEFAULT_MAXIMUM_RECORDS
-    max_streams = command_config.get(MAX_STREAMS_KEY) or DEFAULT_MAXIMUM_STREAMS
-    return TestLimits(max_records, max_pages_per_slice, max_slices, max_streams)
 
 
 def should_migrate_manifest(config: Mapping[str, Any]) -> bool:
@@ -75,21 +62,30 @@ def should_normalize_manifest(config: Mapping[str, Any]) -> bool:
     return config.get("__should_normalize", False)
 
 
-def create_source(config: Mapping[str, Any], limits: TestLimits) -> ManifestDeclarativeSource:
+def create_source(
+    config: Mapping[str, Any],
+    limits: TestLimits,
+    catalog: Optional[ConfiguredAirbyteCatalog],
+    state: Optional[List[AirbyteStateMessage]],
+) -> ConcurrentDeclarativeSource[Optional[List[AirbyteStateMessage]]]:
     manifest = config["__injected_declarative_manifest"]
-    return ManifestDeclarativeSource(
+
+    # We enforce a concurrency level of 1 so that the stream is processed on a single thread
+    # to retain ordering for the grouping of the builder message responses.
+    if "concurrency_level" in manifest:
+        manifest["concurrency_level"]["default_concurrency"] = 1
+    else:
+        manifest["concurrency_level"] = {"type": "ConcurrencyLevel", "default_concurrency": 1}
+
+    return ConcurrentDeclarativeSource(
+        catalog=catalog,
         config=config,
-        emit_connector_builder_messages=True,
+        state=state,
         source_config=manifest,
+        emit_connector_builder_messages=True,
         migrate_manifest=should_migrate_manifest(config),
         normalize_manifest=should_normalize_manifest(config),
-        component_factory=ModelToComponentFactory(
-            emit_connector_builder_messages=True,
-            limit_pages_fetched_per_slice=limits.max_pages_per_slice,
-            limit_slices_fetched=limits.max_slices,
-            disable_retries=True,
-            disable_cache=True,
-        ),
+        limits=limits,
     )
 
 

{airbyte_cdk-6.60.5 → airbyte_cdk-6.60.7}/airbyte_cdk/connector_builder/main.py

@@ -91,12 +91,12 @@ def handle_connector_builder_request(
 def handle_request(args: List[str]) -> str:
     command, config, catalog, state = get_config_and_catalog_from_args(args)
     limits = get_limits(config)
-    source = create_source(config, limits)
-    return orjson.dumps(
+    source = create_source(config=config, limits=limits, catalog=catalog, state=state)
+    return orjson.dumps(  # type: ignore[no-any-return] # Serializer.dump() always returns AirbyteMessage
         AirbyteMessageSerializer.dump(
             handle_connector_builder_request(source, command, config, catalog, state, limits)
         )
-    ).decode()  # type: ignore[no-any-return] # Serializer.dump() always returns AirbyteMessage
+    ).decode()
 
 
 if __name__ == "__main__":

{airbyte_cdk-6.60.5 → airbyte_cdk-6.60.7}/airbyte_cdk/connector_builder/test_reader/helpers.py

@@ -5,7 +5,7 @@
 import json
 from copy import deepcopy
 from json import JSONDecodeError
-from typing import Any, Dict, List, Mapping, Optional
+from typing import Any, Dict, List, Mapping, Optional, Union
 
 from airbyte_cdk.connector_builder.models import (
     AuxiliaryRequest,
@@ -17,6 +17,8 @@ from airbyte_cdk.connector_builder.models import (
 from airbyte_cdk.models import (
     AirbyteLogMessage,
     AirbyteMessage,
+    AirbyteStateBlob,
+    AirbyteStateMessage,
     OrchestratorType,
     TraceType,
 )
@@ -466,7 +468,7 @@ def handle_current_slice(
     return StreamReadSlices(
         pages=current_slice_pages,
         slice_descriptor=current_slice_descriptor,
-        state=[latest_state_message] if latest_state_message else [],
+        state=[convert_state_blob_to_mapping(latest_state_message)] if latest_state_message else [],
         auxiliary_requests=auxiliary_requests if auxiliary_requests else [],
     )
 
@@ -718,3 +720,23 @@ def get_auxiliary_request_type(stream: dict, http: dict) -> str:  # type: ignore
     Determines the type of the auxiliary request based on the stream and HTTP properties.
     """
     return "PARENT_STREAM" if stream.get("is_substream", False) else str(http.get("type", None))
+
+
+def convert_state_blob_to_mapping(
+    state_message: Union[AirbyteStateMessage, Dict[str, Any]],
+) -> Dict[str, Any]:
+    """
+    The AirbyteStreamState stores state as an AirbyteStateBlob which deceivingly is not
+    a dictionary, but rather a list of kwargs fields. This in turn causes it to not be
+    properly turned into a dictionary when translating this back into response output
+    by the connector_builder_handler using asdict()
+    """
+
+    if isinstance(state_message, AirbyteStateMessage) and state_message.stream:
+        state_value = state_message.stream.stream_state
+        if isinstance(state_value, AirbyteStateBlob):
+            state_value_mapping = {k: v for k, v in state_value.__dict__.items()}
+            state_message.stream.stream_state = state_value_mapping  # type: ignore  # we intentionally set this as a Dict so that StreamReadSlices is translated properly in the resulting HTTP response
+        return state_message  # type: ignore  # See above, but when this is an AirbyteStateMessage we must convert AirbyteStateBlob to a Dict
+    else:
+        return state_message  # type: ignore  # This is guaranteed to be a Dict since we check isinstance AirbyteStateMessage above

{airbyte_cdk-6.60.5 → airbyte_cdk-6.60.7}/airbyte_cdk/sources/concurrent_source/concurrent_read_processor.py

@@ -95,11 +95,14 @@ class ConcurrentReadProcessor:
         """
         stream_name = partition.stream_name()
         self._streams_to_running_partitions[stream_name].add(partition)
+        cursor = self._stream_name_to_instance[stream_name].cursor
         if self._slice_logger.should_log_slice_message(self._logger):
             self._message_repository.emit_message(
                 self._slice_logger.create_slice_log_message(partition.to_slice())
             )
-        self._thread_pool_manager.submit(self._partition_reader.process_partition, partition)
+        self._thread_pool_manager.submit(
+            self._partition_reader.process_partition, partition, cursor
+        )
 
     def on_partition_complete_sentinel(
         self, sentinel: PartitionCompleteSentinel
@@ -112,26 +115,16 @@ class ConcurrentReadProcessor:
         """
         partition = sentinel.partition
 
-        try:
-            if sentinel.is_successful:
-                stream = self._stream_name_to_instance[partition.stream_name()]
-                stream.cursor.close_partition(partition)
-        except Exception as exception:
-            self._flag_exception(partition.stream_name(), exception)
-            yield AirbyteTracedException.from_exception(
-                exception, stream_descriptor=StreamDescriptor(name=partition.stream_name())
-            ).as_sanitized_airbyte_message()
-        finally:
-            partitions_running = self._streams_to_running_partitions[partition.stream_name()]
-            if partition in partitions_running:
-                partitions_running.remove(partition)
-                # If all partitions were generated and this was the last one, the stream is done
-                if (
-                    partition.stream_name() not in self._streams_currently_generating_partitions
-                    and len(partitions_running) == 0
-                ):
-                    yield from self._on_stream_is_done(partition.stream_name())
-            yield from self._message_repository.consume_queue()
+        partitions_running = self._streams_to_running_partitions[partition.stream_name()]
+        if partition in partitions_running:
+            partitions_running.remove(partition)
+            # If all partitions were generated and this was the last one, the stream is done
+            if (
+                partition.stream_name() not in self._streams_currently_generating_partitions
+                and len(partitions_running) == 0
+            ):
+                yield from self._on_stream_is_done(partition.stream_name())
+        yield from self._message_repository.consume_queue()
 
     def on_record(self, record: Record) -> Iterable[AirbyteMessage]:
         """

{airbyte_cdk-6.60.5 → airbyte_cdk-6.60.7}/airbyte_cdk/sources/concurrent_source/concurrent_source.py

@@ -4,7 +4,7 @@
 import concurrent
 import logging
 from queue import Queue
-from typing import Iterable, Iterator, List
+from typing import Iterable, Iterator, List, Optional
 
 from airbyte_cdk.models import AirbyteMessage
 from airbyte_cdk.sources.concurrent_source.concurrent_read_processor import ConcurrentReadProcessor
@@ -16,7 +16,7 @@ from airbyte_cdk.sources.concurrent_source.thread_pool_manager import ThreadPool
 from airbyte_cdk.sources.message import InMemoryMessageRepository, MessageRepository
 from airbyte_cdk.sources.streams.concurrent.abstract_stream import AbstractStream
 from airbyte_cdk.sources.streams.concurrent.partition_enqueuer import PartitionEnqueuer
-from airbyte_cdk.sources.streams.concurrent.partition_reader import PartitionReader
+from airbyte_cdk.sources.streams.concurrent.partition_reader import PartitionLogger, PartitionReader
 from airbyte_cdk.sources.streams.concurrent.partitions.partition import Partition
 from airbyte_cdk.sources.streams.concurrent.partitions.types import (
     PartitionCompleteSentinel,
@@ -43,6 +43,7 @@ class ConcurrentSource:
         logger: logging.Logger,
         slice_logger: SliceLogger,
         message_repository: MessageRepository,
+        queue: Optional[Queue[QueueItem]] = None,
         timeout_seconds: int = DEFAULT_TIMEOUT_SECONDS,
     ) -> "ConcurrentSource":
         is_single_threaded = initial_number_of_partitions_to_generate == 1 and num_workers == 1
@@ -59,12 +60,13 @@ class ConcurrentSource:
             logger,
         )
         return ConcurrentSource(
-            threadpool,
-            logger,
-            slice_logger,
-            message_repository,
-            initial_number_of_partitions_to_generate,
-            timeout_seconds,
+            threadpool=threadpool,
+            logger=logger,
+            slice_logger=slice_logger,
+            queue=queue,
+            message_repository=message_repository,
+            initial_number_partitions_to_generate=initial_number_of_partitions_to_generate,
+            timeout_seconds=timeout_seconds,
         )
 
     def __init__(
@@ -72,6 +74,7 @@ class ConcurrentSource:
         threadpool: ThreadPoolManager,
         logger: logging.Logger,
         slice_logger: SliceLogger = DebugSliceLogger(),
+        queue: Optional[Queue[QueueItem]] = None,
         message_repository: MessageRepository = InMemoryMessageRepository(),
         initial_number_partitions_to_generate: int = 1,
         timeout_seconds: int = DEFAULT_TIMEOUT_SECONDS,
@@ -91,25 +94,28 @@ class ConcurrentSource:
         self._initial_number_partitions_to_generate = initial_number_partitions_to_generate
         self._timeout_seconds = timeout_seconds
 
+        # We set a maxsize to for the main thread to process record items when the queue size grows. This assumes that there are less
+        # threads generating partitions that than are max number of workers. If it weren't the case, we could have threads only generating
+        # partitions which would fill the queue. This number is arbitrarily set to 10_000 but will probably need to be changed given more
+        # information and might even need to be configurable depending on the source
+        self._queue = queue or Queue(maxsize=10_000)
+
     def read(
         self,
         streams: List[AbstractStream],
     ) -> Iterator[AirbyteMessage]:
         self._logger.info("Starting syncing")
-
-        # We set a maxsize to for the main thread to process record items when the queue size grows. This assumes that there are less
-        # threads generating partitions that than are max number of workers. If it weren't the case, we could have threads only generating
-        # partitions which would fill the queue. This number is arbitrarily set to 10_000 but will probably need to be changed given more
-        # information and might even need to be configurable depending on the source
-        queue: Queue[QueueItem] = Queue(maxsize=10_000)
         concurrent_stream_processor = ConcurrentReadProcessor(
             streams,
-            PartitionEnqueuer(queue, self._threadpool),
+            PartitionEnqueuer(self._queue, self._threadpool),
             self._threadpool,
             self._logger,
             self._slice_logger,
             self._message_repository,
-            PartitionReader(queue),
+            PartitionReader(
+                self._queue,
+                PartitionLogger(self._slice_logger, self._logger, self._message_repository),
+            ),
         )
 
         # Enqueue initial partition generation tasks
@@ -117,7 +123,7 @@ class ConcurrentSource:
 
         # Read from the queue until all partitions were generated and read
         yield from self._consume_from_queue(
-            queue,
+            self._queue,
             concurrent_stream_processor,
         )
         self._threadpool.check_for_errors_and_shutdown()
@@ -141,7 +147,10 @@ class ConcurrentSource:
                 airbyte_message_or_record_or_exception,
                 concurrent_stream_processor,
             )
-            if concurrent_stream_processor.is_done() and queue.empty():
+            # In the event that a partition raises an exception, anything remaining in
+            # the queue will be missed because is_done() can raise an exception and exit
+            # out of this loop before remaining items are consumed
+            if queue.empty() and concurrent_stream_processor.is_done():
                 # all partitions were generated and processed. we're done here
                 break
 
@@ -161,5 +170,7 @@ class ConcurrentSource:
             yield from concurrent_stream_processor.on_partition_complete_sentinel(queue_item)
         elif isinstance(queue_item, Record):
             yield from concurrent_stream_processor.on_record(queue_item)
+        elif isinstance(queue_item, AirbyteMessage):
+            yield queue_item
         else:
             raise ValueError(f"Unknown queue item type: {type(queue_item)}")

{airbyte_cdk-6.60.5 → airbyte_cdk-6.60.7}/airbyte_cdk/sources/declarative/concurrent_declarative_source.py

@@ -3,7 +3,11 @@
 #
 
 import logging
-from typing import Any, Generic, Iterator, List, Mapping, MutableMapping, Optional, Tuple
+from dataclasses import dataclass, field
+from queue import Queue
+from typing import Any, ClassVar, Generic, Iterator, List, Mapping, MutableMapping, Optional, Tuple
+
+from airbyte_protocol_dataclasses.models import Level
 
 from airbyte_cdk.models import (
     AirbyteCatalog,
@@ -48,6 +52,8 @@ from airbyte_cdk.sources.declarative.stream_slicers.declarative_partition_genera
     StreamSlicerPartitionGenerator,
 )
 from airbyte_cdk.sources.declarative.types import ConnectionDefinition
+from airbyte_cdk.sources.message.concurrent_repository import ConcurrentMessageRepository
+from airbyte_cdk.sources.message.repository import InMemoryMessageRepository, MessageRepository
 from airbyte_cdk.sources.source import TState
 from airbyte_cdk.sources.streams import Stream
 from airbyte_cdk.sources.streams.concurrent.abstract_stream import AbstractStream
@@ -58,6 +64,22 @@ from airbyte_cdk.sources.streams.concurrent.availability_strategy import (
 from airbyte_cdk.sources.streams.concurrent.cursor import ConcurrentCursor, FinalStateCursor
 from airbyte_cdk.sources.streams.concurrent.default_stream import DefaultStream
 from airbyte_cdk.sources.streams.concurrent.helpers import get_primary_key_from_stream
+from airbyte_cdk.sources.streams.concurrent.partitions.types import QueueItem
+
+
+@dataclass
+class TestLimits:
+    __test__: ClassVar[bool] = False  # Tell Pytest this is not a Pytest class, despite its name
+
+    DEFAULT_MAX_PAGES_PER_SLICE: ClassVar[int] = 5
+    DEFAULT_MAX_SLICES: ClassVar[int] = 5
+    DEFAULT_MAX_RECORDS: ClassVar[int] = 100
+    DEFAULT_MAX_STREAMS: ClassVar[int] = 100
+
+    max_records: int = field(default=DEFAULT_MAX_RECORDS)
+    max_pages_per_slice: int = field(default=DEFAULT_MAX_PAGES_PER_SLICE)
+    max_slices: int = field(default=DEFAULT_MAX_SLICES)
+    max_streams: int = field(default=DEFAULT_MAX_STREAMS)
 
 
 class ConcurrentDeclarativeSource(ManifestDeclarativeSource, Generic[TState]):
@@ -73,7 +95,9 @@ class ConcurrentDeclarativeSource(ManifestDeclarativeSource, Generic[TState]):
         source_config: ConnectionDefinition,
         debug: bool = False,
         emit_connector_builder_messages: bool = False,
-        component_factory: Optional[ModelToComponentFactory] = None,
+        migrate_manifest: bool = False,
+        normalize_manifest: bool = False,
+        limits: Optional[TestLimits] = None,
         config_path: Optional[str] = None,
         **kwargs: Any,
     ) -> None:
@@ -81,22 +105,40 @@ class ConcurrentDeclarativeSource(ManifestDeclarativeSource, Generic[TState]):
         #  no longer needs to store the original incoming state. But maybe there's an edge case?
         self._connector_state_manager = ConnectorStateManager(state=state)  # type: ignore  # state is always in the form of List[AirbyteStateMessage]. The ConnectorStateManager should use generics, but this can be done later
 
+        # We set a maxsize to for the main thread to process record items when the queue size grows. This assumes that there are less
+        # threads generating partitions that than are max number of workers. If it weren't the case, we could have threads only generating
+        # partitions which would fill the queue. This number is arbitrarily set to 10_000 but will probably need to be changed given more
+        # information and might even need to be configurable depending on the source
+        queue: Queue[QueueItem] = Queue(maxsize=10_000)
+        message_repository = InMemoryMessageRepository(
+            Level.DEBUG if emit_connector_builder_messages else Level.INFO
+        )
+
         # 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(
+        component_factory = ModelToComponentFactory(
             emit_connector_builder_messages=emit_connector_builder_messages,
             disable_resumable_full_refresh=True,
+            message_repository=ConcurrentMessageRepository(queue, message_repository),
             connector_state_manager=self._connector_state_manager,
             max_concurrent_async_job_count=source_config.get("max_concurrent_async_job_count"),
+            limit_pages_fetched_per_slice=limits.max_pages_per_slice if limits else None,
+            limit_slices_fetched=limits.max_slices if limits else None,
+            disable_retries=True if limits else False,
+            disable_cache=True if limits else False,
         )
 
+        self._limits = limits
+
         super().__init__(
             source_config=source_config,
             config=config,
             debug=debug,
             emit_connector_builder_messages=emit_connector_builder_messages,
+            migrate_manifest=migrate_manifest,
+            normalize_manifest=normalize_manifest,
             component_factory=component_factory,
             config_path=config_path,
         )
@@ -126,6 +168,7 @@ class ConcurrentDeclarativeSource(ManifestDeclarativeSource, Generic[TState]):
             initial_number_of_partitions_to_generate=initial_number_of_partitions_to_generate,
             logger=self.logger,
             slice_logger=self._slice_logger,
+            queue=queue,
             message_repository=self.message_repository,
         )
 
@@ -287,6 +330,9 @@ class ConcurrentDeclarativeSource(ManifestDeclarativeSource, Generic[TState]):
                                 self.message_repository,
                             ),
                             stream_slicer=declarative_stream.retriever.stream_slicer,
+                            slice_limit=self._limits.max_slices
+                            if self._limits
+                            else None,  # technically not needed because create_declarative_stream() -> create_simple_retriever() will apply the decorator. But for consistency and depending how we build create_default_stream, this may be needed later
                         )
                     else:
                         if (
@@ -318,6 +364,7 @@ class ConcurrentDeclarativeSource(ManifestDeclarativeSource, Generic[TState]):
                                 self.message_repository,
                             ),
                             stream_slicer=cursor,
+                            slice_limit=self._limits.max_slices if self._limits else None,
                         )
 
                     concurrent_streams.append(
@@ -349,6 +396,9 @@ class ConcurrentDeclarativeSource(ManifestDeclarativeSource, Generic[TState]):
                             self.message_repository,
                         ),
                         declarative_stream.retriever.stream_slicer,
+                        slice_limit=self._limits.max_slices
+                        if self._limits
+                        else None,  # technically not needed because create_declarative_stream() -> create_simple_retriever() will apply the decorator. But for consistency and depending how we build create_default_stream, this may be needed later
                     )
 
                     final_state_cursor = FinalStateCursor(
@@ -410,6 +460,7 @@ class ConcurrentDeclarativeSource(ManifestDeclarativeSource, Generic[TState]):
                             self.message_repository,
                         ),
                         perpartition_cursor,
+                        slice_limit=self._limits.max_slices if self._limits else None,
                     )
 
                     concurrent_streams.append(

{airbyte_cdk-6.60.5 → airbyte_cdk-6.60.7}/airbyte_cdk/sources/declarative/parsers/model_to_component_factory.py

@@ -622,6 +622,10 @@ SCHEMA_TRANSFORMER_TYPE_MAPPING = {
     SchemaNormalizationModel.Default: TransformConfig.DefaultSchemaNormalization,
 }
 
+# Ideally this should use the value defined in ConcurrentDeclarativeSource, but
+# this would be a circular import
+MAX_SLICES = 5
+
 
 class ModelToComponentFactory:
     EPOCH_DATETIME_FORMAT = "%s"
@@ -2112,6 +2116,15 @@ class ModelToComponentFactory:
         stream_slicer: Optional[PartitionRouter],
         config: Config,
     ) -> Optional[StreamSlicer]:
+        state_transformations = (
+            [
+                self._create_component_from_model(state_migration, config, declarative_stream=model)
+                for state_migration in model.state_migrations
+            ]
+            if model.state_migrations
+            else []
+        )
+
         if model.incremental_sync and stream_slicer:
             if model.retriever.type == "AsyncRetriever":
                 stream_name = model.name or ""
@@ -2119,16 +2132,6 @@ class ModelToComponentFactory:
                 stream_state = self._connector_state_manager.get_stream_state(
                     stream_name, stream_namespace
                 )
-                state_transformations = (
-                    [
-                        self._create_component_from_model(
-                            state_migration, config, declarative_stream=model
-                        )
-                        for state_migration in model.state_migrations
-                    ]
-                    if model.state_migrations
-                    else []
-                )
 
                 return self.create_concurrent_cursor_from_perpartition_cursor(  # type: ignore # This is a known issue that we are creating and returning a ConcurrentCursor which does not technically implement the (low-code) StreamSlicer. However, (low-code) StreamSlicer and ConcurrentCursor both implement StreamSlicer.stream_slices() which is the primary method needed for checkpointing
                     state_manager=self._connector_state_manager,
@@ -2172,7 +2175,7 @@ class ModelToComponentFactory:
                     stream_name=model.name or "",
                     stream_namespace=None,
                     config=config or {},
-                    stream_state_migrations=model.state_migrations,
+                    stream_state_migrations=state_transformations,
                 )
             return self._create_component_from_model(model=model.incremental_sync, config=config)  # type: ignore[no-any-return]  # Will be created Cursor as stream_slicer_model is model.incremental_sync
         return None
@@ -2187,19 +2190,15 @@ class ModelToComponentFactory:
             stream_name=model.name or "", namespace=None
         )
 
-        if model.incremental_sync and stream_slicer:
-            # FIXME there is a discrepancy where this logic is applied on the create_*_cursor methods for
-            #   ConcurrentCursor but it is applied outside of create_concurrent_cursor_from_perpartition_cursor
-            if model.state_migrations:
-                state_transformations = [
-                    self._create_component_from_model(
-                        state_migration, config, declarative_stream=model
-                    )
-                    for state_migration in model.state_migrations
-                ]
-            else:
-                state_transformations = []
+        if model.state_migrations:
+            state_transformations = [
+                self._create_component_from_model(state_migration, config, declarative_stream=model)
+                for state_migration in model.state_migrations
+            ]
+        else:
+            state_transformations = []
 
+        if model.incremental_sync and stream_slicer:
             return self.create_concurrent_cursor_from_perpartition_cursor(  # type: ignore # This is a known issue that we are creating and returning a ConcurrentCursor which does not technically implement the (low-code) StreamSlicer. However, (low-code) StreamSlicer and ConcurrentCursor both implement StreamSlicer.stream_slices() which is the primary method needed for checkpointing
                 state_manager=self._connector_state_manager,
                 model_type=DatetimeBasedCursorModel,
@@ -2220,7 +2219,7 @@ class ModelToComponentFactory:
                     stream_name=model.name or "",
                     stream_namespace=None,
                     config=config or {},
-                    stream_state_migrations=model.state_migrations,
+                    stream_state_migrations=state_transformations,
                 )
             elif type(model.incremental_sync) == DatetimeBasedCursorModel:
                 return self.create_concurrent_cursor_from_datetime_based_cursor(  # type: ignore # This is a known issue that we are creating and returning a ConcurrentCursor which does not technically implement the (low-code) StreamSlicer. However, (low-code) StreamSlicer and ConcurrentCursor both implement StreamSlicer.stream_slices() which is the primary method needed for checkpointing
@@ -2229,7 +2228,7 @@ class ModelToComponentFactory:
                     stream_name=model.name or "",
                     stream_namespace=None,
                     config=config or {},
-                    stream_state_migrations=model.state_migrations,
+                    stream_state_migrations=state_transformations,
                     attempt_to_create_cursor_if_not_provided=True,
                 )
             else:

{airbyte_cdk-6.60.5 → airbyte_cdk-6.60.7}/airbyte_cdk/sources/declarative/requesters/http_requester.py

@@ -168,7 +168,13 @@ class HttpRequester(Requester):
             next_page_token=next_page_token,
         )
 
-        full_url = self._join_url(url_base, path) if url_base else url + path if path else url
+        full_url = (
+            self._join_url(url_base, path)
+            if url_base
+            else self._join_url(url, path)
+            if path
+            else url
+        )
 
         return full_url
 

{airbyte_cdk-6.60.5 → airbyte_cdk-6.60.7}/airbyte_cdk/sources/declarative/stream_slicers/declarative_partition_generator.py

@@ -1,8 +1,11 @@
-# Copyright (c) 2024 Airbyte, Inc., all rights reserved.
+# Copyright (c) 2025 Airbyte, Inc., all rights reserved.
 
-from typing import Any, Iterable, Mapping, Optional
+from typing import Any, Iterable, Mapping, Optional, cast
 
 from airbyte_cdk.sources.declarative.retrievers import Retriever
+from airbyte_cdk.sources.declarative.stream_slicers.stream_slicer_test_read_decorator import (
+    StreamSlicerTestReadDecorator,
+)
 from airbyte_cdk.sources.message import MessageRepository
 from airbyte_cdk.sources.streams.concurrent.partitions.partition import Partition
 from airbyte_cdk.sources.streams.concurrent.partitions.partition_generator import PartitionGenerator
@@ -83,10 +86,23 @@ class DeclarativePartition(Partition):
 
 class StreamSlicerPartitionGenerator(PartitionGenerator):
     def __init__(
-        self, partition_factory: DeclarativePartitionFactory, stream_slicer: StreamSlicer
+        self,
+        partition_factory: DeclarativePartitionFactory,
+        stream_slicer: StreamSlicer,
+        slice_limit: Optional[int] = None,
     ) -> None:
         self._partition_factory = partition_factory
-        self._stream_slicer = stream_slicer
+
+        if slice_limit:
+            self._stream_slicer = cast(
+                StreamSlicer,
+                StreamSlicerTestReadDecorator(
+                    wrapped_slicer=stream_slicer,
+                    maximum_number_of_slices=slice_limit,
+                ),
+            )
+        else:
+            self._stream_slicer = stream_slicer
 
     def generate(self) -> Iterable[Partition]:
         for stream_slice in self._stream_slicer.stream_slices():

{airbyte_cdk-6.60.5 → airbyte_cdk-6.60.7}/airbyte_cdk/sources/declarative/stream_slicers/stream_slicer_test_read_decorator.py

@@ -4,10 +4,10 @@
 
 from dataclasses import dataclass
 from itertools import islice
-from typing import Any, Iterable, Mapping, Optional, Union
+from typing import Any, Iterable
 
 from airbyte_cdk.sources.streams.concurrent.partitions.stream_slicer import StreamSlicer
-from airbyte_cdk.sources.types import StreamSlice, StreamState
+from airbyte_cdk.sources.types import StreamSlice
 
 
 @dataclass