airbyte-cdk 6.34.0.dev0__py3-none-any.whl → 6.34.0.dev1__py3-none-any.whl

airbyte_cdk/entrypoint.py

@@ -22,7 +22,7 @@ from requests import PreparedRequest, Response, Session
 
 from airbyte_cdk.connector import TConfig
 from airbyte_cdk.exception_handler import init_uncaught_exception_handler
-from airbyte_cdk.logger import PRINT_BUFFER, init_logger
+from airbyte_cdk.logger import init_logger
 from airbyte_cdk.models import (
     AirbyteConnectionStatus,
     AirbyteMessage,
@@ -337,11 +337,11 @@ def launch(source: Source, args: List[str]) -> None:
     parsed_args = source_entrypoint.parse_args(args)
     # temporarily removes the PrintBuffer because we're seeing weird print behavior for concurrent syncs
     # Refer to: https://github.com/airbytehq/oncall/issues/6235
-    with PRINT_BUFFER:
-        for message in source_entrypoint.run(parsed_args):
-            # simply printing is creating issues for concurrent CDK as Python uses different two instructions to print: one for the message and
-            # the other for the break line. Adding `\n` to the message ensure that both are printed at the same time
-            print(f"{message}\n", end="")
+    # with PrintBuffer():
+    for message in source_entrypoint.run(parsed_args):
+        # simply printing is creating issues for concurrent CDK as Python uses different two instructions to print: one for the message and
+        # the other for the break line. Adding `\n` to the message ensure that both are printed at the same time
+        print(f"{message}\n", end="", flush=True)
 
 
 def _init_internal_request_filter() -> None:

airbyte_cdk/logger.py

@@ -16,11 +16,8 @@ from airbyte_cdk.models import (
     Level,
     Type,
 )
-from airbyte_cdk.utils import PrintBuffer
 from airbyte_cdk.utils.airbyte_secrets_utils import filter_secrets
 
-PRINT_BUFFER = PrintBuffer(flush_interval=0.1)
-
 LOGGING_CONFIG = {
     "version": 1,
     "disable_existing_loggers": False,
@@ -30,7 +27,7 @@ LOGGING_CONFIG = {
     "handlers": {
         "console": {
             "class": "logging.StreamHandler",
-            "stream": PRINT_BUFFER,
+            "stream": "ext://sys.stdout",
             "formatter": "airbyte",
         },
     },

airbyte_cdk/sources/declarative/incremental/concurrent_partition_cursor.py

@@ -5,6 +5,7 @@
 import copy
 import logging
 import threading
+import time
 from collections import OrderedDict
 from copy import deepcopy
 from datetime import timedelta
@@ -58,7 +59,8 @@ class ConcurrentPerPartitionCursor(Cursor):
     CurrentPerPartitionCursor expects the state of the ConcurrentCursor to follow the format {cursor_field: cursor_value}.
     """
 
-    DEFAULT_MAX_PARTITIONS_NUMBER = 10000
+    DEFAULT_MAX_PARTITIONS_NUMBER = 25_000
+    SWITCH_TO_GLOBAL_LIMIT = 10_000
     _NO_STATE: Mapping[str, Any] = {}
     _NO_CURSOR_STATE: Mapping[str, Any] = {}
     _GLOBAL_STATE_KEY = "state"
@@ -99,9 +101,11 @@ class ConcurrentPerPartitionCursor(Cursor):
         self._new_global_cursor: Optional[StreamState] = None
         self._lookback_window: int = 0
         self._parent_state: Optional[StreamState] = None
-        self._over_limit: int = 0
+        self._number_of_partitions: int = 0
         self._use_global_cursor: bool = False
         self._partition_serializer = PerPartitionKeySerializer()
+        # Track the last time a state message was emitted
+        self._last_emission_time: float = 0.0
 
         self._set_initial_state(stream_state)
 
@@ -141,21 +145,16 @@ class ConcurrentPerPartitionCursor(Cursor):
             raise ValueError("stream_slice cannot be None")
 
         partition_key = self._to_partition_key(stream_slice.partition)
-        self._cursor_per_partition[partition_key].close_partition(partition=partition)
         with self._lock:
             self._semaphore_per_partition[partition_key].acquire()
-            cursor = self._cursor_per_partition[partition_key]
-            if (
-                partition_key in self._finished_partitions
-                and self._semaphore_per_partition[partition_key]._value == 0
-            ):
+            if not self._use_global_cursor:
+                self._cursor_per_partition[partition_key].close_partition(partition=partition)
+                cursor = self._cursor_per_partition[partition_key]
                 if (
-                    self._new_global_cursor is None
-                    or self._new_global_cursor[self.cursor_field.cursor_field_key]
-                    < cursor.state[self.cursor_field.cursor_field_key]
+                    partition_key in self._finished_partitions
+                    and self._semaphore_per_partition[partition_key]._value == 0
                 ):
-                    self._new_global_cursor = copy.deepcopy(cursor.state)
-            if not self._use_global_cursor:
+                    self._update_global_cursor(cursor.state[self.cursor_field.cursor_field_key])
                 self._emit_state_message()
 
     def ensure_at_least_one_state_emitted(self) -> None:
@@ -169,9 +168,23 @@ class ConcurrentPerPartitionCursor(Cursor):
             self._global_cursor = self._new_global_cursor
             self._lookback_window = self._timer.finish()
             self._parent_state = self._partition_router.get_stream_state()
-        self._emit_state_message()
+        self._emit_state_message(throttle=False)
 
-    def _emit_state_message(self) -> None:
+    def _throttle_state_message(self) -> Optional[float]:
+        """
+        Throttles the state message emission to once every 60 seconds.
+        """
+        current_time = time.time()
+        if current_time - self._last_emission_time <= 60:
+            return None
+        return current_time
+
+    def _emit_state_message(self, throttle: bool = True) -> None:
+        if throttle:
+            current_time = self._throttle_state_message()
+            if current_time is None:
+                return
+            self._last_emission_time = current_time
         self._connector_state_manager.update_state_for_stream(
             self._stream_name,
             self._stream_namespace,
@@ -202,6 +215,7 @@ class ConcurrentPerPartitionCursor(Cursor):
                 self._lookback_window if self._global_cursor else 0,
             )
             with self._lock:
+                self._number_of_partitions += 1
                 self._cursor_per_partition[self._to_partition_key(partition.partition)] = cursor
             self._semaphore_per_partition[self._to_partition_key(partition.partition)] = (
                 threading.Semaphore(0)
@@ -232,9 +246,15 @@ class ConcurrentPerPartitionCursor(Cursor):
         - Logs a warning each time a partition is removed, indicating whether it was finished
           or removed due to being the oldest.
         """
+        if not self._use_global_cursor and self.limit_reached():
+            logger.info(
+                f"Exceeded the 'SWITCH_TO_GLOBAL_LIMIT' of {self.SWITCH_TO_GLOBAL_LIMIT}. "
+                f"Switching to global cursor for {self._stream_name}."
+            )
+            self._use_global_cursor = True
+
         with self._lock:
             while len(self._cursor_per_partition) > self.DEFAULT_MAX_PARTITIONS_NUMBER - 1:
-                self._over_limit += 1
                 # Try removing finished partitions first
                 for partition_key in list(self._cursor_per_partition.keys()):
                     if (
@@ -245,7 +265,7 @@ class ConcurrentPerPartitionCursor(Cursor):
                             partition_key
                         )  # Remove the oldest partition
                         logger.warning(
-                            f"The maximum number of partitions has been reached. Dropping the oldest finished partition: {oldest_partition}. Over limit: {self._over_limit}."
+                            f"The maximum number of partitions has been reached. Dropping the oldest finished partition: {oldest_partition}. Over limit: {self._number_of_partitions - self.DEFAULT_MAX_PARTITIONS_NUMBER}."
                         )
                         break
                 else:
@@ -254,7 +274,7 @@ class ConcurrentPerPartitionCursor(Cursor):
                         1
                     ]  # Remove the oldest partition
                     logger.warning(
-                        f"The maximum number of partitions has been reached. Dropping the oldest partition: {oldest_partition}. Over limit: {self._over_limit}."
+                        f"The maximum number of partitions has been reached. Dropping the oldest partition: {oldest_partition}. Over limit: {self._number_of_partitions - self.DEFAULT_MAX_PARTITIONS_NUMBER}."
                     )
 
     def _set_initial_state(self, stream_state: StreamState) -> None:
@@ -314,6 +334,7 @@ class ConcurrentPerPartitionCursor(Cursor):
             self._lookback_window = int(stream_state.get("lookback_window", 0))
 
             for state in stream_state.get(self._PERPARTITION_STATE_KEY, []):
+                self._number_of_partitions += 1
                 self._cursor_per_partition[self._to_partition_key(state["partition"])] = (
                     self._create_cursor(state["cursor"])
                 )
@@ -354,16 +375,26 @@ class ConcurrentPerPartitionCursor(Cursor):
             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
-
         if not record.associated_slice:
             raise ValueError(
                 "Invalid state as stream slices that are emitted should refer to an existing cursor"
             )
-        self._cursor_per_partition[
-            self._to_partition_key(record.associated_slice.partition)
-        ].observe(record)
+
+        record_cursor = self._connector_state_converter.output_format(
+            self._connector_state_converter.parse_value(self._cursor_field.extract_value(record))
+        )
+        self._update_global_cursor(record_cursor)
+        if not self._use_global_cursor:
+            self._cursor_per_partition[
+                self._to_partition_key(record.associated_slice.partition)
+            ].observe(record)
+
+    def _update_global_cursor(self, value: Any) -> None:
+        if (
+            self._new_global_cursor is None
+            or self._new_global_cursor[self.cursor_field.cursor_field_key] < value
+        ):
+            self._new_global_cursor = {self.cursor_field.cursor_field_key: copy.deepcopy(value)}
 
     def _to_partition_key(self, partition: Mapping[str, Any]) -> str:
         return self._partition_serializer.to_partition_key(partition)
@@ -397,4 +428,4 @@ class ConcurrentPerPartitionCursor(Cursor):
         return cursor
 
     def limit_reached(self) -> bool:
-        return self._over_limit > self.DEFAULT_MAX_PARTITIONS_NUMBER
+        return self._number_of_partitions > self.SWITCH_TO_GLOBAL_LIMIT

airbyte_cdk/sources/declarative/incremental/datetime_based_cursor.py

@@ -21,6 +21,7 @@ from airbyte_cdk.sources.declarative.requesters.request_option import (
 )
 from airbyte_cdk.sources.message import MessageRepository
 from airbyte_cdk.sources.types import Config, Record, StreamSlice, StreamState
+from airbyte_cdk.utils.mapping_helpers import _validate_component_request_option_paths
 
 
 @dataclass
@@ -122,6 +123,10 @@ class DatetimeBasedCursor(DeclarativeCursor):
         if not self.cursor_datetime_formats:
             self.cursor_datetime_formats = [self.datetime_format]
 
+        _validate_component_request_option_paths(
+            self.config, self.start_time_option, self.end_time_option
+        )
+
     def get_stream_state(self) -> StreamState:
         return {self.cursor_field.eval(self.config): self._cursor} if self._cursor else {}  # type: ignore  # cursor_field is converted to an InterpolatedString in __post_init__
 

airbyte_cdk/sources/declarative/requesters/paginators/default_paginator.py

@@ -23,6 +23,9 @@ from airbyte_cdk.sources.declarative.requesters.request_option import (
 )
 from airbyte_cdk.sources.declarative.requesters.request_path import RequestPath
 from airbyte_cdk.sources.types import Config, Record, StreamSlice, StreamState
+from airbyte_cdk.utils.mapping_helpers import (
+    _validate_component_request_option_paths,
+)
 
 
 @dataclass
@@ -113,6 +116,13 @@ class DefaultPaginator(Paginator):
         if isinstance(self.url_base, str):
             self.url_base = InterpolatedString(string=self.url_base, parameters=parameters)
 
+        if self.page_token_option and not isinstance(self.page_token_option, RequestPath):
+            _validate_component_request_option_paths(
+                self.config,
+                self.page_size_option,
+                self.page_token_option,
+            )
+
     def get_initial_token(self) -> Optional[Any]:
         """
         Return the page token that should be used for the first request of a stream

airbyte_cdk/sources/file_based/config/abstract_file_based_spec.py

@@ -11,6 +11,7 @@ from pydantic.v1 import AnyUrl, BaseModel, Field
 
 from airbyte_cdk import OneOfOptionConfig
 from airbyte_cdk.sources.file_based.config.file_based_stream_config import FileBasedStreamConfig
+from airbyte_cdk.sources.specs.transfer_modes import DeliverPermissions
 from airbyte_cdk.sources.utils import schema_helpers
 
 
@@ -65,7 +66,7 @@ class AbstractFileBasedSpec(BaseModel):
         order=10,
     )
 
-    delivery_method: Union[DeliverRecords, DeliverRawFiles] = Field(
+    delivery_method: Union[DeliverRecords, DeliverRawFiles, DeliverPermissions] = Field(
         title="Delivery Method",
         discriminator="delivery_type",
         type="object",

airbyte_cdk/sources/file_based/config/validate_config_transfer_modes.py

@@ -0,0 +1,81 @@
+#
+# Copyright (c) 2025 Airbyte, Inc., all rights reserved.
+#
+
+from airbyte_cdk.sources.file_based.config.abstract_file_based_spec import (
+    AbstractFileBasedSpec,
+    DeliverRawFiles,
+)
+from airbyte_cdk.sources.specs.transfer_modes import DeliverPermissions
+
+DELIVERY_TYPE_KEY = "delivery_type"
+DELIVERY_TYPE_PERMISSION_TRANSFER_MODE_VALUE = "use_permissions_transfer"
+DELIVERY_TYPE_FILES_TRANSFER_MODE_VALUE = "use_file_transfer"
+PRESERVE_DIRECTORY_STRUCTURE_KEY = "preserve_directory_structure"
+INCLUDE_IDENTITIES_STREAM_KEY = "include_identities_stream"
+
+
+def use_file_transfer(parsed_config: AbstractFileBasedSpec) -> bool:
+    """Returns `True` if the configuration uses file transfer mode."""
+    return (
+        hasattr(parsed_config.delivery_method, DELIVERY_TYPE_KEY)
+        and parsed_config.delivery_method.delivery_type == DELIVERY_TYPE_FILES_TRANSFER_MODE_VALUE
+    )
+
+
+def preserve_directory_structure(parsed_config: AbstractFileBasedSpec) -> bool:
+    """
+    Determines whether to preserve directory structure during file transfer.
+
+    When enabled, files maintain their subdirectory paths in the destination.
+    When disabled, files are flattened to the root of the destination.
+
+    Args:
+        parsed_config: The parsed configuration containing delivery method settings
+
+    Returns:
+        True if directory structure should be preserved (default), False otherwise
+    """
+    if (
+        use_file_transfer(parsed_config)
+        and hasattr(parsed_config.delivery_method, PRESERVE_DIRECTORY_STRUCTURE_KEY)
+        and isinstance(parsed_config.delivery_method, DeliverRawFiles)
+    ):
+        return parsed_config.delivery_method.preserve_directory_structure
+    return True
+
+
+def use_permissions_transfer(parsed_config: AbstractFileBasedSpec) -> bool:
+    """
+    Determines whether to use permissions transfer to sync ACLs and Identities
+
+    Args:
+        parsed_config: The parsed configuration containing delivery method settings
+
+    Returns:
+        True if permissions transfer should be enabled, False otherwise
+    """
+    return (
+        hasattr(parsed_config.delivery_method, DELIVERY_TYPE_KEY)
+        and parsed_config.delivery_method.delivery_type
+        == DELIVERY_TYPE_PERMISSION_TRANSFER_MODE_VALUE
+    )
+
+
+def include_identities_stream(parsed_config: AbstractFileBasedSpec) -> bool:
+    """
+    There are scenarios where user may not have access to identities but still is valuable to get ACLs
+
+    Args:
+        parsed_config: The parsed configuration containing delivery method settings
+
+    Returns:
+        True if we should include Identities stream.
+    """
+    if (
+        use_permissions_transfer(parsed_config)
+        and hasattr(parsed_config.delivery_method, INCLUDE_IDENTITIES_STREAM_KEY)
+        and isinstance(parsed_config.delivery_method, DeliverPermissions)
+    ):
+        return parsed_config.delivery_method.include_identities_stream
+    return False

airbyte_cdk/sources/file_based/file_based_source.py

@@ -33,6 +33,12 @@ from airbyte_cdk.sources.file_based.config.file_based_stream_config import (
     FileBasedStreamConfig,
     ValidationPolicy,
 )
+from airbyte_cdk.sources.file_based.config.validate_config_transfer_modes import (
+    include_identities_stream,
+    preserve_directory_structure,
+    use_file_transfer,
+    use_permissions_transfer,
+)
 from airbyte_cdk.sources.file_based.discovery_policy import (
     AbstractDiscoveryPolicy,
     DefaultDiscoveryPolicy,
@@ -49,7 +55,12 @@ from airbyte_cdk.sources.file_based.schema_validation_policies import (
     DEFAULT_SCHEMA_VALIDATION_POLICIES,
     AbstractSchemaValidationPolicy,
 )
-from airbyte_cdk.sources.file_based.stream import AbstractFileBasedStream, DefaultFileBasedStream
+from airbyte_cdk.sources.file_based.stream import (
+    AbstractFileBasedStream,
+    DefaultFileBasedStream,
+    FileIdentitiesStream,
+    PermissionsFileBasedStream,
+)
 from airbyte_cdk.sources.file_based.stream.concurrent.adapters import FileBasedStreamFacade
 from airbyte_cdk.sources.file_based.stream.concurrent.cursor import (
     AbstractConcurrentFileBasedCursor,
@@ -66,6 +77,7 @@ from airbyte_cdk.utils.traced_exception import AirbyteTracedException
 DEFAULT_CONCURRENCY = 100
 MAX_CONCURRENCY = 100
 INITIAL_N_PARTITIONS = MAX_CONCURRENCY // 2
+IDENTITIES_STREAM = "identities"
 
 
 class FileBasedSource(ConcurrentSourceAdapter, ABC):
@@ -157,13 +169,20 @@ class FileBasedSource(ConcurrentSourceAdapter, ABC):
         errors = []
         tracebacks = []
         for stream in streams:
+            if isinstance(stream, FileIdentitiesStream):
+                identity = next(iter(stream.load_identity_groups()))
+                if not identity:
+                    errors.append(
+                        "Unable to get identities for current configuration, please check your credentials"
+                    )
+                continue
             if not isinstance(stream, AbstractFileBasedStream):
                 raise ValueError(f"Stream {stream} is not a file-based stream.")
             try:
                 parsed_config = self._get_parsed_config(config)
                 availability_method = (
                     stream.availability_strategy.check_availability
-                    if self._use_file_transfer(parsed_config)
+                    if use_file_transfer(parsed_config) or use_permissions_transfer(parsed_config)
                     else stream.availability_strategy.check_availability_and_parsability
                 )
                 (
@@ -239,7 +258,7 @@ class FileBasedSource(ConcurrentSourceAdapter, ABC):
                         message_repository=self.message_repository,
                     )
                     stream = FileBasedStreamFacade.create_from_stream(
-                        stream=self._make_default_stream(
+                        stream=self._make_file_based_stream(
                             stream_config=stream_config,
                             cursor=cursor,
                             parsed_config=parsed_config,
@@ -270,7 +289,7 @@ class FileBasedSource(ConcurrentSourceAdapter, ABC):
                         CursorField(DefaultFileBasedStream.ab_last_mod_col),
                     )
                     stream = FileBasedStreamFacade.create_from_stream(
-                        stream=self._make_default_stream(
+                        stream=self._make_file_based_stream(
                             stream_config=stream_config,
                             cursor=cursor,
                             parsed_config=parsed_config,
@@ -282,13 +301,17 @@ class FileBasedSource(ConcurrentSourceAdapter, ABC):
                     )
                 else:
                     cursor = self.cursor_cls(stream_config)
-                    stream = self._make_default_stream(
+                    stream = self._make_file_based_stream(
                         stream_config=stream_config,
                         cursor=cursor,
                         parsed_config=parsed_config,
                     )
 
                 streams.append(stream)
+
+            if include_identities_stream(parsed_config):
+                identities_stream = self._make_identities_stream()
+                streams.append(identities_stream)
             return streams
 
         except ValidationError as exc:
@@ -310,8 +333,48 @@ class FileBasedSource(ConcurrentSourceAdapter, ABC):
             validation_policy=self._validate_and_get_validation_policy(stream_config),
             errors_collector=self.errors_collector,
             cursor=cursor,
-            use_file_transfer=self._use_file_transfer(parsed_config),
-            preserve_directory_structure=self._preserve_directory_structure(parsed_config),
+            use_file_transfer=use_file_transfer(parsed_config),
+            preserve_directory_structure=preserve_directory_structure(parsed_config),
+        )
+
+    def _make_permissions_stream(
+        self, stream_config: FileBasedStreamConfig, cursor: Optional[AbstractFileBasedCursor]
+    ) -> AbstractFileBasedStream:
+        return PermissionsFileBasedStream(
+            config=stream_config,
+            catalog_schema=self.stream_schemas.get(stream_config.name),
+            stream_reader=self.stream_reader,
+            availability_strategy=self.availability_strategy,
+            discovery_policy=self.discovery_policy,
+            parsers=self.parsers,
+            validation_policy=self._validate_and_get_validation_policy(stream_config),
+            errors_collector=self.errors_collector,
+            cursor=cursor,
+        )
+
+    def _make_file_based_stream(
+        self,
+        stream_config: FileBasedStreamConfig,
+        cursor: Optional[AbstractFileBasedCursor],
+        parsed_config: AbstractFileBasedSpec,
+    ) -> AbstractFileBasedStream:
+        """
+        Creates different streams depending on the type of the transfer mode selected
+        """
+        if use_permissions_transfer(parsed_config):
+            return self._make_permissions_stream(stream_config, cursor)
+        # we should have a stream for File transfer mode to decouple from DefaultFileBasedStream
+        else:
+            return self._make_default_stream(stream_config, cursor, parsed_config)
+
+    def _make_identities_stream(
+        self,
+    ) -> Stream:
+        return FileIdentitiesStream(
+            catalog_schema=self.stream_schemas.get(FileIdentitiesStream.IDENTITIES_STREAM_NAME),
+            stream_reader=self.stream_reader,
+            discovery_policy=self.discovery_policy,
+            errors_collector=self.errors_collector,
         )
 
     def _get_stream_from_catalog(
@@ -378,33 +441,3 @@ class FileBasedSource(ConcurrentSourceAdapter, ABC):
                 "`input_schema` and `schemaless` options cannot both be set",
                 model=FileBasedStreamConfig,
             )
-
-    @staticmethod
-    def _use_file_transfer(parsed_config: AbstractFileBasedSpec) -> bool:
-        use_file_transfer = (
-            hasattr(parsed_config.delivery_method, "delivery_type")
-            and parsed_config.delivery_method.delivery_type == "use_file_transfer"
-        )
-        return use_file_transfer
-
-    @staticmethod
-    def _preserve_directory_structure(parsed_config: AbstractFileBasedSpec) -> bool:
-        """
-        Determines whether to preserve directory structure during file transfer.
-
-        When enabled, files maintain their subdirectory paths in the destination.
-        When disabled, files are flattened to the root of the destination.
-
-        Args:
-            parsed_config: The parsed configuration containing delivery method settings
-
-        Returns:
-            True if directory structure should be preserved (default), False otherwise
-        """
-        if (
-            FileBasedSource._use_file_transfer(parsed_config)
-            and hasattr(parsed_config.delivery_method, "preserve_directory_structure")
-            and parsed_config.delivery_method.preserve_directory_structure is not None
-        ):
-            return parsed_config.delivery_method.preserve_directory_structure
-        return True