airbyte-cdk 6.13.1.dev4101__py3-none-any.whl → 6.13.1.dev4103__py3-none-any.whl

airbyte_cdk/entrypoint.py

@@ -5,6 +5,7 @@
 import argparse
 import importlib
 import ipaddress
+import json
 import logging
 import os.path
 import socket
@@ -46,6 +47,7 @@ logger = init_logger("airbyte")
 
 VALID_URL_SCHEMES = ["https"]
 CLOUD_DEPLOYMENT_MODE = "cloud"
+_HAS_LOGGED_FOR_SERIALIZATION_ERROR = False
 
 
 class AirbyteEntrypoint(object):
@@ -291,7 +293,17 @@ class AirbyteEntrypoint(object):
 
     @staticmethod
     def airbyte_message_to_string(airbyte_message: AirbyteMessage) -> str:
-        return orjson.dumps(AirbyteMessageSerializer.dump(airbyte_message)).decode()
+        global _HAS_LOGGED_FOR_SERIALIZATION_ERROR
+        serialized_message = AirbyteMessageSerializer.dump(airbyte_message)
+        try:
+            return orjson.dumps(serialized_message).decode()
+        except Exception as exception:
+            if not _HAS_LOGGED_FOR_SERIALIZATION_ERROR:
+                logger.warning(
+                    f"There was an error during the serialization of an AirbyteMessage: `{exception}`. This might impact the sync performances."
+                )
+                _HAS_LOGGED_FOR_SERIALIZATION_ERROR = True
+            return json.dumps(serialized_message)
 
     @classmethod
     def extract_state(cls, args: List[str]) -> Optional[Any]:

airbyte_cdk/sources/declarative/concurrent_declarative_source.py

@@ -3,7 +3,7 @@
 #
 
 import logging
-from typing import Any, Callable, Generic, Iterator, List, Mapping, Optional, Tuple, Union
+from typing import Any, Generic, Iterator, List, Mapping, Optional, Tuple
 
 from airbyte_cdk.models import (
     AirbyteCatalog,
@@ -28,15 +28,11 @@ from airbyte_cdk.sources.declarative.models.declarative_component_schema import
 from airbyte_cdk.sources.declarative.models.declarative_component_schema import (
     DatetimeBasedCursor as DatetimeBasedCursorModel,
 )
-from airbyte_cdk.sources.declarative.models.declarative_component_schema import (
-    DeclarativeStream as DeclarativeStreamModel,
-)
 from airbyte_cdk.sources.declarative.parsers.model_to_component_factory import (
-    ComponentDefinition,
     ModelToComponentFactory,
 )
 from airbyte_cdk.sources.declarative.requesters import HttpRequester
-from airbyte_cdk.sources.declarative.retrievers import Retriever, SimpleRetriever
+from airbyte_cdk.sources.declarative.retrievers import SimpleRetriever
 from airbyte_cdk.sources.declarative.stream_slicers.declarative_partition_generator import (
     DeclarativePartitionFactory,
     StreamSlicerPartitionGenerator,
@@ -52,7 +48,6 @@ from airbyte_cdk.sources.streams.concurrent.availability_strategy import (
 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
-from airbyte_cdk.sources.types import Config, StreamState
 
 
 class ConcurrentDeclarativeSource(ManifestDeclarativeSource, Generic[TState]):
@@ -194,10 +189,11 @@ class ConcurrentDeclarativeSource(ManifestDeclarativeSource, Generic[TState]):
             # Some low-code sources use a combination of DeclarativeStream and regular Python streams. We can't inspect
             # these legacy Python streams the way we do low-code streams to determine if they are concurrent compatible,
             # so we need to treat them as synchronous
-            if (
-                isinstance(declarative_stream, DeclarativeStream)
-                and name_to_stream_mapping[declarative_stream.name]["retriever"]["type"]
+            if isinstance(declarative_stream, DeclarativeStream) and (
+                name_to_stream_mapping[declarative_stream.name]["retriever"]["type"]
                 == "SimpleRetriever"
+                or name_to_stream_mapping[declarative_stream.name]["retriever"]["type"]
+                == "AsyncRetriever"
             ):
                 incremental_sync_component_definition = name_to_stream_mapping[
                     declarative_stream.name
@@ -234,15 +230,27 @@ class ConcurrentDeclarativeSource(ManifestDeclarativeSource, Generic[TState]):
                         stream_state=stream_state,
                     )
 
+                    retriever = declarative_stream.retriever
+
+                    # This is an optimization so that we don't invoke any cursor or state management flows within the
+                    # low-code framework because state management is handled through the ConcurrentCursor.
+                    if declarative_stream and isinstance(retriever, SimpleRetriever):
+                        # Also a temporary hack. In the legacy Stream implementation, as part of the read,
+                        # set_initial_state() is called to instantiate incoming state on the cursor. Although we no
+                        # longer rely on the legacy low-code cursor for concurrent checkpointing, low-code components
+                        # like StopConditionPaginationStrategyDecorator and ClientSideIncrementalRecordFilterDecorator
+                        # still rely on a DatetimeBasedCursor that is properly initialized with state.
+                        if retriever.cursor:
+                            retriever.cursor.set_initial_state(stream_state=stream_state)
+                        # We zero it out here, but since this is a cursor reference, the state is still properly
+                        # instantiated for the other components that reference it
+                        retriever.cursor = None
+
                     partition_generator = StreamSlicerPartitionGenerator(
                         DeclarativePartitionFactory(
                             declarative_stream.name,
                             declarative_stream.get_json_schema(),
-                            self._retriever_factory(
-                                name_to_stream_mapping[declarative_stream.name],
-                                config,
-                                stream_state,
-                            ),
+                            retriever,
                             self.message_repository,
                         ),
                         cursor,
@@ -272,11 +280,7 @@ class ConcurrentDeclarativeSource(ManifestDeclarativeSource, Generic[TState]):
                         DeclarativePartitionFactory(
                             declarative_stream.name,
                             declarative_stream.get_json_schema(),
-                            self._retriever_factory(
-                                name_to_stream_mapping[declarative_stream.name],
-                                config,
-                                {},
-                            ),
+                            declarative_stream.retriever,
                             self.message_repository,
                         ),
                         declarative_stream.retriever.stream_slicer,
@@ -415,34 +419,3 @@ class ConcurrentDeclarativeSource(ManifestDeclarativeSource, Generic[TState]):
                 if stream.stream.name not in concurrent_stream_names
             ]
         )
-
-    def _retriever_factory(
-        self, stream_config: ComponentDefinition, source_config: Config, stream_state: StreamState
-    ) -> Callable[[], Retriever]:
-        def _factory_method() -> Retriever:
-            declarative_stream: DeclarativeStream = self._constructor.create_component(
-                DeclarativeStreamModel,
-                stream_config,
-                source_config,
-                emit_connector_builder_messages=self._emit_connector_builder_messages,
-            )
-
-            # This is an optimization so that we don't invoke any cursor or state management flows within the
-            # low-code framework because state management is handled through the ConcurrentCursor.
-            if (
-                declarative_stream
-                and declarative_stream.retriever
-                and isinstance(declarative_stream.retriever, SimpleRetriever)
-            ):
-                # Also a temporary hack. In the legacy Stream implementation, as part of the read, set_initial_state() is
-                # called to instantiate incoming state on the cursor. Although we no longer rely on the legacy low-code cursor
-                # for concurrent checkpointing, low-code components like StopConditionPaginationStrategyDecorator and
-                # ClientSideIncrementalRecordFilterDecorator still rely on a DatetimeBasedCursor that is properly initialized
-                # with state.
-                if declarative_stream.retriever.cursor:
-                    declarative_stream.retriever.cursor.set_initial_state(stream_state=stream_state)
-                declarative_stream.retriever.cursor = None
-
-            return declarative_stream.retriever
-
-        return _factory_method

airbyte_cdk/sources/declarative/declarative_component_schema.yaml

@@ -667,6 +667,28 @@ definitions:
       $parameters:
         type: object
         additionalProperties: true
+  CustomSchemaNormalization:
+    title: Custom Schema Normalization
+    description: Schema normalization component whose behavior is derived from a custom code implementation of the connector.
+    type: object
+    additionalProperties: true
+    required:
+      - type
+      - class_name
+    properties:
+      type:
+        type: string
+        enum: [ CustomSchemaNormalization ]
+      class_name:
+        title: Class Name
+        description: Fully-qualified name of the class that will be implementing the custom normalization. The format is `source_<name>.<package>.<class_name>`.
+        type: string
+        additionalProperties: true
+        examples:
+          - "source_amazon_seller_partner.components.LedgerDetailedViewReportsTypeTransformer"
+      $parameters:
+        type: object
+        additionalProperties: true
   CustomStateMigration:
     title: Custom State Migration
     description: Apply a custom transformation on the input state.
@@ -1241,6 +1263,7 @@ definitions:
             - "$ref": "#/definitions/KeysToLower"
             - "$ref": "#/definitions/KeysToSnakeCase"
             - "$ref": "#/definitions/FlattenFields"
+            - "$ref": "#/definitions/KeysReplace"
       state_migrations:
         title: State Migrations
         description: Array of state migrations to be applied on the input state
@@ -1785,6 +1808,7 @@ definitions:
             - "$ref": "#/definitions/KeysToLower"
             - "$ref": "#/definitions/KeysToSnakeCase"
             - "$ref": "#/definitions/FlattenFields"
+            - "$ref": "#/definitions/KeysReplace"
       schema_type_identifier:
         "$ref": "#/definitions/SchemaTypeIdentifier"
       $parameters:
@@ -1883,6 +1907,49 @@ definitions:
       $parameters:
         type: object
         additionalProperties: true
+  KeysReplace:
+    title: Keys Replace
+    description: A transformation that replaces symbols in keys.
+    type: object
+    required:
+      - type
+      - old
+      - new
+    properties:
+      type:
+        type: string
+        enum: [KeysReplace]
+      old:
+        type: string
+        title: Old value
+        description: Old value to replace.
+        examples:
+          - " "
+          - "{{ record.id }}"
+          - "{{ config['id'] }}"
+          - "{{ stream_slice['id'] }}"
+        interpolation_context:
+          - config
+          - record
+          - stream_state
+          - stream_slice
+      new:
+        type: string
+        title: New value
+        description: New value to set.
+        examples:
+          - "_"
+          - "{{ record.id }}"
+          - "{{ config['id'] }}"
+          - "{{ stream_slice['id'] }}"
+        interpolation_context:
+          - config
+          - record
+          - stream_state
+          - stream_slice
+      $parameters:
+        type: object
+        additionalProperties: true
   IterableDecoder:
     title: Iterable Decoder
     description: Use this if the response consists of strings separated by new lines (`\n`). The Decoder will wrap each row into a JSON object with the `record` key.
@@ -2555,7 +2622,11 @@ definitions:
           - "$ref": "#/definitions/CustomRecordFilter"
           - "$ref": "#/definitions/RecordFilter"
       schema_normalization:
-        "$ref": "#/definitions/SchemaNormalization"
+        title: Schema Normalization
+        description: Responsible for normalization according to the schema.
+        anyOf:
+          - "$ref": "#/definitions/SchemaNormalization"
+          - "$ref": "#/definitions/CustomSchemaNormalization"
         default: None
       $parameters:
         type: object

airbyte_cdk/sources/declarative/extractors/__init__.py

@@ -9,8 +9,10 @@ from airbyte_cdk.sources.declarative.extractors.record_selector import RecordSel
 from airbyte_cdk.sources.declarative.extractors.response_to_file_extractor import (
     ResponseToFileExtractor,
 )
+from airbyte_cdk.sources.declarative.extractors.type_transformer import TypeTransformer
 
 __all__ = [
+    "TypeTransformer",
     "HttpSelector",
     "DpathExtractor",
     "RecordFilter",

airbyte_cdk/sources/declarative/extractors/record_selector.py

@@ -10,16 +10,14 @@ import requests
 from airbyte_cdk.sources.declarative.extractors.http_selector import HttpSelector
 from airbyte_cdk.sources.declarative.extractors.record_extractor import RecordExtractor
 from airbyte_cdk.sources.declarative.extractors.record_filter import RecordFilter
+from airbyte_cdk.sources.declarative.extractors.type_transformer import (
+    TypeTransformer as DeclarativeTypeTransformer,
+)
 from airbyte_cdk.sources.declarative.interpolation import InterpolatedString
 from airbyte_cdk.sources.declarative.models import SchemaNormalization
 from airbyte_cdk.sources.declarative.transformations import RecordTransformation
 from airbyte_cdk.sources.types import Config, Record, StreamSlice, StreamState
-from airbyte_cdk.sources.utils.transform import TransformConfig, TypeTransformer
-
-SCHEMA_TRANSFORMER_TYPE_MAPPING = {
-    SchemaNormalization.None_: TransformConfig.NoTransform,
-    SchemaNormalization.Default: TransformConfig.DefaultSchemaNormalization,
-}
+from airbyte_cdk.sources.utils.transform import TypeTransformer
 
 
 @dataclass
@@ -38,7 +36,7 @@ class RecordSelector(HttpSelector):
     extractor: RecordExtractor
     config: Config
     parameters: InitVar[Mapping[str, Any]]
-    schema_normalization: TypeTransformer
+    schema_normalization: Union[TypeTransformer, DeclarativeTypeTransformer]
     name: str
     _name: Union[InterpolatedString, str] = field(init=False, repr=False, default="")
     record_filter: Optional[RecordFilter] = None

airbyte_cdk/sources/declarative/extractors/type_transformer.py

@@ -0,0 +1,55 @@
+#
+# Copyright (c) 2025 Airbyte, Inc., all rights reserved.
+#
+
+from abc import ABC, abstractmethod
+from dataclasses import dataclass
+from typing import Any, Dict, Mapping
+
+
+@dataclass
+class TypeTransformer(ABC):
+    """
+    Abstract base class for implementing type transformation logic.
+
+    This class provides a blueprint for defining custom transformations
+    on data records based on a provided schema. Implementing classes
+    must override the `transform` method to specify the transformation
+    logic.
+
+    Attributes:
+        None explicitly defined, as this is a dataclass intended to be
+        subclassed.
+
+    Methods:
+        transform(record: Dict[str, Any], schema: Mapping[str, Any]) -> None:
+            Abstract method that must be implemented by subclasses.
+            It performs a transformation on a given data record based
+            on the provided schema.
+
+    Usage:
+        To use this class, create a subclass that implements the
+        `transform` method with the desired transformation logic.
+    """
+
+    @abstractmethod
+    def transform(
+        self,
+        record: Dict[str, Any],
+        schema: Mapping[str, Any],
+    ) -> None:
+        """
+        Perform a transformation on a data record based on a given schema.
+
+        Args:
+            record (Dict[str, Any]): The data record to be transformed.
+            schema (Mapping[str, Any]): The schema that dictates how
+                the record should be transformed.
+
+        Returns:
+            None
+
+        Raises:
+            NotImplementedError: If the method is not implemented
+                by a subclass.
+        """

airbyte_cdk/sources/declarative/interpolation/macros.py

@@ -94,6 +94,26 @@ def max(*args: typing.Any) -> typing.Any:
     return builtins.max(*args)
 
 
+def min(*args: typing.Any) -> typing.Any:
+    """
+    Returns smallest object of an iterable, or two or more arguments.
+
+    min(iterable, *[, default=obj, key=func]) -> value
+    min(arg1, arg2, *args, *[, key=func]) -> value
+
+    Usage:
+    `"{{ min(2,3) }}"
+
+    With a single iterable argument, return its smallest item. The
+    default keyword-only argument specifies an object to return if
+    the provided iterable is empty.
+    With two or more arguments, return the smallest argument.
+    :param args: args to compare
+    :return: smallest argument
+    """
+    return builtins.min(*args)
+
+
 def day_delta(num_days: int, format: str = "%Y-%m-%dT%H:%M:%S.%f%z") -> str:
     """
     Returns datetime of now() + num_days
@@ -147,6 +167,7 @@ _macros_list = [
     today_utc,
     timestamp,
     max,
+    min,
     day_delta,
     duration,
     format_datetime,

airbyte_cdk/sources/declarative/models/declarative_component_schema.py

@@ -268,6 +268,22 @@ class CustomSchemaLoader(BaseModel):
     parameters: Optional[Dict[str, Any]] = Field(None, alias="$parameters")
 
 
+class CustomSchemaNormalization(BaseModel):
+    class Config:
+        extra = Extra.allow
+
+    type: Literal["CustomSchemaNormalization"]
+    class_name: str = Field(
+        ...,
+        description="Fully-qualified name of the class that will be implementing the custom normalization. The format is `source_<name>.<package>.<class_name>`.",
+        examples=[
+            "source_amazon_seller_partner.components.LedgerDetailedViewReportsTypeTransformer"
+        ],
+        title="Class Name",
+    )
+    parameters: Optional[Dict[str, Any]] = Field(None, alias="$parameters")
+
+
 class CustomStateMigration(BaseModel):
     class Config:
         extra = Extra.allow
@@ -721,6 +737,23 @@ class KeysToSnakeCase(BaseModel):
     parameters: Optional[Dict[str, Any]] = Field(None, alias="$parameters")
 
 
+class KeysReplace(BaseModel):
+    type: Literal["KeysReplace"]
+    old: str = Field(
+        ...,
+        description="Old value to replace.",
+        examples=[" ", "{{ record.id }}", "{{ config['id'] }}", "{{ stream_slice['id'] }}"],
+        title="Old value",
+    )
+    new: str = Field(
+        ...,
+        description="New value to set.",
+        examples=["_", "{{ record.id }}", "{{ config['id'] }}", "{{ stream_slice['id'] }}"],
+        title="New value",
+    )
+    parameters: Optional[Dict[str, Any]] = Field(None, alias="$parameters")
+
+
 class FlattenFields(BaseModel):
     type: Literal["FlattenFields"]
     parameters: Optional[Dict[str, Any]] = Field(None, alias="$parameters")
@@ -1513,7 +1546,11 @@ class RecordSelector(BaseModel):
         description="Responsible for filtering records to be emitted by the Source.",
         title="Record Filter",
     )
-    schema_normalization: Optional[SchemaNormalization] = SchemaNormalization.None_
+    schema_normalization: Optional[Union[SchemaNormalization, CustomSchemaNormalization]] = Field(
+        SchemaNormalization.None_,
+        description="Responsible for normalization according to the schema.",
+        title="Schema Normalization",
+    )
     parameters: Optional[Dict[str, Any]] = Field(None, alias="$parameters")
 
 
@@ -1701,6 +1738,7 @@ class DeclarativeStream(BaseModel):
                 KeysToLower,
                 KeysToSnakeCase,
                 FlattenFields,
+                KeysReplace,
             ]
         ]
     ] = Field(
@@ -1875,6 +1913,7 @@ class DynamicSchemaLoader(BaseModel):
                 KeysToLower,
                 KeysToSnakeCase,
                 FlattenFields,
+                KeysReplace,
             ]
         ]
     ] = Field(

airbyte_cdk/sources/declarative/parsers/model_to_component_factory.py

@@ -82,9 +82,6 @@ from airbyte_cdk.sources.declarative.extractors import (
 from airbyte_cdk.sources.declarative.extractors.record_filter import (
     ClientSideIncrementalRecordFilterDecorator,
 )
-from airbyte_cdk.sources.declarative.extractors.record_selector import (
-    SCHEMA_TRANSFORMER_TYPE_MAPPING,
-)
 from airbyte_cdk.sources.declarative.incremental import (
     ChildPartitionResumableFullRefreshCursor,
     CursorFactory,
@@ -100,7 +97,9 @@ from airbyte_cdk.sources.declarative.interpolation.interpolated_mapping import I
 from airbyte_cdk.sources.declarative.migrations.legacy_to_per_partition_state_migration import (
     LegacyToPerPartitionStateMigration,
 )
-from airbyte_cdk.sources.declarative.models import CustomStateMigration
+from airbyte_cdk.sources.declarative.models import (
+    CustomStateMigration,
+)
 from airbyte_cdk.sources.declarative.models.declarative_component_schema import (
     AddedFieldDefinition as AddedFieldDefinitionModel,
 )
@@ -185,6 +184,9 @@ from airbyte_cdk.sources.declarative.models.declarative_component_schema import
 from airbyte_cdk.sources.declarative.models.declarative_component_schema import (
     CustomSchemaLoader as CustomSchemaLoader,
 )
+from airbyte_cdk.sources.declarative.models.declarative_component_schema import (
+    CustomSchemaNormalization as CustomSchemaNormalizationModel,
+)
 from airbyte_cdk.sources.declarative.models.declarative_component_schema import (
     CustomTransformation as CustomTransformationModel,
 )
@@ -254,6 +256,9 @@ from airbyte_cdk.sources.declarative.models.declarative_component_schema import
 from airbyte_cdk.sources.declarative.models.declarative_component_schema import (
     JwtPayload as JwtPayloadModel,
 )
+from airbyte_cdk.sources.declarative.models.declarative_component_schema import (
+    KeysReplace as KeysReplaceModel,
+)
 from airbyte_cdk.sources.declarative.models.declarative_component_schema import (
     KeysToLower as KeysToLowerModel,
 )
@@ -308,6 +313,9 @@ from airbyte_cdk.sources.declarative.models.declarative_component_schema import
 from airbyte_cdk.sources.declarative.models.declarative_component_schema import (
     ResponseToFileExtractor as ResponseToFileExtractorModel,
 )
+from airbyte_cdk.sources.declarative.models.declarative_component_schema import (
+    SchemaNormalization as SchemaNormalizationModel,
+)
 from airbyte_cdk.sources.declarative.models.declarative_component_schema import (
     SchemaTypeIdentifier as SchemaTypeIdentifierModel,
 )
@@ -417,6 +425,9 @@ from airbyte_cdk.sources.declarative.transformations.add_fields import AddedFiel
 from airbyte_cdk.sources.declarative.transformations.flatten_fields import (
     FlattenFields,
 )
+from airbyte_cdk.sources.declarative.transformations.keys_replace_transformation import (
+    KeysReplaceTransformation,
+)
 from airbyte_cdk.sources.declarative.transformations.keys_to_lower_transformation import (
     KeysToLowerTransformation,
 )
@@ -439,6 +450,11 @@ from airbyte_cdk.sources.utils.transform import TransformConfig, TypeTransformer
 
 ComponentDefinition = Mapping[str, Any]
 
+SCHEMA_TRANSFORMER_TYPE_MAPPING = {
+    SchemaNormalizationModel.None_: TransformConfig.NoTransform,
+    SchemaNormalizationModel.Default: TransformConfig.DefaultSchemaNormalization,
+}
+
 
 class ModelToComponentFactory:
     EPOCH_DATETIME_FORMAT = "%s"
@@ -487,6 +503,7 @@ class ModelToComponentFactory:
             CustomRequesterModel: self.create_custom_component,
             CustomRetrieverModel: self.create_custom_component,
             CustomSchemaLoader: self.create_custom_component,
+            CustomSchemaNormalizationModel: self.create_custom_component,
             CustomStateMigration: self.create_custom_component,
             CustomPaginationStrategyModel: self.create_custom_component,
             CustomPartitionRouterModel: self.create_custom_component,
@@ -509,6 +526,7 @@ class ModelToComponentFactory:
             GzipParserModel: self.create_gzip_parser,
             KeysToLowerModel: self.create_keys_to_lower_transformation,
             KeysToSnakeCaseModel: self.create_keys_to_snake_transformation,
+            KeysReplaceModel: self.create_keys_replace_transformation,
             FlattenFieldsModel: self.create_flatten_fields,
             IterableDecoderModel: self.create_iterable_decoder,
             XmlDecoderModel: self.create_xml_decoder,
@@ -630,6 +648,13 @@ class ModelToComponentFactory:
     ) -> KeysToSnakeCaseTransformation:
         return KeysToSnakeCaseTransformation()
 
+    def create_keys_replace_transformation(
+        self, model: KeysReplaceModel, config: Config, **kwargs: Any
+    ) -> KeysReplaceTransformation:
+        return KeysReplaceTransformation(
+            old=model.old, new=model.new, parameters=model.parameters or {}
+        )
+
     def create_flatten_fields(
         self, model: FlattenFieldsModel, config: Config, **kwargs: Any
     ) -> FlattenFields:
@@ -1560,7 +1585,12 @@ class ModelToComponentFactory:
         )
 
     def create_http_requester(
-        self, model: HttpRequesterModel, decoder: Decoder, config: Config, *, name: str
+        self,
+        model: HttpRequesterModel,
+        config: Config,
+        decoder: Decoder = JsonDecoder(parameters={}),
+        *,
+        name: str,
     ) -> HttpRequester:
         authenticator = (
             self._create_component_from_model(
@@ -1976,12 +2006,11 @@ class ModelToComponentFactory:
         config: Config,
         *,
         name: str,
-        transformations: List[RecordTransformation],
-        decoder: Optional[Decoder] = None,
-        client_side_incremental_sync: Optional[Dict[str, Any]] = None,
+        transformations: List[RecordTransformation] | None = None,
+        decoder: Decoder | None = None,
+        client_side_incremental_sync: Dict[str, Any] | None = None,
         **kwargs: Any,
     ) -> RecordSelector:
-        assert model.schema_normalization is not None  # for mypy
         extractor = self._create_component_from_model(
             model=model.extractor, decoder=decoder, config=config
         )
@@ -1999,8 +2028,10 @@ class ModelToComponentFactory:
                 else None,
                 **client_side_incremental_sync,
             )
-        schema_normalization = TypeTransformer(
-            SCHEMA_TRANSFORMER_TYPE_MAPPING[model.schema_normalization]
+        schema_normalization = (
+            TypeTransformer(SCHEMA_TRANSFORMER_TYPE_MAPPING[model.schema_normalization])
+            if isinstance(model.schema_normalization, SchemaNormalizationModel)
+            else self._create_component_from_model(model.schema_normalization, config=config)  # type: ignore[arg-type] # custom normalization model expected here
         )
 
         return RecordSelector(
@@ -2008,7 +2039,7 @@ class ModelToComponentFactory:
             name=name,
             config=config,
             record_filter=record_filter,
-            transformations=transformations,
+            transformations=transformations or [],
             schema_normalization=schema_normalization,
             parameters=model.parameters or {},
         )