Add schema filter

Author: lazebnyi

airbyte_cdk/sources/declarative/declarative_component_schema.yaml

@@ -2318,6 +2318,12 @@ definitions:
           - "$ref": "#/definitions/AsyncRetriever"
           - "$ref": "#/definitions/CustomRetriever"
           - "$ref": "#/definitions/SimpleRetriever"
+      schema_filter:
+        title: Schema Filter
+        description: placeholder
+        anyOf:
+          - "$ref": "#/definitions/RecordFilter"
+          - "$ref": "#/definitions/CustomRecordFilter"
       schema_transformations:
         title: Schema Transformations
         description: A list of transformations to be applied to the schema.

airbyte_cdk/sources/declarative/models/declarative_component_schema.py

@@ -1,5 +1,3 @@
-# Copyright (c) 2025 Airbyte, Inc., all rights reserved.
-
 # generated by datamodel-codegen:
 #   filename:  declarative_component_schema.yaml
 
@@ -620,7 +618,9 @@ class OAuthAuthenticator(BaseModel):
     scopes: Optional[List[str]] = Field(
         None,
         description="List of scopes that should be granted to the access token.",
-        examples=[["crm.list.read", "crm.objects.contacts.read", "crm.schema.contacts.read"]],
+        examples=[
+            ["crm.list.read", "crm.objects.contacts.read", "crm.schema.contacts.read"]
+        ],
         title="Scopes",
     )
     token_expiry_date: Optional[str] = Field(
@@ -1126,24 +1126,28 @@ class OAuthConfigSpecification(BaseModel):
     class Config:
         extra = Extra.allow
 
-    oauth_user_input_from_connector_config_specification: Optional[Dict[str, Any]] = Field(
-        None,
-        description="OAuth specific blob. This is a Json Schema used to validate Json configurations used as input to OAuth.\nMust be a valid non-nested JSON that refers to properties from ConnectorSpecification.connectionSpecification\nusing special annotation 'path_in_connector_config'.\nThese are input values the user is entering through the UI to authenticate to the connector, that might also shared\nas inputs for syncing data via the connector.\nExamples:\nif no connector values is shared during oauth flow, oauth_user_input_from_connector_config_specification=[]\nif connector values such as 'app_id' inside the top level are used to generate the API url for the oauth flow,\n  oauth_user_input_from_connector_config_specification={\n    app_id: {\n      type: string\n      path_in_connector_config: ['app_id']\n    }\n  }\nif connector values such as 'info.app_id' nested inside another object are used to generate the API url for the oauth flow,\n  oauth_user_input_from_connector_config_specification={\n    app_id: {\n      type: string\n      path_in_connector_config: ['info', 'app_id']\n    }\n  }",
-        examples=[
-            {"app_id": {"type": "string", "path_in_connector_config": ["app_id"]}},
-            {
-                "app_id": {
-                    "type": "string",
-                    "path_in_connector_config": ["info", "app_id"],
-                }
-            },
-        ],
-        title="OAuth user input",
+    oauth_user_input_from_connector_config_specification: Optional[Dict[str, Any]] = (
+        Field(
+            None,
+            description="OAuth specific blob. This is a Json Schema used to validate Json configurations used as input to OAuth.\nMust be a valid non-nested JSON that refers to properties from ConnectorSpecification.connectionSpecification\nusing special annotation 'path_in_connector_config'.\nThese are input values the user is entering through the UI to authenticate to the connector, that might also shared\nas inputs for syncing data via the connector.\nExamples:\nif no connector values is shared during oauth flow, oauth_user_input_from_connector_config_specification=[]\nif connector values such as 'app_id' inside the top level are used to generate the API url for the oauth flow,\n  oauth_user_input_from_connector_config_specification={\n    app_id: {\n      type: string\n      path_in_connector_config: ['app_id']\n    }\n  }\nif connector values such as 'info.app_id' nested inside another object are used to generate the API url for the oauth flow,\n  oauth_user_input_from_connector_config_specification={\n    app_id: {\n      type: string\n      path_in_connector_config: ['info', 'app_id']\n    }\n  }",
+            examples=[
+                {"app_id": {"type": "string", "path_in_connector_config": ["app_id"]}},
+                {
+                    "app_id": {
+                        "type": "string",
+                        "path_in_connector_config": ["info", "app_id"],
+                    }
+                },
+            ],
+            title="OAuth user input",
+        )
     )
-    oauth_connector_input_specification: Optional[OauthConnectorInputSpecification] = Field(
-        None,
-        description='The DeclarativeOAuth specific blob.\nPertains to the fields defined by the connector relating to the OAuth flow.\n\nInterpolation capabilities:\n- The variables placeholders are declared as `{{my_var}}`.\n- The nested resolution variables like `{{ {{my_nested_var}} }}` is allowed as well.\n\n- The allowed interpolation context is:\n  + base64Encoder - encode to `base64`, {{ {{my_var_a}}:{{my_var_b}} | base64Encoder }}\n  + base64Decorer - decode from `base64` encoded string, {{ {{my_string_variable_or_string_value}} | base64Decoder }}\n  + urlEncoder - encode the input string to URL-like format, {{ https://test.host.com/endpoint | urlEncoder}}\n  + urlDecorer - decode the input url-encoded string into text format, {{ urlDecoder:https%3A%2F%2Fairbyte.io | urlDecoder}}\n  + codeChallengeS256 - get the `codeChallenge` encoded value to provide additional data-provider specific authorisation values, {{ {{state_value}} | codeChallengeS256 }}\n\nExamples:\n  - The TikTok Marketing DeclarativeOAuth spec:\n  {\n    "oauth_connector_input_specification": {\n      "type": "object",\n      "additionalProperties": false,\n      "properties": {\n          "consent_url": "https://ads.tiktok.com/marketing_api/auth?{{client_id_key}}={{client_id_value}}&{{redirect_uri_key}}={{ {{redirect_uri_value}} | urlEncoder}}&{{state_key}}={{state_value}}",\n          "access_token_url": "https://business-api.tiktok.com/open_api/v1.3/oauth2/access_token/",\n          "access_token_params": {\n              "{{ auth_code_key }}": "{{ auth_code_value }}",\n              "{{ client_id_key }}": "{{ client_id_value }}",\n              "{{ client_secret_key }}": "{{ client_secret_value }}"\n          },\n          "access_token_headers": {\n              "Content-Type": "application/json",\n              "Accept": "application/json"\n          },\n          "extract_output": ["data.access_token"],\n          "client_id_key": "app_id",\n          "client_secret_key": "secret",\n          "auth_code_key": "auth_code"\n      }\n    }\n  }',
-        title="DeclarativeOAuth Connector Specification",
+    oauth_connector_input_specification: Optional[OauthConnectorInputSpecification] = (
+        Field(
+            None,
+            description='The DeclarativeOAuth specific blob.\nPertains to the fields defined by the connector relating to the OAuth flow.\n\nInterpolation capabilities:\n- The variables placeholders are declared as `{{my_var}}`.\n- The nested resolution variables like `{{ {{my_nested_var}} }}` is allowed as well.\n\n- The allowed interpolation context is:\n  + base64Encoder - encode to `base64`, {{ {{my_var_a}}:{{my_var_b}} | base64Encoder }}\n  + base64Decorer - decode from `base64` encoded string, {{ {{my_string_variable_or_string_value}} | base64Decoder }}\n  + urlEncoder - encode the input string to URL-like format, {{ https://test.host.com/endpoint | urlEncoder}}\n  + urlDecorer - decode the input url-encoded string into text format, {{ urlDecoder:https%3A%2F%2Fairbyte.io | urlDecoder}}\n  + codeChallengeS256 - get the `codeChallenge` encoded value to provide additional data-provider specific authorisation values, {{ {{state_value}} | codeChallengeS256 }}\n\nExamples:\n  - The TikTok Marketing DeclarativeOAuth spec:\n  {\n    "oauth_connector_input_specification": {\n      "type": "object",\n      "additionalProperties": false,\n      "properties": {\n          "consent_url": "https://ads.tiktok.com/marketing_api/auth?{{client_id_key}}={{client_id_value}}&{{redirect_uri_key}}={{ {{redirect_uri_value}} | urlEncoder}}&{{state_key}}={{state_value}}",\n          "access_token_url": "https://business-api.tiktok.com/open_api/v1.3/oauth2/access_token/",\n          "access_token_params": {\n              "{{ auth_code_key }}": "{{ auth_code_value }}",\n              "{{ client_id_key }}": "{{ client_id_value }}",\n              "{{ client_secret_key }}": "{{ client_secret_value }}"\n          },\n          "access_token_headers": {\n              "Content-Type": "application/json",\n              "Accept": "application/json"\n          },\n          "extract_output": ["data.access_token"],\n          "client_id_key": "app_id",\n          "client_secret_key": "secret",\n          "auth_code_key": "auth_code"\n      }\n    }\n  }',
+            title="DeclarativeOAuth Connector Specification",
+        )
     )
     complete_oauth_output_specification: Optional[Dict[str, Any]] = Field(
         None,
@@ -1161,7 +1165,9 @@ class Config:
     complete_oauth_server_input_specification: Optional[Dict[str, Any]] = Field(
         None,
         description="OAuth specific blob. This is a Json Schema used to validate Json configurations persisted as Airbyte Server configurations.\nMust be a valid non-nested JSON describing additional fields configured by the Airbyte Instance or Workspace Admins to be used by the\nserver when completing an OAuth flow (typically exchanging an auth code for refresh token).\nExamples:\n    complete_oauth_server_input_specification={\n      client_id: {\n        type: string\n      },\n      client_secret: {\n        type: string\n      }\n    }",
-        examples=[{"client_id": {"type": "string"}, "client_secret": {"type": "string"}}],
+        examples=[
+            {"client_id": {"type": "string"}, "client_secret": {"type": "string"}}
+        ],
         title="OAuth input specification",
     )
     complete_oauth_server_output_specification: Optional[Dict[str, Any]] = Field(
@@ -1887,7 +1893,9 @@ class RecordSelector(BaseModel):
         description="Responsible for filtering records to be emitted by the Source.",
         title="Record Filter",
     )
-    schema_normalization: Optional[Union[SchemaNormalization, CustomSchemaNormalization]] = Field(
+    schema_normalization: Optional[
+        Union[SchemaNormalization, CustomSchemaNormalization]
+    ] = Field(
         None,
         description="Responsible for normalization according to the schema.",
         title="Schema Normalization",
@@ -2133,7 +2141,9 @@ class Config:
         extra = Extra.allow
 
     type: Literal["DeclarativeStream"]
-    name: Optional[str] = Field("", description="The stream name.", example=["Users"], title="Name")
+    name: Optional[str] = Field(
+        "", description="The stream name.", example=["Users"], title="Name"
+    )
     retriever: Union[SimpleRetriever, AsyncRetriever, CustomRetriever] = Field(
         ...,
         description="Component used to coordinate how records are extracted across stream slices and request pages.",
@@ -2164,7 +2174,7 @@ class Config:
         ]
     ] = Field(
         None,
-        description="Component used to retrieve the schema for the current stream.",
+        description="One or many schema loaders can be used to retrieve the schema for the current stream. When multiple schema loaders are defined, schema properties will be merged together. Schema loaders defined first taking precedence in the event of a conflict.",
         title="Schema Loader",
     )
     transformations: Optional[
@@ -2311,18 +2321,20 @@ class HttpRequester(BaseModelWithDeprecations):
         description="Allows for retrieving a dynamic set of properties from an API endpoint which can be injected into outbound request using the stream_partition.extra_fields.",
         title="Fetch Properties from Endpoint",
     )
-    request_parameters: Optional[Union[Dict[str, Union[str, QueryProperties]], str]] = Field(
-        None,
-        description="Specifies the query parameters that should be set on an outgoing HTTP request given the inputs.",
-        examples=[
-            {"unit": "day"},
-            {
-                "query": 'last_event_time BETWEEN TIMESTAMP "{{ stream_interval.start_time }}" AND TIMESTAMP "{{ stream_interval.end_time }}"'
-            },
-            {"searchIn": "{{ ','.join(config.get('search_in', [])) }}"},
-            {"sort_by[asc]": "updated_at"},
-        ],
-        title="Query Parameters",
+    request_parameters: Optional[Union[Dict[str, Union[str, QueryProperties]], str]] = (
+        Field(
+            None,
+            description="Specifies the query parameters that should be set on an outgoing HTTP request given the inputs.",
+            examples=[
+                {"unit": "day"},
+                {
+                    "query": 'last_event_time BETWEEN TIMESTAMP "{{ stream_interval.start_time }}" AND TIMESTAMP "{{ stream_interval.end_time }}"'
+                },
+                {"searchIn": "{{ ','.join(config.get('search_in', [])) }}"},
+                {"sort_by[asc]": "updated_at"},
+            ],
+            title="Query Parameters",
+        )
     )
     request_headers: Optional[Union[Dict[str, str], str]] = Field(
         None,
@@ -2414,6 +2426,9 @@ class DynamicSchemaLoader(BaseModel):
         description="Component used to coordinate how records are extracted across stream slices and request pages.",
         title="Retriever",
     )
+    schema_filter: Optional[Union[RecordFilter, CustomRecordFilter]] = Field(
+        None, description="placeholder", title="Schema Filter"
+    )
     schema_transformations: Optional[
         List[
             Union[
@@ -2512,7 +2527,9 @@ class QueryProperties(BaseModel):
 
 class StateDelegatingStream(BaseModel):
     type: Literal["StateDelegatingStream"]
-    name: str = Field(..., description="The stream name.", example=["Users"], title="Name")
+    name: str = Field(
+        ..., description="The stream name.", example=["Users"], title="Name"
+    )
     full_refresh_stream: DeclarativeStream = Field(
         ...,
         description="Component used to coordinate how records are extracted across stream slices and request pages when the state is empty or not provided.",
@@ -2601,7 +2618,9 @@ class AsyncRetriever(BaseModel):
     )
     download_extractor: Optional[
         Union[DpathExtractor, CustomRecordExtractor, ResponseToFileExtractor]
-    ] = Field(None, description="Responsible for fetching the records from provided urls.")
+    ] = Field(
+        None, description="Responsible for fetching the records from provided urls."
+    )
     creation_requester: Union[HttpRequester, CustomRequester] = Field(
         ...,
         description="Requester component that describes how to prepare HTTP requests to send to the source API to create the async server-side job.",
@@ -2741,10 +2760,12 @@ class DynamicDeclarativeStream(BaseModel):
     stream_template: DeclarativeStream = Field(
         ..., description="Reference to the stream template.", title="Stream Template"
     )
-    components_resolver: Union[HttpComponentsResolver, ConfigComponentsResolver] = Field(
-        ...,
-        description="Component resolve and populates stream templates with components values.",
-        title="Components Resolver",
+    components_resolver: Union[HttpComponentsResolver, ConfigComponentsResolver] = (
+        Field(
+            ...,
+            description="Component resolve and populates stream templates with components values.",
+            title="Components Resolver",
+        )
     )
 
 

airbyte_cdk/sources/declarative/parsers/model_to_component_factory.py

@@ -2432,10 +2432,14 @@ def create_dynamic_schema_loader(
         schema_type_identifier = self._create_component_from_model(
             model.schema_type_identifier, config=config, parameters=model.parameters or {}
         )
+        schema_filter = self._create_component_from_model(
+            model.schema_filter, config=config, parameters=model.parameters or {}
+        )
         return DynamicSchemaLoader(
             retriever=retriever,
             config=config,
             schema_transformations=schema_transformations,
+            schema_filter=schema_filter,
             schema_type_identifier=schema_type_identifier,
             parameters=model.parameters or {},
         )

airbyte_cdk/sources/declarative/resolvers/components_resolver.py

@@ -22,6 +22,7 @@ class ComponentMappingDefinition:
     value: Union["InterpolatedString", str]
     value_type: Optional[Type[Any]]
     parameters: InitVar[Mapping[str, Any]]
+    create_or_update: Optional[bool] = False
 
 
 @dataclass(frozen=True)
@@ -34,6 +35,7 @@ class ResolvedComponentMappingDefinition:
     value: "InterpolatedString"
     value_type: Optional[Type[Any]]
     parameters: InitVar[Mapping[str, Any]]
+    create_or_update: Optional[bool] = False
 
 
 @deprecated("This class is experimental. Use at your own risk.", category=ExperimentalClassWarning)

airbyte_cdk/sources/declarative/schema/dynamic_schema_loader.py

@@ -14,6 +14,7 @@
 from airbyte_cdk.sources.declarative.interpolation.interpolated_string import InterpolatedString
 from airbyte_cdk.sources.declarative.retrievers.retriever import Retriever
 from airbyte_cdk.sources.declarative.schema.schema_loader import SchemaLoader
+from airbyte_cdk.sources.declarative.extractors.record_filter import RecordFilter
 from airbyte_cdk.sources.declarative.transformations import RecordTransformation
 from airbyte_cdk.sources.source import ExperimentalClassWarning
 from airbyte_cdk.sources.types import Config, StreamSlice, StreamState
@@ -126,6 +127,7 @@ class DynamicSchemaLoader(SchemaLoader):
     parameters: InitVar[Mapping[str, Any]]
     schema_type_identifier: SchemaTypeIdentifier
     schema_transformations: List[RecordTransformation] = field(default_factory=lambda: [])
+    schema_filter: Optional[RecordFilter] = None
 
     def get_json_schema(self) -> Mapping[str, Any]:
         """
@@ -151,20 +153,18 @@ def get_json_schema(self) -> Mapping[str, Any]:
             )
             properties[key] = value
 
-        transformed_properties = self._transform(properties, {})
+        filtered_transformed_properties = self._transform(self._filter(properties))
 
         return {
             "$schema": "https://json-schema.org/draft-07/schema#",
             "type": "object",
             "additionalProperties": True,
-            "properties": transformed_properties,
+            "properties": filtered_transformed_properties,
         }
 
     def _transform(
         self,
         properties: Mapping[str, Any],
-        stream_state: StreamState,
-        stream_slice: Optional[StreamSlice] = None,
     ) -> Mapping[str, Any]:
         for transformation in self.schema_transformations:
             transformation.transform(
@@ -173,6 +173,21 @@ def _transform(
             )
         return properties
 
+    def _filter(
+            self,
+            properties: Mapping[str, Any],
+    ) -> Mapping[str, Any]:
+        if not self.schema_filter:
+            return properties
+
+        filtered_properties = {}
+        for item in self.schema_filter.filter_records(
+                ({k: v} for k, v in properties.items()),
+                {},
+        ):
+            filtered_properties.update(item)
+        return filtered_properties
+
     def _get_key(
         self,
         raw_schema: MutableMapping[str, Any],