Add dynamic stream name to stream model

Author: lazebnyi

airbyte_cdk/sources/declarative/declarative_component_schema.yaml

@@ -1427,6 +1427,13 @@ definitions:
             - "$ref": "#/definitions/LegacyToPerPartitionStateMigration"
             - "$ref": "#/definitions/CustomStateMigration"
         default: []
+      dynamic_stream_name:
+        title: Dynamic Stream Name
+        description: The dynamic stream name that create current stream, if None is static stream.
+        type: string
+        default: None
+        example:
+          - "Tables"
       $parameters:
         type: object
         additional_properties: true
@@ -3766,6 +3773,13 @@ definitions:
       type:
         type: string
         enum: [DynamicDeclarativeStream]
+      name:
+        title: Name
+        description: The dynamic stream name.
+        type: string
+        default: ""
+        example:
+          - "Tables"
       stream_template:
         title: Stream Template
         description: Reference to the stream template.

airbyte_cdk/sources/declarative/manifest_declarative_source.py

@@ -343,6 +343,7 @@ def _stream_configs(self, manifest: Mapping[str, Any]) -> List[Dict[str, Any]]:
         # This has a warning flag for static, but after we finish part 4 we'll replace manifest with self._source_config
         stream_configs: List[Dict[str, Any]] = manifest.get("streams", [])
         for s in stream_configs:
+            s["dynamic_stream_name"] = None
             if "type" not in s:
                 s["type"] = "DeclarativeStream"
         return stream_configs
@@ -354,7 +355,7 @@ def _dynamic_stream_configs(
         dynamic_stream_configs: List[Dict[str, Any]] = []
         seen_dynamic_streams: Set[str] = set()
 
-        for dynamic_definition in dynamic_stream_definitions:
+        for dynamic_definition_index, dynamic_definition in enumerate(dynamic_stream_definitions):
             components_resolver_config = dynamic_definition["components_resolver"]
 
             if not components_resolver_config:
@@ -393,6 +394,8 @@ def _dynamic_stream_configs(
                 # Ensure that each stream is created with a unique name
                 name = dynamic_stream.get("name")
 
+                dynamic_stream["dynamic_stream_name"] = dynamic_definition.get("name", f"dynamic_stream_{dynamic_definition_index}")
+
                 if not isinstance(name, str):
                     raise ValueError(
                         f"Expected stream name {name} to be a string, got {type(name)}."

airbyte_cdk/sources/declarative/models/declarative_component_schema.py

@@ -609,7 +609,9 @@ 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(
@@ -1083,24 +1085,28 @@ 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,
@@ -1118,7 +1124,9 @@ 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(
@@ -1781,7 +1789,9 @@ 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(
         SchemaNormalization.None_,
         description="Responsible for normalization according to the schema.",
         title="Schema Normalization",
@@ -2008,7 +2018,9 @@ class Config:
         description="Component used to fetch data incrementally based on a time field in the data.",
         title="Incremental Sync",
     )
-    name: Optional[str] = Field("", description="The stream name.", example=["Users"], title="Name")
+    name: Optional[str] = Field(
+        "", description="The stream name.", example=["Users"], title="Name"
+    )
     primary_key: Optional[PrimaryKey] = Field(
         "", description="The primary key of the stream.", title="Primary Key"
     )
@@ -2049,6 +2061,12 @@ class Config:
         description="Array of state migrations to be applied on the input state",
         title="State Migrations",
     )
+    dynamic_stream_name: Optional[str] = Field(
+        "None",
+        description="The dynamic stream name that create current stream, if None is static stream.",
+        example=["Tables"],
+        title="Dynamic Stream Name",
+    )
     parameters: Optional[Dict[str, Any]] = Field(None, alias="$parameters")
 
 
@@ -2266,7 +2284,9 @@ class ParentStreamConfig(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.",
@@ -2355,7 +2375,9 @@ class AsyncRetriever(BaseModel):
     )
     download_extractor: Optional[
         Union[CustomRecordExtractor, DpathExtractor, 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[CustomRequester, HttpRequester] = Field(
         ...,
         description="Requester component that describes how to prepare HTTP requests to send to the source API to create the async server-side job.",
@@ -2489,13 +2511,18 @@ class HttpComponentsResolver(BaseModel):
 
 class DynamicDeclarativeStream(BaseModel):
     type: Literal["DynamicDeclarativeStream"]
+    name: Optional[str] = Field(
+        "", description="The dynamic stream name.", example=["Tables"], title="Name"
+    )
     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",
+        )
     )
 
 

unit_tests/sources/declarative/resolvers/test_http_components_resolver.py

@@ -57,6 +57,7 @@ def to_configured_catalog(
     "dynamic_streams": [
         {
             "type": "DynamicDeclarativeStream",
+            "name": "TestDynamicStream",
             "stream_template": {
                 "type": "DeclarativeStream",
                 "name": "",
@@ -487,6 +488,9 @@ def test_dynamic_streams_read_with_http_components_resolver():
         ]
         configured_catalog = to_configured_catalog(configured_streams)
 
+        for stream in source._dynamic_stream_configs(source.resolved_manifest, _CONFIG):
+            assert stream["dynamic_stream_name"] == "TestDynamicStream"
+
         records = [
             message.record
             for message in source.read(MagicMock(), _CONFIG, configured_catalog)
@@ -572,6 +576,9 @@ def test_dynamic_streams_with_http_components_resolver_retriever_with_parent_str
         ]
         configured_catalog = to_configured_catalog(configured_streams)
 
+        for stream in source._dynamic_stream_configs(source.resolved_manifest, _CONFIG):
+            assert stream["dynamic_stream_name"] == "dynamic_stream_0"
+
         records = [
             message.record
             for message in source.read(MagicMock(), _CONFIG, configured_catalog)

unit_tests/sources/declarative/test_concurrent_declarative_source.py

@@ -709,6 +709,9 @@ def test_group_streams():
     )
     concurrent_streams, synchronous_streams = source._group_streams(config=_CONFIG)
 
+    for stream in source._stream_configs(source.resolved_manifest):
+        assert stream["dynamic_stream_name"] is None
+
     # 1 full refresh stream, 3 incremental streams, 1 substream w/o incremental, 1 list based substream w/o incremental
     # 1 async job stream, 1 substream w/ incremental
     assert len(concurrent_streams) == 8