PoC for file upload

Author: maxi297

airbyte_cdk/sources/declarative/concurrent_declarative_source.py

@@ -25,6 +25,7 @@
     PerPartitionWithGlobalCursor,
 )
 from airbyte_cdk.sources.declarative.manifest_declarative_source import ManifestDeclarativeSource
+from airbyte_cdk.sources.declarative.models import FileUploader
 from airbyte_cdk.sources.declarative.models.declarative_component_schema import (
     ConcurrencyLevel as ConcurrencyLevelModel,
 )
@@ -206,6 +207,14 @@ def _group_streams(
             # 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
 
+            file_uploader = None
+            if isinstance(declarative_stream, DeclarativeStream):
+                file_uploader = self._constructor.create_component(
+                    model_type=FileUploader,
+                    component_definition=name_to_stream_mapping[declarative_stream.name]["file_uploader"],
+                    config=config,
+                ) if "file_uploader" in name_to_stream_mapping[declarative_stream.name] else None
+
             if (
                 isinstance(declarative_stream, DeclarativeStream)
                 and name_to_stream_mapping[declarative_stream.name]["type"]
@@ -273,6 +282,7 @@ def _group_streams(
                                 declarative_stream.get_json_schema(),
                                 retriever,
                                 self.message_repository,
+                                file_uploader,
                             ),
                             stream_slicer=declarative_stream.retriever.stream_slicer,
                         )
@@ -303,6 +313,7 @@ def _group_streams(
                                 declarative_stream.get_json_schema(),
                                 retriever,
                                 self.message_repository,
+                                file_uploader,
                             ),
                             stream_slicer=cursor,
                         )
@@ -333,6 +344,7 @@ def _group_streams(
                             declarative_stream.get_json_schema(),
                             declarative_stream.retriever,
                             self.message_repository,
+                            file_uploader,
                         ),
                         declarative_stream.retriever.stream_slicer,
                     )
@@ -392,6 +404,7 @@ def _group_streams(
                             declarative_stream.get_json_schema(),
                             retriever,
                             self.message_repository,
+                            file_uploader,
                         ),
                         perpartition_cursor,
                     )

airbyte_cdk/sources/declarative/declarative_component_schema.yaml

@@ -1422,6 +1422,33 @@ definitions:
             - "$ref": "#/definitions/LegacyToPerPartitionStateMigration"
             - "$ref": "#/definitions/CustomStateMigration"
         default: []
+      file_uploader:
+        title: File Uploader
+        description: (experimental) Describes how to fetch a file
+        type: object
+        required:
+          - type
+          - requester
+          - download_target_extractor
+        properties:
+          type:
+            type: string
+            enum: [ FileUploader ]
+          requester:
+            description: Requester component that describes how to prepare HTTP requests to send to the source API.
+            anyOf:
+              - "$ref": "#/definitions/CustomRequester"
+              - "$ref": "#/definitions/HttpRequester"
+          download_target_extractor:
+            description: Responsible for fetching the url where the file is located. This is applied on each records and not on the HTTP response
+            anyOf:
+              - "$ref": "#/definitions/CustomRecordExtractor"
+              - "$ref": "#/definitions/DpathExtractor"
+          file_extractor:
+            description: Responsible for fetching the content of the file. If not defined, the assumption is that the whole response body is the file content
+            anyOf:
+              - "$ref": "#/definitions/CustomRecordExtractor"
+              - "$ref": "#/definitions/DpathExtractor"
       $parameters:
         type: object
         additional_properties: true

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",
@@ -2006,7 +2016,9 @@ 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"
     )
@@ -2264,7 +2276,9 @@ 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.",
@@ -2278,6 +2292,22 @@ class StateDelegatingStream(BaseModel):
     parameters: Optional[Dict[str, Any]] = Field(None, alias="$parameters")
 
 
+class FileUploader(BaseModel):
+    type: Literal["FileUploader"]
+    requester: Union[CustomRequester, HttpRequester] = Field(
+        ...,
+        description="Requester component that describes how to prepare HTTP requests to send to the source API.",
+    )
+    download_target_extractor: Union[CustomRecordExtractor, DpathExtractor] = Field(
+        ...,
+        description="Responsible for fetching the url where the file is located. This is applied on each records and not on the HTTP response",
+    )
+    file_extractor: Optional[Union[CustomRecordExtractor, DpathExtractor]] = Field(
+        None,
+        description="Responsible for fetching the content of the file. If not defined, the assumption is that the whole response body is the file content",
+    )
+
+
 class SimpleRetriever(BaseModel):
     type: Literal["SimpleRetriever"]
     record_selector: RecordSelector = Field(
@@ -2301,13 +2331,22 @@ class SimpleRetriever(BaseModel):
             CustomPartitionRouter,
             ListPartitionRouter,
             SubstreamPartitionRouter,
-            List[Union[CustomPartitionRouter, ListPartitionRouter, SubstreamPartitionRouter]],
+            List[
+                Union[
+                    CustomPartitionRouter, ListPartitionRouter, SubstreamPartitionRouter
+                ]
+            ],
         ]
     ] = Field(
         [],
         description="PartitionRouter component that describes how to partition the stream, enabling incremental syncs and checkpointing.",
         title="Partition Router",
     )
+    file_uploader: Optional[FileUploader] = Field(
+        None,
+        description="(experimental) Describes how to fetch a file",
+        title="File Uploader",
+    )
     decoder: Optional[
         Union[
             CustomDecoder,
@@ -2345,7 +2384,9 @@ 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.",
@@ -2383,7 +2424,11 @@ class AsyncRetriever(BaseModel):
             CustomPartitionRouter,
             ListPartitionRouter,
             SubstreamPartitionRouter,
-            List[Union[CustomPartitionRouter, ListPartitionRouter, SubstreamPartitionRouter]],
+            List[
+                Union[
+                    CustomPartitionRouter, ListPartitionRouter, SubstreamPartitionRouter
+                ]
+            ],
         ]
     ] = Field(
         [],
@@ -2451,10 +2496,12 @@ 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",
+        )
     )