airbyte-cdk 6.56.7.post1.dev15934698385__tar.gz → 6.57.0__tar.gz

{airbyte_cdk-6.56.7.post1.dev15934698385 → airbyte_cdk-6.57.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: airbyte-cdk
-Version: 6.56.7.post1.dev15934698385
+Version: 6.57.0
 Summary: A framework for writing Airbyte Connectors.
 Home-page: https://airbyte.com
 License: MIT

{airbyte_cdk-6.56.7.post1.dev15934698385 → airbyte_cdk-6.57.0}/airbyte_cdk/__init__.py

@@ -32,6 +32,16 @@ The Airbyte CDK provides a set of classes that help you work with the Airbyte pr
 - `airbyte_cdk.models.airbyte_protocol`
 - `airbyte_cdk.models.airbyte_protocol_serializers`
 
+## Using the CLI (`airbyte_cdk.cli`)
+
+The Airbyte CDK provides two command-line interfaces (CLIs) for interacting with the framework.
+
+- `airbyte-cdk`: This is the main CLI for the Airbyte CDK. It provides commands for building
+  and testing connectors, as well as other utilities. See the `airbyte_cdk.cli.airbyte_cdk` module
+  for more details.
+- `source-declarative-manifest`: This command allows you to run declarative manifests directly.
+  See the `airbyte_cdk.cli.source_declarative_manifest` module for more details.
+
 ---
 
 API Reference
@@ -129,7 +139,6 @@ from .sources.declarative.schema import JsonFileSchemaLoader
 from .sources.declarative.transformations.add_fields import AddedFieldDefinition, AddFields
 from .sources.declarative.transformations.transformation import RecordTransformation
 from .sources.declarative.types import FieldPointer
-from .sources.declarative.validators import DpathValidator, PredicateValidator, ValidateAdheresToSchema, ValidationStrategy, Validator
 from .sources.declarative.yaml_declarative_source import YamlDeclarativeSource
 from .sources.message import InMemoryMessageRepository, MessageRepository
 from .sources.source import TState
@@ -340,12 +349,6 @@ __all__ = [
     "Record",
     "Source",
     "StreamSlice",
-    # Validators
-    "Validator",
-    "DpathValidator",
-    "PredicateValidator",
-    "ValidationStrategy",
-    "ValidateAdheresToSchema",
 ]
 
 __version__: str

airbyte_cdk-6.57.0/airbyte_cdk/cli/source_declarative_manifest/README.md

@@ -0,0 +1,65 @@
+# Source Declarative Manifest CLI
+
+The source-declarative-manifest CLI is included in the airbyte-cdk package.
+This CLI enables connector interfaces to be run locally on manifest-only connectors,
+much like we already do with Python connectors.
+
+## Installation
+
+The airbyte-cdk library can be installed globally using pipx:
+
+```bash
+pipx install airbyte-cdk
+```
+
+If you are using a cloned airbyte-python-cdk repo locally,
+you can also create a virtual environment to enable the CLI.
+From the root directory of airbyte-python-cdk:
+
+```bash
+python -m venv .venv
+source .venv/bin/activate
+pip install -e .
+```
+
+## Usage
+
+### Options
+
+--help: displays the list of available commands
+
+### Commands
+
+- spec: Outputs the JSON configuration specification. NOTE: This currently just outputs the base source-declarative-manifest spec
+- check: Runs a connection_check to verify a connection can be made with the passed config
+- discover: Outputs a catalog describing the source's schema
+- read: Reads the source using the passed config and catalog, and outputs messages to STDOUT
+
+### Command options
+
+- --config: The relative path to the config to inject into SDM.
+- --catalog: The relative path to the configured catalog.
+- --state: The relative path to the state object to pass. Only used when running an incremental read.
+- --manifest-path: The relative path to the local YAML manifest to inject into SDM.
+- --components-path: The relative path to the custom components to mount, if they exist.
+
+| Option              | spec | check    | discover | read     |
+| ------------------- | ---- | -------- | -------- | -------- |
+| `--config`          | ❌   | required | required | required |
+| `--catalog`         | ❌   | ❌       | required | required |
+| `--state`           | ❌   | ❌       | ❌       | optional |
+| `--manifest-path`   | ❌   | required | required | required |
+| `--components-path` | ❌   | optional | optional | optional |
+
+### Examples
+
+Here are some basic examples of how to run source-declarative-manifest commands locally.
+Note that the paths are relative. These examples assume the user is currently at the root level of a connector dir:
+
+```bash
+source-declarative-manifest check --config secrets/config.json --manifest-path manifest.yaml
+```
+
+```bash
+source-declarative-manifest read --config secrets/config.json --catalog integration_tests/configured_catalog.json --manifest-path manifest.yaml --components-path components.py
+```

{airbyte_cdk-6.56.7.post1.dev15934698385 → airbyte_cdk-6.57.0}/airbyte_cdk/cli/source_declarative_manifest/__init__.py

@@ -1,3 +1,8 @@
+"""
+.. include:: ./README.md
+   :start-line: 2
+"""
+
 from airbyte_cdk.cli.source_declarative_manifest._run import run
 
 __all__ = [

{airbyte_cdk-6.56.7.post1.dev15934698385 → airbyte_cdk-6.57.0}/airbyte_cdk/cli/source_declarative_manifest/_run.py

@@ -299,5 +299,9 @@ def _register_components_from_file(filepath: str) -> None:
 
 
 def run() -> None:
+    """Run the `source-declarative-manifest` CLI.
+
+    Args are detected from the command line, and the appropriate command is executed.
+    """
     args: list[str] = sys.argv[1:]
     handle_command(args)

{airbyte_cdk-6.56.7.post1.dev15934698385 → airbyte_cdk-6.57.0}/airbyte_cdk/manifest_migrations/migrations/http_requester_request_body_json_data_to_request_body.py

@@ -54,7 +54,7 @@ class HttpRequesterRequestBodyJsonDataToRequestBody(ManifestMigration):
         if isinstance(manifest[key], str):
             self._migrate_value(manifest, key, text_type)
         elif isinstance(manifest[key], dict):
-            if manifest[key].get(query_key) is not None:
+            if isinstance(manifest[key].get(query_key), str):
                 self._migrate_value(manifest, key, graph_ql_type)
             else:
                 self._migrate_value(manifest, key, json_object_type)

{airbyte_cdk-6.56.7.post1.dev15934698385 → airbyte_cdk-6.57.0}/airbyte_cdk/sources/declarative/declarative_component_schema.yaml

@@ -2140,30 +2140,6 @@ definitions:
           - stream_interval
           - stream_partition
           - stream_slice
-        examples:
-          - type: RequestBodyJsonObject
-            value:
-              sort_order: "ASC"
-              sort_field: "CREATED_AT"
-          - type: RequestBodyJsonObject
-            value:
-              key: "{{ config['value'] }}"
-          - type: RequestBodyJsonObject
-            value:
-              sort:
-                field: "updated_at"
-                order: "ascending"
-          - type: RequestBodyPlainText
-            value: "plain_text_body"
-          - type: RequestBodyUrlEncodedForm
-            value:
-              param1: "value1"
-              param2: "{{ config['param2_value'] }}"
-          - type: RequestBodyGraphQL
-            value:
-              query:
-                param1: "value1"
-                param2: "{{ config['param2_value'] }}"
       error_handler:
         title: Error Handler
         description: Error handler component that defines how to handle errors.
@@ -3145,15 +3121,12 @@ definitions:
       type:
         type: string
         enum: [ParentStreamConfig]
-      lazy_read_pointer:
-        title: Lazy Read Pointer
-        description: If set, this will enable lazy reading, using the initial read of parent records to extract child records.
-        type: array
-        default: []
-        items:
-          type: string
-        interpolation_context:
-          - config
+      stream:
+        title: Parent Stream
+        description: Reference to the parent stream.
+        anyOf:
+          - "$ref": "#/definitions/DeclarativeStream"
+          - "$ref": "#/definitions/StateDelegatingStream"
       parent_key:
         title: Parent Key
         description: The primary key of records from the parent stream that will be used during the retrieval of records for the current substream. This parent identifier field is typically a characteristic of the child records being extracted from the source API.
@@ -3161,12 +3134,6 @@ definitions:
         examples:
           - "id"
           - "{{ config['parent_record_id'] }}"
-      stream:
-        title: Parent Stream
-        description: Reference to the parent stream.
-        anyOf:
-          - "$ref": "#/definitions/DeclarativeStream"
-          - "$ref": "#/definitions/StateDelegatingStream"
       partition_field:
         title: Current Parent Key Value Identifier
         description: While iterating over parent records during a sync, the parent_key value can be referenced by using this field.
@@ -3183,6 +3150,15 @@ definitions:
         description: Indicates whether the parent stream should be read incrementally based on updates in the child stream.
         type: boolean
         default: false
+      lazy_read_pointer:
+        title: Lazy Read Pointer
+        description: If set, this will enable lazy reading, using the initial read of parent records to extract child records.
+        type: array
+        default: []
+        items:
+          type: string
+        interpolation_context:
+          - config
       extra_fields:
         title: Extra Fields
         description: Array of field paths to include as additional fields in the stream slice. Each path is an array of strings representing keys to access fields in the respective parent record. Accessible via `stream_slice.extra_fields`. Missing fields are set to `None`.
@@ -3616,18 +3592,18 @@ definitions:
         default: false
       partition_router:
         title: Partition Router
-        description: PartitionRouter component that describes how to partition the stream, enabling incremental syncs and checkpointing.
-        default: []
+        description: Used to iteratively execute requests over a set of values, such as a parent stream's records or a list of constant values.
         anyOf:
-          - "$ref": "#/definitions/ListPartitionRouter"
           - "$ref": "#/definitions/SubstreamPartitionRouter"
+          - "$ref": "#/definitions/ListPartitionRouter"
           - "$ref": "#/definitions/GroupingPartitionRouter"
           - "$ref": "#/definitions/CustomPartitionRouter"
           - type: array
+            title: Multiple Partition Routers
             items:
               anyOf:
-                - "$ref": "#/definitions/ListPartitionRouter"
                 - "$ref": "#/definitions/SubstreamPartitionRouter"
+                - "$ref": "#/definitions/ListPartitionRouter"
                 - "$ref": "#/definitions/GroupingPartitionRouter"
                 - "$ref": "#/definitions/CustomPartitionRouter"
       $parameters:
@@ -4259,6 +4235,11 @@ definitions:
           - "$ref": "#/definitions/HttpComponentsResolver"
           - "$ref": "#/definitions/ConfigComponentsResolver"
           - "$ref": "#/definitions/ParametrizedComponentsResolver"
+      use_parent_parameters:
+        title: Use Parent Parameters
+        description: Whether or not to prioritize parent parameters over component parameters when constructing dynamic streams. Defaults to true for backward compatibility.
+        type: boolean
+        default: true
     required:
       - type
       - stream_template
@@ -4326,10 +4307,9 @@ definitions:
       - query
     properties:
       query:
-        type: object
-        additionalProperties: true
+        type: string
         description: The GraphQL query to be executed
-        default: {}
+        default: "query {\n    \n}"
     additionalProperties: true
   DpathValidator:
     title: Dpath Validator

{airbyte_cdk-6.56.7.post1.dev15934698385 → airbyte_cdk-6.57.0}/airbyte_cdk/sources/declarative/interpolation/macros.py

@@ -4,6 +4,7 @@
 
 import builtins
 import datetime
+import re
 import typing
 from typing import Optional, Union
 from urllib.parse import quote_plus
@@ -194,6 +195,18 @@ def sanitize_url(value: str) -> str:
     return sanitization_strategy(value)
 
 
+def camel_case_to_snake_case(value: str) -> str:
+    """
+     Converts CamelCase strings to snake_case format
+
+     Usage:
+    `"{{ camel_case_to_snake_case('CamelCase') }}"`
+     :param value: string to convert from CamelCase to snake_case
+     :return: snake_case formatted string
+    """
+    return re.sub(r"(?<!^)(?=[A-Z])", "_", value).lower()
+
+
 _macros_list = [
     now_utc,
     today_utc,
@@ -206,5 +219,6 @@ _macros_list = [
     today_with_timezone,
     str_to_datetime,
     sanitize_url,
+    camel_case_to_snake_case,
 ]
 macros = {f.__name__: f for f in _macros_list}

{airbyte_cdk-6.56.7.post1.dev15934698385 → airbyte_cdk-6.57.0}/airbyte_cdk/sources/declarative/manifest_declarative_source.py

@@ -553,9 +553,13 @@ class ManifestDeclarativeSource(DeclarativeSource):
             for dynamic_stream in components_resolver.resolve_components(
                 stream_template_config=stream_template_config
             ):
+                # Get the use_parent_parameters configuration from the dynamic definition
+                # Default to True for backward compatibility, since connectors were already using it by default when this param was added
+                use_parent_parameters = dynamic_definition.get("use_parent_parameters", True)
+
                 dynamic_stream = {
                     **ManifestComponentTransformer().propagate_types_and_parameters(
-                        "", dynamic_stream, {}, use_parent_parameters=True
+                        "", dynamic_stream, {}, use_parent_parameters=use_parent_parameters
                     )
                 }
 

{airbyte_cdk-6.56.7.post1.dev15934698385 → airbyte_cdk-6.57.0}/airbyte_cdk/sources/declarative/models/declarative_component_schema.py

@@ -1,3 +1,5 @@
+# Copyright (c) 2025 Airbyte, Inc., all rights reserved.
+
 # generated by datamodel-codegen:
 #   filename:  declarative_component_schema.yaml
 
@@ -1554,7 +1556,7 @@ class RequestBodyGraphQlQuery(BaseModel):
     class Config:
         extra = Extra.allow
 
-    query: Dict[str, Any] = Field(..., description="The GraphQL query to be executed")
+    query: str = Field(..., description="The GraphQL query to be executed")
 
 
 class ValidateAdheresToSchema(BaseModel):
@@ -2595,34 +2597,6 @@ class HttpRequester(BaseModelWithDeprecations):
     ] = Field(
         None,
         description="Specifies how to populate the body of the request with a payload. Can contain nested objects.",
-        examples=[
-            {
-                "type": "RequestBodyJsonObject",
-                "value": {"sort_order": "ASC", "sort_field": "CREATED_AT"},
-            },
-            {
-                "type": "RequestBodyJsonObject",
-                "value": {"key": "{{ config['value'] }}"},
-            },
-            {
-                "type": "RequestBodyJsonObject",
-                "value": {"sort": {"field": "updated_at", "order": "ascending"}},
-            },
-            {"type": "RequestBodyPlainText", "value": "plain_text_body"},
-            {
-                "type": "RequestBodyUrlEncodedForm",
-                "value": {"param1": "value1", "param2": "{{ config['param2_value'] }}"},
-            },
-            {
-                "type": "RequestBodyGraphQL",
-                "value": {
-                    "query": {
-                        "param1": "value1",
-                        "param2": "{{ config['param2_value'] }}",
-                    }
-                },
-            },
-        ],
         title="Request Body",
     )
     error_handler: Optional[
@@ -2676,10 +2650,8 @@ class DynamicSchemaLoader(BaseModel):
 
 class ParentStreamConfig(BaseModel):
     type: Literal["ParentStreamConfig"]
-    lazy_read_pointer: Optional[List[str]] = Field(
-        [],
-        description="If set, this will enable lazy reading, using the initial read of parent records to extract child records.",
-        title="Lazy Read Pointer",
+    stream: Union[DeclarativeStream, StateDelegatingStream] = Field(
+        ..., description="Reference to the parent stream.", title="Parent Stream"
     )
     parent_key: str = Field(
         ...,
@@ -2687,9 +2659,6 @@ class ParentStreamConfig(BaseModel):
         examples=["id", "{{ config['parent_record_id'] }}"],
         title="Parent Key",
     )
-    stream: Union[DeclarativeStream, StateDelegatingStream] = Field(
-        ..., description="Reference to the parent stream.", title="Parent Stream"
-    )
     partition_field: str = Field(
         ...,
         description="While iterating over parent records during a sync, the parent_key value can be referenced by using this field.",
@@ -2706,6 +2675,11 @@ class ParentStreamConfig(BaseModel):
         description="Indicates whether the parent stream should be read incrementally based on updates in the child stream.",
         title="Incremental Dependency",
     )
+    lazy_read_pointer: Optional[List[str]] = Field(
+        [],
+        description="If set, this will enable lazy reading, using the initial read of parent records to extract child records.",
+        title="Lazy Read Pointer",
+    )
     extra_fields: Optional[List[List[str]]] = Field(
         None,
         description="Array of field paths to include as additional fields in the stream slice. Each path is an array of strings representing keys to access fields in the respective parent record. Accessible via `stream_slice.extra_fields`. Missing fields are set to `None`.",
@@ -2800,22 +2774,22 @@ class SimpleRetriever(BaseModel):
     )
     partition_router: Optional[
         Union[
-            ListPartitionRouter,
             SubstreamPartitionRouter,
+            ListPartitionRouter,
             GroupingPartitionRouter,
             CustomPartitionRouter,
             List[
                 Union[
-                    ListPartitionRouter,
                     SubstreamPartitionRouter,
+                    ListPartitionRouter,
                     GroupingPartitionRouter,
                     CustomPartitionRouter,
                 ]
             ],
         ]
     ] = Field(
-        [],
-        description="PartitionRouter component that describes how to partition the stream, enabling incremental syncs and checkpointing.",
+        None,
+        description="Used to iteratively execute requests over a set of values, such as a parent stream's records or a list of constant values.",
         title="Partition Router",
     )
     parameters: Optional[Dict[str, Any]] = Field(None, alias="$parameters")
@@ -2986,6 +2960,11 @@ class DynamicDeclarativeStream(BaseModel):
         description="Component resolve and populates stream templates with components values.",
         title="Components Resolver",
     )
+    use_parent_parameters: Optional[bool] = Field(
+        True,
+        description="Whether or not to prioritize parent parameters over component parameters when constructing dynamic streams. Defaults to true for backward compatibility.",
+        title="Use Parent Parameters",
+    )
 
 
 ComplexFieldType.update_forward_refs()

{airbyte_cdk-6.56.7.post1.dev15934698385 → airbyte_cdk-6.57.0}/airbyte_cdk/sources/declarative/requesters/request_options/interpolated_request_options_provider.py

@@ -100,7 +100,7 @@ class InterpolatedRequestOptionsProvider(RequestOptionsProvider):
             if self.request_body.type == "RequestBodyUrlEncodedForm":
                 self.request_body_data = self.request_body.value
             elif self.request_body.type == "RequestBodyGraphQL":
-                self.request_body_json = {"query": self.request_body.value.query}
+                self.request_body_json = self.request_body.value.dict(exclude_none=True)
             elif self.request_body.type in ("RequestBodyJsonObject", "RequestBodyPlainText"):
                 self.request_body_json = self.request_body.value
             else:

{airbyte_cdk-6.56.7.post1.dev15934698385 → airbyte_cdk-6.57.0}/airbyte_cdk/sources/declarative/retrievers/file_uploader/connector_builder_file_uploader.py

@@ -21,6 +21,4 @@ class ConnectorBuilderFileUploader(FileUploader):
 
     def upload(self, record: Record) -> None:
         self.file_uploader.upload(record=record)
-        for file_reference_key, file_reference_value in record.file_reference.__dict__.items():
-            if not file_reference_key.startswith("_"):
-                record.data[file_reference_key] = file_reference_value  # type: ignore
+        record.data["source_file_relative_path"] = record.file_reference.source_file_relative_path  # type: ignore

{airbyte_cdk-6.56.7.post1.dev15934698385 → airbyte_cdk-6.57.0}/airbyte_cdk/sources/declarative/validators/dpath_validator.py

@@ -53,7 +53,7 @@ class DpathValidator(Validator):
                 for value in values:
                     self.strategy.validate(value)
             except KeyError as e:
-                logger.warning(f"Error validating path. Key not found: {e}")
+                logger.info(f"Validation skipped. Key not found: {e}")
                 return
 
         else:
@@ -61,5 +61,5 @@ class DpathValidator(Validator):
                 value = dpath.get(input_data, path)
                 self.strategy.validate(value)
             except KeyError as e:
-                logger.warning(f"Error validating path. Key not found: {e}")
+                logger.info(f"Validation skipped. Key not found: {e}")
                 return

{airbyte_cdk-6.56.7.post1.dev15934698385 → airbyte_cdk-6.57.0}/pyproject.toml

@@ -23,7 +23,7 @@ classifiers = [
 keywords = ["airbyte", "connector-development-kit", "cdk"]
 
 # Python CDK uses dynamic versioning: https://github.com/mtkennerly/poetry-dynamic-versioning
-version = "6.56.7.post1.dev15934698385"  # Version will be calculated dynamically.
+version = "6.57.0"  # Version will be calculated dynamically.
 
 [tool.poetry-dynamic-versioning]
 enable = false