setup

openml._api.setup

APIBackend

Central backend for accessing all OpenML API resource interfaces.

This class provides a singleton interface to dataset, task, flow, evaluation, run, setup, study, and other resource APIs. It also manages configuration through a nested Config object and allows dynamic retrieval and updating of configuration values.

PARAMETER	DESCRIPTION
config	Optional configuration object. If not provided, a default Config instance is created. TYPE: Config

ATTRIBUTE	DESCRIPTION
dataset	Interface for dataset-related API operations. TYPE: DatasetAPI
task	Interface for task-related API operations. TYPE: TaskAPI
evaluation_measure	Interface for evaluation measure-related API operations. TYPE: EvaluationMeasureAPI
estimation_procedure	Interface for estimation procedure-related API operations. TYPE: EstimationProcedureAPI
evaluation	Interface for evaluation-related API operations. TYPE: EvaluationAPI
flow	Interface for flow-related API operations. TYPE: FlowAPI
study	Interface for study-related API operations. TYPE: StudyAPI
run	Interface for run-related API operations. TYPE: RunAPI
setup	Interface for setup-related API operations. TYPE: SetupAPI

get_instance classmethod

get_instance() -> APIBackend

Get the singleton instance of the APIBackend.

RETURNS	DESCRIPTION
APIBackend	Singleton instance of the backend.

Source code in openml/_api/setup/backend.py

@classmethod
def get_instance(cls) -> APIBackend:
    """
    Get the singleton instance of the APIBackend.

    Returns
    -------
    APIBackend
        Singleton instance of the backend.
    """
    if cls._instance is None:
        cls._instance = cls()
    return cls._instance

APIBackendBuilder

APIBackendBuilder(api_version: APIVersion, fallback_api_version: APIVersion | None = None)

Builder for constructing API backend instances with all resource-specific APIs.

This class organizes resource-specific API objects (datasets, tasks, flows, evaluations, runs, setups, studies, etc.) and provides a centralized access point for both the primary API version and an optional fallback API version.

The constructor automatically initializes:

• HTTPClient for the primary API version
• Optional HTTPClient for a fallback API version
• MinIOClient for file storage operations
• Resource-specific API instances, optionally wrapped with fallback proxies

PARAMETER	DESCRIPTION
api_version	The primary API version to use for all resource APIs and HTTP communication. TYPE: APIVersion
fallback_api_version	Optional fallback API version to wrap resource APIs with a FallbackProxy. TYPE: APIVersion | None DEFAULT: None

ATTRIBUTE	DESCRIPTION
dataset	API interface for dataset resources. TYPE: ResourceAPI | FallbackProxy
task	API interface for task resources. TYPE: ResourceAPI | FallbackProxy
evaluation_measure	API interface for evaluation measure resources. TYPE: ResourceAPI | FallbackProxy
estimation_procedure	API interface for estimation procedure resources. TYPE: ResourceAPI | FallbackProxy
evaluation	API interface for evaluation resources. TYPE: ResourceAPI | FallbackProxy
flow	API interface for flow resources. TYPE: ResourceAPI | FallbackProxy
study	API interface for study resources. TYPE: ResourceAPI | FallbackProxy
run	API interface for run resources. TYPE: ResourceAPI | FallbackProxy
setup	API interface for setup resources. TYPE: ResourceAPI | FallbackProxy
http_client	Client for HTTP communication using the primary API version. TYPE: HTTPClient
fallback_http_client	Client for HTTP communication using the fallback API version, if provided. TYPE: HTTPClient | None
minio_client	Client for file storage operations (MinIO/S3). TYPE: MinIOClient

Source code in openml/_api/setup/builder.py

def __init__(self, api_version: APIVersion, fallback_api_version: APIVersion | None = None):
    # initialize clients and resource APIs in-place
    self._build(api_version, fallback_api_version)