Module madsci.client.data_client

Client for the MADSci Experiment Manager.

Classes¶

DataClient(data_server_url: str | pydantic.networks.AnyUrl | None = None, object_storage_settings: madsci.common.types.datapoint_types.ObjectStorageSettings | None = None, config: madsci.common.types.client_types.DataClientConfig | None = None)

Client for the MADSci Experiment Manager.

Create a new Datapoint Client.

Args: data_server_url: The base URL of the Data Manager. If not provided, it will be taken from the current MadsciContext. object_storage_settings: Configuration for object storage (e.g., MinIO). If not provided, defaults will be used. config: Client configuration for retry and timeout settings. If not provided, uses default DataClientConfig.

Class variables¶

data_server_url: pydantic.networks.AnyUrl | None :

Methods¶

extract_datapoint_ids_from_action_result(self, action_result: Any) ‑> list[str]

Extract all datapoint IDs from an ActionResult.

Args: action_result: ActionResult object to extract IDs from

Returns: List of unique datapoint ULID strings

get_datapoint(self, datapoint_id: str | ulid.ULID, timeout: float | None = None) ‑> madsci.common.types.datapoint_types.DataPoint

Get a datapoint’s metadata by ID, either from local storage or server.

Args: datapoint_id: The ID of the datapoint to get. timeout: Optional timeout override in seconds. If None, uses config.timeout_default.

get_datapoint_metadata(self, datapoint_id: str) ‑> dict[str, typing.Any]

Get basic metadata for a datapoint without fetching the full data.

Useful for UI display where you need labels, types, timestamps, etc. without loading large file contents or values.

Args: datapoint_id: ULID string of the datapoint

Returns: Dictionary with metadata fields like label, data_type, data_timestamp

get_datapoint_value(self, datapoint_id: str | ulid.ULID, timeout: float | None = None) ‑> Any

Get a datapoint value by ID. If the datapoint is JSON, returns the JSON data. Otherwise, returns the raw data as bytes.

Args: datapoint_id: The ID of the datapoint to get. timeout: Optional timeout override in seconds. If None, uses config.timeout_data_operations.

get_datapoints(self, number: int = 10, timeout: float | None = None) ‑> list[madsci.common.types.datapoint_types.DataPoint]

Get a list of the latest datapoints.

Args: number: Number of datapoints to retrieve. timeout: Optional timeout override in seconds. If None, uses config.timeout_default.

get_datapoints_by_ids(self, datapoint_ids: list[str]) ‑> dict[str, madsci.common.types.datapoint_types.DataPoint]

Fetch multiple datapoints by their IDs in a batch operation.

This method enables just-in-time fetching of datapoints when only IDs are stored in workflows, following the principle of efficient datapoint management.

Args: datapoint_ids: List of datapoint ULID strings to fetch

Returns: Dictionary mapping datapoint IDs to DataPoint objects

Raises: Exception: If any datapoint cannot be fetched

get_datapoints_metadata(self, datapoint_ids: list[str]) ‑> dict[str, dict[str, typing.Any]]

Get metadata for multiple datapoints efficiently.

Args: datapoint_ids: List of datapoint ULID strings

Returns: Dictionary mapping datapoint IDs to metadata dictionaries

query_datapoints(self, selector: Any, timeout: float | None = None) ‑> dict[str, madsci.common.types.datapoint_types.DataPoint]

Query datapoints based on a selector.

Args: selector: Query selector for filtering datapoints. timeout: Optional timeout override in seconds. If None, uses config.timeout_default.

save_datapoint_value(self, datapoint_id: str | ulid.ULID, output_filepath: str, timeout: float | None = None) ‑> None

Get an datapoint value by ID.

Args: datapoint_id: The ID of the datapoint to save. output_filepath: Path where the datapoint value should be saved. timeout: Optional timeout override in seconds. If None, uses config.timeout_data_operations.

submit_datapoint(self, datapoint: madsci.common.types.datapoint_types.DataPoint, timeout: float | None = None) ‑> madsci.common.types.datapoint_types.DataPoint

Submit a Datapoint object.

If object storage is configured and the datapoint is a file type, the file will be automatically uploaded to object storage instead of being sent to the Data Manager server.

Args: datapoint: The datapoint to submit. timeout: Optional timeout override in seconds. If None, uses config.timeout_data_operations.

Returns: The submitted datapoint with server-assigned IDs if applicable