vertebrae.datasets.identity
Explicit identities for dataset cache and reproducibility keys.
Classes
An explicit, serializable policy for identifying a dataset. |
Module Contents
- class vertebrae.datasets.identity.DatasetIdentity[source]
An explicit, serializable policy for identifying a dataset.
Use one of the named constructors instead of instantiating this class directly. Content identities read all identity-bearing dataset values when first resolved. Datasets must be treated as immutable after their identity is resolved or derived.
- classmethod declared(dataset_id, revision)[source]
Identify a caller-managed dataset revision without inspecting its content.
- Parameters:
dataset_id (str)
revision (str)
- Return type:
- classmethod from_manifest(dataset_id, manifest)[source]
Identify a dataset from a complete caller-provided manifest.
- Parameters:
dataset_id (str)
manifest (Mapping[str, Any])
- Return type:
- classmethod from_content()[source]
Explicitly identify a dataset by lazily hashing all of its content.
- Return type:
- classmethod ephemeral()[source]
Create an explicit run-local identity that remains stable when serialized.
- Return type:
- classmethod derived(parent_key, operation, recipe)[source]
Create an identity for a deterministic transformation of another dataset.
- Parameters:
parent_key (str)
operation (str)
recipe (Any)
- Return type: