Skip to main content

Crate dataforge

Crate dataforge 

Source
Expand description

Fast dataset discovery and archive helpers for thesa and other Rust tools.

Made by Trevor Knott for Knott Dynamics.

DataForge normalizes dataset targets, queries public dataset catalogs, and archives dataset files with bounded concurrent downloads. The first-class provider is Hugging Face Datasets; Zenodo is included for DOI/record-backed research datasets.

assert_eq!(
    dataforge::parse_dataset_target_for_provider(
        "https://huggingface.co/datasets/squad",
        dataforge::DatasetProvider::Hf,
    )
    .unwrap(),
    dataforge::DatasetTarget::Dataset("squad".to_string())
);
assert_eq!(
    dataforge::DatasetProvider::from_slug("hugging-face"),
    Some(dataforge::DatasetProvider::Hf)
);

Structs§

ArchiveFailure
Failed dataset or file archive entry.
ArchiveOptions
Options controlling archive writes.
ArchiveSummary
Summary returned after an archive run.
ArchivedFile
One archived or skipped file.
DataForge
Blocking dataset discovery and archive client.
DatasetDetails
Full dataset metadata plus files to archive.
DatasetFile
One downloadable file in a dataset archive.
DatasetSummary
Lightweight dataset metadata returned by discovery calls.

Enums§

DataForgeError
Errors returned by DataForge.
DatasetProvider
Supported dataset providers.
DatasetTarget
Normalized dataset target requested by a user or CLI.

Functions§

parse_dataset_target
Parse a Hugging Face-style dataset target.
parse_dataset_target_for_provider
Parse a dataset target using provider-specific rules.

Type Aliases§

Result
Result type used by DataForge APIs.