Command-line reference

Verified research dataset artifacts

Usage: datamonger [OPTIONS] <COMMAND>

Options

• --selector <SELECTOR>

  Read a registry selector from this file instead of discovering a project selector

• --cache-dir <CACHE_DIR>

  Use this directory for the CLI's private cache

• --offline

  Use only bundled or cached registry indexes and cached artifacts

• --json

  Print structured JSON results and report errors as JSON on stderr

datamonger list

List datasets in the selected registry

Usage: datamonger list [OPTIONS]

Options

• --source <SOURCE>

  Include only datasets from this source, such as uci or libsvm

datamonger info

Show a dataset's metadata, provenance, and registered artifacts

Usage: datamonger info <DATASET>

Arguments

• <DATASET>

  Dataset identifier in SOURCE:NAME or SOURCE:NAME@VERSION form

datamonger fetch

Download artifacts and verify their registered sizes and SHA-256 hashes.

Fetches every artifact unless --artifact selects one. Files stay in the private cache unless --output-dir also exports them. Registered compression is preserved; datasets are not decoded or checked for canonical logical values.

Usage: datamonger fetch [OPTIONS] <DATASET>

Arguments

• <DATASET>

  Dataset identifier in SOURCE:NAME or SOURCE:NAME@VERSION form

Options

• --artifact <ARTIFACT>

  Fetch only the named artifact, such as train or test

• --output-dir <OUTPUT_DIR>

  Also export verified bytes under SOURCE/NAME/VERSION/ARTIFACT.FORMAT.

  Compressed artifacts retain their compression suffix. Existing exports are never overwritten.

datamonger registry

Inspect the selected registry or resolve a release through an HTTPS catalog

Usage: datamonger registry <COMMAND>

datamonger registry show

Show the active registry selector without downloading its index

Usage: datamonger registry show

datamonger registry resolve

Resolve a release to a selector with a fixed index SHA-256.

Requires network access. Prints the selector as JSON and optionally writes it to a file. A checksum fixes the index bytes but does not authenticate the catalog's publisher.

Usage: datamonger registry resolve [OPTIONS] <RELEASE>

Arguments

• <RELEASE>

  Registry release to resolve, such as 1.1.0

Options

• --catalog-url <CATALOG_URL>

  HTTPS catalog from which to resolve the release

  Default value: https://raw.githubusercontent.com/jolars/datamonger/main/registry/catalog.json

• --output <OUTPUT>

  Also write the selector to this file, such as .datamonger/selector.json

datamonger cache

Inspect or clean the CLI's private cache

Usage: datamonger cache <COMMAND>

datamonger cache list

List cached registries and artifacts, including their sizes and integrity

Usage: datamonger cache list

datamonger cache clean

Remove cached objects, skipping any objects currently in use.

Requires either --dataset or --all. Dataset cleanup removes artifacts associated with the exact version in a bundled or cached registry.

Usage: datamonger cache clean [OPTIONS]

Options

• --dataset <DATASET>

  Remove artifacts for a dataset in SOURCE:NAME@VERSION form

• --all

  Remove all cached registry indexes and artifacts