rs-metadata

This repository generates and defines the LUMC CodeMeta profile with CodeMeta 3.1 vocabulary. It also checks consistency across the metadata in a repository. The project supports LUMC Research Software guidelines on metadata.

The documentation is available at lumc-dcc.github.io/rs-metadata.

Quick start

pip install rs-metadata
rs-metadata init

init creates missing codemeta.json, CITATION.cff, and CI workflow files. It uses metadata files and the git remote where possible, and it never overwrites existing files.

rs-metadata validate

tells you if you still missing real values.

If you already have the metadata files, you can add just one workflow:

name: Validate software metadata
on: [push, pull_request]

jobs:
  metadata:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - uses: LUMC-DCC/rs-metadata@v1

What it checks

• Profile conformance. Every field the profile includes, in a valid shape, with a valid JSON-LD context. Values are checked against their declared CodeMeta/schema.org types.
• Value sanity. E.g., we check if identifiers (ORCID and ROR) and dates are of valid shape, if the licenses exist, and whether some placeholder values remain.
• Cross-file consistency. codemeta.json is the anchor metadata; every other metadata file is compared against it.

flowchart LR
    S0[CITATION.cff]:::src --> N([normalized CodeMeta concepts]):::mid
    S1[github-repo.json]:::src --> N
    S2[.zenodo.json]:::src --> N
    S3[biotools.json]:::src --> N
    S4[pyproject.toml]:::src --> N
    S5[package.json]:::src --> N
    S6[DESCRIPTION]:::src --> N
    S7[Cargo.toml]:::src --> N
    S8[Project.toml]:::src --> N
    S9[Dockerfile]:::src --> N
    CM[codemeta.json<br/><i>anchor</i>]:::anchor --> N

    S0 ~~~ S1 ~~~ S2 ~~~ S3 ~~~ S4
    S5 ~~~ S6 ~~~ S7 ~~~ S8 ~~~ S9
    CM

    classDef src fill:#eef4fb,stroke:#5b8db8
    classDef anchor fill:#fdf1e2,stroke:#c98b34
    classDef mid fill:#f2f2f2,stroke:#888

Comparison is semantic. https://spdx.org/licenses/Apache-2.0 equals Apache-2.0, git+https://github.com/o/t.git equals https://github.com/o/t, J. Carberry matches Josiah Carberry, >=1.24 equals >=1.24.0, and list order never matters.

It does not treat naming conventions as conflicts: a PyPI package named my-tool can match a citation title of MyTool. A dependency in only one file is incomplete metadata, not a contradiction.

Example output

ERROR [consistency.mismatch]
Property: version

  codemeta.json:10
    version = 1.4.0

  CITATION.cff:8
    version = 1.2.0

  version differs between codemeta.json and CITATION.cff.

  Suggested fix:
    Update whichever file is stale so both describe the same release.

In CI, the same finding becomes an inline annotation, a job summary, and a machine-readable report.

Commands

Exit status is 0 when the run passes, 1 when it fails, 2 for a usage error. Full reference: Running rs-metadata.

Contributing

Contributions are welcome! Please read CONTRIBUTING.md and Architecture.

License

This project is licensed under Apache 2.0. See the LICENSE file for details.