fairscape-conversion

Your metadata is already written down — as a datasheet, a datapackage, or a workflow engine's run output. Convert it into a FAIRSCAPE/EVI RO-Crate (ro-crate-metadata.json) instead of re-entering it. One command per format.

pip install -e .        # from this directory; installs fairscape-conversion

The fairscape import / fairscape export commands come with the fairscape CLI. Without it, every conversion also runs as python -m fairscape_conversion.core.cli convert <format> <import|export> IN [OUT].

Import — get an RO-Crate

Pick the row that matches what you have.

Export — from an RO-Crate

Try it — no data needed

Every conversion above has a runnable example in examples/, on real input that ships with the package. They need nothing installed:

python examples/run_all.py            # every one, with a pass/fail table
python examples/import_d4d.py         # or just the one you care about

Each prints what went in, what came out, the provenance edges it created, and whether the result still matches the plugin's reviewed golden file.

examples/mlflow/mlflow_to_rocrate.ipynb is the end-to-end walk-through: it trains a scikit-learn model with MLflow tracking, converts the store it just created, validates the crate, and renders its datasheet and provenance graph inline. A three-step pipeline with usedMLModel lives next to it in examples/mlflow/pipeline/. Both are checked in with their outputs.

The inputs and expected outputs the examples run on live inside each plugin:

Linked crates — when one run's inputs were another run's outputs

Two independently converted crates do not know about each other. Pass the upstream crate to the conversion and the new crate reuses its identifiers for whatever it consumed from it, as a stub that points back:

fairscape import mlflow ./mlruns --experiment NAME -o ./crate --link-crate /path/to/upstream-crate
fairscape_conversion link ./crate --link-crate /path/to/upstream-crate    # or any crate, after the fact

The match is by file path, then md5, then containing directory; it runs after every importer, so it is not specific to any format. fairscape-artifacts follows the pointer and draws one evidence graph across both crates. The convention and two runnable pairs (Nextflow → MLflow, Snakemake → Snakemake) are in examples/linked-crates/.

From Python

import yaml
from fairscape_conversion.plugins import d4d

crate = d4d.convert("import", yaml.safe_load(open("datasheet.yaml")))

Same shape for every format: wrroc, c2m2, cromwell, snakemake, mlflow, cpm, redcap, frictionless, galaxy (convert("import", ...)), and d4d/wrroc/croissant/cpm/frictionless (convert("export", crate)). cpm takes the crate directory rather than a parsed document, because it reads the PROV bundle files registered in the metadata alongside it.

More

• Your format isn't listed → docs/NEW-PLUGIN.md — a converter is a folder of two CSVs plus a small plugin class.
• How the engine works, what's tested → docs/INTERNALS.md and MAPPING-SCHEMA.md.