Cleaner test inputs for RO-Crate Validator

[douglowe]

Currently the profile files for ensuring that invalid crates are caught by the validator are created as duplicates of the original (valid) crate. This means that multiple, almost identical, files will have to be edited everytime the crate profile is changed.

Instead it makes sense to programatically modify the valid crate to create invalid crates for each test, saving the test file in pytest's temporary directory for ease of checking these.

ChatGPT suggests this style of structure.

Have a reusable fixture for a valid base input

# conftest.py or in your test file
import pytest
import copy

@pytest.fixture
def base_jsonld():
    return {
        "@context": [
            "https://w3id.org/ro/crate/1.1/context",
            "https://w3id.org/ro/terms/test"
        ],
        "@graph": [
            {
                "@id": "ro-crate-metadata.json",
                "@type": "CreativeWork",
                "about": {"@id": "./"},
                "conformsTo": [
                    {"@id": "https://w3id.org/ro/crate/1.1"},
                    {"@id": "https://w3id.org/workflowhub/workflow-ro-crate/1.0"}
                ]
            }
        ]
    }

Define error variants as transformations

def missing_context(data):
    data = copy.deepcopy(data)
    del data["@context"]
    return data

def missing_graph_entry(data):
    data = copy.deepcopy(data)
    data["@graph"].clear()
    return data

def malformed_conforms_to(data):
    data = copy.deepcopy(data)
    data["@graph"][0]["conformsTo"] = "not-a-list"
    return data

Parameterize the tests

import pytest
import json
from mymodule import process_jsonld  # your processing function

@pytest.mark.parametrize("modifier,expected_exception", [
    (missing_context, KeyError),
    (missing_graph_entry, ValueError),
    (malformed_conforms_to, TypeError),
])
def test_jsonld_processing_errors(base_jsonld, modifier, expected_exception, tmp_path):
    modified = modifier(base_jsonld)
    input_file = tmp_path / "test.json"
    input_file.write_text(json.dumps(modified, indent=2))

    with pytest.raises(expected_exception):
        process_jsonld(input_file)

[douglowe]

Discussions with Eli and Oliver - we think that SPARQL will be a cleaner way to do this. e.g.

DELETE
    ./ schema:sourceOrganization ?

See also https://github.com/workflowhub-eu/workflowhub-graph/blob/e25e18a764da61437e5ad687876457928b792f9d/workflowhub_graph/workflowhub_graph/consolidate.py#L4 for inspiration