Proposal: make a new dataclass for ontological validation

[liamhuber]

I currently have snippets like this running around:

if (
    len(validation["missing_triples"]) > 0
    or len(validation["incompatible_connections"]) > 0
    or len(validation["distinct_units"]) > 0
):

Or, less egregiously, annoying string references like

self.assertEqual(validate_values(graph)["incompatible_connections"], [])

Also, since #255, they type return on ontology.validate_values is extremely vague:

semantikon/semantikon/ontology.py

Line 329 in e173ae1

) -> dict[str, Any]:

Let's make a new dataclass to hold the validation like

@dataclass
class Validation:
    missing_triples: list
    incompatible_connections: list
    distinct_units: dict[URIRef, list[URIRef]]

    def valid():
        return not any([
           self.missing_triples,
           self.incompatible_connections,
           self.distinct_units
       ])  # We could iterate over fields, but no need to get fancy yet

    def report():
        return as_dict(self)  # Or a JSON or whatever

Actually, that's basically it, but I want to make sure @samwaseda doesn't hate it before I bother writing tests and adapting all the existing tests.

[samwaseda]

Nice. That sounds good to me.

[liamhuber]

Here are some methods I found helpful in pyiron_workflow that might have a good home right on a validation object:

https://github.com/pyiron/pyiron_workflow/blob/8661600c263949f399d4cf5e53a6f39c7744420c/pyiron_workflow/knowledge.py#L252-L303