feat: add language and datatype support when converting rdflib.terms.… #356
Conversation
…Literal to JSON-LD
|
of course, this PR is broken if we consider that only However, it would still fix the missing Any thoughts ? |
|
@ticapix I have moved the work for this PR from the contributor fork to the linkml-runtime branch https://github.com/linkml/linkml-runtime/pull/new/add_rdf_lang_type_support . These branches will be lifted over in the merge to the mono-repo so you won't lose you work. You will need to open a new PR against the linkml monorepo after the move. I'd prefer to get this in but if we can't please open a new PR when you resume work. Thanks! |
There was a problem hiding this comment.
The core changes in yamlutils.py to serialize langstrings looks solid. The biggest change needed is how the test is constructed. I gave my suggestion on how I would handle it, but it'd probably be good to check with people who have stronger opinions on how tests in here are presented.
| from jsonasobj2 import JsonObj, JsonObjTypes, as_dict, as_json, items | ||
| from rdflib import Graph, URIRef | ||
| from jsonasobj2 import JsonObj, as_json, as_dict, JsonObjTypes, items | ||
| import jsonasobj2 |
| from typing import ( | ||
| Any, | ||
| ClassVar, | ||
| Dict, |
There was a problem hiding this comment.
Since support for Python 3.8 was dropped earlier this year, these imports should be removed, and plain old dict and list should be used.
There was a problem hiding this comment.
Both of these entries should be removed-- html5lib should be added to the [dependencies] section in pyproject.toml.
There was a problem hiding this comment.
I'm not sure about the strategy here to define a schema inline-- if this test were merged, the definition would potentially have to be kept up to date with changes in behavior of linkml runtime. (There's already one here, as dataclasses_init_fn_with_kwargs has been deprecated).
There's already a Person schema defined in tests. Can that be reused here to define a test case? (See these lines).
It might be good to add a new test case to that file instead of adding a new file here. (Although others may have a different opinion). There's another test there that is doing something very similar to what this test is doing:
# see https://github.com/linkml/linkml/issues/429
SCHEMA_429 = INPUT_PATH / "personinfo_test_issue_429.yaml"
DATA_429 = INPUT_PATH / "example_personinfo_test_issue_429_data.yaml"
OUT_429 = OUTPUT_PATH / "example_personinfo_test_issue_429_data.ttl"
...
@pytest.fixture
def issue_429_graph():
"""Create RDF graph for issue 429 testing."""
view = SchemaView(str(SCHEMA_429))
container = yaml_loader.load(str(DATA_429), target_class=Container_429)
rdflib_dumper.dump(container, schemaview=view, to_file=str(OUT_429))
g = Graph()
g.parse(str(OUT_429), format="ttl")
return g
|
@ticapix I have moved the work on this PR into the repo branch https://github.com/linkml/linkml-runtime/tree/add_rdf_lang_type_support . If you would like to continue this work please re-open the PR on the mono-repo after the merge from that branch. We'll lift it over for you so it should take less effort. |
4 participants
Proposal to address linkml/linkml#2475
Without this, it's impossible to convert a Python Instance of a LinkML gen-python class to RDF without loosing language and datatype information.