Running the tests

Ontospy’s tests use pytest. They live in ontospy/tests, together with the RDF files they load (ontospy/tests/rdf).

Setup

From a clone of the repository, in a virtual environment:

$ pip install -e .
$ pip install pytest

Install with -e (editable): the RDF files in ontospy/tests/rdf aren’t included in the package, so the tests and quick tests only find them in a source checkout.

Running the tests

From the repository root:

$ pytest                       # all tests
$ pytest -m "not remote"       # skip the tests that need network access
$ pytest -m remote             # only the tests that need network access

Some useful variations:

$ pytest ontospy/tests/test_methods.py            # one file
$ pytest ontospy/tests/test_methods.py -k test2   # one test
$ pytest -s                                       # show the tests' printed output
$ pytest -W ignore::DeprecationWarning            # hide rdflib deprecation warnings

Each test file can also be run on its own with unittest, eg python -m ontospy.tests.test_methods.

The local tests take about a minute, mostly spent loading the ontologies in ontospy/tests/rdf.

What the tests cover

File	What it checks	Network
`test_load_local.py`	Every ontology in `ontospy/tests/rdf` loads	No
`test_methods.py`	Specific `Ontospy` methods, preferred titles and languages, custom entity classes	No
`test_shapes.py`	SHACL shapes are extracted (SciGraph ontology)	No
`test_gendocs.py`	Every documentation type builds for an ontology with individuals (`pizza.ttl`)	No
`test_load_remote.py`	A sample online ontology loads	Yes
`test_sparql.py`	Connecting to and querying the DBpedia SPARQL endpoint	Yes

The network-dependent tests are tagged with the remote marker in ontospy/tests/conftest.py. They can fail for reasons outside Ontospy, eg when DBpedia is down or slow.

Continuous integration

On every push and pull request to master, GitHub Actions runs pytest -m "not remote" on Python 3.10 to 3.14. See .github/workflows/ci.yml.

Quick tests

ontospy/tests/quick.py is a scratchpad for trying things out before they become proper tests. It is not run by pytest. Run a numbered quick test with:

$ ontospy_quicktest 4

or python -m ontospy.tests.quick 4.

Number	What it does
1	Loads `paper.jsonld`
2	Queries the DBpedia SPARQL endpoint
3	Navigates DBpedia triples using rdflib’s `SPARQLStore` directly
4	Lists the individuals in `pizza.ttl` and the classes they are instances of

Local-only tests

Tests that shouldn’t be committed, eg against a private SPARQL endpoint that needs credentials, can go in ontospy/tests/local/:

the folder is gitignored
it has no __init__.py, so it is never included in the package
pytest still picks the tests up, and marks them as remote

Read credentials from environment variables rather than writing them in the test files, and skip the tests when they are missing, eg:

import os
import pytest

@pytest.fixture(scope="module")
def credentials():
    user, pwd = os.environ.get("MY_SPARQL_USER"), os.environ.get("MY_SPARQL_PASSWORD")
    if not (user and pwd):
        pytest.skip("Set MY_SPARQL_USER and MY_SPARQL_PASSWORD to run these tests")
    return (user, pwd)