Testing

The recognizer test suite.

Run the tests

make test

Runs uv run pytest against the tests in src/sombra/tests/.

What the tests cover

• Pattern matching — each SOMBRA_* entity is matched against valid and invalid samples.
• Context boosting — confidence rises when a context word appears nearby.
• Parity between code and YAML — the same recognizer defined both ways behaves identically.

Writing a new test

Follow the existing structure in src/sombra/tests/: build a recognizer, run it over a sample text with analyzer.analyze(), and assert the expected entity type and score threshold.

def test_cpf_recognizer(analyzer):
    results = analyzer.analyze(
        text="Meu CPF é 123.456.789-00.",
        language="pt",
    )
    assert any(r.entity_type == "SOMBRA_CPF" for r in results)
    assert any(r.score >= 0.9 for r in results)

Next

Ready to smoke-test the stack end to end? See Quick tests.