def test_raises_exception_on_invalid_file() -> None:
with pytest.raises(Exception) as exception_info:
# The python test file is not a valid transducer.
TransducerFile(__file__)
exception_messages = " ".join(exception_info.value.args)
assert "wrong or corrupt file?" in exception_messages
def test_raises_exception_on_missing_file() -> None:
with pytest.raises(Exception) as exception_info:
TransducerFile("/does-not-exist.hfstol")
exception_messages = " ".join(exception_info.value.args)
assert "Transducer not found" in exception_messages
assert "‘/does-not-exist.hfstol’" in exception_messages
def relaxed_analyzer():
return TransducerFile(FST_DIR / settings.RELAXED_ANALYZER_FST_FILENAME)
def test_symbol_lookup1(fst: TransducerFile, surface: str,
deep: list[list[str]]) -> None:
assert fst.lookup_symbols(surface) == deep
def test_limit(fst: TransducerFile) -> None:
assert fst.lookup("môswa", limit=1) == ["môswa+N+A+Sg"] # type: ignore
def test_create_from_path_obj() -> None:
fst = TransducerFile(Path(TEST_FST))
assert fst.lookup("itwêwina") == ["itwêwin+N+I+Pl"]
def test_bulk_lookup_with_iterator(fst: TransducerFile) -> None:
assert (fst.bulk_lookup(w for w in EXPECTED_BULK_LOOKUP_RESULT_1.keys()) ==
EXPECTED_BULK_LOOKUP_RESULT_1)
def test_bulk_lookup(fst: TransducerFile) -> None:
assert (fst.bulk_lookup(
EXPECTED_BULK_LOOKUP_RESULT_1.keys()) == EXPECTED_BULK_LOOKUP_RESULT_1)
def strict_generator_with_morpheme_boundaries():
return TransducerFile(
FST_DIR / "crk-strict-generator-with-morpheme-boundaries.hfstol"
)
def test_valid_lookup_with_invalid_symbol(fst: TransducerFile) -> None:
assert fst.lookup("avocado") == []
assert fst.lookup("navajo") == []
def test_subsequent_lookups(fst: TransducerFile) -> None:
assert fst.lookup("itwêwina") == ["itwêwin+N+I+Pl"]
assert fst.lookup("nikî-nipân") == ["PV/ki+nipâw+V+AI+Ind+1Sg"]
def test_symbol_count() -> None:
# If this returned a non-number, we’d get a TypeError here.
assert TransducerFile(TEST_FST).symbol_count() > 0
def test_lookup_lemma_with_affixes(fst: TransducerFile, surface: str,
example: Analysis) -> None:
analyses = fst.lookup_lemma_with_affixes(surface)
assert example in analyses
def fst() -> TransducerFile:
return TransducerFile(TEST_FST)
def strict_analyzer():
return TransducerFile(FST_DIR / settings.STRICT_ANALYZER_FST_FILENAME)
def test_multiple_analyses(fst: TransducerFile) -> None:
assert fst.lookup("môswa") == ["môswa+N+A+Sg", "môswa+N+A+Obv"]
def strict_generator():
return TransducerFile(FST_DIR / settings.STRICT_GENERATOR_FST_FILENAME)
def cmro_transcriptor():
return TransducerFile(settings.BASE_DIR / "resources" / "fst" /
"default-to-cmro.hfstol")