def test_frequency_none():
drop_caches()
with settings.file_override("FREQUENCY"):
with open(settings.FREQUENCY, "wb") as f:
pickle.dump({"dog": 2, "cat": 3}, f)
assert set(SubstitutionFeaturesMixin._frequency()) == {"dog", "cat"}
def test_frequency_none_with_computed():
drop_caches()
# Lemmas are all lowercase.
for word in SubstitutionFeaturesMixin._frequency():
assert word.islower() or is_int(word[0]) or is_int(word[-1]) or word in ["%", "!"]
def test_frequency():
drop_caches()
assert SubstitutionFeaturesMixin._frequency("dog") == 7865
assert SubstitutionFeaturesMixin._frequency("play") == 45848
assert np.isnan(SubstitutionFeaturesMixin._frequency("wickiup"))
