def test_container_feature_lookup():
arbitrary_attribute = "other"
messages = [
Message(data={TEXT: "A"}, features=[_dummy_features(1, TEXT)]),
Message(
data={
INTENT: "B",
arbitrary_attribute: "C"
},
features=[_dummy_features(2, arbitrary_attribute)],
),
Message(data={TEXT: "A2"}, features=[_dummy_features(3, TEXT)]),
Message(
data={
INTENT: "B2",
arbitrary_attribute: "C2"
},
features=[_dummy_features(4, arbitrary_attribute)],
),
]
table = MessageContainerForCoreFeaturization()
table.add_all(messages)
# If we don't specify a list of attributes, the resulting features dictionary will
# only contain those attributes for which there are features.
sub_state = {TEXT: "A", INTENT: "B", arbitrary_attribute: "C"}
features = table.collect_features(sub_state=sub_state)
for attribute, feature_value in [
(TEXT, 1),
(INTENT, None),
(arbitrary_attribute, 2),
]:
if feature_value is not None:
assert attribute in features
assert len(features[attribute]) == 1
assert feature_value == features[attribute][0].features[0]
else:
assert attribute not in features
# If we query features for `INTENT`, then a key will be there, even if there are
# no features
features = table.collect_features(sub_state=sub_state,
attributes=list(sub_state.keys()))
assert INTENT in features
assert len(features[INTENT]) == 0
# We only get the list of features we want...
features = table.collect_features(sub_state,
attributes=[arbitrary_attribute])
assert TEXT not in features
assert INTENT not in features
assert len(features[arbitrary_attribute]) == 1
# ... even if there are no features:
YET_ANOTHER = "another"
features = table.collect_features(sub_state, attributes=[YET_ANOTHER])
assert len(features[YET_ANOTHER]) == 0
def test_container_feature_lookup_fails_if_different_features_for_same_attribute():
broken_table = MessageContainerForCoreFeaturization()
broken_table._table = {
TEXT: {"A": Message(data={}, features=[_dummy_features(2, TEXT)])},
INTENT: {"B": Message(data={}, features=[_dummy_features(1, TEXT)])},
}
with pytest.raises(
RuntimeError, match=f"Feature for attribute {TEXT} has already been"
):
broken_table.collect_features({TEXT: "A", INTENT: "B"})
def test_container_feature_lookup_works_if_messages_are_broken_but_consistent():
not_broken_but_strange_table = MessageContainerForCoreFeaturization()
not_broken_but_strange_table._table = {
TEXT: {"A": Message(data=dict())},
INTENT: {"B": Message(data=dict(), features=[_dummy_features(1, TEXT)])},
}
features = not_broken_but_strange_table.collect_features({TEXT: "A", INTENT: "B"})
assert TEXT in features and len(features[TEXT]) == 1
def test_container_feature_lookup_fails_without_key_attribute():
table = MessageContainerForCoreFeaturization()
with pytest.raises(ValueError, match="Unknown key"):
table.collect_features({TEXT: "A-unknown"})