def test_data_frequency_total_ordering(valid_frequencies):
assert len(valid_frequencies) == 10
for idx_1, f1 in enumerate(valid_frequencies):
for idx_2, f2 in enumerate(valid_frequencies):
freq1 = DataFrequency(frequency=f1)
freq2 = DataFrequency(frequency=f2)
if idx_1 > idx_2:
assert freq1 > freq2
if idx_1 < idx_2:
assert freq1 < freq2
if idx_1 == idx_2:
assert freq1 == freq2
if idx_1 != idx_2:
assert freq1 != freq2
if idx_1 <= idx_2:
assert freq1 <= freq2
if idx_1 >= idx_2:
assert freq1 >= freq2
def data_frequency(self, dataset_file: DatasetFile) -> DataFrequency:
"""
Get the data frequency from config
:param dataset_file: The dataset file to use
:return: the desired data frequency
"""
frequency = self.config_item(dataset_file, "Dataset.DataFrequency")
return DataFrequency(frequency)
def test_dataset_frequency_invalid():
with pytest.raises(ValueError) as ex:
DataFrequency("1y")
assert (
str(ex.value) ==
"Invalid frequency. Frequency 1y does not match ^Y|M|W|D|H|30min|15min|10min|5min|1min$"
)
def test_dataset_create_noop_errors(configuration_data, forecast_stub):
config = Config()
config.config = configuration_data
dataset_file = DatasetFile("RetailDemandTRM.csv", "some_bucket")
dataset = config.dataset(dataset_file)
configured_dataset = configuration_data.get("RetailDemandTRM").get(
"Datasets")[2]
params = {
"DatasetType": configured_dataset.get("DatasetType"),
"DatasetName": "RetailDemandTRM",
"Domain": configured_dataset.get("Domain"),
"Schema": configured_dataset.get("Schema"),
"DataFrequency": configured_dataset.get("DataFrequency"),
}
create_params = deepcopy(params)
create_params["Tags"] = [{"Key": "SolutionId", "Value": "SOL0123"}]
forecast_stub.add_response(
"describe_dataset",
params,
)
forecast_stub.add_response("create_dataset", {"DatasetArn": dataset.arn},
create_params)
forecast_stub.add_response(
"describe_dataset",
params,
)
dataset.cli = forecast_stub.client
dataset.create()
# clobber the values to trigger some exceptions
# this is likey caused by a user changing configuration unexpectedly
dataset._dataset_type = DatasetType.RELATED_TIME_SERIES
dataset._dataset_domain = DatasetDomain.WORK_FORCE
dataset._data_frequency = DataFrequency("1min")
dataset._dataset_schema = {}
with pytest.raises(ValueError) as excinfo:
dataset.create()
assert "dataset type" in str(excinfo.value)
assert "dataset domain" in str(excinfo.value)
assert "data frequency" in str(excinfo.value)
assert "dataset schema" in str(excinfo.value)
def test_dataset_create_noop_errors(configuration_data, forecast_stub):
config = Config()
config.config = configuration_data
dataset_file = DatasetFile("RetailDemandTRM.csv", "some_bucket")
dataset = config.dataset(dataset_file)
configured_dataset = configuration_data.get("RetailDemandTRM").get("Datasets")[2]
for i in range(0, 2):
forecast_stub.add_response(
"describe_dataset",
{
"DatasetType": configured_dataset.get("DatasetType"),
"DatasetName": "RetailDemandTRM",
"Domain": configured_dataset.get("Domain"),
"Schema": configured_dataset.get("Schema"),
"DataFrequency": configured_dataset.get("DataFrequency"),
},
)
# should not call anything
dataset.cli = forecast_stub.client
dataset.create()
# clobber the values to trigger some exceptions
# this is likey caused by a user changing configuration unexpectedly
dataset._dataset_type = DatasetType.RELATED_TIME_SERIES
dataset._dataset_domain = DatasetDomain.WORK_FORCE
dataset._data_frequency = DataFrequency("1min")
dataset._dataset_schema = {}
with pytest.raises(ValueError) as excinfo:
dataset.create()
assert "dataset type" in str(excinfo.value)
assert "dataset domain" in str(excinfo.value)
assert "data frequency" in str(excinfo.value)
assert "dataset schema" in str(excinfo.value)
def test_dataset_frequency_str(valid_frequencies):
for frequency in valid_frequencies:
freq = DataFrequency(frequency=frequency)
assert str(freq) == frequency
def test_dataset_frequency_repr(valid_frequencies):
for frequency in valid_frequencies:
freq = DataFrequency(frequency=frequency)
assert repr(freq) == f"DataFrequency(frequency='{frequency}')"
def test_dataset_frequency_valid(valid_frequencies):
for valid_frequency in valid_frequencies:
frequency = DataFrequency(valid_frequency)
assert frequency == valid_frequency
def test_invalid_data_frequency_comparison():
frequency = DataFrequency("1min")
invalid = "not_valid"
with pytest.raises(ValueError):
frequency > invalid