def test_create_dataset_with_name_prefix(self, client, mocker):
entity_name = "myentity"
feature_ids = ["myentity.feature1", "myentity.feature2"]
fs = FeatureSet(entity_name, feature_ids)
start_date = "2018-01-02"
end_date = "2018-12-31"
limit = 100
name_prefix = "feast"
ds_pb = DatasetInfo_pb(name="dataset_name",
tableUrl="project.dataset.table")
mock_dssvc_stub = training.DatasetServiceStub(
grpc.insecure_channel(""))
mocker.patch.object(
mock_dssvc_stub,
"CreateDataset",
return_value=DatasetServiceTypes.CreateDatasetResponse(
datasetInfo=ds_pb))
client._dataset_service_stub = mock_dssvc_stub
ds = client.create_dataset(fs,
start_date,
end_date,
limit=limit,
name_prefix=name_prefix)
assert "dataset_name" == ds.name
assert "project.dataset.table" == ds.full_table_id
mock_dssvc_stub.CreateDataset.assert_called_once_with(
DatasetServiceTypes.CreateDatasetRequest(
featureSet=fs.proto,
startDate=_timestamp_from_datetime(_parse_date(start_date)),
endDate=_timestamp_from_datetime(_parse_date(end_date)),
limit=limit,
namePrefix=name_prefix))
def create_dataset(self,
feature_set,
start_date,
end_date,
limit=None,
name_prefix=None):
"""
Create training dataset for a feature set. The training dataset
will be bounded by event timestamp between start_date and end_date.
Specify limit to limit number of row returned. The training dataset
will reside in a bigquery table specified by destination.
Args:
feature_set (feast.sdk.resources.feature_set.FeatureSet):
feature set representing the data wanted
start_date (str): starting date of the training data in ISO 8601
format (e.g.: "2018-12-31")
end_date (str): end date of training data in ISO 8601 format (e.g.:
"2018-12-31")
limit (int, optional): (default: None) maximum number of row
returned
name_prefix (str, optional): (default: None) name prefix.
:return:
feast.resources.feature_set.DatasetInfo: DatasetInfo containing
the information of training dataset
"""
self._check_create_dataset_args(feature_set, start_date, end_date,
limit)
req = DatasetServiceTypes.CreateDatasetRequest(
featureSet=feature_set.proto,
startDate=_timestamp_from_datetime(_parse_date(start_date)),
endDate=_timestamp_from_datetime(_parse_date(end_date)),
limit=limit,
namePrefix=name_prefix,
)
if self.verbose:
print("creating training dataset for features: " +
str(feature_set.features))
self._connect_core()
resp = self._dataset_service_stub.CreateDataset(req)
if self.verbose:
print("created dataset {}: {}".format(resp.datasetInfo.name,
resp.datasetInfo.tableUrl))
return DatasetInfo(resp.datasetInfo.name, resp.datasetInfo.tableUrl)
