def test_mk_fields_from():
fields_name_error = [{}]
fields_valid = [{
"name": "flight_id",
"type": "INTEGER",
"description": "flight id",
"tags": ["tag1", "tag2"]
}, {
"name": "flight_name",
"type": "VARCHAR",
"description": "flight name",
"tags": ["tag3", "tag4"]
}, {
"name": "flight_date",
"type": "TIMESTAMP",
"description": "flight date"
}]
new_fields_valid = [{
"name": "flight_id",
"type": "INTEGER",
"description": "flight id",
"tags": ["tag1", "tag2"]
}, {
"name": "flight_name",
"type": "VARCHAR",
"description": "flight name",
"tags": ["tag3", "tag4"]
}, {
"name": "flight_date",
"type": "TIMESTAMP",
"description": "flight date"
}]
assert Utils.mk_fields_from(fields=fields_valid) == new_fields_valid
with pytest.raises(ValueError):
Utils.mk_fields_from(fields=fields_name_error)
def create_dataset(self,
namespace_name,
dataset_name,
dataset_type,
physical_name,
source_name,
description=None,
run_id=None,
schema_location=None,
fields=None,
tags=None):
Utils.check_name_length(namespace_name, 'namespace_name')
Utils.check_name_length(dataset_name, 'dataset_name')
Utils.is_instance_of(dataset_type, DatasetType)
if dataset_type == DatasetType.STREAM:
MarquezClient._is_none(schema_location, 'schema_location')
Utils.check_name_length(physical_name, 'physical_name')
Utils.check_name_length(source_name, 'source_name')
payload = {
'type': dataset_type.value,
'physicalName': physical_name,
'sourceName': source_name,
}
if description:
payload['description'] = description
if run_id:
payload['runId'] = run_id
if fields:
payload['fields'] = Utils.mk_fields_from(fields)
if tags:
payload['tags'] = tags
if schema_location:
payload['schemaLocation'] = schema_location
return self._put(self._url('/namespaces/{0}/datasets/{1}',
namespace_name, dataset_name),
payload=payload)