def instance(source_uri, random_seed=0):
relation_helper = RelationHelper(source_uri)
private_manager = kernel.PrivateManager(relation_helper.filename,
relation_helper.config,
random_seed=random_seed)
kernel_service = pservice.KernelService(private_manager)
return ProtectedDataSource(kernel_service,
DataManager(transformation.Null()))
def __init__(self,
source_uri,
data_config,
file_delimiter=',',
random_seed=0,
budget=1.0):
self.source_uri = source_uri
self.data_config = data_config
self.file_delimiter = file_delimiter
self.random_seed = random_seed
self._budget = budget
self.prng = np.random.RandomState(random_seed)
self._data_manager = DataManager(transformation.Null())
self._tgraph = TransformationGraph()
self._data = None
data_node = self._data_manager._dgraph.root
self._tgraph.insert_transformation(data_node, transformation.Null())
def partition(self, operator, after):
assert util.contains_superclass(operator.__class__, 'SplitByPartition')
data_nodes = self._data_manager.partition(operator, self.prng,
after.data_node)
meta_node = self._tgraph.insert_transformation(after.data_node,
operator, after)
child_nodes = []
for data_node in data_nodes:
child_node = self._tgraph.insert_transformation(
data_node, transformation.Null(), meta_node)
child_nodes.append(child_node)
return child_nodes
def __init__(self,
kernel_service,
data_manager=DataManager(transformation.Null()),
private_node_id=None,
data_node=None,
random_seed=None):
self.kernel_service = kernel_service
self.data_manager = data_manager
self.prng = np.random.RandomState(random_seed)
if private_node_id is not None:
self.private_node_id = private_node_id
else:
self.private_node_id = self.kernel_service.get_root_id()
if data_node is not None:
self.data_node = data_node
else:
self.data_node = self.data_manager._dgraph.root