def _align_data_index(self, data_instance, flowid, data_application=None):
header = data_instance.schema.get('header')
if data_application is None:
LOGGER.warning("not data_application!")
return
transfer_variable = CrossValidationTransferVariable()
if data_application == consts.TRAIN_DATA:
transfer_id = transfer_variable.train_sid
elif data_application == consts.TEST_DATA:
transfer_id = transfer_variable.test_sid
else:
LOGGER.warning("data_application error!")
return
if self.role == consts.GUEST:
data_sid = data_instance.mapValues(lambda v: 1)
transfer_id.remote(data_sid,
role=consts.HOST,
idx=-1,
suffix=(flowid, ))
LOGGER.info("remote {} to host".format(data_application))
return None
elif self.role == consts.HOST:
data_sid = transfer_id.get(idx=0, suffix=(flowid, ))
LOGGER.info("get {} from guest".format(data_application))
join_data_insts = data_sid.join(data_instance, lambda s, d: d)
join_data_insts.schema['header'] = header
return join_data_insts
def _align_data_index(self, data_instance, flowid, data_application=None):
schema = data_instance.schema
if data_application is None:
# LOGGER.warning("not data_application!")
# return
raise ValueError(
"In _align_data_index, data_application should be provided.")
transfer_variable = CrossValidationTransferVariable()
if data_application == consts.TRAIN_DATA:
transfer_id = transfer_variable.train_sid
elif data_application == consts.TEST_DATA:
transfer_id = transfer_variable.test_sid
else:
raise ValueError(
"In _align_data_index, data_application should be provided.")
if self.role == consts.GUEST:
data_sid = data_instance.mapValues(lambda v: 1)
transfer_id.remote(data_sid,
role=consts.HOST,
idx=-1,
suffix=(flowid, ))
LOGGER.info("remote {} to host".format(data_application))
return data_instance
elif self.role == consts.HOST:
data_sid = transfer_id.get(idx=0, suffix=(flowid, ))
LOGGER.info("get {} from guest".format(data_application))
join_data_insts = data_sid.join(data_instance, lambda s, d: d)
join_data_insts.schema = schema
return join_data_insts