def retrieve_portal_manifest(etcd, portal_name):
etcd_key = os.path.join(portal_name, 'manifest')
raw_data = etcd.get_data(etcd_key)
if raw_data is None:
raise ValueError("the manifest of {} should be stored "\
"in etcd".format(portal_name))
return text_format.Parse(raw_data, common_pb.DataJoinPortalManifest())
def _setUpPortalManifest(self):
self._portal_name = 'test_portal'
self._etcd_l.delete_prefix(self._portal_name)
self._etcd_f.delete_prefix(self._portal_name)
self._portal_manifest_l = common_pb.DataJoinPortalManifest(
name=self._portal_name,
input_partition_num=4,
output_partition_num=2,
input_data_base_dir='./portal_input_l',
output_data_base_dir='./portal_output_l',
begin_timestamp=common.trim_timestamp_by_hourly(
common.convert_datetime_to_timestamp(datetime.now())))
self._portal_manifest_f = common_pb.DataJoinPortalManifest(
name=self._portal_name,
input_partition_num=2,
output_partition_num=2,
input_data_base_dir='./portal_input_f',
output_data_base_dir='./portal_output_f',
begin_timestamp=common.trim_timestamp_by_hourly(
common.convert_datetime_to_timestamp(datetime.now())))
common.commit_portal_manifest(self._etcd_l, self._portal_manifest_l)
common.commit_portal_manifest(self._etcd_f, self._portal_manifest_f)
def _update_portal_commited_timestamp(self, new_committed_datetime):
new_manifest = None
with self._lock:
old_committed_datetime = common.convert_timestamp_to_datetime(
common.trim_timestamp_by_hourly(
self._portal_manifest.committed_timestamp))
assert new_committed_datetime > old_committed_datetime
new_manifest = common_pb.DataJoinPortalManifest()
new_manifest.MergeFrom(self._portal_manifest)
assert new_manifest is not None
new_manifest.committed_timestamp.MergeFrom(
common.trim_timestamp_by_hourly(
common.convert_datetime_to_timestamp(new_committed_datetime)))
common.commit_portal_manifest(self._etcd, new_manifest)
return new_manifest
def _prepare_test(self):
self._portal_manifest = common_pb.DataJoinPortalManifest(
name='test_portal',
input_partition_num=4,
output_partition_num=8,
input_data_base_dir='./portal_input',
output_data_base_dir='./portal_output')
self._portal_options = dj_pb.DataJoinPotralOptions(
example_validator=dj_pb.ExampleValidatorOptions(
example_validator='EXAMPLE_VALIDATOR',
validate_event_time=True,
),
reducer_buffer_size=128,
raw_data_options=dj_pb.RawDataOptions(raw_data_iter='TF_RECORD'),
use_mock_etcd=True)
self._date_time = common.convert_timestamp_to_datetime(
common.trim_timestamp_by_hourly(
common.convert_datetime_to_timestamp(datetime.now())))
self._generate_input_data()
