def test_workflow_schema(self):
state = None
wf = Workflow(data=WorkflowData('test-workflow', 'ABC123', state=state))
obj_before = wf.to_dict()
wf = default_and_validate(wf, workflow_schema())
# state should be defaulted to INACTIVE
self.assertNotEqual(obj_before, wf.to_dict())
def test_workflow_schema(self):
state = None
wf = Workflow(
data=WorkflowData('test-workflow', 'ABC123', state=state))
obj_before = wf.to_dict()
wf = default_and_validate(wf, workflow_schema())
# state should be defaulted to INACTIVE
self.assertNotEqual(obj_before, wf.to_dict())
def setUp(self):
dart = Dart(host='localhost', port=5000)
""" :type dart: dart.client.python.dart_client.Dart """
self.dart = dart
dst_args = {'action_sleep_time_in_seconds': 0}
dst0 = Datastore(data=DatastoreData('test-datastore0',
'no_op_engine',
args=dst_args,
state=DatastoreState.TEMPLATE))
self.datastore0 = self.dart.save_datastore(dst0)
dst1 = Datastore(data=DatastoreData('test-datastore1',
'no_op_engine',
args=dst_args,
state=DatastoreState.TEMPLATE))
self.datastore1 = self.dart.save_datastore(dst1)
wf0 = Workflow(data=WorkflowData(
'test-workflow0', self.datastore0.id, state=WorkflowState.ACTIVE))
self.workflow0 = self.dart.save_workflow(wf0, self.datastore0.id)
wf1 = Workflow(data=WorkflowData(
'test-workflow1', self.datastore1.id, state=WorkflowState.ACTIVE))
self.workflow1 = self.dart.save_workflow(wf1, self.datastore1.id)
a00 = Action(data=ActionData(NoOpActionTypes.action_that_succeeds.name,
NoOpActionTypes.action_that_succeeds.name,
state=ActionState.TEMPLATE))
a01 = Action(data=ActionData(NoOpActionTypes.action_that_succeeds.name,
NoOpActionTypes.action_that_succeeds.name,
state=ActionState.TEMPLATE))
self.action00, self.action01 = self.dart.save_actions(
[a00, a01], workflow_id=self.workflow0.id)
a10 = Action(data=ActionData(NoOpActionTypes.action_that_succeeds.name,
NoOpActionTypes.action_that_succeeds.name,
state=ActionState.TEMPLATE))
a11 = Action(data=ActionData(NoOpActionTypes.action_that_succeeds.name,
NoOpActionTypes.action_that_succeeds.name,
state=ActionState.TEMPLATE))
self.action10, self.action11 = self.dart.save_actions(
[a10, a11], workflow_id=self.workflow1.id)
tr_args = {'completed_workflow_id': self.workflow0.id}
tr = Trigger(data=TriggerData('test-trigger', 'workflow_completion',
None, tr_args, TriggerState.ACTIVE))
self.trigger = self.dart.save_trigger(tr)
st_args = {
'fire_after': 'ALL',
'completed_trigger_ids': [self.trigger.id]
}
st = Trigger(data=TriggerData('test-super-trigger', 'super',
[self.workflow1.id], st_args,
TriggerState.ACTIVE))
self.super_trigger = self.dart.save_trigger(st)
def put_workflow(workflow):
""" :type workflow: dart.model.workflow.Workflow """
updated_workflow = Workflow.from_dict(request.get_json())
if updated_workflow.data.state not in [WorkflowState.ACTIVE, WorkflowState.INACTIVE]:
return {'results': 'ERROR', 'error_message': 'state must be ACTIVE or INACTIVE'}, 400, None
workflow = workflow_service().update_workflow_state(workflow, updated_workflow.data.state)
return {'results': workflow.to_dict()}
def setUp(self):
self.dart = Dart(host='localhost', port=5000)
args = {'action_sleep_time_in_seconds': 0}
dst = Datastore(data=DatastoreData('test-datastore', 'no_op_engine', args=args, state=DatastoreState.TEMPLATE))
self.datastore = self.dart.save_datastore(dst)
wf = Workflow(data=WorkflowData('test-workflow', self.datastore.id, state=WorkflowState.ACTIVE))
self.workflow = self.dart.save_workflow(wf, self.datastore.id)
def test_crud(self):
wf = Workflow(data=WorkflowData(name='test-workflow',
datastore_id=self.datastore.id,
engine_name='no_op_engine'))
posted_wf = self.dart.save_workflow(wf, self.datastore.id)
self.assertEqual(posted_wf.data.to_dict(), wf.data.to_dict())
workflow = self.dart.get_workflow(posted_wf.id)
self.assertEqual(posted_wf.to_dict(), workflow.to_dict())
workflow.data.concurrency = 2
workflow.data.state = WorkflowState.ACTIVE
put_workflow = self.dart.save_workflow(workflow)
self.assertEqual(put_workflow.data.concurrency, 2)
self.assertEqual(put_workflow.data.state, WorkflowState.ACTIVE)
self.assertNotEqual(posted_wf.to_dict(), put_workflow.to_dict())
self.dart.delete_workflow(workflow.id)
try:
self.dart.get_workflow(workflow.id)
except DartRequestException as e:
self.assertEqual(e.response.status_code, 404)
return
self.fail('workflow should have been missing after delete!')
def test_workflow_schema_invalid(self):
with self.assertRaises(DartValidationException) as context:
name = None
wf = Workflow(data=WorkflowData(name, 'ABC123'))
# should fail because the name is missing
default_and_validate(wf, workflow_schema())
self.assertTrue(isinstance(context.exception, DartValidationException))
def post_workflow(datastore):
""" :type datastore: dart.model.datastore.Datastore """
workflow = Workflow.from_dict(request.get_json())
workflow.data.datastore_id = datastore.id
workflow.data.engine_name = datastore.data.engine_name
if datastore.data.state == DatastoreState.ACTIVE:
# only templated datastores can use concurrencies > 1
workflow.data.concurrency = 1
workflow = workflow_service().save_workflow(workflow)
return {'results': workflow.to_dict()}
def post_workflow(datastore):
""" :type datastore: dart.model.datastore.Datastore """
workflow = Workflow.from_dict(request.get_json())
workflow.data.datastore_id = datastore.id
workflow.data.engine_name = datastore.data.engine_name
if datastore.data.state == DatastoreState.ACTIVE:
# only templated datastores can use concurrencies > 1
workflow.data.concurrency = 1
workflow = workflow_service().save_workflow(workflow)
return {'results': workflow.to_dict()}
def _resolve_and_save_workflow(self, entity_id, entity_map, actual_entities_by_node_id, actual_entities_by_unsaved_id):
actual_id, unsaved_id = self._resolve(EntityType.workflow, entity_id, entity_map, actual_entities_by_unsaved_id)
if actual_id:
return actual_id
node_id = self._node_id(EntityType.workflow, unsaved_id)
workflow = Workflow.from_dict(entity_map['unsaved_entities'][node_id])
assert isinstance(workflow, Workflow)
workflow.data.datastore_id = self._resolve_and_save_datastore(workflow.data.datastore_id, entity_map, actual_entities_by_node_id, actual_entities_by_unsaved_id)
workflow = self._workflow_service.save_workflow(workflow, commit=False, flush=True)
actual_entities_by_node_id[node_id] = workflow
actual_entities_by_unsaved_id[unsaved_id] = workflow
return workflow.id
def _resolve_and_save_workflow(self, entity_id, entity_map, actual_entities_by_node_id, actual_entities_by_unsaved_id):
actual_id, unsaved_id = self._resolve(EntityType.workflow, entity_id, entity_map, actual_entities_by_unsaved_id)
if actual_id:
return actual_id
node_id = self._node_id(EntityType.workflow, unsaved_id)
workflow = Workflow.from_dict(entity_map['unsaved_entities'][node_id])
assert isinstance(workflow, Workflow)
workflow.data.datastore_id = self._resolve_and_save_datastore(workflow.data.datastore_id, entity_map, actual_entities_by_node_id, actual_entities_by_unsaved_id)
workflow = self._workflow_service.save_workflow(workflow, commit=False, flush=True)
actual_entities_by_node_id[node_id] = workflow
actual_entities_by_unsaved_id[unsaved_id] = workflow
return workflow.id
def put_workflow(workflow):
""" :type workflow: dart.model.workflow.Workflow """
updated_workflow = Workflow.from_dict(request.get_json())
if updated_workflow.data.state not in [
WorkflowState.ACTIVE, WorkflowState.INACTIVE
]:
return {
'results': 'ERROR',
'error_message': 'state must be ACTIVE or INACTIVE'
}, 400, None
workflow = workflow_service().update_workflow_state(
workflow, updated_workflow.data.state)
return {'results': workflow.to_dict()}
def setUp(self):
self.dart = Dart(host='localhost', port=5000)
args = {'action_sleep_time_in_seconds': 0}
dst = Datastore(data=DatastoreData(name='test-datastore',
engine_name='no_op_engine',
args=args,
state=DatastoreState.TEMPLATE))
self.datastore = self.dart.save_datastore(dst)
wf = Workflow(data=WorkflowData(name='test-workflow',
datastore_id=self.datastore.id))
self.workflow = self.dart.save_workflow(workflow=wf,
datastore_id=self.datastore.id)
self.maxDiff = 99999
def setUp(self):
dart = Dart(host='localhost', port=5000)
""" :type dart: dart.client.python.dart_client.Dart """
self.dart = dart
cs = [
Column('c1', DataType.VARCHAR, 50),
Column('c2', DataType.BIGINT)
]
df = DataFormat(FileFormat.TEXTFILE, RowFormat.DELIMITED)
dataset_data = DatasetData(
name='test-dataset',
table_name='test_dataset_table',
load_type=LoadType.INSERT,
location=('s3://' + os.environ['DART_TEST_BUCKET'] + '/impala'),
data_format=df,
columns=cs,
tags=[])
self.dataset = self.dart.save_dataset(Dataset(data=dataset_data))
start = 's3://' + os.environ['DART_TEST_BUCKET'] + '/impala/impala'
end = 's3://' + os.environ['DART_TEST_BUCKET'] + '/impala/install'
regex = '.*\\.rpm'
ds = Subscription(data=SubscriptionData(
'test-subscription', self.dataset.id, start, end, regex))
self.subscription = self.dart.save_subscription(ds)
dst_args = {'action_sleep_time_in_seconds': 0}
dst = Datastore(data=DatastoreData('test-datastore',
'no_op_engine',
args=dst_args,
state=DatastoreState.TEMPLATE))
self.datastore = self.dart.save_datastore(dst)
wf = Workflow(data=WorkflowData(
'test-workflow', self.datastore.id, state=WorkflowState.ACTIVE))
self.workflow = self.dart.save_workflow(wf, self.datastore.id)
a_args = {'subscription_id': self.subscription.id}
a0 = Action(data=ActionData(NoOpActionTypes.action_that_succeeds.name,
NoOpActionTypes.action_that_succeeds.name,
state=ActionState.TEMPLATE))
a1 = Action(data=ActionData(NoOpActionTypes.consume_subscription.name,
NoOpActionTypes.consume_subscription.name,
a_args,
state=ActionState.TEMPLATE))
self.action0, self.action1 = self.dart.save_actions(
[a0, a1], workflow_id=self.workflow.id)
def setUp(self):
dart = Dart(host='localhost', port=5000)
""" :type dart: dart.client.python.dart_client.Dart """
self.dart = dart
dst_args = {'action_sleep_time_in_seconds': 0}
dst = Datastore(data=DatastoreData('test-datastore',
'no_op_engine',
args=dst_args,
state=DatastoreState.ACTIVE))
self.datastore = self.dart.save_datastore(dst)
wf = Workflow(data=WorkflowData(
'test-workflow', self.datastore.id, state=WorkflowState.ACTIVE))
self.workflow = self.dart.save_workflow(wf, self.datastore.id)
a = Action(data=ActionData(NoOpActionTypes.action_that_fails.name,
NoOpActionTypes.action_that_fails.name,
state=ActionState.TEMPLATE))
self.dart.save_actions([a], workflow_id=self.workflow.id)
def add_no_op_engine_sub_graphs(config):
engine_config = config['engines']['no_op_engine']
opts = engine_config['options']
dart = Dart(opts['dart_host'], opts['dart_port'], opts['dart_api_version'])
assert isinstance(dart, Dart)
_logger.info('saving no_op_engine sub_graphs')
engine_id = None
for e in dart.get_engines():
if e.data.name == 'no_op_engine':
engine_id = e.id
if not engine_id:
raise
subgraph_definitions = [
SubGraphDefinition(data=SubGraphDefinitionData(
name='workflow chaining demo',
description='demonstrate workflow chaining',
engine_name='no_op_engine',
related_type=EntityType.datastore,
related_is_a=Relationship.PARENT,
workflows=[
Workflow(id=Ref.workflow(1),
data=WorkflowData(
name='no-op-workflow-chaining-wf1',
datastore_id=Ref.parent(),
engine_name='no_op_engine',
state=WorkflowState.ACTIVE,
)),
Workflow(id=Ref.workflow(2),
data=WorkflowData(
name='no-op-workflow-chaining-wf2',
datastore_id=Ref.parent(),
engine_name='no_op_engine',
state=WorkflowState.ACTIVE,
)),
],
actions=[
Action(id=Ref.action(1),
data=ActionData(
name=NoOpActionTypes.action_that_succeeds.name,
engine_name='no_op_engine',
action_type_name=NoOpActionTypes.
action_that_succeeds.name,
workflow_id=Ref.workflow(1),
order_idx=1,
state=ActionState.TEMPLATE,
)),
Action(id=Ref.action(2),
data=ActionData(
name=NoOpActionTypes.action_that_succeeds.name,
action_type_name=NoOpActionTypes.
action_that_succeeds.name,
engine_name='no_op_engine',
workflow_id=Ref.workflow(1),
order_idx=2,
state=ActionState.TEMPLATE,
)),
Action(id=Ref.action(3),
data=ActionData(
name=NoOpActionTypes.action_that_succeeds.name,
action_type_name=NoOpActionTypes.
action_that_succeeds.name,
engine_name='no_op_engine',
workflow_id=Ref.workflow(1),
order_idx=3,
state=ActionState.TEMPLATE,
)),
Action(id=Ref.action(4),
data=ActionData(
name=NoOpActionTypes.action_that_succeeds.name,
action_type_name=NoOpActionTypes.
action_that_succeeds.name,
engine_name='no_op_engine',
workflow_id=Ref.workflow(1),
order_idx=4,
state=ActionState.TEMPLATE,
)),
Action(id=Ref.action(5),
data=ActionData(
name=NoOpActionTypes.action_that_succeeds.name,
action_type_name=NoOpActionTypes.
action_that_succeeds.name,
engine_name='no_op_engine',
workflow_id=Ref.workflow(2),
order_idx=1,
state=ActionState.TEMPLATE,
)),
Action(id=Ref.action(6),
data=ActionData(
name=NoOpActionTypes.action_that_succeeds.name,
action_type_name=NoOpActionTypes.
action_that_succeeds.name,
engine_name='no_op_engine',
workflow_id=Ref.workflow(2),
order_idx=2,
state=ActionState.TEMPLATE,
)),
Action(id=Ref.action(7),
data=ActionData(
name=NoOpActionTypes.action_that_fails.name,
action_type_name=NoOpActionTypes.action_that_fails.
name,
engine_name='no_op_engine',
workflow_id=Ref.workflow(2),
order_idx=3,
state=ActionState.TEMPLATE,
)),
],
triggers=[
Trigger(id=Ref.trigger(1),
data=TriggerData(
name='no-op-trigger-workflow-completion',
trigger_type_name=workflow_completion_trigger.name,
workflow_ids=[Ref.workflow(2)],
state=TriggerState.ACTIVE,
args={'completed_workflow_id': Ref.workflow(1)})),
],
))
]
for e in subgraph_definitions:
s = dart.save_subgraph_definition(e, engine_id)
_logger.info('created subgraph_definition: %s' % s.id)
def _get_static_subgraphs_by_related_type(engine, graph_entity_service):
engine_name = engine.data.name
sub_graph_map = {EntityType.workflow: []}
for action_type in engine.data.supported_action_types:
entity_models = graph_entity_service.to_entity_models_with_randomized_ids(
[
Action(
id=Ref.action(1),
data=ActionData(
name=action_type.name,
action_type_name=action_type.name,
engine_name=engine_name,
workflow_id=Ref.parent(),
state=ActionState.TEMPLATE,
args={} if action_type.params_json_schema else None))
])
sub_graph_map[EntityType.workflow].append(
SubGraph(
name=action_type.name,
description=action_type.description,
related_type=EntityType.workflow,
related_is_a=Relationship.PARENT,
graph=graph_entity_service.to_graph(None, entity_models),
entity_map=graph_entity_service.to_entity_map(entity_models),
icon='●',
))
entity_models = graph_entity_service.to_entity_models_with_randomized_ids([
Trigger(id=Ref.trigger(1),
data=TriggerData(
name='%s_trigger' % workflow_completion_trigger.name,
trigger_type_name=workflow_completion_trigger.name,
state=TriggerState.INACTIVE,
workflow_ids=[],
args={'completed_workflow_id': Ref.parent()}))
])
sub_graph_map[EntityType.workflow].extend([
SubGraph(
name='workflow completion trigger',
description='create a new workflow_completion trigger entity',
related_type=EntityType.workflow,
related_is_a=Relationship.PARENT,
graph=graph_entity_service.to_graph(None, entity_models),
entity_map=graph_entity_service.to_entity_map(entity_models),
icon='▼',
),
])
entity_models = graph_entity_service.to_entity_models_with_randomized_ids([
Trigger(id=Ref.trigger(1),
data=TriggerData(
name='%s_trigger' % scheduled_trigger.name,
trigger_type_name=scheduled_trigger.name,
state=TriggerState.INACTIVE,
workflow_ids=[Ref.child()],
))
])
sub_graph_map[EntityType.workflow].extend([
SubGraph(
name='scheduled trigger',
description='create a new scheduled trigger entity',
related_type=EntityType.workflow,
related_is_a=Relationship.CHILD,
graph=graph_entity_service.to_graph(None, entity_models),
entity_map=graph_entity_service.to_entity_map(entity_models),
icon='▼',
),
])
entity_models = graph_entity_service.to_entity_models_with_randomized_ids([
Trigger(id=Ref.trigger(1),
data=TriggerData(
name='%s_trigger' % super_trigger.name,
trigger_type_name=super_trigger.name,
state=TriggerState.INACTIVE,
workflow_ids=[Ref.child()],
))
])
sub_graph_map[EntityType.workflow].extend([
SubGraph(
name='super trigger',
description='create a new super trigger entity',
related_type=EntityType.workflow,
related_is_a=Relationship.CHILD,
graph=graph_entity_service.to_graph(None, entity_models),
entity_map=graph_entity_service.to_entity_map(entity_models),
icon='▼',
),
])
entity_models = graph_entity_service.to_entity_models_with_randomized_ids([
Workflow(id=Ref.workflow(1),
data=WorkflowData(name='workflow',
datastore_id=Ref.parent(),
engine_name=engine_name,
state=WorkflowState.INACTIVE))
])
sub_graph_map[EntityType.datastore] = [
SubGraph(
name='workflow',
description='create a new workflow entity',
related_type=EntityType.datastore,
related_is_a=Relationship.PARENT,
graph=graph_entity_service.to_graph(None, entity_models),
entity_map=graph_entity_service.to_entity_map(entity_models),
icon='◆',
)
]
for action_type in engine.data.supported_action_types:
entity_models = graph_entity_service.to_entity_models_with_randomized_ids(
[
Action(
id=Ref.action(1),
data=ActionData(
name=action_type.name,
action_type_name=action_type.name,
engine_name=engine_name,
datastore_id=Ref.parent(),
state=ActionState.HAS_NEVER_RUN,
args={} if action_type.params_json_schema else None))
])
sub_graph_map[EntityType.datastore].append(
SubGraph(
name=action_type.name,
description=action_type.description,
related_type=EntityType.datastore,
related_is_a=Relationship.PARENT,
graph=graph_entity_service.to_graph(None, entity_models),
entity_map=graph_entity_service.to_entity_map(entity_models),
icon='●',
))
return sub_graph_map
datastore = dart.save_datastore(Datastore(
data=DatastoreData(
'beacon_native_app_impala',
'emr_engine',
state=DatastoreState.TEMPLATE,
args={'data_to_freespace_ratio': 0.25}
)
))
print 'created datastore: %s' % datastore.id
workflow = dart.save_workflow(Workflow(
data=WorkflowData(
'load_beacon_native_app_impala',
datastore.id,
state=WorkflowState.ACTIVE,
on_failure_email=['*****@*****.**'],
on_success_email=['*****@*****.**'],
on_started_email=['*****@*****.**'],
)
), datastore.id)
print 'created workflow: %s' % workflow.id
a0, a1 = dart.save_actions([
Action(data=ActionData('start_datastore', 'start_datastore', state=ActionState.TEMPLATE)),
Action(data=ActionData('load_dataset', 'load_dataset', state=ActionState.TEMPLATE, args={
'dataset_id': dataset.id,
's3_path_start_prefix_inclusive': 's3://example-bucket/prd/beacon/native_app/v2/parquet/snappy/createdpartition=2015-06-27',
})),
], workflow_id=workflow.id)
print 'created action: %s' % a0.id
print 'created action: %s' % a1.id
'dataset_id': dataset.id,
's3_path_end_prefix_exclusive':
's3://example-bucket/prd/inbound/overlord/eu-all-events/2015/08/05/',
'target_file_format': FileFormat.PARQUET,
'target_row_format': RowFormat.NONE,
'target_compression': Compression.SNAPPY,
})),
],
datastore_id=datastore.id)
print 'created action: %s' % a0.id
print 'created action: %s' % a1.id
workflow = dart.save_workflow(workflow=Workflow(data=WorkflowData(
name='owen_eu_parquet_workflow_DW-3213_v3',
datastore_id=datastore.id,
state=WorkflowState.ACTIVE,
on_failure_email=['*****@*****.**'],
on_success_email=['*****@*****.**'],
on_started_email=['*****@*****.**'],
)),
datastore_id=datastore.id)
print 'created workflow: %s' % workflow.id
a2 = dart.save_actions(actions=[
Action(data=ActionData('consume_subscription',
'consume_subscription',
state=ActionState.TEMPLATE,
args={
'subscription_id': subscription.id,
'target_file_format': FileFormat.PARQUET,
'target_row_format': RowFormat.NONE,
'target_compression': Compression.SNAPPY,
def put_workflow(workflow):
""" :type workflow: dart.model.workflow.Workflow """
return update_workflow(workflow, Workflow.from_dict(request.get_json()))
'dataset_id': '34HWJLF5N9',
's3_path_end_prefix_exclusive':
's3://example-bucket/prd/inbound/overlord/raw/rmndirect/2015/08/18/',
'target_file_format': FileFormat.PARQUET,
'target_row_format': RowFormat.NONE,
'target_compression': Compression.SNAPPY,
})),
],
datastore_id=datastore.id)
print 'created action: %s' % actions[0].id
print 'created action: %s' % actions[1].id
workflow = dart.save_workflow(workflow=Workflow(data=WorkflowData(
name='rmn_direct_workflow_DW-3307',
datastore_id=datastore.id,
state=WorkflowState.ACTIVE,
on_failure_email=['*****@*****.**'],
on_success_email=['*****@*****.**'],
on_started_email=['*****@*****.**'],
)),
datastore_id=datastore.id)
print 'created workflow: %s' % workflow.id
wf_actions = dart.save_actions(actions=[
Action(data=ActionData('consume_subscription',
'consume_subscription',
state=ActionState.TEMPLATE,
args={
'subscription_id': subscription.id,
'target_file_format': FileFormat.PARQUET,
'target_row_format': RowFormat.NONE,
'target_compression': Compression.SNAPPY,
def patch_workflow(workflow):
""" :type workflow: dart.model.workflow.Workflow """
p = JsonPatch(request.get_json())
return update_workflow(workflow,
Workflow.from_dict(p.apply(workflow.to_dict())))
def put_workflow(workflow):
""" :type workflow: dart.model.workflow.Workflow """
return update_workflow(workflow, Workflow.from_dict(request.get_json()))
def patch_workflow(workflow):
""" :type workflow: dart.model.workflow.Workflow """
p = JsonPatch(request.get_json())
return update_workflow(workflow, Workflow.from_dict(p.apply(workflow.to_dict())))
def setUp(self):
dart = Dart(host='localhost', port=5000)
""" :type dart: dart.client.python.dart_client.Dart """
self.dart = dart
cs = [Column('c1', DataType.VARCHAR, 50), Column('c2', DataType.BIGINT)]
df = DataFormat(FileFormat.PARQUET, RowFormat.NONE)
dataset_data = DatasetData('test-dataset0', 'test_dataset_table0', 's3://test/dataset/0/%s' + random_id(), df, cs)
self.dataset0 = self.dart.save_dataset(Dataset(data=dataset_data))
cs = [Column('c1', DataType.VARCHAR, 50), Column('c2', DataType.BIGINT)]
df = DataFormat(FileFormat.PARQUET, RowFormat.NONE)
dataset1_location = 's3://test/dataset/1/%s' + random_id()
dataset_data = DatasetData('test-dataset1', 'test_dataset_table1', dataset1_location, df, cs)
self.dataset1 = self.dart.save_dataset(Dataset(data=dataset_data))
cs = [Column('c1', DataType.VARCHAR, 50), Column('c2', DataType.BIGINT)]
df = DataFormat(FileFormat.PARQUET, RowFormat.NONE)
dataset_data = DatasetData('test-dataset2-no-show', 'test_dataset_table2', 's3://test/dataset/2/%s' + random_id(), df, cs)
self.dataset2 = self.dart.save_dataset(Dataset(data=dataset_data))
s = Subscription(data=SubscriptionData('test-subscription0', self.dataset0.id))
self.subscription0 = self.dart.save_subscription(s)
s = Subscription(data=SubscriptionData('test-subscription2-no-show', self.dataset2.id))
self.subscription2 = self.dart.save_subscription(s)
dst_args = {'action_sleep_time_in_seconds': 0}
dst = Datastore(data=DatastoreData('test-datastore0', 'no_op_engine', args=dst_args, state=DatastoreState.TEMPLATE))
self.datastore0 = self.dart.save_datastore(dst)
dst = Datastore(data=DatastoreData('test-datastore1', 'no_op_engine', args=dst_args, state=DatastoreState.TEMPLATE))
self.datastore1 = self.dart.save_datastore(dst)
dst = Datastore(data=DatastoreData('test-datastore2-no-show', 'no_op_engine', args=dst_args, state=DatastoreState.ACTIVE))
self.datastore2 = self.dart.save_datastore(dst)
wf0 = Workflow(data=WorkflowData('test-workflow0', self.datastore0.id, state=WorkflowState.ACTIVE))
self.workflow0 = self.dart.save_workflow(wf0, self.datastore0.id)
wf1 = Workflow(data=WorkflowData('test-workflow1', self.datastore1.id, state=WorkflowState.ACTIVE))
self.workflow1 = self.dart.save_workflow(wf1, self.datastore1.id)
wf2 = Workflow(data=WorkflowData('test-workflow2-no-show', self.datastore2.id, state=WorkflowState.ACTIVE))
self.workflow2 = self.dart.save_workflow(wf2, self.datastore2.id)
a_args = {'source_hdfs_path': 'hdfs:///user/hive/warehouse/test', 'destination_s3_path': dataset1_location}
a00 = Action(data=ActionData(NoOpActionTypes.action_that_succeeds.name, NoOpActionTypes.action_that_succeeds.name, state=ActionState.TEMPLATE))
a01 = Action(data=ActionData(NoOpActionTypes.consume_subscription.name, NoOpActionTypes.consume_subscription.name, {'subscription_id': self.subscription0.id}, state=ActionState.TEMPLATE))
a02 = Action(data=ActionData(NoOpActionTypes.action_that_succeeds.name, NoOpActionTypes.action_that_succeeds.name, state=ActionState.TEMPLATE))
a03 = Action(data=ActionData(NoOpActionTypes.copy_hdfs_to_s3_action.name, NoOpActionTypes.copy_hdfs_to_s3_action.name, a_args, state=ActionState.TEMPLATE))
a04 = Action(data=ActionData(NoOpActionTypes.action_that_succeeds.name, NoOpActionTypes.action_that_succeeds.name, state=ActionState.TEMPLATE))
self.action00, self.action01, self.action02, self.action03, self.action04 = \
self.dart.save_actions([a00, a01, a02, a03, a04], workflow_id=self.workflow0.id)
a10 = Action(data=ActionData(NoOpActionTypes.load_dataset.name, NoOpActionTypes.load_dataset.name, {'dataset_id': self.dataset1.id}, state=ActionState.TEMPLATE))
self.action10 = self.dart.save_actions([a10], workflow_id=self.workflow1.id)
a20 = Action(data=ActionData(NoOpActionTypes.action_that_succeeds.name, NoOpActionTypes.action_that_succeeds.name, state=ActionState.HAS_NEVER_RUN))
a21 = Action(data=ActionData(NoOpActionTypes.load_dataset.name, NoOpActionTypes.load_dataset.name, {'dataset_id': self.dataset2.id}, state=ActionState.TEMPLATE))
self.action20 = self.dart.save_actions([a20], datastore_id=self.datastore2.id)
self.action21 = self.dart.save_actions([a21], workflow_id=self.workflow2.id)
self.event1 = self.dart.save_event(Event(data=EventData('test-event1', state=EventState.ACTIVE)))
self.event2 = self.dart.save_event(Event(data=EventData('test-event2-no-show', state=EventState.ACTIVE)))
tr_args = {'event_id': self.event1.id}
tr = Trigger(data=TriggerData('test-event-trigger1', 'event', [self.workflow1.id], tr_args, TriggerState.ACTIVE))
self.event_trigger1 = self.dart.save_trigger(tr)
tr_args = {'event_id': self.event2.id}
tr = Trigger(data=TriggerData('test-event-trigger2-no-show', 'event', [self.workflow2.id], tr_args, TriggerState.ACTIVE))
self.event_trigger2 = self.dart.save_trigger(tr)
st_args = {'fire_after': 'ALL', 'completed_trigger_ids': [self.event_trigger1.id]}
st = Trigger(data=TriggerData('test-super-trigger1', 'super', None, st_args, TriggerState.ACTIVE))
self.super_trigger1 = self.dart.save_trigger(st)
st_args = {'fire_after': 'ANY', 'completed_trigger_ids': [self.super_trigger1.id]}
st = Trigger(data=TriggerData('test-super-trigger2', 'super', [self.workflow1.id], st_args, TriggerState.ACTIVE))
self.super_trigger2 = self.dart.save_trigger(st)
assert subscription.data.state == SubscriptionState.ACTIVE
print 'done.'
datastore = dart.save_datastore(
Datastore(data=DatastoreData(name='weblogs_rmn_legacy',
engine_name='emr_engine',
state=DatastoreState.TEMPLATE,
args={
'data_to_freespace_ratio': 0.50,
})))
print 'created datastore: %s' % datastore.id
workflow = dart.save_workflow(workflow=Workflow(data=WorkflowData(
name='weblogs_rmn_legacy_parse_to_delimited',
datastore_id=datastore.id,
state=WorkflowState.ACTIVE,
on_failure_email=['*****@*****.**', '*****@*****.**'],
on_success_email=['*****@*****.**', '*****@*****.**'],
on_started_email=['*****@*****.**', '*****@*****.**'],
)),
datastore_id=datastore.id)
print 'created workflow: %s' % workflow.id
a2 = dart.save_actions(actions=[
Action(data=ActionData('consume_subscription',
'consume_subscription',
state=ActionState.TEMPLATE,
args={
'subscription_id': subscription.id,
'target_file_format': FileFormat.TEXTFILE,
'target_row_format': RowFormat.DELIMITED,
'target_compression': Compression.GZIP,
