def test_workflow_events(self):
wf = Workflow(self.gbdx)
workflow_id = '4347109104758907277'
events = wf.events(workflow_id)
assert len(events) > 0
assert isinstance(events, list)
for event in events:
assert 'task' in event.keys()
assert 'state' in event.keys()
assert 'event' in event.keys()
assert 'timestamp' in event.keys()
assert 'when' in event.keys()
assert 'note' in event.keys()
assert event['state'] in ['pending','running','complete']
assert event['event'] in ['submitted','scheduled','rescheduling','started','succeeded','failed','timedout']
def test_workflow_events():
wf = Workflow(gbdx)
workflow_id = '4347109104758907277'
events = wf.events(workflow_id)
assert len(events) > 0
assert isinstance(events, list)
for event in events:
assert 'task' in event.keys()
assert 'state' in event.keys()
assert 'event' in event.keys()
assert 'timestamp' in event.keys()
assert 'when' in event.keys()
assert 'note' in event.keys()
assert event['state'] in ['pending','running','complete']
assert event['event'] in ['submitted','scheduled','rescheduling','started','succeeded','failed','timedout']
class Workflow(object):
def __init__(self, tasks, **kwargs):
self.workflow = WF()
self.name = kwargs.get('name', str(uuid.uuid4())[:8] )
self.id = None
self.callback = kwargs.get('callback', None )
self.definition = None
self.tasks = tasks
batch_values = []
for task in self.tasks:
if task.batch_values:
batch_values.extend(task.batch_values)
if batch_values:
self.batch_values = batch_values
else:
self.batch_values = None
def savedata(self, output, location=None):
'''
Save output data from any task in this workflow to S3
Args:
output: Reference task output (e.g. task.inputs.output1).
location (optional): Subfolder under which the output will be saved.
It will be placed under the account directory in gbd-customer-data bucket:
s3://gbd-customer-data/{account_id}/{location}
Leave blank to save to: workflow_output/{workflow_id}/{task_name}/{port_name}
Returns:
None
'''
output.persist = True
if location:
output.persist_location = location
def workflow_skeleton(self):
return {
"tasks": [],
"name": self.name
}
def list_workflow_outputs(self):
'''
Get a list of outputs from the workflow that are saved to S3. To get resolved locations call workflow status.
Args:
None
Returns:
list
'''
workflow_outputs = []
for task in self.tasks:
for output_port_name in task.outputs._portnames:
if task.outputs.__getattribute__(output_port_name).persist:
workflow_outputs.append(task.name + ':' + output_port_name)
return workflow_outputs
def generate_workflow_description(self):
'''
Generate workflow json for launching the workflow against the gbdx api
Args:
None
Returns:
json string
'''
if not self.tasks:
raise WorkflowError('Workflow contains no tasks, and cannot be executed.')
self.definition = self.workflow_skeleton()
if self.batch_values:
self.definition["batch_values"] = self.batch_values
all_input_port_values = [t.inputs.__getattribute__(input_port_name).value for t in self.tasks for
input_port_name in t.inputs._portnames]
for task in self.tasks:
# only include multiplex output ports in this task if other tasks refer to them in their inputs.
# 1. find the multplex output port_names in this task
# 2. see if they are referred to in any other tasks inputs
# 3. If not, exclude them from the workflow_def
output_multiplex_ports_to_exclude = []
multiplex_output_port_names = [portname for portname in task.outputs._portnames if
task.outputs.__getattribute__(portname).is_multiplex]
for p in multiplex_output_port_names:
output_port_reference = 'source:' + task.name + ':' + p
if output_port_reference not in all_input_port_values:
output_multiplex_ports_to_exclude.append(p)
task_def = task.generate_task_workflow_json(
output_multiplex_ports_to_exclude=output_multiplex_ports_to_exclude)
self.definition['tasks'].append(task_def)
if self.callback:
self.definition['callback'] = self.callback
return self.definition
def execute(self):
'''
Execute the workflow.
Args:
None
Returns:
Workflow_id
'''
# if not self.tasks:
# raise WorkflowError('Workflow contains no tasks, and cannot be executed.')
# for task in self.tasks:
# self.definition['tasks'].append( task.generate_task_workflow_json() )
self.generate_workflow_description()
# hit batch workflow endpoint if batch values
if self.batch_values:
self.id = self.workflow.launch_batch_workflow(self.definition)
# use regular workflow endpoint if no batch values
else:
self.id = self.workflow.launch(self.definition)
return self.id
@property
def task_ids(self):
'''
Get the task IDs of a running workflow
Args:
None
Returns:
List of task IDs
'''
if not self.id:
raise WorkflowError('Workflow is not running. Cannot get task IDs.')
if self.batch_values:
raise NotImplementedError("Query Each Workflow Id within the Batch Workflow for task IDs.")
wf = self.workflow.get(self.id)
return [task['id'] for task in wf['tasks']]
@task_ids.setter
def task_ids(self, value):
raise NotImplementedError("Cannot set workflow task_ids, readonly.")
def cancel(self):
'''
Cancel a running workflow.
Args:
None
Returns:
None
'''
if not self.id:
raise WorkflowError('Workflow is not running. Cannot cancel.')
if self.batch_values:
self.workflow.batch_workflow_cancel(self.id)
else:
self.workflow.cancel(self.id)
@property
def status(self):
if not self.id:
raise WorkflowError('Workflow is not running. Cannot check status.')
if self.batch_values:
status = self.workflow.batch_workflow_status(self.id)
else:
status = self.workflow.status(self.id)
return status
@status.setter
def status(self, value):
raise NotImplementedError("Cannot set workflow status, readonly.")
@property
def events(self):
if not self.id:
raise WorkflowError('Workflow is not running. Cannot check status.')
if self.batch_values:
raise NotImplementedError("Query Each Workflow Id within the Batch Workflow for Events")
return self.workflow.events(self.id)
@events.setter
def events(self, value):
raise NotImplementedError("Cannot set workflow events, readonly.")
@property
def complete(self):
if not self.id:
return False
# check if all sub workflows are either done, failed, or timedout
if self.batch_values:
return all(workflow.get("state") in ["succeeded", "failed", "timedout"] for workflow in
self.status['workflows'])
else:
return self.status['state'] == 'complete'
@complete.setter
def complete(self, value):
raise NotImplementedError("Cannot set workflow complete, readonly.")
@property
def failed(self):
if not self.id:
return False
if self.batch_values:
raise NotImplementedError("Query Each Workflow Id within the Batch Workflow for Current State")
status = self.status
return status['state'] == 'complete' and status['event'] == 'failed'
@failed.setter
def failed(self, value):
raise NotImplementedError("Cannot set workflow failed, readonly.")
@property
def canceled(self):
if not self.id:
return False
if self.batch_values:
raise NotImplementedError("Query Each Workflow Id within the Batch Workflow for Current State")
status = self.status
return status['state'] == 'complete' and status['event'] == 'canceled'
@canceled.setter
def canceled(self, value):
raise NotImplementedError("Cannot set workflow canceled, readonly.")
@property
def succeeded(self):
if not self.id:
return False
# check if all sub workflows are succeeded
if self.batch_values:
return all(workflow.get("state") == "succeeded" for workflow in self.status['workflows'])
status = self.status
return status['state'] == 'complete' and status['event'] == 'succeeded'
@succeeded.setter
def succeeded(self, value):
raise NotImplementedError("Cannot set workflow succeeded, readonly.")
@property
def running(self):
if not self.id:
return False
if self.batch_values:
# check if any sub workflows are running
return any(workflow.get("state") not in ["succeeded", "failed", "timedout"] for workflow in
self.status['workflows'])
status = self.status
return status['state'] == 'running' and status['event'] == 'started'
@running.setter
def running(self, value):
raise NotImplementedError("Cannot set workflow running, readonly.")
@property
def timedout(self):
if not self.id:
return False
if self.batch_values:
raise NotImplementedError("Query Each Workflow Id within the Batch Workflow for Current State")
status = self.status
return status['state'] == 'complete' and status['event'] == 'timedout'
@timedout.setter
def timedout(self, value):
raise NotImplementedError("Cannot set workflow timedout, readonly.")
@property
def stdout(self):
'''
Get stdout from all the tasks of a workflow.
Args:
None
Returns:
List of tasks with their stdout, formatted like this:
[
{
"id": "4488895771403082552",
"taskType": "AOP_Strip_Processor",
"name": "Task1",
"stdout": "............"
}
]
'''
if not self.id:
raise WorkflowError('Workflow is not running. Cannot get stdout.')
if self.batch_values:
raise NotImplementedError("Query Each Workflow Id within the Batch Workflow for stdout.")
wf = self.workflow.get(self.id)
stdout_list = []
for task in wf['tasks']:
stdout_list.append(
{
'id': task['id'],
'taskType': task['taskType'],
'name': task['name'],
'stdout': self.workflow.get_stdout(self.id, task['id'])
}
)
return stdout_list
@stdout.setter
def stdout(self, value):
raise NotImplementedError("Cannot set workflow stdout, readonly.")
@property
def stderr(self):
'''
Get stderr from all the tasks of a workflow.
Args:
None
Returns:
List of tasks with their stderr, formatted like this:
[
{
"id": "4488895771403082552",
"taskType": "AOP_Strip_Processor",
"name": "Task1",
"stderr": "............"
}
]
'''
if not self.id:
raise WorkflowError('Workflow is not running. Cannot get stderr.')
if self.batch_values:
raise NotImplementedError("Query Each Workflow Id within the Batch Workflow for stderr.")
wf = self.workflow.get(self.id)
stderr_list = []
for task in wf['tasks']:
stderr_list.append(
{
'id': task['id'],
'taskType': task['taskType'],
'name': task['name'],
'stderr': self.workflow.get_stderr(self.id, task['id'])
}
)
return stderr_list
@stderr.setter
def stderr(self, value):
raise NotImplementedError("Cannot set workflow stderr, readonly.")