def testTaleImportBinderFail(self):
with mock.patch("girder.plugins.wholetale.lib.pids_to_entities") as mock_pids:
mock_pids.side_effect = ValueError
resp = self.request(
path="/tale/import",
method="POST",
user=self.user,
params={
"url": "http://use.yt/upload/ef4cd901",
"spawn": False,
"imageId": self.image["_id"],
"asTale": True,
"taleKwargs": json.dumps({"title": "tale should fail"}),
},
)
self.assertStatusOk(resp)
tale = resp.json
job = Job().findOne({"type": "wholetale.import_binder"})
self.assertEqual(
json.loads(job["kwargs"])["taleId"]["$oid"], tale["_id"]
)
for i in range(300):
if job["status"] in {JobStatus.SUCCESS, JobStatus.ERROR}:
break
time.sleep(0.1)
job = Job().load(job["_id"], force=True)
self.assertEqual(job["status"], JobStatus.ERROR)
Job().remove(job)
tale = Tale().load(tale["_id"], force=True)
self.assertEqual(tale["status"], TaleStatus.ERROR)
Tale().remove(tale)
def _onUpload(event):
"""
Look at uploads containing references related to this plugin. If found,
they are used to link item task outputs back to a job document.
"""
try:
ref = json.loads(event.info.get('reference', ''))
except ValueError:
return
if isinstance(ref, dict) and ref.get('type') == 'item_tasks.output':
jobModel = Job()
tokenModel = Token()
token = event.info['currentToken']
if tokenModel.hasScope(token, 'item_tasks.job_write:%s' % ref['jobId']):
job = jobModel.load(ref['jobId'], force=True, exc=True)
else:
job = jobModel.load(
ref['jobId'], level=AccessType.WRITE, user=event.info['currentUser'], exc=True)
file = event.info['file']
item = Item().load(file['itemId'], force=True)
# Add link to job model to the output item
jobModel.updateJob(job, otherFields={
'itemTaskBindings.outputs.%s.itemId' % ref['id']: item['_id']
})
# Also a link in the item to the job that created it
item['createdByJob'] = job['_id']
Item().save(item)
def scheduleThumbnailJob(file,
attachToType,
attachToId,
user,
width=0,
height=0,
crop=True):
"""
Schedule a local thumbnail creation job and return it.
"""
job = Job().createLocalJob(title='Generate thumbnail for %s' %
file['name'],
user=user,
type='thumbnails.create',
public=False,
module='girder.plugins.thumbnails.worker',
kwargs={
'fileId': str(file['_id']),
'width': width,
'height': height,
'crop': crop,
'attachToType': attachToType,
'attachToId': str(attachToId)
})
Job().scheduleJob(job)
return job
def testWorkerStatusEndpoint(self):
# Create a job to be handled by the worker plugin
from girder.plugins.jobs.models.job import Job
job = Job().createJob(title='title',
type='foo',
handler='worker_handler',
user=self.admin,
public=False,
args=(),
kwargs={})
job['kwargs'] = {
'jobInfo':
utils.jobInfoSpec(job),
'inputs':
[utils.girderInputSpec(self.adminFolder, resourceType='folder')],
'outputs':
[utils.girderOutputSpec(self.adminFolder, token=self.adminToken)]
}
job = Job().save(job)
self.assertEqual(job['status'], JobStatus.INACTIVE)
# Schedule the job
with mock.patch('celery.Celery') as celeryMock:
instance = celeryMock.return_value
instance.send_task.return_value = FakeAsyncResult()
Job().scheduleJob(job)
# Call the worker status endpoint
resp = self.request('/worker/status', method='GET', user=self.admin)
self.assertStatusOk(resp)
self.assertHasKeys(resp.json,
['report', 'stats', 'ping', 'active', 'reserved'])
def schedule(event):
"""
This is bound to the "jobs.schedule" event, and will be triggered any time
a job is scheduled. This handler will process any job that has the
handler field set to "worker_handler".
"""
job = event.info
if job['handler'] == 'worker_handler':
task = job.get('celeryTaskName', 'girder_worker.run')
# Set the job status to queued
Job().updateJob(job, status=JobStatus.QUEUED)
# Send the task to celery
asyncResult = getCeleryApp().send_task(
task, job['args'], job['kwargs'], queue=job.get('celeryQueue'), headers={
'jobInfoSpec': jobInfoSpec(job, job.get('token', None)),
'apiUrl': getWorkerApiUrl()
})
# Record the task ID from celery.
Job().updateJob(job, otherFields={
'celeryTaskId': asyncResult.task_id
})
# Stop event propagation since we have taken care of scheduling.
event.stopPropagation()
def createThumbnails(self, params):
self.requireParams(['spec'], params)
try:
spec = json.loads(params['spec'])
if not isinstance(spec, list):
raise ValueError()
except ValueError:
raise RestException('The spec parameter must be a JSON list.')
maxThumbnailFiles = int(Setting().get(
constants.PluginSettings.LARGE_IMAGE_MAX_THUMBNAIL_FILES))
if maxThumbnailFiles <= 0:
raise RestException('Thumbnail files are not enabled.')
jobKwargs = {'spec': spec}
if params.get('logInterval') is not None:
jobKwargs['logInterval'] = float(params['logInterval'])
job = Job().createLocalJob(
module='girder.plugins.large_image.rest.large_image',
function='createThumbnailsJob',
kwargs=jobKwargs,
title='Create large image thumbnail files.',
type='large_image_create_thumbnails',
user=self.getCurrentUser(),
public=True,
async=True,
)
Job().scheduleJob(job)
return job
def testCopyWorkspaceFail(self):
tale = Tale().createTale(
self.image,
[],
creator=self.admin,
title="tale one",
public=True,
config={"memLimit": "2g"},
)
job = Job().createLocalJob(
title='Copy "{title}" workspace'.format(**tale),
user=self.user,
type="wholetale.copy_workspace",
public=False,
async=True,
module="girder.plugins.wholetale.tasks.copy_workspace",
args=(tale["workspaceId"], "non_existing"),
kwargs={"user": self.user, "tale": tale},
)
Job().scheduleJob(job)
for i in range(300):
if job["status"] in {JobStatus.SUCCESS, JobStatus.ERROR}:
break
time.sleep(0.1)
job = Job().load(job["_id"], force=True)
self.assertEqual(job["status"], JobStatus.ERROR)
Job().remove(job)
tale = Tale().load(tale["_id"], force=True)
self.assertEqual(tale["status"], TaleStatus.ERROR)
Tale().remove(tale)
def testBuildFail(self):
from girder.plugins.jobs.models.job import Job
resp = self.request(path='/instance',
method='POST',
user=self.user,
params={
'taleId': str(self.tale_one['_id']),
'name': 'tale that will fail',
'spawn': False
})
self.assertStatusOk(resp)
instance = resp.json
job = Job().createJob(title='Fake build job',
type='celery',
handler='worker_handler',
user=self.user,
public=False,
args=[str(self.tale_one['_id']), False],
kwargs={},
otherFields={
'wt_notification_id': 'nonexisting',
'instance_id': instance['_id']
})
job = Job().save(job)
self.assertEqual(job['status'], JobStatus.INACTIVE)
Job().updateJob(job, log='job queued', status=JobStatus.QUEUED)
Job().updateJob(job, log='job running', status=JobStatus.RUNNING)
Job().updateJob(job, log='job failed', status=JobStatus.ERROR)
instance = Instance().load(instance['_id'], force=True)
self.assertEqual(instance['status'], InstanceStatus.ERROR)
Instance().remove(instance)
def _waitForJobToBeRunning(self, job):
from girder.plugins.jobs.constants import JobStatus
from girder.plugins.jobs.models.job import Job
job = Job().load(id=job['_id'], force=True)
while job['status'] != JobStatus.RUNNING:
time.sleep(0.01)
job = Job().load(id=job['_id'], force=True)
return job
def execGraph(yaml_graph, username):
# Write YAML graph to a file
#yaml_path = "graph.yaml";f = open(yaml_path,"w");f.write(yaml_graph);f.close()
# TODO: How do we cache/lookup this execution again later? Hash the command?
# Give our job a unique name
job_name = username + "-" + str(
datetime.datetime.now().strftime('%Y%m%d-%H%M%S-%f'))
job_type = 'k8s.io/yggdrasil'
# Specify the Docker image and command(s) to run
docker_image = "cropsinsilico/jupyterlab:latest"
init_command = "mkdir -p /pvc/" + job_name + " && cp -R /pvc/models/* /pvc/" + job_name + " && chown -R 1000:100 /pvc/" + job_name
command = "echo '" + str(
yaml_graph
) + "' > graph.yml && echo Running in $(pwd): && ls -al && yggrun graph.yml"
# Encode our username with Jupyter's special homebrew recipe
username = jupyterUserEncode(username)
# Job must run in same namespace as the PVC
namespace = "hub"
# Specify some arbitrary limits
timeout = 300
num_cpus = 2
max_ram_mb = 8384
# Create a record in the Job database
jobModel = JobModel()
job_model = jobModel.createJob(job_name,
job_type,
async=True,
kwargs={
'name': job_name,
'type': job_type,
'namespace': namespace,
'username': username,
'init_command': init_command,
'command': command,
'image': docker_image,
'timeout': timeout,
'num_cpus': num_cpus,
'max_ram_mb': max_ram_mb,
})
jobModel.save(job_model)
# Create and run the job
k8s_job = KubernetesJob(username, job_name, namespace, timeout,
init_command, command, docker_image, num_cpus,
max_ram_mb)
if not k8s_job.is_running():
jobModel.scheduleJob(job_model)
k8s_job.submit()
return job_name
def setUp(self):
base.TestCase.setUp(self)
self.users = [User().createUser(
'usr' + str(n), 'passwd', 'tst', 'usr', '*****@*****.**' % n)
for n in range(3)]
from girder.plugins.jobs.models.job import Job
self.jobModel = Job()
def testWorkerDifferentTask(self):
# Test the settings
resp = self.request('/system/setting', method='PUT', params={
'key': worker.PluginSettings.API_URL,
'value': 'bad value'
}, user=self.admin)
self.assertStatus(resp, 400)
self.assertEqual(resp.json['message'], 'API URL must start with http:// or https://.')
resp = self.request('/system/setting', method='PUT', params={
'list': json.dumps([{
'key': worker.PluginSettings.BROKER,
'value': 'amqp://[email protected]'
}, {
'key': worker.PluginSettings.BACKEND,
'value': 'amqp://[email protected]'
}])
}, user=self.admin)
self.assertStatusOk(resp)
# Create a job to be handled by the worker plugin
from girder.plugins.jobs.models.job import Job
jobModel = Job()
job = jobModel.createJob(
title='title', type='foo', handler='worker_handler',
user=self.admin, public=False, args=(), kwargs={},
otherFields={
'celeryTaskName': 'some_other.task',
'celeryQueue': 'my_other_q'
})
job['kwargs'] = {
'jobInfo': utils.jobInfoSpec(job),
'inputs': [
utils.girderInputSpec(self.adminFolder, resourceType='folder')
],
'outputs': [
utils.girderOutputSpec(self.adminFolder, token=self.adminToken)
]
}
job = jobModel.save(job)
# Schedule the job, make sure it is sent to celery
with mock.patch('celery.Celery') as celeryMock:
instance = celeryMock.return_value
instance.send_task.return_value = FakeAsyncResult()
jobModel.scheduleJob(job)
sendTaskCalls = celeryMock.return_value.send_task.mock_calls
self.assertEqual(len(sendTaskCalls), 1)
self.assertEqual(sendTaskCalls[0][1], (
'some_other.task', job['args'], job['kwargs']))
self.assertIn('queue', sendTaskCalls[0][2])
self.assertEqual(sendTaskCalls[0][2]['queue'], 'my_other_q')
def _postTileViaHttp(self, itemId, fileId, jobAction=None):
"""
When we know we need to process a job, we have to use an actual http
request rather than the normal simulated request to cherrypy. This is
required because cherrypy needs to know how it was reached so that
girder_worker can reach it when done.
:param itemId: the id of the item with the file to process.
:param fileId: the id of the file that should be processed.
:param jobAction: if 'delete', delete the job immediately.
:returns: metadata from the tile if the conversion was successful,
False if it converted but didn't result in useable tiles, and
None if it failed.
"""
from girder.plugins.jobs.models.job import Job
headers = [('Accept', 'application/json')]
self._buildHeaders(headers, None, self.admin, None, None, None)
headers = {header[0]: header[1] for header in headers}
req = requests.post('http://127.0.0.1:%d/api/v1/item/%s/tiles' %
(int(os.environ['GIRDER_PORT']), itemId),
headers=headers,
data={'fileId': fileId})
self.assertEqual(req.status_code, 200)
# If we ask to create the item again right away, we should be told that
# either there is already a job running or the item has already been
# added
req = requests.post('http://127.0.0.1:%d/api/v1/item/%s/tiles' %
(int(os.environ['GIRDER_PORT']), itemId),
headers=headers,
data={'fileId': fileId})
self.assertEqual(req.status_code, 400)
self.assertTrue('Item already has' in req.json()['message']
or 'Item is scheduled' in req.json()['message'])
if jobAction == 'delete':
Job().remove(Job().find({}, sort=[('_id', SortDir.DESCENDING)])[0])
starttime = time.time()
resp = None
while time.time() - starttime < 30:
try:
resp = self.request(path='/item/%s/tiles' % itemId,
user=self.admin)
self.assertStatusOk(resp)
break
except AssertionError as exc:
if 'didn\'t meet requirements' in exc.args[0]:
return False
if 'No large image file' in exc.args[0]:
return None
self.assertIn('is still pending creation', exc.args[0])
time.sleep(0.1)
self.assertStatusOk(resp)
return resp.json
def testTaleCopy(self):
from girder.plugins.wholetale.models.tale import Tale
from girder.plugins.wholetale.constants import TaleStatus
from girder.plugins.jobs.models.job import Job
from girder.plugins.jobs.constants import JobStatus
tale = Tale().createTale(self.image, [],
creator=self.admin,
public=True)
workspace = Tale().createWorkspace(tale)
# Below workarounds a bug, it will be addressed elsewhere.
workspace = Folder().setPublic(workspace, True, save=True)
adapter = assetstore_utilities.getAssetstoreAdapter(self.ws_assetstore)
size = 101
data = BytesIO(b' ' * size)
files = []
files.append(Upload().uploadFromFile(data,
size,
'file01.txt',
parentType='folder',
parent=workspace,
assetstore=self.ws_assetstore))
fullPath = adapter.fullPath(files[0])
# Create a copy
resp = self.request(path='/tale/{_id}/copy'.format(**tale),
method='POST',
user=self.user)
self.assertStatusOk(resp)
new_tale = resp.json
self.assertFalse(new_tale['public'])
self.assertEqual(new_tale['dataSet'], tale['dataSet'])
self.assertEqual(new_tale['copyOfTale'], str(tale['_id']))
self.assertEqual(new_tale['imageId'], str(tale['imageId']))
self.assertEqual(new_tale['creatorId'], str(self.user['_id']))
self.assertEqual(new_tale['status'], TaleStatus.PREPARING)
copied_file_path = re.sub(workspace['name'], new_tale['_id'], fullPath)
job = Job().findOne({'type': 'wholetale.copy_workspace'})
for i in range(10):
job = Job().load(job['_id'], force=True)
if job['status'] == JobStatus.SUCCESS:
break
time.sleep(0.1)
self.assertTrue(os.path.isfile(copied_file_path))
resp = self.request(path='/tale/{_id}'.format(**new_tale),
method='GET',
user=self.user)
self.assertStatusOk(resp)
new_tale = resp.json
self.assertEqual(new_tale['status'], TaleStatus.READY)
Tale().remove(new_tale)
Tale().remove(tale)
def testWorkerWithParent(self):
from girder.plugins.jobs.models.job import Job
jobModel = Job()
parentJob = jobModel.createJob(
title='title', type='foo', handler='worker_handler',
user=self.admin, public=False, otherFields={'celeryTaskId': '1234'})
childJob = jobModel.createJob(
title='title', type='foo', handler='worker_handler',
user=self.admin, public=False, otherFields={'celeryTaskId': '5678',
'celeryParentTaskId': '1234'})
self.assertEqual(parentJob['_id'], childJob['parentId'])
def testLocalJob(self):
# Make sure local jobs still work
from girder.plugins.jobs.models.job import Job
job = Job().createLocalJob(title='local',
type='local',
user=self.users[0],
module='plugin_tests.worker_test',
function='local_job')
Job().scheduleJob(job)
job = Job().load(job['_id'], force=True, includeLog=True)
self.assertIn('job ran', job['log'])
def testWorkerCancel(self):
from girder.plugins.jobs.models.job import Job
jobModel = Job()
job = jobModel.createJob(
title='title', type='foo', handler='worker_handler',
user=self.admin, public=False, args=(), kwargs={})
job['kwargs'] = {
'jobInfo': utils.jobInfoSpec(job),
'inputs': [
utils.girderInputSpec(self.adminFolder, resourceType='folder')
],
'outputs': [
utils.girderOutputSpec(self.adminFolder, token=self.adminToken)
]
}
job = jobModel.save(job)
self.assertEqual(job['status'], JobStatus.INACTIVE)
# Schedule the job, make sure it is sent to celery
with mock.patch('celery.Celery') as celeryMock, \
mock.patch('girder.plugins.worker.AsyncResult') as asyncResult:
instance = celeryMock.return_value
instance.send_task.return_value = FakeAsyncResult()
jobModel.scheduleJob(job)
jobModel.cancelJob(job)
asyncResult.assert_called_with('fake_id', app=mock.ANY)
# Check we called revoke
asyncResult.return_value.revoke.assert_called_once()
job = jobModel.load(job['_id'], force=True)
self.assertEqual(job['status'], CustomJobStatus.CANCELING)
def run(job):
jobModel = Job()
jobModel.updateJob(job, status=JobStatus.RUNNING)
src_workspace_id, dest_workspace_id = job["args"]
user = job["kwargs"]["user"]
tale = job["kwargs"]["tale"]
try:
parent = Folder().load(src_workspace_id,
user=user,
exc=True,
level=AccessType.READ)
workspace = Folder().load(dest_workspace_id, user=user, exc=True)
Folder().copyFolderComponents(parent, workspace, user, None)
tale["status"] = TaleStatus.READY
Tale().updateTale(tale)
jobModel.updateJob(job,
status=JobStatus.SUCCESS,
log="Copying finished")
except Exception:
tale["status"] = TaleStatus.ERROR
Tale().updateTale(tale)
t, val, tb = sys.exc_info()
log = "%s: %s\n%s" % (t.__name__, repr(val), traceback.extract_tb(tb))
jobModel.updateJob(job, status=JobStatus.ERROR, log=log)
raise
def delete(self, item, skipFileIds=None):
deleted = False
if 'largeImage' in item:
job = None
if 'jobId' in item['largeImage']:
try:
job = Job().load(item['largeImage']['jobId'],
force=True,
exc=True)
except ValidationException:
# The job has been deleted, but we still need to clean up
# the rest of the tile information
pass
if (item['largeImage'].get('expected') and job
and job.get('status')
in (JobStatus.QUEUED, JobStatus.RUNNING)):
# cannot cleanly remove the large image, since a conversion
# job is currently in progress
# TODO: cancel the job
# TODO: return a failure error code
return False
# If this file was created by the worker job, delete it
if 'jobId' in item['largeImage']:
if job:
# TODO: does this eliminate all traces of the job?
# TODO: do we want to remove the original job?
Job().remove(job)
del item['largeImage']['jobId']
if 'originalId' in item['largeImage']:
# The large image file should not be the original file
assert item['largeImage']['originalId'] != \
item['largeImage'].get('fileId')
if ('fileId' in item['largeImage'] and
(not skipFileIds
or item['largeImage']['fileId'] not in skipFileIds)):
file = File().load(id=item['largeImage']['fileId'],
force=True)
if file:
File().remove(file)
del item['largeImage']['originalId']
del item['largeImage']
item = self.save(item)
deleted = True
self.removeThumbnailFiles(item)
return deleted
def executeTask(self, item, jobTitle, includeJobInfo, inputs, outputs):
user = self.getCurrentUser()
if jobTitle is None:
jobTitle = item['name']
task, handler = self._validateTask(item)
if task.get('mode') == 'girder_worker':
return runCeleryTask(item['meta']['itemTaskImport'], inputs)
jobModel = self.model('job', 'jobs')
jobModel = Job()
job = jobModel.createJob(
title=jobTitle, type='item_task', handler=handler, user=user)
# If this is a user auth token, we make an IO-enabled token
token = self.getCurrentToken()
tokenModel = Token()
if tokenModel.hasScope(token, TokenScope.USER_AUTH):
token = tokenModel.createToken(
user=user, days=7, scope=(TokenScope.DATA_READ, TokenScope.DATA_WRITE))
job['itemTaskTempToken'] = token['_id']
token = tokenModel.addScope(token, 'item_tasks.job_write:%s' % job['_id'])
job.update({
'itemTaskId': item['_id'],
'itemTaskBindings': {
'inputs': inputs,
'outputs': outputs
},
'kwargs': {
'task': task,
'inputs': self._transformInputs(inputs, token),
'outputs': self._transformOutputs(outputs, token, job, task, item['_id']),
'validate': False,
'auto_convert': False,
'cleanup': True
}
})
if includeJobInfo:
job['kwargs']['jobInfo'] = utils.jobInfoSpec(job)
if 'itemTaskCeleryQueue' in item.get('meta', {}):
job['celeryQueue'] = item['meta']['itemTaskCeleryQueue']
job = jobModel.save(job)
jobModel.scheduleJob(job)
return job
def copyTale(self, tale):
user = self.getCurrentUser()
image = self.model('image', 'wholetale').load(tale['imageId'],
user=user,
level=AccessType.READ,
exc=True)
default_author = ' '.join((user['firstName'], user['lastName']))
new_tale = self._model.createTale(
image,
tale['dataSet'],
creator=user,
save=True,
title=tale.get('title'),
description=tale.get('description'),
public=False,
config=tale.get('config'),
icon=image.get('icon', ('https://raw.githubusercontent.com/'
'whole-tale/dashboard/master/public/'
'images/whole_tale_logo.png')),
illustration=tale.get('illustration',
('https://raw.githubusercontent.com/'
'whole-tale/dashboard/master/public/'
'images/demo-graph2.jpg')),
authors=tale.get('authors', default_author),
category=tale.get('category', 'science'),
narrative=tale.get('narrative'),
licenseSPDX=tale.get('licenseSPDX'),
status=TaleStatus.PREPARING,
relatedIdentifiers=tale.get('relatedIdentifiers'),
)
new_tale['copyOfTale'] = tale['_id']
new_tale = self._model.save(new_tale)
# asynchronously copy the workspace of a source Tale
tale_workspaceId = self._model.createWorkspace(tale)['_id']
new_tale_workspaceId = self._model.createWorkspace(new_tale)['_id']
job = Job().createLocalJob(
title='Copy "{title}" workspace'.format(**tale),
user=user,
type='wholetale.copy_workspace',
public=False,
async=True,
module='girder.plugins.wholetale.tasks.copy_workspace',
args=(tale_workspaceId, new_tale_workspaceId),
kwargs={
'user': user,
'tale': new_tale
})
Job().scheduleJob(job)
return new_tale
def runSlicerCliTasksDescriptionForFolder(self, folder, image, args, pullImage, params):
jobModel = Job()
token = Token().createToken(
days=3, scope='item_task.set_task_spec.%s' % folder['_id'], user=self.getCurrentUser())
job = jobModel.createJob(
title='Read docker task specs: %s' % image, type='folder.item_task_slicer_cli_description',
handler='worker_handler', user=self.getCurrentUser())
if args[-1:] == ['--xml']:
args = args[:-1]
jobOptions = {
'itemTaskId': folder['_id'],
'kwargs': {
'task': {
'mode': 'docker',
'docker_image': image,
'container_args': args + ['--xml'],
'pull_image': pullImage,
'outputs': [{
'id': '_stdout',
'format': 'text'
}],
},
'outputs': {
'_stdout': {
'mode': 'http',
'method': 'POST',
'format': 'text',
'url': '/'.join((utils.getWorkerApiUrl(), 'folder', str(folder['_id']),
'item_task_slicer_cli_xml')),
'headers': {'Girder-Token': token['_id']},
'params': {
'image': image,
'args': json.dumps(args),
'pullImage': pullImage
}
}
},
'jobInfo': utils.jobInfoSpec(job),
'validate': False,
'auto_convert': False
}
}
job.update(jobOptions)
job = jobModel.save(job)
jobModel.scheduleJob(job)
return job
def wait_for_status(user, job, status):
"""
Utility to wait for a job model to move into a particular state.
:param job: The job model to wait on
:param status: The state to wait for.
:returns: True if the job model moved into the requested state, False otherwise.
"""
retries = 0
jobModel = Job()
while retries < 10:
job = jobModel.load(job['_id'], user=user)
if job['status'] == status:
return True
return False
def attachParentJob(event):
"""Attach parentJob before a model is saved."""
job = event.info
if job.get('celeryParentTaskId'):
celeryParentTaskId = job['celeryParentTaskId']
parentJob = Job().findOne({'celeryTaskId': celeryParentTaskId})
event.info['parentId'] = parentJob['_id']
def jobInfoSpec(job, token=None, logPrint=True):
"""
Build the jobInfo specification for a task to write status and log output
back to a Girder job.
:param job: The job document representing the worker task.
:type job: dict
:param token: The token to use. Creates a job token if not passed.
:type token: str or dict
:param logPrint: Whether standard output from the job should be
"""
if token is None:
token = Job().createJobToken(job)
if isinstance(token, dict):
token = token['_id']
return {
'method': 'PUT',
'url': '/'.join((getWorkerApiUrl(), 'job', str(job['_id']))),
'reference': str(job['_id']),
'headers': {
'Girder-Token': token
},
'logPrint': logPrint
}
def _createThumbnails(self, spec, cancel=False):
from girder.plugins.jobs.constants import JobStatus
from girder.plugins.jobs.models.job import Job
params = {'spec': json.dumps(spec)}
if cancel:
params['logInterval'] = 0
resp = self.request(method='PUT',
path='/large_image/thumbnails',
user=self.admin,
params=params)
self.assertStatusOk(resp)
job = resp.json
if cancel:
job = self._waitForJobToBeRunning(job)
job = Job().cancelJob(job)
starttime = time.time()
while True:
self.assertTrue(time.time() - starttime < 30)
resp = self.request('/job/%s' % str(job['_id']))
self.assertStatusOk(resp)
if resp.json.get('status') == JobStatus.SUCCESS:
return True
if resp.json.get('status') == JobStatus.ERROR:
return False
if resp.json.get('status') == JobStatus.CANCELED:
return 'canceled'
time.sleep(0.1)
def startRun(self, run, entrypoint):
user = self.getCurrentUser()
if not entrypoint:
entrypoint = "run.sh"
runRoot = Folder().load(run['parentId'],
user=user,
level=AccessType.WRITE)
tale = Tale().load(runRoot['meta']['taleId'],
user=user,
level=AccessType.READ)
resource = {
'type': 'wt_recorded_run',
'tale_id': tale['_id'],
'tale_title': tale['title']
}
token = Token().createToken(user=user, days=0.5)
notification = init_progress(resource, user, 'Recorded run',
'Initializing', RECORDED_RUN_STEP_TOTAL)
rrTask = recorded_run.signature(
args=[str(run['_id']),
str(tale['_id']), entrypoint],
girder_job_other_fields={
'wt_notification_id': str(notification['_id']),
},
girder_client_token=str(token['_id']),
).apply_async()
return Job().filter(rrTask.job, user=user)
def cancel(event):
"""
This is bound to the "jobs.cancel" event, and will be triggered any time
a job is canceled. This handler will process any job that has the
handler field set to "worker_handler".
"""
job = event.info
if job['handler'] in ['worker_handler', 'celery_handler']:
# Stop event propagation and prevent default, we are using a custom state
event.stopPropagation().preventDefault()
celeryTaskId = job.get('celeryTaskId')
if celeryTaskId is None:
msg = ("Unable to cancel Celery task. Job '%s' doesn't have a Celery task id."
% job['_id'])
logger.warn(msg)
return
if job['status'] not in [CustomJobStatus.CANCELING, JobStatus.CANCELED,
JobStatus.SUCCESS, JobStatus.ERROR]:
# Set the job status to canceling
Job().updateJob(job, status=CustomJobStatus.CANCELING)
# Send the revoke request.
asyncResult = AsyncResult(celeryTaskId, app=getCeleryApp())
asyncResult.revoke()
def load(info):
events.bind('jobs.schedule', 'worker', schedule)
events.bind('jobs.status.validate', 'worker', validateJobStatus)
events.bind('jobs.status.validTransitions', 'worker', validTransitions)
events.bind('jobs.cancel', 'worker', cancel)
events.bind('model.job.save.after', 'worker', attachJobInfoSpec)
events.bind('model.job.save', 'worker', attachParentJob)
Job().exposeFields(AccessType.SITE_ADMIN, {'celeryTaskId', 'celeryQueue'})
def testTaleImportZip(self):
image = self.model("image", "wholetale").createImage(
name="Jupyter Classic",
creator=self.user,
public=True,
config=dict(
template="base.tpl",
buildpack="PythonBuildPack",
user="******",
port=8888,
urlPath="",
),
)
with mock.patch("fs.copy.copy_fs") as mock_copy:
with open(
os.path.join(DATA_PATH, "5c92fbd472a9910001fbff72.zip"), "rb"
) as fp:
resp = self.request(
path="/tale/import",
method="POST",
user=self.user,
type="application/zip",
body=fp.read(),
)
self.assertStatusOk(resp)
tale = resp.json
from girder.plugins.jobs.models.job import Job
job = Job().findOne({"type": "wholetale.import_tale"})
self.assertEqual(
json.loads(job["kwargs"])["taleId"]["$oid"], tale["_id"]
)
for i in range(300):
if job["status"] in {JobStatus.SUCCESS, JobStatus.ERROR}:
break
time.sleep(0.1)
job = Job().load(job["_id"], force=True)
self.assertEqual(job["status"], JobStatus.SUCCESS)
mock_copy.assert_called_once()
# TODO: make it more extensive...
self.assertTrue(
self.model("tale", "wholetale").findOne({"title": "Water Tale"}) is not None
)
self.model("image", "wholetale").remove(image)
def _updateJob(event):
"""
Called when a job is saved, updated, or removed. If this is a large image
job and it is ended, clean up after it.
"""
from girder.plugins.jobs.constants import JobStatus
from girder.plugins.jobs.models.job import Job
job = event.info[
'job'] if event.name == 'jobs.job.update.after' else event.info
meta = job.get('meta', {})
if (meta.get('creator') != 'large_image' or not meta.get('itemId')
or meta.get('task') != 'createImageItem'):
return
status = job['status']
if event.name == 'model.job.remove' and status not in (JobStatus.ERROR,
JobStatus.CANCELED,
JobStatus.SUCCESS):
status = JobStatus.CANCELED
if status not in (JobStatus.ERROR, JobStatus.CANCELED, JobStatus.SUCCESS):
return
item = Item().load(meta['itemId'], force=True)
if not item or 'largeImage' not in item:
return
if item.get('largeImage', {}).get('expected'):
# We can get a SUCCESS message before we get the upload message, so
# don't clear the expected status on success.
if status != JobStatus.SUCCESS:
del item['largeImage']['expected']
notify = item.get('largeImage', {}).get('notify')
msg = None
if notify:
del item['largeImage']['notify']
if status == JobStatus.SUCCESS:
msg = 'Large image created'
elif status == JobStatus.CANCELED:
msg = 'Large image creation canceled'
else: # ERROR
msg = 'FAILED: Large image creation failed'
msg += ' for item %s' % item['name']
if (status in (JobStatus.ERROR, JobStatus.CANCELED)
and 'largeImage' in item):
del item['largeImage']
Item().save(item)
if msg and event.name != 'model.job.remove':
Job().updateJob(job, progressMessage=msg)
if notify:
Notification().createNotification(
type='large_image.finished_image_item',
data={
'job_id': job['_id'],
'item_id': item['_id'],
'success': status == JobStatus.SUCCESS,
'status': status
},
user={'_id': job.get('userId')},
expires=datetime.datetime.utcnow() +
datetime.timedelta(seconds=30))
def setUp(self):
base.TestCase.setUp(self)
self.users = [User().createUser(
'usr' + str(n), 'passwd', 'tst', 'usr', '*****@*****.**' % n)
for n in range(3)]
from girder.plugins.jobs.models.job import Job
self.jobModel = Job()
def testTaleImportZipFail(self):
image = Image().createImage(
name="Jupyter Classic",
creator=self.user,
public=True,
config=dict(
template="base.tpl",
buildpack="PythonBuildPack",
user="******",
port=8888,
urlPath="",
),
)
with mock.patch("girder.plugins.wholetale.lib.pids_to_entities") as mock_pids:
mock_pids.side_effect = ValueError
with open(
os.path.join(DATA_PATH, "5c92fbd472a9910001fbff72.zip"), "rb"
) as fp:
resp = self.request(
path="/tale/import",
method="POST",
user=self.user,
type="application/zip",
body=fp.read(),
)
self.assertStatusOk(resp)
tale = resp.json
job = Job().findOne({"type": "wholetale.import_tale"})
self.assertEqual(
json.loads(job["kwargs"])["taleId"]["$oid"], tale["_id"]
)
for i in range(300):
if job["status"] in {JobStatus.SUCCESS, JobStatus.ERROR}:
break
time.sleep(0.1)
job = Job().load(job["_id"], force=True)
self.assertEqual(job["status"], JobStatus.ERROR)
Job().remove(job)
tale = Tale().load(tale["_id"], force=True)
self.assertEqual(tale["status"], TaleStatus.ERROR)
Tale().remove(tale)
Image().remove(image)
def testWorkerWithParent(self):
from girder.plugins.jobs.models.job import Job
jobModel = Job()
parentJob = jobModel.createJob(title='title',
type='foo',
handler='worker_handler',
user=self.admin,
public=False,
otherFields={'celeryTaskId': '1234'})
childJob = jobModel.createJob(title='title',
type='foo',
handler='worker_handler',
user=self.admin,
public=False,
otherFields={
'celeryTaskId': '5678',
'celeryParentTaskId': '1234'
})
self.assertEqual(parentJob['_id'], childJob['parentId'])
def run(job):
jobModel = Job()
jobModel.updateJob(job, status=JobStatus.RUNNING)
try:
newFile = createThumbnail(**job['kwargs'])
log = 'Created thumbnail file %s.' % newFile['_id']
jobModel.updateJob(job, status=JobStatus.SUCCESS, log=log)
except Exception:
t, val, tb = sys.exc_info()
log = '%s: %s\n%s' % (t.__name__, repr(val), traceback.extract_tb(tb))
jobModel.updateJob(job, status=JobStatus.ERROR, log=log)
raise
def setMultiscaleMetaData(jobId, inputFolderId, outputFolderId):
"""Set the multiscale meta data for the jobId.
Currently, we use this to keep track of the input and output
folders.
Returns the updated job.
"""
# We want to update the job with some multiscale settings.
# We will put it in the meta data.
job = Job().findOne({'_id': jobId})
multiscale_io = {
'meta': {
'multiscale_settings': {
'inputFolderId': inputFolderId,
'outputFolderId': outputFolderId
}
}
}
return Job().updateJob(job, otherFields=multiscale_io)
def buildImage(self, image, params):
user = self.getCurrentUser()
recipe = self.model('recipe', 'wholetale').load(image['recipeId'],
user=user,
level=AccessType.READ,
exc=True)
url = '{}/archive/{}.tar.gz'.format(recipe['url'], recipe['commitId'])
jobTitle = 'Building image %s' % image['fullName']
jobModel = Job()
# Create a job to be handled by the worker plugin
job = jobModel.createJob(
title=jobTitle,
type='build_image',
handler='worker_handler',
user=user,
public=False,
args=(str(image['_id']), image['fullName'], url),
kwargs={},
otherFields={'celeryTaskName': 'gwvolman.tasks.build_image'})
jobModel.scheduleJob(job)
return job
def testWorker(self):
# Test the settings
resp = self.request('/system/setting', method='PUT', params={
'list': json.dumps([{
'key': worker.PluginSettings.BROKER,
'value': 'amqp://[email protected]'
}, {
'key': worker.PluginSettings.BACKEND,
'value': 'amqp://[email protected]'
}])
}, user=self.admin)
self.assertStatusOk(resp)
# Create a job to be handled by the worker plugin
from girder.plugins.jobs.models.job import Job
jobModel = Job()
job = jobModel.createJob(
title='title', type='foo', handler='worker_handler',
user=self.admin, public=False, args=(), kwargs={})
job['kwargs'] = {
'jobInfo': utils.jobInfoSpec(job),
'inputs': [
utils.girderInputSpec(self.adminFolder, resourceType='folder')
],
'outputs': [
utils.girderOutputSpec(self.adminFolder, token=self.adminToken)
]
}
job = jobModel.save(job)
self.assertEqual(job['status'], JobStatus.INACTIVE)
# Schedule the job, make sure it is sent to celery
with mock.patch('celery.Celery') as celeryMock:
instance = celeryMock.return_value
instance.send_task.return_value = FakeAsyncResult()
jobModel.scheduleJob(job)
# Make sure we sent the job to celery
self.assertEqual(len(celeryMock.mock_calls), 2)
self.assertEqual(celeryMock.mock_calls[0][1], ('girder_worker',))
self.assertEqual(celeryMock.mock_calls[0][2], {
'broker': 'amqp://[email protected]',
'backend': 'amqp://[email protected]'
})
sendTaskCalls = celeryMock.return_value.send_task.mock_calls
self.assertEqual(len(sendTaskCalls), 1)
self.assertEqual(sendTaskCalls[0][1], (
'girder_worker.run', job['args'], job['kwargs']))
self.assertTrue('headers' in sendTaskCalls[0][2])
self.assertTrue('jobInfoSpec' in sendTaskCalls[0][2]['headers'])
# Make sure we got and saved the celery task id
job = jobModel.load(job['_id'], force=True)
self.assertEqual(job['celeryTaskId'], 'fake_id')
self.assertEqual(job['status'], JobStatus.QUEUED)
def run(job):
jobModel = Job()
jobModel.updateJob(job, status=JobStatus.RUNNING)
try:
newFile = createThumbnail(**job['kwargs'])
log = 'Created thumbnail file %s.' % newFile['_id']
jobModel.updateJob(job, status=JobStatus.SUCCESS, log=log)
except Exception:
t, val, tb = sys.exc_info()
log = '%s: %s\n%s' % (t.__name__, repr(val), traceback.extract_tb(tb))
jobModel.updateJob(job, status=JobStatus.ERROR, log=log)
raise
def runSlicerCliTasksDescriptionForItem(
self, item, image, args, setName, setDescription, pullImage, params):
if 'meta' not in item:
item['meta'] = {}
if image is None:
image = item.get('meta', {}).get('itemTaskSpec', {}).get('docker_image')
if not image:
raise RestException(
'You must pass an image parameter, or set the itemTaskSpec.docker_image '
'field of the item.')
jobModel = Job()
token = Token().createToken(
days=3, scope='item_task.set_task_spec.%s' % item['_id'])
job = jobModel.createJob(
title='Read docker Slicer CLI: %s' % image, type='item.item_task_slicer_cli_description',
handler='worker_handler', user=self.getCurrentUser())
if args[-1:] == ['--xml']:
args = args[:-1]
job.update({
'itemTaskId': item['_id'],
'kwargs': {
'task': {
'mode': 'docker',
'docker_image': image,
'container_args': args + ['--xml'],
'pull_image': pullImage,
'outputs': [{
'id': '_stdout',
'format': 'text'
}],
},
'outputs': {
'_stdout': {
'mode': 'http',
'method': 'PUT',
'format': 'text',
'url': '/'.join((utils.getWorkerApiUrl(), 'item', str(item['_id']),
'item_task_slicer_cli_xml')),
'params': {
'setName': setName,
'setDescription': setDescription
},
'headers': {'Girder-Token': token['_id']}
}
},
'jobInfo': utils.jobInfoSpec(job),
'validate': False,
'auto_convert': False
}
})
item['meta']['itemTaskSpec'] = {
'mode': 'docker',
'docker_image': image
}
if args:
item['meta']['itemTaskSlicerCliArgs'] = args
Item().save(item)
job = jobModel.save(job)
jobModel.scheduleJob(job)
return job
def testConfigureItemTaskFromSlicerCli(self):
# Create a new item that will become a task
item = Item().createItem(name='placeholder', creator=self.admin, folder=self.privateFolder)
# Create task to introspect container
with mock.patch('girder.plugins.jobs.models.job.Job.scheduleJob') as scheduleMock:
resp = self.request(
'/item/%s/item_task_slicer_cli_description' % item['_id'], method='POST', params={
'image': 'johndoe/foo:v5',
'args': json.dumps(['--foo', 'bar'])
}, user=self.admin)
self.assertStatusOk(resp)
self.assertEqual(resp.json['_modelType'], 'job')
self.assertEqual(len(scheduleMock.mock_calls), 1)
job = scheduleMock.mock_calls[0][1][0]
self.assertEqual(job['handler'], 'worker_handler')
self.assertEqual(job['itemTaskId'], item['_id'])
self.assertEqual(job['kwargs']['outputs']['_stdout']['method'], 'PUT')
self.assertTrue(job['kwargs']['outputs']['_stdout']['url'].endswith(
'item/%s/item_task_slicer_cli_xml' % item['_id']))
token = job['kwargs']['outputs']['_stdout']['headers']['Girder-Token']
# Task should not be registered until we get the callback
resp = self.request('/item_task', user=self.admin)
self.assertStatusOk(resp)
self.assertEqual(resp.json, [])
# Image and args should be stored in the item metadata
item = Item().load(item['_id'], force=True)
self.assertEqual(item['meta']['itemTaskSpec']['docker_image'], 'johndoe/foo:v5')
self.assertEqual(item['meta']['itemTaskSlicerCliArgs'], ['--foo', 'bar'])
# Simulate callback from introspection job
with open(os.path.join(os.path.dirname(__file__), 'slicer_cli.xml')) as f:
xml = f.read()
resp = self.request(
'/item/%s/item_task_slicer_cli_xml' % item['_id'], method='PUT', params={
'setName': True,
'setDescription': True
}, token=token, body=xml, type='application/xml')
self.assertStatusOk(resp)
# We should only be able to see tasks we have read access on
resp = self.request('/item_task')
self.assertStatusOk(resp)
self.assertEqual(resp.json, [])
resp = self.request('/item_task', user=self.admin)
self.assertStatusOk(resp)
self.assertEqual(len(resp.json), 1)
self.assertEqual(resp.json[0]['_id'], str(item['_id']))
item = Item().load(item['_id'], force=True)
self.assertEqual(item['name'], 'PET phantom detector CLI')
self.assertEqual(
item['description'],
u'**Description**: Detects positions of PET/CT pocket phantoms in PET image.\n\n'
u'**Author(s)**: Girder Developers\n\n**Version**: 1.0\n\n'
u'**License**: Apache 2.0\n\n**Acknowledgements**: *none*\n\n'
u'*This description was auto-generated from the Slicer CLI XML specification.*'
)
self.assertTrue(item['meta']['isItemTask'])
self.assertEqual(item['meta']['itemTaskSpec'], {
'mode': 'docker',
'docker_image': 'johndoe/foo:v5',
'container_args': [
'--foo', 'bar', '--InputImage=$input{--InputImage}',
'--MaximumLineStraightnessDeviation=$input{--MaximumLineStraightnessDeviation}',
'--MaximumRadius=$input{--MaximumRadius}',
'--MaximumSphereDistance=$input{--MaximumSphereDistance}',
'--MinimumRadius=$input{--MinimumRadius}',
'--MinimumSphereActivity=$input{--MinimumSphereActivity}',
'--MinimumSphereDistance=$input{--MinimumSphereDistance}',
'--SpheresPerPhantom=$input{--SpheresPerPhantom}', '$flag{--StrictSorting}',
'--DetectedPoints=$output{--DetectedPoints}'
],
'inputs': [{
'description': 'Input image to be analysed.',
'format': 'image',
'name': 'InputImage', 'type': 'image', 'id': '--InputImage',
'target': 'filepath'
}, {
'description': 'Used for eliminating detections which are not in a straight line. '
'Unit: multiples of geometric average of voxel spacing',
'format': 'number',
'default': {'data': 1.0},
'type': 'number',
'id': '--MaximumLineStraightnessDeviation',
'name': 'MaximumLineStraightnessDeviation'
}, {
'description': 'Used for eliminating too big blobs. Unit: millimeter [mm]',
'format': 'number', 'default': {'data': 20.0},
'type': 'number',
'id': '--MaximumRadius',
'name': 'MaximumRadius'
}, {
'description': 'Signifies maximum distance between adjacent sphere centers [mm]. '
'Used to separate phantoms from tumors.',
'format': 'number', 'default': {'data': 40.0},
'type': 'number',
'id': '--MaximumSphereDistance',
'name': 'MaximumSphereDistance'
}, {
'description': 'Used for eliminating too small blobs. Unit: millimeter [mm]',
'format': 'number',
'default': {'data': 3.0},
'type': 'number',
'id': '--MinimumRadius',
'name': 'MinimumRadius'
}, {
'description': 'Used for thresholding in blob detection. '
'Unit: becquerels per milliliter [Bq/ml]',
'format': 'number', 'default': {'data': 5000.0},
'type': 'number',
'id': '--MinimumSphereActivity',
'name': 'MinimumSphereActivity'
}, {
'description': 'Signifies minimum distance between adjacent sphere centers [mm]. '
'Used to separate phantoms from tumors.',
'format': 'number',
'default': {'data': 30.0},
'type': 'number',
'id': '--MinimumSphereDistance',
'name': 'MinimumSphereDistance'
}, {
'description': 'What kind of phantom are we working with here?',
'format': 'number-enumeration',
'default': {'data': 3},
'type': 'number-enumeration',
'id': '--SpheresPerPhantom',
'name': 'SpheresPerPhantom',
'values': [2, 3]
}, {
'description': 'Controls whether spheres within a phantom must have descending '
'activities. If OFF, they can have approximately same activities '
'(within 15%).',
'format': 'boolean',
'default': {'data': False},
'type': 'boolean',
'id': '--StrictSorting',
'name': 'StrictSorting'
}],
'outputs': [{
'description': 'Fiducial points, one for each detected sphere. '
'Will be multiple of 3.',
'format': 'new-file',
'name': 'DetectedPoints',
'type': 'new-file',
'id': '--DetectedPoints',
'target': 'filepath'
}]
})
# Shouldn't be able to run the task if we don't have execute permission flag
Folder().setUserAccess(
self.privateFolder, user=self.user, level=AccessType.READ, save=True)
resp = self.request(
'/item_task/%s/execution' % item['_id'], method='POST', user=self.user)
self.assertStatus(resp, 403)
# Grant the user permission, and run the task
from girder.plugins.item_tasks.constants import ACCESS_FLAG_EXECUTE_TASK
Folder().setUserAccess(
self.privateFolder, user=self.user, level=AccessType.WRITE,
flags=ACCESS_FLAG_EXECUTE_TASK, currentUser=self.admin, save=True)
inputs = {
'--InputImage': {
'mode': 'girder',
'resource_type': 'item',
'id': str(item['_id'])
},
'--MaximumLineStraightnessDeviation': {
'mode': 'inline',
'data': 1
},
'--MaximumRadius': {
'mode': 'inline',
'data': 20
},
'--MaximumSphereDistance': {
'mode': 'inline',
'data': 40
},
'--MinimumRadius': {
'mode': 'inline',
'data': 3
},
'--MinimumSphereActivity': {
'mode': 'inline',
'data': 5000
},
'--MinimumSphereDistance': {
'mode': 'inline',
'data': 30
},
'--SpheresPerPhantom': {
'mode': 'inline',
'data': 3},
'--StrictSorting': {
'mode': 'inline',
'data': False
}
}
outputs = {
'--DetectedPoints': {
'mode': 'girder',
'parent_id': str(self.privateFolder['_id']),
'parent_type': 'folder',
'name': 'test.txt'
}
}
# Ensure task was scheduled
with mock.patch('girder.plugins.jobs.models.job.Job.scheduleJob') as scheduleMock:
resp = self.request(
'/item_task/%s/execution' % item['_id'], method='POST', user=self.user, params={
'inputs': json.dumps(inputs),
'outputs': json.dumps(outputs)
})
self.assertEqual(len(scheduleMock.mock_calls), 1)
self.assertStatusOk(resp)
job = resp.json
self.assertEqual(job['_modelType'], 'job')
self.assertNotIn('kwargs', job) # ordinary user can't see kwargs
from girder.plugins.jobs.models.job import Job
jobModel = Job()
job = jobModel.load(job['_id'], force=True)
output = job['kwargs']['outputs']['--DetectedPoints']
# Simulate output from the worker
contents = b'Hello world'
resp = self.request(
path='/file', method='POST', token=output['token'], params={
'parentType': output['parent_type'],
'parentId': output['parent_id'],
'name': output['name'],
'size': len(contents),
'mimeType': 'text/plain',
'reference': output['reference']
})
self.assertStatusOk(resp)
uploadId = resp.json['_id']
fields = [('offset', 0), ('uploadId', uploadId)]
files = [('chunk', output['name'], contents)]
resp = self.multipartRequest(
path='/file/chunk', fields=fields, files=files, token=output['token'])
self.assertStatusOk(resp)
file = resp.json
self.assertEqual(file['_modelType'], 'file')
self.assertEqual(file['size'], 11)
self.assertEqual(file['mimeType'], 'text/plain')
file = File().load(file['_id'], force=True)
# Make sure temp token is removed once we change job status to final state
job = jobModel.load(job['_id'], force=True)
self.assertIn('itemTaskTempToken', job)
from girder.plugins.jobs.constants import JobStatus
# Transition through states to SUCCESS
job = jobModel.updateJob(job, status=JobStatus.QUEUED)
job = jobModel.updateJob(job, status=JobStatus.RUNNING)
job = jobModel.updateJob(job, status=JobStatus.SUCCESS)
self.assertNotIn('itemTaskTempToken', job)
self.assertIn('itemTaskBindings', job)
# Wait for async data.process event to bind output provenance
start = time.time()
while time.time() - start < 15:
job = jobModel.load(job['_id'], force=True)
if 'itemId' in job['itemTaskBindings']['outputs']['--DetectedPoints']:
break
else:
time.sleep(0.2)
else:
raise Exception('Output binding did not occur in time')
self.assertEqual(
job['itemTaskBindings']['outputs']['--DetectedPoints']['itemId'], file['itemId'])
class JobsTestCase(base.TestCase):
def setUp(self):
base.TestCase.setUp(self)
self.users = [User().createUser(
'usr' + str(n), 'passwd', 'tst', 'usr', '*****@*****.**' % n)
for n in range(3)]
from girder.plugins.jobs.models.job import Job
self.jobModel = Job()
def testJobs(self):
self.job = None
def schedule(event):
self.job = event.info
if self.job['handler'] == 'my_handler':
self.job['status'] = JobStatus.RUNNING
self.job = self.jobModel.save(self.job)
self.assertEqual(self.job['args'], ('hello', 'world'))
self.assertEqual(self.job['kwargs'], {'a': 'b'})
events.bind('jobs.schedule', 'test', schedule)
# Create a job
job = self.jobModel.createJob(
title='Job Title', type='my_type', args=('hello', 'world'),
kwargs={'a': 'b'}, user=self.users[1], handler='my_handler',
public=False)
self.assertEqual(self.job, None)
self.assertEqual(job['status'], JobStatus.INACTIVE)
# Schedule the job, make sure our handler was invoked
self.jobModel.scheduleJob(job)
self.assertEqual(self.job['_id'], job['_id'])
self.assertEqual(self.job['status'], JobStatus.RUNNING)
# Since the job is not public, user 2 should not have access
path = '/job/%s' % job['_id']
resp = self.request(path, user=self.users[2])
self.assertStatus(resp, 403)
resp = self.request(path, user=self.users[2], method='PUT')
self.assertStatus(resp, 403)
resp = self.request(path, user=self.users[2], method='DELETE')
self.assertStatus(resp, 403)
# If no user is specified, we should get a 401 error
resp = self.request(path, user=None)
self.assertStatus(resp, 401)
# Make sure user who created the job can see it
resp = self.request(path, user=self.users[1])
self.assertStatusOk(resp)
# We should be able to update the job as the user who created it
resp = self.request(path, method='PUT', user=self.users[1], params={
'log': 'My log message\n'
})
self.assertStatusOk(resp)
# We should be able to create a job token and use that to update it too
token = self.jobModel.createJobToken(job)
resp = self.request(path, method='PUT', params={
'log': 'append message',
'token': token['_id']
})
self.assertStatusOk(resp)
# We shouldn't get the log back in this case
self.assertNotIn('log', resp.json)
# Do a fetch on the job itself to get the log
resp = self.request(path, user=self.users[1])
self.assertStatusOk(resp)
self.assertEqual(
resp.json['log'], ['My log message\n', 'append message'])
# Test overwriting the log and updating status
resp = self.request(path, method='PUT', params={
'log': 'overwritten log',
'overwrite': 'true',
'status': JobStatus.SUCCESS,
'token': token['_id']
})
self.assertStatusOk(resp)
self.assertNotIn('log', resp.json)
self.assertEqual(resp.json['status'], JobStatus.SUCCESS)
job = self.jobModel.load(job['_id'], force=True, includeLog=True)
self.assertEqual(job['log'], ['overwritten log'])
# We should be able to delete the job as the user who created it
resp = self.request(path, user=self.users[1], method='DELETE')
self.assertStatusOk(resp)
job = self.jobModel.load(job['_id'], force=True)
self.assertIsNone(job)
def testLegacyLogBehavior(self):
# Force save a job with a string log to simulate a legacy job record
job = self.jobModel.createJob(
title='legacy', type='legacy', user=self.users[1], save=False)
job['log'] = 'legacy log'
job = self.jobModel.save(job, validate=False)
self.assertEqual(job['log'], 'legacy log')
# Load the record, we should now get the log as a list
job = self.jobModel.load(job['_id'], force=True, includeLog=True)
self.assertEqual(job['log'], ['legacy log'])
def testListJobs(self):
job = self.jobModel.createJob(title='A job', type='t', user=self.users[1], public=False)
anonJob = self.jobModel.createJob(title='Anon job', type='t')
# Ensure timestamp for public job is strictly higher (ms resolution)
time.sleep(0.1)
publicJob = self.jobModel.createJob(
title='Anon job', type='t', public=True)
# User 1 should be able to see their own jobs
resp = self.request('/job', user=self.users[1], params={
'userId': self.users[1]['_id']
})
self.assertStatusOk(resp)
self.assertEqual(len(resp.json), 1)
self.assertEqual(resp.json[0]['_id'], str(job['_id']))
# User 2 should not see user 1's jobs in the list
resp = self.request('/job', user=self.users[2], params={
'userId': self.users[1]['_id']
})
self.assertEqual(resp.json, [])
# Omitting a userId should assume current user
resp = self.request('/job', user=self.users[1])
self.assertStatusOk(resp)
self.assertEqual(len(resp.json), 1)
self.assertEqual(resp.json[0]['_id'], str(job['_id']))
# Explicitly passing "None" should show anonymous jobs
resp = self.request('/job', user=self.users[0], params={
'userId': 'none'
})
self.assertStatusOk(resp)
self.assertEqual(len(resp.json), 2)
self.assertEqual(resp.json[0]['_id'], str(publicJob['_id']))
self.assertEqual(resp.json[1]['_id'], str(anonJob['_id']))
# Non-admins should only see public anon jobs
resp = self.request('/job', params={'userId': 'none'})
self.assertStatusOk(resp)
self.assertEqual(len(resp.json), 1)
self.assertEqual(resp.json[0]['_id'], str(publicJob['_id']))
def testListAllJobs(self):
self.jobModel.createJob(title='user 0 job', type='t', user=self.users[0], public=False)
self.jobModel.createJob(title='user 1 job', type='t', user=self.users[1], public=False)
self.jobModel.createJob(title='user 1 job', type='t', user=self.users[1], public=True)
self.jobModel.createJob(title='user 2 job', type='t', user=self.users[2])
self.jobModel.createJob(title='anonymous job', type='t')
self.jobModel.createJob(title='anonymous public job', type='t2', public=True)
# User 0, as a site admin, should be able to see all jobs
resp = self.request('/job/all', user=self.users[0])
self.assertStatusOk(resp)
self.assertEqual(len(resp.json), 6)
# Test deprecated listAll method
jobs = list(self.jobModel.listAll(limit=0, offset=0, sort=None, currentUser=self.users[0]))
self.assertEqual(len(jobs), 6)
# get with filter
resp = self.request('/job/all', user=self.users[0], params={
'types': json.dumps(['t']),
'statuses': json.dumps([0])
})
self.assertStatusOk(resp)
self.assertEqual(len(resp.json), 5)
# get with unmet filter conditions
resp = self.request('/job/all', user=self.users[0], params={
'types': json.dumps(['nonexisttype'])
})
self.assertStatusOk(resp)
self.assertEqual(len(resp.json), 0)
# User 1, as non site admin, should encounter http 403 (Forbidden)
resp = self.request('/job/all', user=self.users[1])
self.assertStatus(resp, 403)
# Not authenticated user should encounter http 401 (unauthorized)
resp = self.request('/job/all')
self.assertStatus(resp, 401)
def testFiltering(self):
job = self.jobModel.createJob(title='A job', type='t', user=self.users[1], public=True)
job['_some_other_field'] = 'foo'
job = self.jobModel.save(job)
resp = self.request('/job/%s' % job['_id'])
self.assertStatusOk(resp)
self.assertTrue('created' in resp.json)
self.assertTrue('_some_other_field' not in resp.json)
self.assertTrue('kwargs' not in resp.json)
self.assertTrue('args' not in resp.json)
resp = self.request('/job/%s' % job['_id'], user=self.users[0])
self.assertTrue('kwargs' in resp.json)
self.assertTrue('args' in resp.json)
self.jobModel.exposeFields(level=AccessType.READ, fields={'_some_other_field'})
self.jobModel.hideFields(level=AccessType.READ, fields={'created'})
resp = self.request('/job/%s' % job['_id'])
self.assertStatusOk(resp)
self.assertEqual(resp.json['_some_other_field'], 'foo')
self.assertTrue('created' not in resp.json)
def testJobProgressAndNotifications(self):
job = self.jobModel.createJob(title='a job', type='t', user=self.users[1], public=True)
path = '/job/%s' % job['_id']
resp = self.request(path)
self.assertEqual(resp.json['progress'], None)
self.assertEqual(resp.json['timestamps'], [])
resp = self.request(path, method='PUT', user=self.users[1], params={
'progressTotal': 100,
'progressCurrent': 3,
'progressMessage': 'Started',
'notify': 'false',
'status': JobStatus.QUEUED
})
self.assertStatusOk(resp)
self.assertEqual(resp.json['progress'], {
'total': 100,
'current': 3,
'message': 'Started',
'notificationId': None
})
# The status update should make it so we now have a timestamp
self.assertEqual(len(resp.json['timestamps']), 1)
self.assertEqual(resp.json['timestamps'][0]['status'], JobStatus.QUEUED)
self.assertIn('time', resp.json['timestamps'][0])
# If the status does not change on update, no timestamp should be added
resp = self.request(path, method='PUT', user=self.users[1], params={
'status': JobStatus.QUEUED
})
self.assertStatusOk(resp)
self.assertEqual(len(resp.json['timestamps']), 1)
self.assertEqual(resp.json['timestamps'][0]['status'], JobStatus.QUEUED)
# We passed notify=false, so we should only have the job creation notification
resp = self.request(path='/notification/stream', method='GET',
user=self.users[1], isJson=False,
params={'timeout': 0})
messages = self.getSseMessages(resp)
self.assertEqual(len(messages), 1)
# Update progress with notify=true (the default)
resp = self.request(path, method='PUT', user=self.users[1], params={
'progressCurrent': 50,
'progressMessage': 'Something bad happened',
'status': JobStatus.ERROR
})
self.assertStatusOk(resp)
self.assertNotEqual(resp.json['progress']['notificationId'], None)
# We should now see three notifications (job created + job status + progress)
resp = self.request(path='/notification/stream', method='GET',
user=self.users[1], isJson=False,
params={'timeout': 0})
messages = self.getSseMessages(resp)
job = self.jobModel.load(job['_id'], force=True)
self.assertEqual(len(messages), 3)
creationNotify = messages[0]
progressNotify = messages[1]
statusNotify = messages[2]
self.assertEqual(creationNotify['type'], 'job_created')
self.assertEqual(creationNotify['data']['_id'], str(job['_id']))
self.assertEqual(statusNotify['type'], 'job_status')
self.assertEqual(statusNotify['data']['_id'], str(job['_id']))
self.assertEqual(int(statusNotify['data']['status']), JobStatus.ERROR)
self.assertNotIn('kwargs', statusNotify['data'])
self.assertNotIn('log', statusNotify['data'])
self.assertEqual(progressNotify['type'], 'progress')
self.assertEqual(progressNotify['data']['title'], job['title'])
self.assertEqual(progressNotify['data']['current'], float(50))
self.assertEqual(progressNotify['data']['state'], 'error')
self.assertEqual(progressNotify['_id'], str(job['progress']['notificationId']))
def testDotsInKwargs(self):
kwargs = {
'$key.with.dots': 'value',
'foo': [{
'moar.dots': True
}]
}
job = self.jobModel.createJob(title='dots', type='x', user=self.users[0], kwargs=kwargs)
# Make sure we can update a job and notification creation works
self.jobModel.updateJob(job, status=JobStatus.QUEUED, notify=True)
self.assertEqual(job['kwargs'], kwargs)
resp = self.request('/job/%s' % job['_id'], user=self.users[0])
self.assertStatusOk(resp)
self.assertEqual(resp.json['kwargs'], kwargs)
job = self.jobModel.load(job['_id'], force=True)
self.assertEqual(job['kwargs'], kwargs)
job = self.jobModel.filter(job, self.users[0])
self.assertEqual(job['kwargs'], kwargs)
job = self.jobModel.filter(job, self.users[1])
self.assertFalse('kwargs' in job)
def testLocalJob(self):
job = self.jobModel.createLocalJob(
title='local', type='local', user=self.users[0], kwargs={
'hello': 'world'
}, module='plugin_tests.local_job_impl')
self.jobModel.scheduleJob(job)
job = self.jobModel.load(job['_id'], force=True, includeLog=True)
self.assertEqual(job['log'], ['job ran!'])
job = self.jobModel.createLocalJob(
title='local', type='local', user=self.users[0], kwargs={
'hello': 'world'
}, module='plugin_tests.local_job_impl', function='fail')
self.jobModel.scheduleJob(job)
job = self.jobModel.load(job['_id'], force=True, includeLog=True)
self.assertEqual(job['log'], ['job failed'])
def testValidateCustomStatus(self):
job = self.jobModel.createJob(title='test', type='x', user=self.users[0])
def validateStatus(event):
if event.info == 1234:
event.preventDefault().addResponse(True)
def validTransitions(event):
if event.info['status'] == 1234:
event.preventDefault().addResponse([JobStatus.INACTIVE])
with self.assertRaises(ValidationException):
self.jobModel.updateJob(job, status=1234) # Should fail
with events.bound('jobs.status.validate', 'test', validateStatus), \
events.bound('jobs.status.validTransitions', 'test', validTransitions):
self.jobModel.updateJob(job, status=1234) # Should work
with self.assertRaises(ValidationException):
self.jobModel.updateJob(job, status=4321) # Should fail
def testValidateCustomStrStatus(self):
job = self.jobModel.createJob(title='test', type='x', user=self.users[0])
def validateStatus(event):
states = ['a', 'b', 'c']
if event.info in states:
event.preventDefault().addResponse(True)
def validTransitions(event):
if event.info['status'] == 'a':
event.preventDefault().addResponse([JobStatus.INACTIVE])
with self.assertRaises(ValidationException):
self.jobModel.updateJob(job, status='a')
with events.bound('jobs.status.validate', 'test', validateStatus), \
events.bound('jobs.status.validTransitions', 'test', validTransitions):
self.jobModel.updateJob(job, status='a')
self.assertEqual(job['status'], 'a')
with self.assertRaises(ValidationException), \
events.bound('jobs.status.validate', 'test', validateStatus):
self.jobModel.updateJob(job, status='foo')
def testUpdateOtherFields(self):
job = self.jobModel.createJob(title='test', type='x', user=self.users[0])
job = self.jobModel.updateJob(job, otherFields={'other': 'fields'})
self.assertEqual(job['other'], 'fields')
def testCancelJob(self):
job = self.jobModel.createJob(title='test', type='x', user=self.users[0])
# add to the log
job = self.jobModel.updateJob(job, log='entry 1\n')
# Reload without the log
job = self.jobModel.load(id=job['_id'], force=True)
self.assertEqual(len(job.get('log', [])), 0)
# Cancel
job = self.jobModel.cancelJob(job)
self.assertEqual(job['status'], JobStatus.CANCELED)
# Reloading should still have the log and be canceled
job = self.jobModel.load(id=job['_id'], force=True, includeLog=True)
self.assertEqual(job['status'], JobStatus.CANCELED)
self.assertEqual(len(job.get('log', [])), 1)
def testCancelJobEndpoint(self):
job = self.jobModel.createJob(title='test', type='x', user=self.users[0])
# Ensure requires write perms
jobCancelUrl = '/job/%s/cancel' % job['_id']
resp = self.request(jobCancelUrl, user=self.users[1], method='PUT')
self.assertStatus(resp, 403)
# Try again with the right user
jobCancelUrl = '/job/%s/cancel' % job['_id']
resp = self.request(jobCancelUrl, user=self.users[0], method='PUT')
self.assertStatusOk(resp)
self.assertEqual(resp.json['status'], JobStatus.CANCELED)
def testJobsTypesAndStatuses(self):
self.jobModel.createJob(title='user 0 job', type='t1', user=self.users[0], public=False)
self.jobModel.createJob(title='user 1 job', type='t2', user=self.users[1], public=False)
self.jobModel.createJob(title='user 1 job', type='t3', user=self.users[1], public=True)
self.jobModel.createJob(title='user 2 job', type='t4', user=self.users[2])
self.jobModel.createJob(title='anonymous job', type='t5')
self.jobModel.createJob(title='anonymous public job', type='t6', public=True)
# User 1, as non site admin, should encounter http 403 (Forbidden)
resp = self.request('/job/typeandstatus/all', user=self.users[1])
self.assertStatus(resp, 403)
# Admin user gets all types and statuses
resp = self.request('/job/typeandstatus/all', user=self.users[0])
self.assertStatusOk(resp)
self.assertEqual(len(resp.json['types']), 6)
self.assertEqual(len(resp.json['statuses']), 1)
# standard user gets types and statuses of its own jobs
resp = self.request('/job/typeandstatus', user=self.users[1])
self.assertStatusOk(resp)
self.assertEqual(len(resp.json['types']), 2)
self.assertEqual(len(resp.json['statuses']), 1)
def testDefaultParentId(self):
job = self.jobModel.createJob(title='Job', type='Job', user=self.users[0])
# If not specified parentId should be None
self.assertEquals(job['parentId'], None)
def testIsParentIdCorrect(self):
parentJob = self.jobModel.createJob(
title='Parent Job', type='Parent Job', user=self.users[0])
childJob = self.jobModel.createJob(
title='Child Job', type='Child Job', user=self.users[0], parentJob=parentJob)
# During initialization parent job should be set correctly
self.assertEqual(childJob['parentId'], parentJob['_id'])
def testSetParentCorrectly(self):
parentJob = self.jobModel.createJob(
title='Parent Job', type='Parent Job', user=self.users[0])
childJob = self.jobModel.createJob(title='Child Job', type='Child Job', user=self.users[0])
self.jobModel.setParentJob(childJob, parentJob)
# After setParentJob method is called parent job should be set correctly
self.assertEqual(childJob['parentId'], parentJob['_id'])
def testParentCannotBeEqualToChild(self):
childJob = self.jobModel.createJob(title='Child Job', type='Child Job', user=self.users[0])
# Cannot set a job as it's own parent
with self.assertRaises(ValidationException):
self.jobModel.setParentJob(childJob, childJob)
def testParentIdCannotBeOverridden(self):
parentJob = self.jobModel.createJob(
title='Parent Job', type='Parent Job', user=self.users[0])
anotherParentJob = self.jobModel.createJob(
title='Another Parent Job', type='Parent Job', user=self.users[0])
childJob = self.jobModel.createJob(
title='Child Job', type='Child Job', user=self.users[0], parentJob=parentJob)
with self.assertRaises(ValidationException):
# If parent job is set, cannot be overridden
self.jobModel.setParentJob(childJob, anotherParentJob)
def testListChildJobs(self):
parentJob = self.jobModel.createJob(
title='Parent Job', type='Parent Job', user=self.users[0])
childJob = self.jobModel.createJob(
title='Child Job', type='Child Job', user=self.users[0], parentJob=parentJob)
self.jobModel.createJob(
title='Another Child Job', type='Child Job', user=self.users[0], parentJob=parentJob)
# Should return a list with 2 jobs
self.assertEquals(len(list(self.jobModel.listChildJobs(parentJob))), 2)
# Should return an empty list
self.assertEquals(len(list(self.jobModel.listChildJobs(childJob))), 0)
def testListChildJobsRest(self):
parentJob = self.jobModel.createJob(
title='Parent Job', type='Parent Job', user=self.users[0])
childJob = self.jobModel.createJob(
title='Child Job', type='Child Job', user=self.users[0], parentJob=parentJob)
self.jobModel.createJob(
title='Another Child Job', type='Child Job', user=self.users[0], parentJob=parentJob)
resp = self.request('/job', user=self.users[0],
params={'parentId': str(parentJob['_id'])})
resp2 = self.request('/job', user=self.users[0],
params={'parentId': str(childJob['_id'])})
self.assertStatusOk(resp)
self.assertStatusOk(resp2)
# Should return a list with 2 jobs
self.assertEquals(len(resp.json), 2)
# Should return an empty list
self.assertEquals(len(resp2.json), 0)
def testCreateJobRest(self):
resp = self.request('/job', method='POST',
user=self.users[0],
params={'title': 'job', 'type': 'job'})
# If user does not have the necessary token status is 403
self.assertStatus(resp, 403)
token = Token().createToken(scope=REST_CREATE_JOB_TOKEN_SCOPE)
resp2 = self.request(
'/job', method='POST', token=token, params={'title': 'job', 'type': 'job'})
# If user has the necessary token status is 200
self.assertStatusOk(resp2)
def testJobStateTransitions(self):
job = self.jobModel.createJob(
title='user 0 job', type='t1', user=self.users[0], public=False)
# We can't move straight to SUCCESS
with self.assertRaises(ValidationException):
job = self.jobModel.updateJob(job, status=JobStatus.SUCCESS)
self.jobModel.updateJob(job, status=JobStatus.QUEUED)
self.jobModel.updateJob(job, status=JobStatus.RUNNING)
self.jobModel.updateJob(job, status=JobStatus.ERROR)
# We shouldn't be able to move backwards
with self.assertRaises(ValidationException):
self.jobModel.updateJob(job, status=JobStatus.QUEUED)
with self.assertRaises(ValidationException):
self.jobModel.updateJob(job, status=JobStatus.RUNNING)
with self.assertRaises(ValidationException):
self.jobModel.updateJob(job, status=JobStatus.INACTIVE)
def testJobSaveEventModification(self):
def customSave(event):
kwargs = json_util.loads(event.info['kwargs'])
kwargs['key2'] = 'newvalue'
event.info['kwargs'] = json_util.dumps(kwargs)
job = self.jobModel.createJob(title='A job', type='t', user=self.users[1], public=True)
job['kwargs'] = {'key1': 'value1', 'key2': 'value2'}
with events.bound('model.job.save', 'test', customSave):
job = self.jobModel.save(job)
self.assertEqual(job['kwargs']['key2'], 'newvalue')
