def upload_zipped_flat_media_files(
gc: GirderClient,
manager: JobManager,
folderId: str,
working_directory: Path,
create_subfolder=False,
):
"""Takes a flat folder of media files and/or annotation and generates a dataset from it"""
listOfFileNames = os.listdir(working_directory)
validation = gc.sendRestRequest('POST',
'/dive_dataset/validate_files',
json=listOfFileNames)
root_folderId = folderId
default_fps = gc.getFolder(root_folderId).get(
f"meta.{constants.FPSMarker}", -1)
if validation.get('ok', False):
manager.write(f"Annotations: {validation['annotations']}\n")
manager.write(f"Media: {validation['media']}\n")
dataset_type = validation['type']
manager.write(f"Type: {dataset_type}\n")
if create_subfolder != '':
sub_folder = gc.createFolder(
folderId,
create_subfolder,
reuseExisting=True,
)
root_folderId = str(sub_folder["_id"])
# Upload all resulting items back into the root folder
manager.updateStatus(JobStatus.PUSHING_OUTPUT)
# create a source folder to place the zipFile inside of
gc.upload(f'{working_directory}/*', root_folderId)
if dataset_type == constants.ImageSequenceType and default_fps == -1:
default_fps = 1
gc.addMetadataToFolder(
str(root_folderId),
{
constants.TypeMarker: dataset_type,
constants.FPSMarker: default_fps
},
)
# After uploading the default files we do a the postprocess for video conversion now
gc.sendRestRequest("POST",
f"/dive_rpc/postprocess/{str(root_folderId)}")
else:
manager.write(f"Message: {validation['message']}\n")
raise Exception("Could not Validate media Files")
def test_upload_zip_data(dataset: dict):
user = zipUser
client = GirderClient(apiUrl='http://localhost:8010/api/v1')
client.authenticate(username=user['login'], password=user['password'])
dsPath = localDataRoot / str(dataset['path'])
privateFolder = getTestFolder(client)
newDatasetFolder = client.createFolder(
privateFolder['_id'],
dataset['name'],
metadata={
'fps': dataset['fps'],
'type': dataset['type'],
},
)
if Path(dsPath).is_file():
client.uploadFileToFolder(newDatasetFolder['_id'], str(dsPath))
client.post(f'dive_rpc/postprocess/{newDatasetFolder["_id"]}')
wait_for_jobs(client,
max_wait_timeout=30,
expected_status=dataset['job_status'])
resultFolder = client.getFolder(newDatasetFolder['_id'])
# verify sub datasets if they exist
if dataset.get('subDatasets', False):
folders = list(client.listFolder(newDatasetFolder['_id']))
for item in dataset["subDatasets"]:
matches = [x for x in folders if x["name"] == item["name"]]
if len(matches) > 0:
meta = matches[0].get("meta", {})
assert meta.get("fps", -1) == item["fps"]
assert meta.get("type", "") == item["type"]
assert meta.get("annotate", False)
elif dataset['job_status'] == JobStatus.SUCCESS:
assert resultFolder['meta'].get("annotate", False)
assert type(resultFolder['meta'].get("fps")) in [int, float]
assert type(resultFolder['meta'].get("type")) == str
else:
assert resultFolder['meta'].get("annotate", None) is None
class CumulusClient():
'''Application interface to cumulus-based client for HPC systems
supporting NEWT API.
Note: the methods must be called in a specific order!
create_cluster()
create_omega3p_script()
create_job()
upload_inputs()
submit_job()
Then optionally:
monitor_job()
download_results()
release_resources()
'''
# ---------------------------------------------------------------------
def __init__(self, girder_url, newt_sessionid):
'''
'''
self._client = None
self._cluster_id = None
self._girder_url = girder_url
self._input_folder_id = None
self._job_folder_id = None
self._job_id = None
self._output_folder_id = None
self._private_folder_id = None
self._script_id = None
self._session = requests.Session()
# Authenticate with Girder using the newt session id
url = '%s/api/v1/newt/authenticate/%s' % \
(self._girder_url, newt_sessionid)
r = self._session.put(url)
if r.status_code != 200:
raise HttpError(r.status_code, r.text, r.url, r.request.method)
# Instantiate Girder client
url = '%s/api/v1' % self._girder_url
self._client = GirderClient(apiUrl=url)
self._client.token = self._session.cookies['girderToken']
user = self._client.get('user/me')
#print 'user', user
user_id = user['_id']
r = self._client.listFolder(user_id, 'user', name='Private')
if len(r) != 1:
raise Exception('Wrong number of users; should be 1 got %s' %
len(r))
self._private_folder_id = r[0]['_id']
print 'private_folder_id', self._private_folder_id
# ---------------------------------------------------------------------
def job_id(self):
'''Returns current job id (which may be None)
'''
return self._job_id
# ---------------------------------------------------------------------
def create_cluster(self, machine_name, cluster_name=None):
'''
'''
if cluster_name is None:
user = self._client.get('user/me')
user_name = user.get('firstName', 'user')
cluster_name = '%s.%s' % (machine_name, user_name)
cluster = None
cluster_list = self._client.get('clusters')
for extant_cluster in cluster_list:
if extant_cluster['name'] == cluster_name:
cluster = extant_cluster
self._cluster_id = extant_cluster['_id']
break
if not cluster:
body = {
'config': {
'host': machine_name
},
'name': cluster_name,
'type': 'newt'
}
r = self._client.post('clusters', data=json.dumps(body))
self._cluster_id = r['_id']
print 'cluster_id', self._cluster_id
# Reset the state of the cluster
body = {'status': 'created'}
r = self._client.patch('clusters/%s' % self._cluster_id,
data=json.dumps(body))
# Now test the connection
r = self._client.put('clusters/%s/start' % self._cluster_id)
sleeps = 0
while True:
time.sleep(1)
r = self._client.get('clusters/%s/status' % self._cluster_id)
if r['status'] == 'running':
break
elif r['status'] == 'error':
r = self._client.get('clusters/%s/log' % self._cluster_id)
print r
raise Exception('ERROR creating cluster')
if sleeps > 9:
raise Exception('Cluster never moved into running state')
sleeps += 1
# ---------------------------------------------------------------------
def create_omega3p_script(self,
omega3p_filename,
name=None,
number_of_tasks=1):
'''Creates script to submit omega3p job
'''
command = 'srun -n %s /project/projectdirs/ace3p/{{machine}}/omega3p %s' % \
(number_of_tasks, omega3p_filename)
if name is None:
name = omega3p_filename
body = {'commands': [command], 'name': name}
r = self._client.post('scripts', data=json.dumps(body))
self._script_id = r['_id']
print 'script_id', self._script_id
# ---------------------------------------------------------------------
def create_input(self, input_paths, folder_name='input_files'):
'''DEPRECATED Uploads input files
'''
folder_id = self.get_folder(self._private_folder_id, folder_name)
if folder_id is None:
return
print 'input_folder_id', folder_id
self._input_folder_id = folder_id
def upload_file(path):
name = os.path.basename(path)
size = os.path.getsize(path)
with open(path, 'rb') as fp:
self._client.uploadFile(self._input_folder_id,
fp,
name,
size,
parentType='folder')
for input_path in input_paths:
if not input_path or not os.path.exists(input_path):
raise Exception('Input file not found: %s' % input_path)
upload_file(input_path)
# ---------------------------------------------------------------------
def create_output_folder(self, folder_name='output_files'):
'''DEPRECATED
'''
folder_id = self.get_folder(self._private_folder_id, folder_name)
print 'output_folder_id', folder_id
self._output_folder_id = folder_id
# ---------------------------------------------------------------------
def create_job(self, job_name, tail=None):
'''
'''
# Create job folders
folder_name = uuid.uuid4().hex # unique name
self._job_folder_id = self.get_folder(self._private_folder_id,
folder_name)
print 'Created job folder', folder_name
self._input_folder_id = self.get_folder(self._job_folder_id,
'input_files')
self._output_folder_id = self.get_folder(self._job_folder_id,
'output_files')
# Make sure job_name isn't null
if not job_name:
job_name = 'CumulusJob'
# Create job spec
body = {
'name': job_name,
'scriptId': self._script_id,
'output': [{
'folderId': self._output_folder_id,
'path': '.'
}],
'input': [{
'folderId': self._input_folder_id,
'path': '.'
}]
}
if tail:
body['output'].append({"path": tail, "tail": True})
job = self._client.post('jobs', data=json.dumps(body))
self._job_id = job['_id']
print 'Created job_id', self._job_id
# ---------------------------------------------------------------------
def upload_inputs(self, input_paths):
'''Uploads input files to input folder
'''
if not self._input_folder_id:
raise Exception('Input folder missing')
def upload_file(path):
name = os.path.basename(path)
size = os.path.getsize(path)
with open(path, 'rb') as fp:
self._client.uploadFile(self._input_folder_id,
fp,
name,
size,
parentType='folder')
for input_path in input_paths:
if not input_path or not os.path.exists(input_path):
raise Exception('Input file not found: %s' % input_path)
upload_file(input_path)
# ---------------------------------------------------------------------
def submit_job(self,
machine,
project_account,
timeout_minutes,
queue='debug',
qos=None,
number_of_nodes=1,
job_output_dir=None):
'''
'''
body = {
'machine': machine,
'account': project_account,
'numberOfNodes': number_of_nodes,
'maxWallTime': {
'hours': 0,
'minutes': timeout_minutes,
'seconds': 0
},
'queue': queue,
}
if qos:
body['qualityOfService'] = qos
#print 'jobOutputDir', job_output_dir
if job_output_dir:
body['jobOutputDir'] = job_output_dir
print 'Setting jobOutputDir', job_output_dir
url = 'clusters/%s/job/%s/submit' % (self._cluster_id, self._job_id)
self._client.put(url, data=json.dumps(body))
print 'Submitted job', self._job_id
# ---------------------------------------------------------------------
def monitor_job(self, tail=None):
'''Periodically monitors job status
'''
log_offset = 0
job_timeout = 60 * timeout_minutes
start = time.time()
while True:
time.sleep(2)
# Provide some feedback at startup
if log_offset == 0:
sys.stdout.write('.')
#print 'Checking status'
r = self._client.get('jobs/%s' % self._job_id)
#print r
if r['status'] in ['error', 'unexpectederror']:
r = self._client.get('jobs/%s/log' % self._job_id)
raise Exception(str(r))
elif r['status'] == 'complete':
break
# Tail log file
if tail:
params = {'offset': log_offset, 'path': tail}
#print 'Checking tail'
r = self._client.get('jobs/%s/output' % self._job_id,
parameters=params)
#print r
output = r['content']
if output and log_offset == 0:
print # end the user feedback dots
log_offset += len(output)
for l in output:
print l
sys.stdout.flush()
if time.time() - start > job_timeout:
raise Exception('Job timeout')
# ---------------------------------------------------------------------
def download_results(self, destination_folder):
'''Downloads all output files to a local directory
'''
if not os.path.exists(destination_folder):
os.makedirs(destination_folder)
self._client.downloadFolderRecursive(self._output_folder_id,
destination_folder)
print 'Downloaded files to %s' % destination_folder
# ---------------------------------------------------------------------
def release_resources(self):
'''Closes/deletes any current resources
'''
resource_info = {
'clusters': [self._cluster_id],
'jobs': [self._job_id],
'scripts': [self._script_id],
'folder': [self._job_folder]
}
for resource_type, id_list in resource_info.items():
for resource_id in id_list:
if resource_id is not None:
url = '%s/%s' % (resource_type, resource_id)
self._client.delete(url)
self._input_folder_id = None
self._job_folder_id = None
self._job_id = None
self._output_folder_id = None
self._script_id = None
# ---------------------------------------------------------------------
def get_folder(self, parent_id, name):
'''Returns folder_id, creating one if needed
'''
# Check if folder already exists
folder_list = self._client.listFolder(parent_id, name=name)
if folder_list:
folder = folder_list[0]
#print 'found folder %s: %s' % (name, str(folder))
return folder['_id']
# (else)
try:
r = self._client.createFolder(parent_id, name)
return r['_id']
except HttpError as e:
print e.responseText
return None
def upload_exported_zipped_dataset(
gc: GirderClient,
manager: JobManager,
folderId: str,
working_directory: Path,
create_subfolder='',
):
"""Uploads a folder that is generated from the export of a zip file and sets metadata"""
listOfFileNames = os.listdir(working_directory)
potential_meta_files = list(
filter(constants.metaRegex.match, listOfFileNames))
if len(potential_meta_files) == 0:
manager.write(
"Could not find meta.json or config.json file within the subdirectroy\n"
)
return
print(listOfFileNames)
# load meta.json to get datatype and verify list of files
meta = {}
for meta_name in potential_meta_files:
with open(f"{working_directory}/{meta_name}") as f:
meta = json.load(f)
type = meta[constants.TypeMarker]
if type == constants.ImageSequenceType:
imageData = meta['imageData']
for image in imageData:
if image["filename"] not in listOfFileNames:
manager.write(
"Could not find {item['filename']} file within the list of files\n"
)
return
elif type == constants.VideoType:
video = meta["video"]
if video["filename"] not in listOfFileNames:
manager.write(
"Could not find {item['filename']} file within the list of files\n"
)
return
# remove the auxilary directory so we don't have to tag them all
if constants.AuxiliaryFolderName in listOfFileNames and os.path.isdir(
f'{working_directory}/{constants.AuxiliaryFolderName}'):
shutil.rmtree(f'{working_directory}/{constants.AuxiliaryFolderName}')
root_folderId = folderId
if create_subfolder != '':
sub_folder = gc.createFolder(
folderId,
create_subfolder,
reuseExisting=True,
)
root_folderId = str(sub_folder['_id'])
manager.updateStatus(JobStatus.PUSHING_OUTPUT)
# create a source folder to place the zipFile inside of
gc.upload(f'{working_directory}/*', root_folderId)
# Now we set all the metadata for the folders and items
all_files = list(gc.listItem(root_folderId))
root_meta = {
"type": type,
"attributes": meta.get("attributes", None),
"customTypeStyling": meta.get("customTypeStyling", None),
"confidenceFilters": meta.get("confidenceFilters", None),
"fps": meta["fps"],
"version": meta["version"],
}
if type == constants.VideoType:
# set transcoded and non-transcoded versions
transcoded_video = list(
gc.listItem(root_folderId, name=video["filename"]))
if len(transcoded_video) == 1:
ffprobe = meta["ffprobe_info"]
avgFpsString = ffprobe["avg_frame_rate"]
dividend, divisor = [int(v) for v in avgFpsString.split('/')]
originalFps = dividend / divisor
transcoded_metadata = {
"codec": "h264",
"originalFps": originalFps,
"originalFpsString": avgFpsString,
"source_video": False,
"transcoder": "ffmpeg",
}
gc.addMetadataToItem(str(transcoded_video[0]['_id']),
transcoded_metadata)
# other video is tagged as the source video
for item in all_files:
if (item["name"].endswith(tuple(constants.validVideoFormats))
and item["name"] != video["filename"]):
source_metadata = {
"codec": ffprobe["codec_name"],
"originalFps": originalFps,
"originalFpsString": avgFpsString,
"source_video": False,
}
gc.addMetadataToItem(str(item['_id']), source_metadata)
root_meta["originalFps"] = originalFps
root_meta["originalFpsString"] = avgFpsString
# Need to tag folder Level data (annotate, and others)
root_meta[constants.DatasetMarker] = True
gc.addMetadataToFolder(root_folderId, root_meta)
gc.post(f'dive_rpc/postprocess/{root_folderId}', data={"skipJobs": True})
class BaseIntegrationTest(unittest.TestCase):
def __init__(self, name, girder_url, girder_user, girder_password, job_timeout=60, cleanup=True):
super(BaseIntegrationTest, self).__init__(name)
self._job_id = None
self._script_id = None
self._output_folder_id = None
self._input_folder_id = None
self._girder_url = girder_url
self._girder_user = girder_user
self._girder_password = girder_password
self._job_timeout = job_timeout
self._data = 'Need more input!'
self._cleanup = cleanup
def setUp(self):
url = '%s/api/v1' % self._girder_url
self._client = GirderClient(apiUrl=url)
self._client.authenticate(self._girder_user,
self._girder_password)
user = self._client.get('user/me')
self._user_id = user['_id']
r = list(self._client.listFolder(self._user_id, 'user', name='Private'))
self.assertEqual(len(r), 1)
self._private_folder_id = r[0]['_id']
def tearDown(self):
if not self._cleanup:
return
if self._job_id:
try:
url = 'jobs/%s' % self._job_id
self._client.delete(url)
except Exception as e:
traceback.print_exc()
if self._script_id:
try:
url = 'scripts/%s' % self._script_id
self._client.delete(url)
except Exception:
traceback.print_exc()
if self._output_folder_id:
try:
url = 'folder/%s' % self._output_folder_id
self._client.delete(url)
except Exception:
traceback.print_exc()
if self._input_folder_id:
try:
url = 'folder/%s' % self._input_folder_id
self._client.delete(url)
except Exception:
traceback.print_exc()
def create_script(self, commands=[
'sleep 10', 'cat CumulusIntegrationTestInput'
]):
body = {
'commands': commands,
'name': 'CumulusIntegrationTestLob'
}
r = self._client.post('scripts', data=json.dumps(body))
self._script_id = r['_id']
def create_input(self, folder_name='CumulusInput'):
r = self._client.createFolder(self._private_folder_id, folder_name)
self._input_folder_id = r['_id']
size = len(self._data)
item = self._client.uploadFile(self._input_folder_id,
StringIO(self._data), 'CumulusIntegrationTestInput', size,
parentType='folder')
self._item_id = item['itemId']
def create_output_folder(self, folder_name='CumulusOutput'):
r = self._client.createFolder(self._private_folder_id, folder_name)
self._output_folder_id = r['_id']
def create_job(self, job_name='CumulusIntegrationTestJob', tail=None):
body = {
'name': job_name,
'scriptId': self._script_id,
'output': [{
'folderId': self._output_folder_id,
'path': '.'
}],
'input': [
{
'folderId': self._input_folder_id,
'path': '.'
}
]
}
if tail:
body['output'].append({
"path": tail,
"tail": True
})
job = self._client.post('jobs', data=json.dumps(body))
self._job_id = job['_id']
def submit_job(self, job_params={}, timeout=None):
url = 'clusters/%s/job/%s/submit' % (self._cluster_id, self._job_id)
self._client.put(url, data=json.dumps(job_params))
start = time.time()
while True:
time.sleep(1)
r = self._client.get('jobs/%s' % self._job_id)
if r['status'] in ['error', 'unexpectederror']:
r = self._client.get('jobs/%s/log' % self._job_id)
self.fail(str(r))
elif r['status'] == 'complete':
break
if time.time() - start > timeout:
self.fail('Job didn\'t complete in timeout')
def assert_output(self):
r = self._client.listItem(self._output_folder_id)
self.assertEqual(len(r), 4)
stdout_item = None
for i in r:
if i['name'].startswith('CumulusIntegrationTestJob-%s.o' % self._job_id):
stdout_item = i
break
self.assertIsNotNone(stdout_item)
r = self._client.get('item/%s/files' % i['_id'])
self.assertEqual(len(r), 1)
path = os.path.join(tempfile.gettempdir(), self._job_id)
try:
self._client.downloadFile(r[0]['_id'], path)
with open(path, 'rb') as fp:
self.assertEqual(fp.read(), self._data)
finally:
if os.path.exists(path):
os.remove(path)
class BaseIntegrationTest(unittest.TestCase):
def __init__(self,
name,
girder_url,
girder_user,
girder_password,
job_timeout=60,
cleanup=True):
super(BaseIntegrationTest, self).__init__(name)
self._job_id = None
self._script_id = None
self._output_folder_id = None
self._input_folder_id = None
self._girder_url = girder_url
self._girder_user = girder_user
self._girder_password = girder_password
self._job_timeout = job_timeout
self._data = 'Need more input!'
self._cleanup = cleanup
def setUp(self):
url = '%s/api/v1' % self._girder_url
self._client = GirderClient(apiUrl=url)
self._client.authenticate(self._girder_user, self._girder_password)
user = self._client.get('user/me')
self._user_id = user['_id']
r = list(self._client.listFolder(self._user_id, 'user',
name='Private'))
self.assertEqual(len(r), 1)
self._private_folder_id = r[0]['_id']
def tearDown(self):
if not self._cleanup:
return
if self._job_id:
try:
url = 'jobs/%s' % self._job_id
self._client.delete(url)
except Exception as e:
traceback.print_exc()
if self._script_id:
try:
url = 'scripts/%s' % self._script_id
self._client.delete(url)
except Exception:
traceback.print_exc()
if self._output_folder_id:
try:
url = 'folder/%s' % self._output_folder_id
self._client.delete(url)
except Exception:
traceback.print_exc()
if self._input_folder_id:
try:
url = 'folder/%s' % self._input_folder_id
self._client.delete(url)
except Exception:
traceback.print_exc()
def create_script(self,
commands=['sleep 10',
'cat CumulusIntegrationTestInput']):
body = {'commands': commands, 'name': 'CumulusIntegrationTestLob'}
r = self._client.post('scripts', data=json.dumps(body))
self._script_id = r['_id']
def create_input(self, folder_name='CumulusInput'):
r = self._client.createFolder(self._private_folder_id, folder_name)
self._input_folder_id = r['_id']
size = len(self._data)
item = self._client.uploadFile(self._input_folder_id,
StringIO(self._data),
'CumulusIntegrationTestInput',
size,
parentType='folder')
self._item_id = item['itemId']
def create_output_folder(self, folder_name='CumulusOutput'):
r = self._client.createFolder(self._private_folder_id, folder_name)
self._output_folder_id = r['_id']
def create_job(self, job_name='CumulusIntegrationTestJob', tail=None):
body = {
'name': job_name,
'scriptId': self._script_id,
'output': [{
'folderId': self._output_folder_id,
'path': '.'
}],
'input': [{
'folderId': self._input_folder_id,
'path': '.'
}]
}
if tail:
body['output'].append({"path": tail, "tail": True})
job = self._client.post('jobs', data=json.dumps(body))
self._job_id = job['_id']
def submit_job(self, job_params={}, timeout=None):
url = 'clusters/%s/job/%s/submit' % (self._cluster_id, self._job_id)
self._client.put(url, data=json.dumps(job_params))
start = time.time()
while True:
time.sleep(1)
r = self._client.get('jobs/%s' % self._job_id)
if r['status'] in ['error', 'unexpectederror']:
r = self._client.get('jobs/%s/log' % self._job_id)
self.fail(str(r))
elif r['status'] == 'complete':
break
if time.time() - start > timeout:
self.fail('Job didn\'t complete in timeout')
def assert_output(self):
r = self._client.listItem(self._output_folder_id)
self.assertEqual(len(r), 4)
stdout_item = None
for i in r:
if i['name'].startswith('CumulusIntegrationTestJob-%s.o' %
self._job_id):
stdout_item = i
break
self.assertIsNotNone(stdout_item)
r = self._client.get('item/%s/files' % i['_id'])
self.assertEqual(len(r), 1)
path = os.path.join(tempfile.gettempdir(), self._job_id)
try:
self._client.downloadFile(r[0]['_id'], path)
with open(path, 'rb') as fp:
self.assertEqual(fp.read(), self._data)
finally:
if os.path.exists(path):
os.remove(path)
