def __fetch_object(self, opts):
"""
Upload a file to Diffgram from an Azure Blob
:param opts: Dictionary with parameters for object fetching.
:return: file obj if file was uploaded, else False
"""
spec_list = [{'bucket_name': str, 'path': str}]
log = regular_log.default()
log, input = regular_input.input_check_many(untrusted_input = opts,
spec_list = spec_list,
log = log)
if len(log["error"].keys()) >= 1:
return {'log': log}
shared_access_signature = BlobSharedAccessSignature(
account_name = self.connection_client.account_name,
account_key = self.connection_client.credential.account_key
)
expiration_offset = 40368000
blob_name = opts['path']
container = opts['bucket_name']
added_seconds = datetime.timedelta(0, expiration_offset)
expiry_time = datetime.datetime.utcnow() + added_seconds
filename = blob_name.split("/")[-1]
sas = shared_access_signature.generate_blob(
container_name = container,
blob_name = blob_name,
start = datetime.datetime.utcnow(),
expiry = expiry_time,
permission = BlobSasPermissions(read = True),
content_disposition = 'attachment; filename=' + filename,
)
sas_url = 'https://{}.blob.core.windows.net/{}/{}?{}'.format(
self.connection_client.account_name,
container,
blob_name,
sas
)
with sessionMaker.session_scope() as session:
project = Project.get_by_string_id(session, self.config_data.get('project_string_id'))
member = session.query(Member).filter(Member.user_id == opts['event_data']['request_user']).first()
# Deduct Media Type:
extension = Path(opts['path']).suffix
extension = extension.lower()
media_type = None
if extension in images_allowed_file_names:
media_type = 'image'
elif extension in videos_allowed_file_names:
media_type = 'video'
else:
# TODO: Decide, do we want to raise an exception? or just do nothing?
log = regular_log.default()
log['error']['invalid_type'] = 'File must type of: {} {}'.format(str(images_allowed_file_names),
str(videos_allowed_file_names))
log['error']['file_name'] = opts['path']
log['opts'] = opts
Event.new(
session = session,
member_id = opts['event_data']['request_user'],
kind = 'microsoft_azure_new_import_warning',
description = 'Skipped import for {}, invalid file type.'.format(opts['path']),
error_log = log,
project_id = project.id,
member = member,
success = False
)
return None
# metadata = self.connection_client.head_object(Bucket=self.config_data['bucket_name'], Key=path)
created_input = packet.enqueue_packet(self.config_data['project_string_id'],
session = session,
media_url = sas_url,
media_type = media_type,
job_id = opts.get('job_id'),
batch_id = opts.get('batch_id'),
file_name = opts.get('path'),
video_split_duration = opts.get('video_split_duration'),
directory_id = opts.get('directory_id'),
extract_labels_from_batch = True)
log = regular_log.default()
log['opts'] = opts
Event.new(
session = session,
member_id = opts['event_data']['request_user'],
kind = 'microsoft_azure_new_import_success',
description = 'New cloud import for {}'.format(opts['path']),
error_log = opts,
project_id = project.id,
member = member,
success = True
)
return created_input
def __send_export(self, opts):
spec_list = [{'project_string_id': dict}]
log = regular_log.default()
log, input = regular_input.input_check_many(untrusted_input = self.config_data,
spec_list = spec_list,
log = log)
if len(log["error"].keys()) >= 1:
return {'log': log}
spec_list = [
{'path': str},
{"format": {
'default': 'JSON',
'kind': str,
'valid_values_list': ['JSON', 'YAML']
}},
{'export_id': str},
{'bucket_name': str},
]
log = regular_log.default()
log, input = regular_input.input_check_many(untrusted_input = opts,
spec_list = spec_list,
log = log,
string_len_not_zero = False)
if len(log["error"].keys()) >= 1:
return {'log': log}
if not opts['path'].endswith('/') and opts['path'] != '':
log['error']['path'] = 'Path on bucket must be a folder, not a filename.'
return log
with sessionMaker.session_scope() as session:
project = Project.get_by_string_id(session, self.config_data['project_string_id'])
member = session.query(Member).filter(Member.user_id == opts['event_data']['request_user']).first()
export = session.query(Export).filter(Export.id == opts['export_id']).first()
# Check perms and export status.
export_check_result = check_export_permissions_and_status(export,
self.config_data['project_string_id'],
session)
if len(export_check_result['error'].keys()) > 1:
return export_check_result
result = export_view_core(
export = export,
format = opts['format'],
return_type = 'bytes')
filename = generate_file_name_from_export(export, session)
if opts['path'] != '':
key = '{}{}.{}'.format(opts['path'], filename, opts['format'].lower())
else:
key = '{}.{}'.format(filename, opts['format'].lower())
file = io.BytesIO(result)
blob_client = self.connection_client.get_blob_client(container = opts['bucket_name'], blob = key)
content_type = mimetypes.guess_type(filename)[0]
my_content_settings = ContentSettings(content_type = content_type)
blob_client.upload_blob(file, content_settings = my_content_settings)
log = regular_log.default()
log['opts'] = opts
Event.new(
session = session,
member_id = opts['event_data']['request_user'],
kind = 'microsoft_azure_new_export_success',
description = 'New cloud export for {}{}'.format(opts['path'], filename),
error_log = opts,
member = member,
project_id = project.id,
success = True
)
return {'result': True}
def job_launch_list_core(session, metadata_proposed, output_mode="serialize"):
"""
Get the job_launch objects based on filters in
metadata_proposed.
"""
meta = default_metadata(metadata_proposed)
output_job_launch_list = []
limit_counter = 0
# CAUTION
# Multiple "modes", for output and trainer builder, maybe more in future
builder_or_trainer_mode = meta['builder_or_trainer']['mode']
# It doesn't really make sense to have this here
# Should be part of some other meta data checking or something.
if builder_or_trainer_mode not in ["builder", "trainer"]:
raise Forbidden("Invalid builder_or_trainer_mode mode.")
query = session.query(JobLaunch).join(Job)
user = User.get(session)
if user.last_builder_or_trainer_mode != builder_or_trainer_mode:
raise Forbidden(
"Invalid user relation to builder_or_trainer_mode mode.")
### START FILTERS ###
if meta["status"]:
if meta["status"] != "All":
query = query.filter(JobLaunch.status == meta["status"])
if meta["date_from"]:
date_from = datetime.datetime.strptime(meta["date_from"], "%Y-%m-%d")
date_from = date_from.replace(hour=0,
minute=0,
second=0,
microsecond=0)
query = query.filter(JobLaunch.time_created >= date_from)
if meta["date_to"]:
date_to = datetime.datetime.strptime(meta["date_to"], "%Y-%m-%d")
date_to = date_to.replace(hour=0, minute=0, second=0, microsecond=0)
query = query.filter(JobLaunch.time_created <= date_to)
if meta["job_ids"]:
query = query.filter(Job.id.in_(meta["job_ids"]))
# Also assumes org is None.
# Actually this should be complimentary still
if meta["project_string_id"]:
project = Project.get_by_string_id(
session=session, project_string_id=meta["project_string_id"])
query = query.filter(Job.project_id == project.id)
#### END FILTERS ###
query = query.order_by(Job.time_created.desc())
query = query.limit(meta["limit"])
query = query.offset(meta["start_index"])
job_launch_list = query.all()
if output_mode == "serialize":
for job_launch in job_launch_list:
serialized = job_launch.serialize_for_list_view(session=session)
output_job_launch_list.append(serialized)
limit_counter += 1
meta['end_index'] = meta['start_index'] + len(output_job_launch_list)
meta['length_current_page'] = len(output_job_launch_list)
if limit_counter == 0:
meta['no_results_match_meta'] = True
return output_job_launch_list, meta
def __fetch_folder(self, opts):
result = []
if self.config_data.get('project_string_id') is None:
return {'result': 'error'}
paths = opts['path']
if type(paths) != list:
paths = [paths]
with sessionMaker.session_scope() as session:
project = Project.get_by_string_id(
session, self.config_data.get('project_string_id'))
member = session.query(Member).filter(
Member.user_id == opts['event_data']['request_user']).first()
for path in paths:
blobs = self.connection_client.list_blobs(opts['bucket_name'],
prefix=path)
for blob in blobs:
# Deduct Media Type:
if blob.name.endswith('/'):
continue
blob_expiry = int(time.time() + (60 * 60 * 24 * 30))
signed_url = blob.generate_signed_url(
expiration=blob_expiry)
extension = Path(blob.path).suffix
media_type = None
if extension in images_allowed_file_names:
media_type = 'image'
elif extension in videos_allowed_file_names:
media_type = 'video'
else:
logging.warn('File: {} must type of: {} {}'.format(
blob.name, str(images_allowed_file_names),
str(videos_allowed_file_names)))
log = regular_log.default()
log['error'][
'invalid_type'] = 'File must type of: {} {}'.format(
str(images_allowed_file_names),
str(videos_allowed_file_names))
log['error']['file_name'] = path
log['opts'] = opts
Event.new(
session=session,
member_id=opts['event_data']['request_user'],
kind='google_cloud_new_import_warning',
description=
'Skipped import for {}, invalid file type.'.format(
blob.name),
error_log=log,
project_id=project.id,
member=member,
success=False)
continue
result = []
# TODO: check Input() table for duplicate file?
created_input = packet.enqueue_packet(
self.config_data['project_string_id'],
session=session,
media_url=signed_url,
media_type=media_type,
job_id=opts.get('job_id'),
batch_id=opts.get('batch_id'),
file_name=path,
video_split_duration=opts.get('video_split_duration'),
directory_id=opts.get('directory_id'),
extract_labels_from_batch=True)
log = regular_log.default()
log['opts'] = opts
Event.new(session=session,
member_id=opts['event_data']['request_user'],
kind='google_cloud_new_import_success',
description='New cloud import for {}'.format(
blob.name),
error_log=opts,
project_id=project.id,
member=member,
success=True)
result.append(created_input)
return result
def __fetch_object(self, opts):
"""Upload a file to diffgram from an S3 bucket
:param s3_file_key: path of file to fetch from
:return: file obj if file was uploaded, else False
"""
spec_list = [{'bucket_name': str, 'path': str}]
log = regular_log.default()
log, input = regular_input.input_check_many(untrusted_input=opts,
spec_list=spec_list,
log=log)
if len(log["error"].keys()) >= 1:
return {'log': log}
# This might be an issue. Currently not supporting urls with no expiration. Biggest time is 1 week.
signed_url = self.connection_client.generate_presigned_url('get_object',
Params={'Bucket': opts['bucket_name'],
'Key': opts['path']},
ExpiresIn=3600 * 24 * 6) # 5 Days.
with sessionMaker.session_scope() as session:
project = Project.get_by_string_id(session, self.config_data.get('project_string_id'))
member = session.query(Member).filter(Member.user_id == opts['event_data']['request_user']).first()
# Deduct Media Type:
extension = Path(opts['path']).suffix
extension = extension.lower()
media_type = None
if extension in images_allowed_file_names:
media_type = 'image'
elif extension in videos_allowed_file_names:
media_type = 'video'
else:
# TODO: Decide, do we want to raise an exception? or just do nothing?
log = regular_log.default()
log['error']['invalid_type'] = 'File must type of: {} {}'.format(str(images_allowed_file_names),
str(videos_allowed_file_names))
log['error']['file_name'] = opts['path']
log['opts'] = opts
Event.new(
session=session,
member_id=opts['event_data']['request_user'],
kind='aws_s3_new_import_warning',
description='Skipped import for {}, invalid file type.'.format(opts['path']),
error_log=log,
project_id=project.id,
member=member,
success=False
)
return None
# print('AAAAA', opts, opts.get('job_id'))
# metadata = self.connection_client.head_object(Bucket=self.config_data['bucket_name'], Key=path)
created_input = packet.enqueue_packet(self.config_data['project_string_id'],
session=session,
media_url=signed_url,
media_type=media_type,
job_id=opts.get('job_id'),
video_split_duration=opts.get('video_split_duration'),
directory_id=opts.get('directory_id'))
log = regular_log.default()
log['opts'] = opts
Event.new(
session=session,
member_id=opts['event_data']['request_user'],
kind='aws_s3_new_import_success',
description='New cloud import for {}'.format(opts['path']),
error_log=opts,
project_id=project.id,
member=member,
success=True
)
return created_input
