def create(self, module):
try:
upload_files, total_file_size = get_files_in_current_directory(
file_type='code')
except OSError:
sys.exit(
"Directory contains too many files to upload. If you have data files in the current directory, "
"please upload them separately using \"floyd data\" command and remove them from here.\n"
"See http://docs.floydhub.com/faqs/job/#i-get-too-many-open-files-error-when-i-run-my-project "
"for more details on how to fix this.")
if total_file_size > self.MAX_UPLOAD_SIZE:
sys.exit((
"Code size too large to sync, please keep it under %s.\n"
"If you have data files in the current directory, please upload them "
"separately using \"floyd data\" command and remove them from here.\n"
"You may find the following documentation useful:\n\n"
"\thttps://docs.floydhub.com/guides/create_and_upload_dataset/\n"
"\thttps://docs.floydhub.com/guides/data/mounting_data/\n"
"\thttps://docs.floydhub.com/guides/floyd_ignore/") %
(sizeof_fmt(self.MAX_UPLOAD_SIZE)))
floyd_logger.info("Creating project run. Total upload size: %s",
sizeof_fmt(total_file_size))
floyd_logger.debug("Creating module. Uploading: %s files",
len(upload_files))
floyd_logger.info("Syncing code ...")
# Add request data
upload_files.append(("json", json.dumps(module.to_dict())))
multipart_encoder = MultipartEncoder(fields=upload_files)
# Attach progress bar
progress_callback, bar = create_progress_callback(multipart_encoder)
multipart_encoder_monitor = MultipartEncoderMonitor(
multipart_encoder, progress_callback)
try:
response = self.request(
"POST",
self.url,
data=multipart_encoder_monitor,
headers={"Content-Type": multipart_encoder.content_type},
timeout=3600)
finally:
# always make sure we clear the console
bar.done()
return response.json().get("id")
def create(self, data):
"""
Create a temporary directory for the tar file that will be removed at the
end of the operation.
"""
with tempfile.TemporaryDirectory() as temp_directory:
floyd_logger.info("Compressing data ...")
compressed_file_path = os.path.join(temp_directory, "data.tar.gz")
# Create tarfile
floyd_logger.debug("Creating tarfile with contents of current directory: {}".format(compressed_file_path))
create_tarfile(source_dir='.', filename=compressed_file_path)
total_file_size = os.path.getsize(compressed_file_path)
floyd_logger.info("Creating data source. Total upload size: {}".format(sizeof_fmt(total_file_size)))
floyd_logger.info("Uploading compressed data ...")
# Add request data
request_data = []
request_data.append(("data", ('data.tar', open(compressed_file_path, 'rb'), 'text/plain')))
request_data.append(("json", json.dumps(data.to_dict())))
multipart_encoder = MultipartEncoder(
fields=request_data
)
# Attach progress bar
progress_callback = create_progress_callback(multipart_encoder)
multipart_encoder_monitor = MultipartEncoderMonitor(multipart_encoder, progress_callback)
response = self.request("POST",
self.url,
data=multipart_encoder_monitor,
headers={"Content-Type": multipart_encoder.content_type},
timeout=3600)
floyd_logger.info("Done")
return response.json().get("id")
def complete_upload(data_config):
data_endpoint = data_config.data_endpoint
data_id = data_config.data_id
tarball_path = data_config.tarball_path
if not data_id:
floyd_logger.error("Corrupted upload state, please start a new one.")
sys.exit(1)
# check for tarball upload, upload to server if not done
if not data_config.resource_id and (tarball_path and data_endpoint):
floyd_logger.debug("Getting fresh upload credentials")
creds = DataClient().new_tus_credentials(data_id)
if not creds:
sys.exit(1)
file_size = os.path.getsize(tarball_path)
# check for upload limit dimension
if file_size > MAX_UPLOAD_SIZE:
try:
floyd_logger.info("Removing compressed data...")
rmtree(os.path.dirname(tarball_path))
except (OSError, TypeError):
pass
sys.exit(("Data size too large to upload, please keep it under %s.\n") %
(sizeof_fmt(MAX_UPLOAD_SIZE)))
floyd_logger.info("Uploading compressed data. Total upload size: %s",
sizeof_fmt(file_size))
tus_client = TusDataClient()
if not tus_client.resume_upload(tarball_path, data_endpoint, auth=creds):
floyd_logger.error("Failed to finish upload!")
return
try:
floyd_logger.info("Removing compressed data...")
rmtree(os.path.dirname(tarball_path))
except (OSError, TypeError):
pass
floyd_logger.debug("Created data with id : %s", data_id)
floyd_logger.info("Upload finished.")
# Update data config
data_config.set_tarball_path(None)
data_config.set_data_endpoint(None)
data_source = DataClient().get(data_id)
data_config.set_resource_id(data_source.resource_id)
DataConfigManager.set_config(data_config)
# data tarball uploaded, check for server untar
if data_config.resource_id:
floyd_logger.info(
"Waiting for server to unpack data.\n"
"You can exit at any time and come back to check the status with:\n"
"\tfloyd data upload -r")
try:
for i in dots(ResourceWaitIter(data_config.resource_id),
label='Waiting for unpack...'):
pass
except WaitTimeoutException:
clint_STREAM.write('\n')
clint_STREAM.flush()
floyd_logger.info(
"Looks like it is going to take longer for Floydhub to unpack "
"your data. Please check back later.")
sys.exit(1)
else:
data_config.set_resource_id(None)
data_config.set_tarball_path(None)
data_config.set_data_endpoint(None)
data_config.set_resource_id(None)
data_config.set_data_id(None)
DataConfigManager.set_config(data_config)
# Print output
table_output = [["NAME"],
[normalize_data_name(data_config.data_name)]]
floyd_logger.info('')
floyd_logger.info(tabulate(table_output, headers="firstrow"))
