def init(dataset_name):
"""
Initialize a new dataset at the current dir.
Then run the upload command to copy all the files in this
directory to FloydHub.
floyd data upload
"""
dataset_obj = DatasetClient().get_by_name(dataset_name)
if not dataset_obj:
namespace, name = get_namespace_from_name(dataset_name)
create_dataset_base_url = "{}/datasets/create".format(
floyd.floyd_web_host)
create_dataset_url = "{}?name={}&namespace={}".format(
create_dataset_base_url, name, namespace)
floyd_logger.info(
("Dataset name does not match your list of datasets. "
"Create your new dataset in the web dashboard:\n\t%s"),
create_dataset_base_url)
webbrowser.open(create_dataset_url)
name = click.prompt(
'Press ENTER to use dataset name "%s" or enter a different name' %
dataset_name,
default=dataset_name,
show_default=False)
dataset_name = name.strip() or dataset_name
dataset_obj = DatasetClient().get_by_name(dataset_name)
if not dataset_obj:
raise FloydException(
'Dataset "%s" does not exist on floydhub.com. Ensure it exists before continuing.'
% dataset_name)
namespace, name = get_namespace_from_name(dataset_name)
data_config = DataConfig(name=name,
namespace=namespace,
family_id=dataset_obj.id)
DataConfigManager.set_config(data_config)
floyd_logger.info(
"Data source \"{}\" initialized in current directory".format(
dataset_name))
floyd_logger.info("""
You can now upload your data to Floyd by:
floyd data upload
""")
def init(project_name):
"""
Initialize new project at the current dir.
After init run your command. Example:
floyd run 'python tensorflow.py > /output/model.1'
"""
project_obj = ProjectClient().get_by_name(project_name)
if not project_obj:
namespace, name = get_namespace_from_name(project_name)
create_project_base_url = "{}/projects/create".format(
floyd.floyd_web_host)
create_project_url = "{}?name={}&namespace={}".format(
create_project_base_url, name, namespace)
floyd_logger.info(('Project name does not yet exist on floydhub.com. '
'Create your new project on floydhub.com:\n\t%s'),
create_project_base_url)
webbrowser.open(create_project_url)
name = click.prompt(
'Press ENTER to use project name "%s" or enter a different name' %
project_name,
default=project_name,
show_default=False)
project_name = name.strip() or project_name
project_obj = ProjectClient().get_by_name(project_name)
if not project_obj:
raise FloydException(
'Project "%s" does not exist on floydhub.com. Ensure it exists before continuing.'
% project_name)
namespace, name = get_namespace_from_name(project_name)
experiment_config = ExperimentConfig(name=name,
namespace=namespace,
family_id=project_obj.id)
ExperimentConfigManager.set_config(experiment_config)
FloydIgnoreManager.init()
yaml_config = read_yaml_config()
if not yaml_config:
copyfile(os.path.join(os.path.dirname(__file__), 'default_floyd.yml'),
'floyd.yml')
floyd_logger.info("Project \"%s\" initialized in current directory",
project_name)
def get_by_name(self, name, namespace=None):
"""
name: can be either <namespace>/<project_name> or just <project_name>
namespace: if specified, will skip name parsing, defaults to current user's username
"""
if not namespace:
namespace, name = get_namespace_from_name(name)
if not namespace:
namespace = AuthConfigManager.get_access_token().username
try:
response = self.request('GET',
'%s/%s/%s' % (self.url, namespace, name))
return Project.from_dict(response.json())
except NotFoundException:
return None
def test_get_namespace_from_name(self, _0):
"""
Test Regex for args of dataset and project initialization
"""
from floyd.cli.utils import get_namespace_from_name
# PATTERN: <dataset_or_project_name>
assert get_namespace_from_name('test') == ('pete', 'test')
# PATTERN: <namespace>/[dataset|project]/<dataset_or_project_name>
assert get_namespace_from_name('fo.o_user/datasets/test_1') == (
'fo.o_user', 'test_1')
assert get_namespace_from_name('fo.o_user/projects/test_1') == (
'fo.o_user', 'test_1')
# PATTERN: <namespace>/<dataset_or_project_name>
assert get_namespace_from_name('fo.o_user/test_1-1') == ('fo.o_user',
'test_1-1')
# BAD PATTERN: Bad argument (/)
with self.assertRaises(SystemExit) as cm:
get_namespace_from_name('/')
error = (
"Argument '/' doesn't match any recognized pattern:\n"
"\tfloyd [data] init <project_or_dataset_name>\n"
"\tfloyd [data] init <namespace>/<project_or_dataset_name>\n"
"\tfloyd [data] init <namespace>/[projects|dataset]/<project_or_dataset_name>\n"
"\n Note: Argument can only contain alphanumeric, hyphen-minus '-' , underscore '_' and dot '.' characters."
)
self.assertEqual(cm.exception.code, error)
# BAD PATTERN: Characters not allowed (:?)
with self.assertRaises(SystemExit) as cm:
get_namespace_from_name('test_:?1-1')
error = (
"Argument 'test_:?1-1' doesn't match any recognized pattern:\n"
"\tfloyd [data] init <project_or_dataset_name>\n"
"\tfloyd [data] init <namespace>/<project_or_dataset_name>\n"
"\tfloyd [data] init <namespace>/[projects|dataset]/<project_or_dataset_name>\n"
"\n Note: Argument can only contain alphanumeric, hyphen-minus '-' , underscore '_' and dot '.' characters."
)
self.assertEqual(cm.exception.code, error)
