def test_token_without_user_path(tmpdir):
config_path = tmpdir / 'substra.json'
config_path.write_text(json.dumps(configuration.DEFAULT_CONFIG), "UTF-8")
client = substra.Client(config_path=config_path, profile_name='default')
assert client._current_profile['token'] == ''
client = substra.Client(config_path=config_path,
profile_name='default',
token='foo')
assert client._current_profile['token'] == 'foo'
def test_client_multi_nodes_dataset(dataset_query):
"""Assert that the owner is gotten from the metadata in debug mode"""
client = substra.Client(debug=True)
dataset_query['metadata'] = {substra.DEBUG_OWNER: 'owner_1'}
key = client.add_dataset(dataset_query)
asset = client.get_dataset(key)
assert asset.owner == 'owner_1'
def main(network):
client = substra.Client()
if network == 'docker':
client.add_profile('default', 'substra', 'p@$swr0d44',
'http://substra-backend.owkin.xyz', '0.0')
elif network == 'skaffold':
client.add_profile('default', 'node-1', 'p@$swr0d44',
'http://substra-backend.node-1.com', '0.0')
else:
raise Exception('Unknow network')
client.login()
client.set_profile('default')
assets = {}
assets['objective'] = json.dumps(client.list_objective(), indent=4)
assets['datamanager'] = json.dumps(client.list_dataset(), indent=4)
assets['algo'] = json.dumps(client.list_algo(), indent=4)
assets['traintuple'] = json.dumps(client.list_traintuple(), indent=4)
assets['testtuple'] = json.dumps(client.list_testtuple(), indent=4)
assets['computeplan'] = json.dumps(client.list_compute_plan(), indent=4)
assets['compositetraintuple'] = json.dumps(
client.list_composite_traintuple(), indent=4)
assets['compositealgo'] = json.dumps(client.list_composite_algo(),
indent=4)
assets['model'] = json.dumps([
res for res in client.client.list('model')
if (('traintuple' in res or 'compositeTraintuple' in res)
and 'testtuple' in res)
],
indent=4)
with open(assets_path, 'w') as f:
f.write('"""\nWARNING\n=======\n\nDO NOT MANUALLY EDIT THIS FILE!\n\n'
'It is generated using substrapp/tests/generate_assets.py\n\n'
'In order to update this file:\n'
'1. start a clean instance of substra\n'
'2. run populate.py\n'
'3. run substrapp/tests/generate_assets.py\n"""\n\n')
for k, v in assets.items():
if network == 'docker':
v = v.replace('substra-backend.owkin.xyz:8000', 'testserver')
v = v.replace('substra-backend.chunantes.xyz:8001',
'testserver')
if network == 'skaffold':
v = v.replace('substra-backend.node-1.com', 'testserver')
v = v.replace('substra-backend.node-2.com', 'testserver')
v = v.replace('true', 'True')
v = v.replace('false', 'False')
v = v.replace('null', 'None')
f.write(f'{k} = {v}')
f.write('\n\n')
def __init__(self, node_name, node_id, address, user, password):
super().__init__()
# session added/modified assets during the session lifetime
self.state = _State()
# node / client
self.node_id = node_id
self._client = substra.Client()
self._client.add_profile(node_name, address, '0.0',
user=user,
password=password)
def test_client_multi_nodes_cp(asset_factory):
"""Assert that there is one CP local folder per node"""
client = substra.Client(debug=True)
dataset_query = asset_factory.create_dataset(
metadata={substra.DEBUG_OWNER: 'owner_1'})
dataset_1_key = client.add_dataset(dataset_query)
dataset_2_query = asset_factory.create_dataset(
metadata={substra.DEBUG_OWNER: 'owner_2'})
dataset_2_key = client.add_dataset(dataset_2_query)
data_sample = asset_factory.create_data_sample(datasets=[dataset_1_key],
test_only=False)
sample_1_key = client.add_data_sample(data_sample)
data_sample = asset_factory.create_data_sample(datasets=[dataset_2_key],
test_only=False)
sample_2_key = client.add_data_sample(data_sample)
algo_query = asset_factory.create_algo()
algo_key = client.add_algo(algo_query)
cp = asset_factory.create_compute_plan()
cp.traintuples = [
substra.sdk.schemas.ComputePlanTraintupleSpec(
algo_key=algo_key,
data_manager_key=dataset_1_key,
traintuple_id=1,
train_data_sample_keys=[sample_1_key],
),
substra.sdk.schemas.ComputePlanTraintupleSpec(
algo_key=algo_key,
data_manager_key=dataset_2_key,
traintuple_id=2,
train_data_sample_keys=[sample_2_key],
)
]
client.add_compute_plan(cp)
path_cp_1 = Path.cwd() / "local-worker" / "compute_plans" / "owner_1"
path_cp_2 = Path.cwd() / "local-worker" / "compute_plans" / "owner_2"
assert path_cp_1.is_dir()
assert path_cp_2.is_dir()
def __init__(self,
debug,
node_id=None,
address=None,
token=None,
user=None,
password=None):
super().__init__()
self.node_id = node_id
self.debug = debug
self._client = substra.Client(debug=debug,
url=address,
insecure=False,
token=token)
if not token:
token = self._client.login(user, password)
self.token = token
def test_compute_plan_add_update(asset_factory):
client = substra.Client(debug=True)
compute_plan = client.add_compute_plan(
substra.sdk.schemas.ComputePlanSpec(
tag=None,
clean_models=False,
metadata=dict(),
))
dataset_query = asset_factory.create_dataset()
dataset_key = client.add_dataset(dataset_query)
data_sample = asset_factory.create_data_sample(datasets=[dataset_key], test_only=False)
data_sample_key = client.add_data_sample(data_sample)
algo_query = asset_factory.create_algo()
algo_key = client.add_algo(algo_query)
traintuple_key = client.add_traintuple(
substra.sdk.schemas.TraintupleSpec(
algo_key=algo_key,
data_manager_key=dataset_key,
train_data_sample_keys=[data_sample_key],
in_models_keys=None,
compute_plan_key=compute_plan.key,
rank=None,
metadata=None,
)
)
traintuple = substra.sdk.schemas.ComputePlanTraintupleSpec(
algo_key=algo_key,
data_manager_key=dataset_key,
train_data_sample_keys=[data_sample_key],
in_models_ids=[traintuple_key],
traintuple_id=0,
)
compute_plan = client.update_compute_plan(
key=compute_plan.key,
data={
"traintuples": [traintuple],
}
)
def main():
client = substra.Client(url='http://substra-backend.node-1.com',
insecure=False)
client.login('node-1', 'p@$swr0d44')
assets = {}
assets['objective'] = json.dumps(client.list_objective(), indent=4)
assets['datamanager'] = json.dumps(client.list_dataset(), indent=4)
assets['algo'] = json.dumps(client.list_algo(), indent=4)
assets['traintuple'] = json.dumps(client.list_traintuple(), indent=4)
assets['testtuple'] = json.dumps(client.list_testtuple(), indent=4)
assets['computeplan'] = json.dumps(client.list_compute_plan(), indent=4)
assets['compositetraintuple'] = json.dumps(
client.list_composite_traintuple(), indent=4)
assets['compositealgo'] = json.dumps(client.list_composite_algo(),
indent=4)
models = client._backend.list(substra.sdk.schemas.Type.Model)
models = [(model, list(model.keys()).pop()) for model in models]
assets['model'] = json.dumps([
client._backend.get(substra.sdk.schemas.Type.Model,
model[mtype]['key']) for model, mtype in models
],
indent=4)
with open(assets_path, 'w') as f:
f.write('"""\nWARNING\n=======\n\nDO NOT MANUALLY EDIT THIS FILE!\n\n'
'It is generated using substrapp/tests/generate_assets.py\n\n'
'In order to update this file:\n'
'1. start a clean instance of substra\n'
'2. run computation on it (with e2e tests for instance)\n'
'3. run substrapp/tests/generate_assets.py\n"""\n\n')
for k, v in assets.items():
v = v.replace('substra-backend.node-1.com', 'testserver')
v = v.replace('substra-backend.node-2.com', 'testserver')
v = v.replace('true', 'True')
v = v.replace('false', 'False')
v = v.replace('null', 'None')
f.write(f'{k} = {v}')
f.write('\n\n')
# See the License for the specific language governing permissions and
# limitations under the License.
import json
import os
import zipfile
import substra
current_directory = os.path.dirname(__file__)
assets_directory = os.path.join(current_directory, '../assets')
with open(os.path.join(current_directory, '../../config.json'), 'r') as f:
config = json.load(f)
client = substra.Client()
client.add_profile(config['profile_name'], config['username'], config['password'], config['url'])
client.login()
ALGO = {
'name': 'Constant death predictor',
'description': os.path.join(assets_directory, 'algo_constant/description.md')
}
ALGO_DOCKERFILE_FILES = [
os.path.join(assets_directory, 'algo_constant/algo.py'),
os.path.join(assets_directory, 'algo_constant/Dockerfile'),
]
########################################################
# Build archive
########################################################
pg = tqdm(total=length)
progress_handler = logging.StreamHandler(SimpleNamespace(write=lambda x: pg.write(x, end='')))
substra_logger.removeHandler(default_stream_handler)
substra_logger.addHandler(progress_handler)
try:
yield pg
finally:
pg.close()
substra_logger.removeHandler(progress_handler)
substra_logger.addHandler(default_stream_handler)
current_directory = os.path.dirname(__file__)
assets_directory = os.path.join(current_directory, '../assets')
client = substra.Client(profile_name="node-1")
DATASET = {
'name': 'deepfake-detection',
'type': 'mp4,npy',
'data_opener': os.path.join(assets_directory, 'dataset/opener.py'),
'description': os.path.join(assets_directory, 'dataset/description.md'),
'permissions': {
'public': False,
'authorized_ids': []
},
}
TEST_DATA_SAMPLES_PATHS = [
os.path.join(assets_directory, 'test_data_samples', path)
for path in os.listdir(os.path.join(assets_directory, 'test_data_samples'))
def test_client_tmp_dir():
"""Test the creation of a temp directory for the debug client"""
client = substra.Client(debug=True)
assert client.temp_directory
def test_login_without_profile(tmpdir):
client = substra.Client()
with pytest.raises(substra.exceptions.SDKException):
client.login('foo', 'bar')
def client(tmpdir):
c = substra.Client(url="http://foo.io")
return c
def test_login_remote_without_url(tmpdir):
with pytest.raises(substra.exceptions.SDKException):
substra.Client()
def client(tmpdir):
config_path = tmpdir / "substra.cfg"
c = substra.Client(config_path=str(config_path))
c.add_profile('test', 'foo', 'password', url="http://foo.io")
return c
substra_logger.addHandler(progress_handler)
try:
yield pg
finally:
pg.close()
substra_logger.removeHandler(progress_handler)
substra_logger.addHandler(default_stream_handler)
# Define the current and asset directories
current_directory = pathlib.Path(__file__).resolve().parents[1]
assets_directory = current_directory.parent / "titanic" / "assets"
algo_directory = current_directory.parent / "compute_plan" / "assets" / "algo_sgd"
# Define the client
client = substra.Client(debug=True)
DATASET = {
"name": "Titanic",
"type": "csv",
"data_opener": str(assets_directory / "dataset" / "opener.py"),
"description": str(assets_directory / "dataset" / "description.md"),
"permissions": {"public": False, "authorized_ids": []},
}
TEST_DATA_SAMPLES_PATHS = [
assets_directory / "test_data_samples" / path
for path in (assets_directory / "test_data_samples").glob("*")
]
TRAIN_DATA_SAMPLES_PATHS = [
