def test_09_add_command_with_metric_for_wrong_entity(self):
repo_type = DATASETS
self.set_up_add()
create_spec(self, repo_type, self.tmp_dir)
workspace = os.path.join(self.tmp_dir, repo_type, DATASET_NAME)
os.makedirs(os.path.join(workspace, 'data'))
create_file(workspace, 'file1', '0')
metrics_options = '--metric Accuracy 1 --metric Recall 2'
self.assertIn(
output_messages['INFO_ADDING_PATH'] % repo_type,
check_output(MLGIT_ADD %
(repo_type, DATASET_NAME, metrics_options)))
index = os.path.join(ML_GIT_DIR, repo_type, 'index', 'metadata',
DATASET_NAME, 'INDEX.yaml')
self._check_index(index, ['data/file1'], [])
with open(os.path.join(workspace, DATASET_NAME + '.spec')) as spec:
spec_file = yaml_processor.load(spec)
spec_key = get_spec_key(repo_type)
metrics = spec_file[spec_key].get('metrics', {})
self.assertTrue(metrics == {})
def set_up_push(self):
os.makedirs(self.workspace)
create_spec(self,
self.repo_type,
self.tmp_dir,
version=1,
mutability='strict',
store_type=self.store_type)
self.assertIn(messages[0], check_output(MLGIT_INIT))
self.assertIn(
messages[2] % (GIT_PATH, self.repo_type),
check_output(MLGIT_REMOTE_ADD % (self.repo_type, GIT_PATH)))
self.assertIn(
messages[87] % (self.store_type, self.bucket),
check_output('ml-git repository store add %s --type=%s' %
(self.bucket, self.store_type)))
self.assertNotIn(ERROR_MESSAGE,
check_output(MLGIT_ENTITY_INIT % 'dataset'))
add_file(self, self.repo_type, '', 'new')
metadata_path = os.path.join(ML_GIT_DIR, 'dataset', 'metadata')
self.assertIn(
messages[17] %
(os.path.join(self.tmp_dir, metadata_path),
os.path.join('computer-vision', 'images', 'dataset-ex')),
check_output(MLGIT_COMMIT % (self.repo_type, 'dataset-ex', '')))
HEAD = os.path.join(ML_GIT_DIR, 'dataset', 'refs', 'dataset-ex',
'HEAD')
self.assertTrue(os.path.exists(HEAD))
def test_04_list_tags_without_similar_tags(self):
self._list_tag_entity('dataset')
entity_type = 'dataset'
similar_entity = 'dataset-ex2'
workspace = os.path.join('dataset', similar_entity)
os.makedirs(workspace, exist_ok=True)
create_spec(self,
'dataset',
self.tmp_dir,
artifact_name=similar_entity)
add_file(self,
'dataset',
'--bumpversion',
'new',
artifact_name=similar_entity)
self.assertIn(
messages[17] %
(os.path.join(self.tmp_dir, ML_GIT_DIR, 'dataset', 'metadata'),
os.path.join('computer-vision', 'images', similar_entity)),
check_output(MLGIT_COMMIT % ('dataset', similar_entity, '')))
check_output(MLGIT_PUSH % ('dataset', similar_entity))
self.assertNotIn(
similar_entity,
check_output(MLGIT_TAG_LIST % (entity_type, entity_type + '-ex')))
self.assertIn(
similar_entity,
check_output(MLGIT_TAG_LIST % (entity_type, similar_entity)))
def test_07_add_command_with_multiple_files(self):
self.set_up_add()
create_spec(self, DATASETS, self.tmp_dir)
workspace = os.path.join(self.tmp_dir, DATASETS, DATASET_NAME)
os.makedirs(os.path.join(workspace, 'data'))
create_file(workspace, 'file1', '0')
create_file(workspace, 'file2', '1')
create_file(workspace, 'file3', '1')
self.assertIn(
output_messages['INFO_ADDING_PATH'] % DATASETS,
check_output(
MLGIT_ADD %
(DATASETS, DATASET_NAME, os.path.join('data', 'file1'))))
index = os.path.join(ML_GIT_DIR, DATASETS, 'index', 'metadata',
DATASET_NAME, 'INDEX.yaml')
self._check_index(index, ['data/file1'], ['data/file2', 'data/file3'])
self.assertIn(
output_messages['INFO_ADDING_PATH'] % DATASETS,
check_output(MLGIT_ADD % (DATASETS, DATASET_NAME, 'data')))
self._check_index(index, ['data/file1', 'data/file2', 'data/file3'],
[])
create_file(workspace, 'file4', '0')
self.assertIn(output_messages['INFO_ADDING_PATH'] % DATASETS,
check_output(MLGIT_ADD % (DATASETS, DATASET_NAME, '')))
self._check_index(
index, ['data/file1', 'data/file2', 'data/file3', 'data/file4'],
[])
def test_10_add_command_with_metric_file(self):
repo_type = MODELS
entity_name = '{}-ex'.format(repo_type)
self.set_up_add(repo_type)
create_spec(self, repo_type, self.tmp_dir)
workspace = os.path.join(self.tmp_dir, repo_type, entity_name)
os.makedirs(os.path.join(workspace, 'data'))
create_file(workspace, 'file1', '0')
csv_file = os.path.join(self.tmp_dir, 'metrics.csv')
self.create_csv_file(csv_file, {'Accuracy': 1, 'Recall': 2})
metrics_options = '--metrics-file="{}"'.format(csv_file)
self.assertIn(
output_messages['INFO_ADDING_PATH'] % repo_type,
check_output(MLGIT_ADD %
(repo_type, entity_name, metrics_options)))
index = os.path.join(ML_GIT_DIR, repo_type, 'index', 'metadata',
entity_name, 'INDEX.yaml')
self._check_index(index, ['data/file1'], [])
with open(os.path.join(workspace, entity_name + '.spec')) as spec:
spec_file = yaml_processor.load(spec)
spec_key = get_spec_key(repo_type)
metrics = spec_file[spec_key].get('metrics', {})
self.assertFalse(metrics == {})
self.assertTrue(metrics['Accuracy'] == 1)
self.assertTrue(metrics['Recall'] == 2)
def test_05_add_command_without_file_added(self):
self.set_up_add()
create_spec(self, 'dataset', self.tmp_dir)
self.assertNotIn(ERROR_MESSAGE, check_output(MLGIT_ADD % ('dataset', 'dataset-ex', '')))
self.assertIn(messages[27], check_output(MLGIT_ADD % ('dataset', 'dataset-ex', '--bumpversion')))
def test_04_commit_command_with_version(self):
init_repository(DATASETS, self)
create_spec(self, DATASETS, self.tmp_dir)
workspace = os.path.join(self.tmp_dir, DATASETS, DATASET_NAME)
os.makedirs(os.path.join(workspace, 'data'))
create_file(workspace, 'file1', '0')
self.assertIn(output_messages['INFO_ADDING_PATH'] % DATASETS,
check_output(MLGIT_ADD % (DATASETS, DATASET_NAME, "")))
self.assertIn(
output_messages['INFO_COMMIT_REPO'] % (os.path.join(
self.tmp_dir, ML_GIT_DIR, DATASETS, 'metadata'), DATASET_NAME),
check_output(MLGIT_COMMIT % (DATASETS, DATASET_NAME, '')))
create_file(workspace, 'file2', '1')
self.assertIn(output_messages['INFO_ADDING_PATH'] % DATASETS,
check_output(MLGIT_ADD % (DATASETS, DATASET_NAME, "")))
self.assertIn(
output_messages['ERROR_INVALID_VALUE_FOR'] % ('--version', '-10'),
check_output(MLGIT_COMMIT %
(DATASETS, DATASET_NAME, ' --version=-10')))
self.assertIn(
output_messages['ERROR_INVALID_VALUE_FOR'] % ('--version', 'test'),
check_output(MLGIT_COMMIT %
(DATASETS, DATASET_NAME, '--version=test')))
self.assertIn(
output_messages['INFO_COMMIT_REPO'] % (os.path.join(
self.tmp_dir, ML_GIT_DIR, DATASETS, 'metadata'), DATASET_NAME),
check_output(MLGIT_COMMIT %
(DATASETS, DATASET_NAME, '--version=2')))
def set_up_test(self, repo_type=MODELS):
self.TAG_TIMES = []
entity_name = '{}-ex'.format(repo_type)
init_repository(repo_type, self)
create_spec(self, repo_type, self.tmp_dir)
metrics_options = '--metric Accuracy 10 --metric Recall 10'
self.assertNotIn(
ERROR_MESSAGE,
check_output(MLGIT_ADD %
(repo_type, entity_name, metrics_options)))
self.assertNotIn(
ERROR_MESSAGE,
check_output(MLGIT_COMMIT % (repo_type, entity_name, '')))
self._git_commit_time()
metrics_options = '--metric Accuracy 20 --metric Recall 20'
workspace = os.path.join(self.tmp_dir, repo_type, entity_name)
os.makedirs(os.path.join(workspace, 'data'))
create_file(workspace, 'file1', '0')
self.assertNotIn(
ERROR_MESSAGE,
check_output(MLGIT_ADD %
(repo_type, entity_name, metrics_options)))
self.assertNotIn(
ERROR_MESSAGE,
check_output(MLGIT_COMMIT %
(repo_type, entity_name, ' --version=2')))
self._git_commit_time()
def set_up_push(self):
os.makedirs(self.workspace)
create_spec(self,
self.repo_type,
self.tmp_dir,
version=1,
mutability=STRICT,
storage_type=self.storage_type)
self.assertIn(
output_messages['INFO_INITIALIZED_PROJECT_IN'] % self.tmp_dir,
check_output(MLGIT_INIT))
self.assertIn(
output_messages['INFO_ADD_REMOTE'] % (GIT_PATH, self.repo_type),
check_output(MLGIT_REMOTE_ADD % (self.repo_type, GIT_PATH)))
self.assertIn(
output_messages['INFO_ADD_STORAGE_WITHOUT_PROFILE'] %
(self.storage_type, self.bucket),
check_output('ml-git repository storage add %s --type=%s' %
(self.bucket, self.storage_type)))
self.assertNotIn(ERROR_MESSAGE,
check_output(MLGIT_ENTITY_INIT % DATASETS))
add_file(self, self.repo_type, '', 'new')
metadata_path = os.path.join(ML_GIT_DIR, DATASETS, 'metadata')
self.assertIn(
output_messages['INFO_COMMIT_REPO'] %
(os.path.join(self.tmp_dir, metadata_path), DATASET_NAME),
check_output(MLGIT_COMMIT % (self.repo_type, DATASET_NAME, '')))
HEAD = os.path.join(ML_GIT_DIR, DATASETS, 'refs', DATASET_NAME, 'HEAD')
self.assertTrue(os.path.exists(HEAD))
def _create_entity_with_mutability(self, entity_type, mutability_type):
init_repository(entity_type, self)
workspace = os.path.join(self.tmp_dir, entity_type,
entity_type + '-ex')
create_spec(self, entity_type, self.tmp_dir, 1, mutability_type)
os.makedirs(os.path.join(workspace, 'data'))
create_file(workspace, 'file1', '0')
self.assertNotIn(
ERROR_MESSAGE,
check_output(MLGIT_ADD % (entity_type, entity_type + '-ex', '')))
self.assertIn(
messages[17] %
(os.path.join(self.tmp_dir, ML_GIT_DIR, entity_type, 'metadata'),
os.path.join('computer-vision', 'images', entity_type + '-ex')),
check_output(MLGIT_COMMIT %
(entity_type, entity_type + '-ex', '')))
self.assertNotIn(
ERROR_MESSAGE,
check_output(MLGIT_PUSH % (entity_type, entity_type + '-ex')))
clear(os.path.join(self.tmp_dir, ML_GIT_DIR))
clear(workspace)
clear(os.path.join(self.tmp_dir, entity_type))
def _create_entity_with_mutability(self, entity_type, mutability_type):
init_repository(entity_type, self)
workspace = os.path.join(self.tmp_dir, entity_type, entity_type + '-ex')
create_spec(self, entity_type, self.tmp_dir, 1, mutability_type)
os.makedirs(os.path.join(workspace, 'data'))
create_file(workspace, 'file1', '0')
self._push_files(entity_type, '')
self._clear_path()
def test_06_commit_with_large_version_number(self):
init_repository(DATASETS, self)
create_spec(self, DATASETS, self.tmp_dir)
self.assertIn(output_messages['ERROR_INVALID_VALUE_FOR'] % ('--version', '9999999999'),
check_output(MLGIT_COMMIT % (DATASETS, DATASET_NAME, ' --version=9999999999')))
self.assertIn(output_messages['ERROR_INVALID_VALUE_FOR'] % ('--version', '9999999999'),
check_output(MLGIT_COMMIT % (MODELS, MODELS + '-ex', ' --version=9999999999')))
self.assertIn(output_messages['ERROR_INVALID_VALUE_FOR'] % ('--version', '9999999999'),
check_output(MLGIT_COMMIT % (LABELS, LABELS + '-ex', ' --version=9999999999')))
def test_03_checkout(self):
os.makedirs(self.workspace)
create_spec(self,
self.repo_type,
self.tmp_dir,
version=1,
mutability='strict',
store_type=self.store_type)
self.assertIn(messages[0], check_output(MLGIT_INIT))
self.assertIn(
messages[2] % (GIT_PATH, self.repo_type),
check_output(MLGIT_REMOTE_ADD % (self.repo_type, GIT_PATH)))
self.assertIn(
messages[87] % (self.store_type, self.bucket),
check_output('ml-git repository store add %s --type=%s' %
(self.bucket, self.store_type)))
self.assertNotIn(ERROR_MESSAGE,
check_output(MLGIT_ENTITY_INIT % 'dataset'))
add_file(self, self.repo_type, '', 'new')
metadata_path = os.path.join(ML_GIT_DIR, 'dataset', 'metadata')
self.assertIn(
messages[17] %
(os.path.join(self.tmp_dir, metadata_path),
os.path.join('computer-vision', 'images', 'dataset-ex')),
check_output(MLGIT_COMMIT % (self.repo_type, 'dataset-ex', '')))
HEAD = os.path.join(ML_GIT_DIR, 'dataset', 'refs', 'dataset-ex',
'HEAD')
self.assertTrue(os.path.exists(HEAD))
self.assertEqual(os.getenv('AZURE_STORAGE_CONNECTION_STRING'),
self.dev_store_account_)
self.assertNotIn(
ERROR_MESSAGE,
check_output(MLGIT_PUSH % (self.repo_type, 'dataset-ex')))
clear(self.workspace)
clear(os.path.join(ML_GIT_DIR, 'dataset'))
self.assertNotIn(ERROR_MESSAGE,
check_output(MLGIT_ENTITY_INIT % self.repo_type))
self.assertEqual(os.getenv('AZURE_STORAGE_CONNECTION_STRING'),
self.dev_store_account_)
self.assertNotIn(
ERROR_MESSAGE,
check_output(
MLGIT_CHECKOUT %
(self.repo_type, 'computer-vision__images__dataset-ex__1')))
ws_path = os.path.join(self.tmp_dir, 'dataset', 'computer-vision',
'images', 'dataset-ex')
self.assertTrue(os.path.isfile(os.path.join(ws_path, 'newfile0')))
self.assertTrue(os.path.isfile(os.path.join(ws_path, 'newfile1')))
self.assertTrue(os.path.isfile(os.path.join(ws_path, 'newfile2')))
self.assertTrue(os.path.isfile(os.path.join(ws_path, 'newfile3')))
self.assertTrue(os.path.isfile(os.path.join(ws_path, 'newfile4')))
def setUp_test(self):
init_repository('dataset', self)
create_spec(self, 'dataset', self.tmp_dir)
self.assertIn(messages[13] % 'dataset',
check_output(MLGIT_ADD % ('dataset', 'dataset-ex', '')))
self.assertIn(
messages[17] %
(os.path.join(self.tmp_dir, ML_GIT_DIR, 'dataset', 'metadata'),
os.path.join('computer-vision', 'images', 'dataset-ex')),
check_output(
MLGIT_COMMIT %
('dataset', 'dataset-ex', '-m ' + self.COMMIT_MESSAGE)))
def test_05_commit_command_with_deprecated_version_number(self):
init_repository(DATASETS, self)
create_spec(self, DATASETS, self.tmp_dir)
workspace = os.path.join(self.tmp_dir, DATASETS, DATASET_NAME)
os.makedirs(os.path.join(workspace, 'data'))
create_file(workspace, 'file1', '0')
self.assertIn(output_messages['INFO_ADDING_PATH'] % DATASETS,
check_output(MLGIT_ADD % (DATASETS, DATASET_NAME, "")))
result = check_output(MLGIT_COMMIT % (DATASETS, DATASET_NAME, '--version-number=2'))
self.assertIn(output_messages['ERROR_NO_SUCH_OPTION'] % '--version-number', result)
def test_03_checkout(self):
os.makedirs(self.workspace)
create_spec(self,
self.repo_type,
self.tmp_dir,
version=1,
mutability=STRICT,
storage_type=self.storage_type)
self.assertIn(
output_messages['INFO_INITIALIZED_PROJECT_IN'] % self.tmp_dir,
check_output(MLGIT_INIT))
self.assertIn(
output_messages['INFO_ADD_REMOTE'] % (GIT_PATH, self.repo_type),
check_output(MLGIT_REMOTE_ADD % (self.repo_type, GIT_PATH)))
self.assertIn(
output_messages['INFO_ADD_STORAGE_WITHOUT_PROFILE'] %
(self.storage_type, self.bucket),
check_output('ml-git repository storage add %s --type=%s' %
(self.bucket, self.storage_type)))
self.assertNotIn(ERROR_MESSAGE,
check_output(MLGIT_ENTITY_INIT % DATASETS))
add_file(self, self.repo_type, '', 'new')
metadata_path = os.path.join(ML_GIT_DIR, DATASETS, 'metadata')
self.assertIn(
output_messages['INFO_COMMIT_REPO'] %
(os.path.join(self.tmp_dir, metadata_path), DATASET_NAME),
check_output(MLGIT_COMMIT % (self.repo_type, DATASET_NAME, '')))
HEAD = os.path.join(ML_GIT_DIR, DATASETS, 'refs', DATASET_NAME, 'HEAD')
self.assertTrue(os.path.exists(HEAD))
self.assertEqual(os.getenv('AZURE_STORAGE_CONNECTION_STRING'),
self.dev_store_account_)
self.assertNotIn(
ERROR_MESSAGE,
check_output(MLGIT_PUSH % (self.repo_type, DATASET_NAME)))
clear(self.workspace)
clear(os.path.join(ML_GIT_DIR, DATASETS))
self.assertNotIn(ERROR_MESSAGE,
check_output(MLGIT_ENTITY_INIT % self.repo_type))
self.assertEqual(os.getenv('AZURE_STORAGE_CONNECTION_STRING'),
self.dev_store_account_)
self.assertNotIn(
ERROR_MESSAGE,
check_output(MLGIT_CHECKOUT % (self.repo_type, DATASET_TAG)))
ws_path = os.path.join(self.tmp_dir, DATASETS, DATASET_NAME)
self.assertTrue(os.path.isfile(os.path.join(ws_path, 'newfile0')))
self.assertTrue(os.path.isfile(os.path.join(ws_path, 'newfile1')))
self.assertTrue(os.path.isfile(os.path.join(ws_path, 'newfile2')))
self.assertTrue(os.path.isfile(os.path.join(ws_path, 'newfile3')))
self.assertTrue(os.path.isfile(os.path.join(ws_path, 'newfile4')))
def test_05_add_command_without_file_added(self):
self.set_up_add()
create_spec(self, DATASETS, self.tmp_dir)
self.assertNotIn(
ERROR_MESSAGE,
check_output(MLGIT_ADD % (DATASETS, DATASET_NAME, '')))
self.assertIn(
output_messages['INFO_NO_NEW_DATA_TO_ADD'],
check_output(MLGIT_ADD %
(DATASETS, DATASET_NAME, '--bumpversion')))
def test_16_add_command_with_metric_file_empty(self):
repo_type = MODELS
entity_name = '{}-ex'.format(repo_type)
self.set_up_add(repo_type)
create_spec(self, repo_type, self.tmp_dir)
workspace = os.path.join(self.tmp_dir, repo_type, entity_name)
os.makedirs(os.path.join(workspace, 'data'))
create_file(workspace, 'file1', '0')
metrics_options = '--metrics-file='
self.assertIn(
output_messages['ERROR_EMPTY_VALUE'],
check_output(MLGIT_ADD %
(repo_type, entity_name, metrics_options)))
def set_up_test(self, repo_type=DATASETS, with_metrics=False):
entity = '{}-ex'.format(repo_type)
init_repository(repo_type, self)
create_spec(self, repo_type, self.tmp_dir)
metrics_options = ''
if with_metrics:
metrics_options = '--metric Accuracy 1 --metric Recall 2'
self.assertIn(
output_messages['INFO_ADDING_PATH'] % repo_type,
check_output(MLGIT_ADD % (repo_type, entity, metrics_options)))
self.assertIn(
output_messages['INFO_COMMIT_REPO'] % (os.path.join(
self.tmp_dir, ML_GIT_DIR, repo_type, 'metadata'), entity),
check_output(MLGIT_COMMIT %
(repo_type, entity, '-m ' + self.COMMIT_MESSAGE)))
def test_17_add_command_with_empty_metric_file(self):
repo_type = MODELS
entity_name = '{}-ex'.format(repo_type)
self.set_up_add(repo_type)
create_spec(self, repo_type, self.tmp_dir)
workspace = os.path.join(self.tmp_dir, repo_type, entity_name)
os.makedirs(os.path.join(workspace, 'data'))
create_file(workspace, 'file1', '0')
csv_file = os.path.join(self.tmp_dir, 'metrics.csv')
with open(csv_file, 'wt') as f:
f.write('')
metrics_options = '--metrics-file="{}"'.format(csv_file)
self.assertIn(
output_messages['ERROR_INVALID_METRICS_FILE'],
check_output(MLGIT_ADD %
(repo_type, entity_name, metrics_options)))
def test_06_commit_with_large_version_number(self):
init_repository('dataset', self)
create_spec(self, 'dataset', self.tmp_dir)
self.assertIn(
messages[96] % '9999999999',
check_output(
MLGIT_COMMIT %
('dataset', 'dataset' + '-ex', ' --version=9999999999')))
self.assertIn(
messages[96] % '9999999999',
check_output(MLGIT_COMMIT %
('model', 'model' + '-ex', ' --version=9999999999')))
self.assertIn(
messages[96] % '9999999999',
check_output(
MLGIT_COMMIT %
('labels', 'labels' + '-ex', ' --version=9999999999')))
def set_up_test(self, entity):
init_repository(entity, self)
workspace = os.path.join(self.tmp_dir, entity, entity + '-ex')
os.makedirs(workspace, exist_ok=True)
create_spec(self, entity, self.tmp_dir, 20, STRICT)
os.makedirs(os.path.join(workspace, 'data'), exist_ok=True)
self.create_file(workspace, 'file1', '0')
self.create_file(workspace, 'file2', '1')
self.create_file(workspace, 'file3', 'a')
self.create_file(workspace, 'file4', 'b')
api.add(entity,
entity + '-ex',
bumpversion=True,
fsck=False,
file_path=['file'])
api.commit(entity, entity + '-ex')
def set_up_unlock(self, entity_type, mutability_type):
init_repository(entity_type, self)
workspace = os.path.join(entity_type, entity_type + '-ex')
create_spec(self,
entity_type,
self.tmp_dir,
1,
mutability=mutability_type)
os.makedirs(os.path.join(workspace, 'data'))
with open(os.path.join(workspace, self.file), 'w') as file:
file.write('0' * 2048)
self.assertNotIn(
ERROR_MESSAGE,
check_output(MLGIT_ADD %
(entity_type, entity_type + '-ex', '--bumpversion')))
def test_05_commit_command_with_deprecated_version_number(self):
init_repository('dataset', self)
create_spec(self, 'dataset', self.tmp_dir)
workspace = os.path.join(self.tmp_dir, 'dataset', 'dataset-ex')
os.makedirs(os.path.join(workspace, 'data'))
create_file(workspace, 'file1', '0')
self.assertIn(messages[13] % 'dataset',
check_output(MLGIT_ADD % ('dataset', 'dataset-ex', "")))
result = check_output(
MLGIT_COMMIT %
('dataset', 'dataset' + '-ex', '--version-number=2'))
self.assertIn(messages[106] % ('--version-number', '--version'),
result)
self.assertIn(
messages[17] %
(os.path.join(self.tmp_dir, ML_GIT_DIR, 'dataset', 'metadata'),
os.path.join('computer-vision', 'images', 'dataset' + '-ex')),
result)
def test_07_add_command_with_multiple_files(self):
self.set_up_add()
create_spec(self, 'dataset', self.tmp_dir)
workspace = os.path.join(self.tmp_dir, 'dataset', 'dataset-ex')
os.makedirs(os.path.join(workspace, 'data'))
create_file(workspace, 'file1', '0')
create_file(workspace, 'file2', '1')
create_file(workspace, 'file3', '1')
self.assertIn(messages[13] % 'dataset', check_output(MLGIT_ADD % ('dataset', 'dataset-ex',
os.path.join('data', 'file1'))))
index = os.path.join(ML_GIT_DIR, 'dataset', 'index', 'metadata', 'dataset-ex', 'INDEX.yaml')
self._check_index(index, ['data/file1'], ['data/file2', 'data/file3'])
self.assertIn(messages[13] % 'dataset', check_output(MLGIT_ADD % ('dataset', 'dataset-ex', 'data')))
self._check_index(index, ['data/file1', 'data/file2', 'data/file3'], [])
create_file(workspace, 'file4', '0')
self.assertIn(messages[13] % 'dataset', check_output(MLGIT_ADD % ('dataset', 'dataset-ex', '')))
self._check_index(index, ['data/file1', 'data/file2', 'data/file3', 'data/file4'], [])
def set_up_push(self, create_know_file=False):
os.makedirs(self.workspace)
create_spec(self,
self.repo_type,
self.tmp_dir,
version=1,
mutability=STRICT,
storage_type=self.storage_type)
self.assertIn(
output_messages['INFO_INITIALIZED_PROJECT_IN'] % self.tmp_dir,
check_output(MLGIT_INIT))
self.assertIn(
output_messages['INFO_ADD_REMOTE'] % (GIT_PATH, self.repo_type),
check_output(MLGIT_REMOTE_ADD % (self.repo_type, GIT_PATH)))
self.assertIn(
output_messages['INFO_ADD_STORAGE_WITHOUT_PROFILE'] %
(self.storage_type, self.bucket),
check_output(
'ml-git repository storage add %s --type=%s' %
('mlgit --username=mlgit_user '
'--endpoint-url=127.0.0.1 --port=9922 --private-key=' +
FAKE_SSH_KEY_PATH, self.storage_type)))
self.assertNotIn(ERROR_MESSAGE,
check_output(MLGIT_ENTITY_INIT % DATASETS))
if create_know_file:
with open(os.path.join(self.repo_type, DATASET_NAME, 'file'),
'wt') as z:
z.write(str('0' * 10011))
add_file(self, self.repo_type, '', 'new')
metadata_path = os.path.join(ML_GIT_DIR, self.repo_type, 'metadata')
self.assertIn(
output_messages['INFO_COMMIT_REPO'] %
(os.path.join(self.tmp_dir, metadata_path), DATASET_NAME),
check_output(MLGIT_COMMIT % (self.repo_type, DATASET_NAME, '')))
HEAD = os.path.join(ML_GIT_DIR, self.repo_type, 'refs', DATASET_NAME,
'HEAD')
self.assertTrue(os.path.exists(HEAD))
def test_04_list_tags_without_similar_tags(self):
self._list_tag_entity(DATASETS)
entity_type = DATASETS
similar_entity = 'datasets-ex2'
workspace = os.path.join(DATASETS, similar_entity)
os.makedirs(workspace, exist_ok=True)
create_spec(self, DATASETS, self.tmp_dir, artifact_name=similar_entity)
add_file(self,
DATASETS,
'--bumpversion',
'new',
artifact_name=similar_entity)
self.assertIn(
output_messages['INFO_COMMIT_REPO'] %
(os.path.join(self.tmp_dir, ML_GIT_DIR, DATASETS,
'metadata'), similar_entity),
check_output(MLGIT_COMMIT % (DATASETS, similar_entity, '')))
check_output(MLGIT_PUSH % (DATASETS, similar_entity))
self.assertNotIn(
similar_entity,
check_output(MLGIT_TAG_LIST % (entity_type, entity_type + '-ex')))
self.assertIn(
similar_entity,
check_output(MLGIT_TAG_LIST % (entity_type, similar_entity)))
def test_04_commit_command_with_version(self):
init_repository('dataset', self)
create_spec(self, 'dataset', self.tmp_dir)
workspace = os.path.join(self.tmp_dir, 'dataset', 'dataset-ex')
os.makedirs(os.path.join(workspace, 'data'))
create_file(workspace, 'file1', '0')
self.assertIn(messages[13] % 'dataset',
check_output(MLGIT_ADD % ('dataset', 'dataset-ex', "")))
self.assertIn(
messages[17] %
(os.path.join(self.tmp_dir, ML_GIT_DIR, 'dataset', 'metadata'),
os.path.join('computer-vision', 'images', 'dataset' + '-ex')),
check_output(MLGIT_COMMIT % ('dataset', 'dataset' + '-ex', '')))
create_file(workspace, 'file2', '1')
self.assertIn(messages[13] % 'dataset',
check_output(MLGIT_ADD % ('dataset', 'dataset-ex', "")))
self.assertIn(
messages[96] % '-10',
check_output(MLGIT_COMMIT %
('dataset', 'dataset' + '-ex', ' --version=-10')))
self.assertIn(
messages[96] % 'test',
check_output(MLGIT_COMMIT %
('dataset', 'dataset' + '-ex', '--version=test')))
self.assertIn(
messages[17] %
(os.path.join(self.tmp_dir, ML_GIT_DIR, 'dataset', 'metadata'),
os.path.join('computer-vision', 'images', 'dataset' + '-ex')),
check_output(MLGIT_COMMIT %
('dataset', 'dataset' + '-ex', '--version=2')))
def test_20_model_related(self):
git_server = os.path.join(self.tmp_dir, GIT_PATH)
self.assertIn(
output_messages['INFO_INITIALIZED_PROJECT_IN'] % self.tmp_dir,
check_output(MLGIT_INIT))
self.assertIn(
output_messages['INFO_ADD_REMOTE'] % (git_server, MODELS),
check_output(MLGIT_REMOTE_ADD % (MODELS, git_server)))
self.assertIn(
output_messages['INFO_ADD_STORAGE'] %
(STORAGE_TYPE, BUCKET_NAME, PROFILE),
check_output(MLGIT_STORAGE_ADD % (BUCKET_NAME, PROFILE)))
self.assertIn(
output_messages['INFO_METADATA_INIT'] %
(git_server,
os.path.join(self.tmp_dir, '.ml-git', MODELS, 'metadata')),
check_output(MLGIT_ENTITY_INIT % MODELS))
edit_config_yaml(os.path.join(self.tmp_dir, '.ml-git'))
workspace_model = os.path.join(MODELS, MODELS + '-ex')
os.makedirs(workspace_model)
version = 1
create_spec(self, MODELS, self.tmp_dir, version)
with open(os.path.join(self.tmp_dir, workspace_model, 'file1'),
'wb') as z:
z.write(b'0' * 1024)
self.assertIn(
output_messages['INFO_ADD_REMOTE'] % (git_server, DATASETS),
check_output(MLGIT_REMOTE_ADD % (DATASETS, git_server)))
self.assertIn(
output_messages['INFO_ADD_STORAGE'] %
(STORAGE_TYPE, BUCKET_NAME, PROFILE),
check_output(MLGIT_STORAGE_ADD % (BUCKET_NAME, PROFILE)))
self.assertIn(
output_messages['INFO_METADATA_INIT'] %
(git_server,
os.path.join(self.tmp_dir, '.ml-git', DATASETS, 'metadata')),
check_output(MLGIT_ENTITY_INIT % DATASETS))
edit_config_yaml(os.path.join(self.tmp_dir, '.ml-git'))
workspace_dataset = os.path.join(DATASETS, DATASETS + '-ex')
os.makedirs(workspace_dataset)
version = 1
create_spec(self, DATASETS, self.tmp_dir, version)
with open(os.path.join(self.tmp_dir, workspace_dataset, 'file1'),
'wb') as z:
z.write(b'0' * 1024)
expected_push_result = '2.00/2.00'
self.assertIn(
output_messages['INFO_ADDING_PATH'] % DATASETS,
check_output(MLGIT_ADD %
(DATASETS, DATASET_NAME, '--bumpversion')))
self.assertIn(
output_messages['INFO_COMMIT_REPO'] % (os.path.join(
self.tmp_dir, '.ml-git', DATASETS, 'metadata'), DATASET_NAME),
check_output(MLGIT_COMMIT % (DATASETS, DATASET_NAME, '')))
self.assertIn(expected_push_result,
check_output(MLGIT_PUSH % (DATASETS, DATASET_NAME)))
self.assertIn(
output_messages['INFO_ADD_REMOTE'] % (git_server, LABELS),
check_output(MLGIT_REMOTE_ADD % (LABELS, git_server)))
self.assertIn(
output_messages['INFO_ADD_STORAGE'] %
(STORAGE_TYPE, BUCKET_NAME, PROFILE),
check_output(MLGIT_STORAGE_ADD % (BUCKET_NAME, PROFILE)))
self.assertIn(
output_messages['INFO_METADATA_INIT'] %
(git_server,
os.path.join(self.tmp_dir, '.ml-git', LABELS, 'metadata')),
check_output(MLGIT_ENTITY_INIT % LABELS))
edit_config_yaml(os.path.join(self.tmp_dir, '.ml-git'))
workspace_labels = os.path.join(LABELS, LABELS + '-ex')
os.makedirs(workspace_labels)
version = 1
create_spec(self, LABELS, self.tmp_dir, version)
with open(os.path.join(self.tmp_dir, workspace_labels, 'file1'),
'wb') as z:
z.write(b'0' * 1024)
self.assertIn(
output_messages['INFO_ADDING_PATH'] % LABELS,
check_output(MLGIT_ADD %
(LABELS, LABELS + '-ex', '--bumpversion')))
self.assertIn(
output_messages['INFO_COMMIT_REPO'] % (os.path.join(
self.tmp_dir, '.ml-git', LABELS, 'metadata'), LABELS + '-ex'),
check_output(MLGIT_COMMIT % (LABELS, LABELS + '-ex', '')))
self.assertIn(expected_push_result,
check_output(MLGIT_PUSH % (LABELS, LABELS + '-ex')))
self.assertIn(
output_messages['INFO_ADDING_PATH'] % MODELS,
check_output(MLGIT_ADD %
(MODELS, MODELS + '-ex', '--bumpversion')))
self.assertIn(
output_messages['INFO_COMMIT_REPO'] % (os.path.join(
self.tmp_dir, '.ml-git', MODELS, 'metadata'), MODELS + '-ex'),
check_output(MLGIT_COMMIT %
(MODELS, MODELS + '-ex', '--dataset=datasets-ex') +
' --labels=labels-ex'))
self.assertIn(expected_push_result,
check_output(MLGIT_PUSH % (MODELS, MODELS + '-ex')))
set_write_read(os.path.join(self.tmp_dir, workspace_model, 'file1'))
set_write_read(os.path.join(self.tmp_dir, workspace_dataset, 'file1'))
set_write_read(os.path.join(self.tmp_dir, workspace_labels, 'file1'))
if not sys.platform.startswith('linux'):
recursive_write_read(os.path.join(self.tmp_dir, '.ml-git'))
clear(os.path.join(self.tmp_dir, MODELS))
clear(os.path.join(self.tmp_dir, DATASETS))
clear(os.path.join(self.tmp_dir, LABELS))
clear(os.path.join(self.tmp_dir, '.ml-git', MODELS))
clear(os.path.join(self.tmp_dir, '.ml-git', DATASETS))
clear(os.path.join(self.tmp_dir, '.ml-git', LABELS))
self.assertIn(
output_messages['INFO_METADATA_INIT'] %
(git_server,
os.path.join(self.tmp_dir, '.ml-git', MODELS, 'metadata')),
check_output(MLGIT_ENTITY_INIT % MODELS))
self.assertNotIn(
ERROR_MESSAGE,
check_output(MLGIT_CHECKOUT %
(MODELS, 'computer-vision__images__models-ex__2') +
' -d -l'))
self.assertTrue(os.path.exists(os.path.join(self.tmp_dir, MODELS)))
self.assertTrue(os.path.exists(os.path.join(self.tmp_dir, DATASETS)))
self.assertTrue(os.path.exists(os.path.join(self.tmp_dir, LABELS)))
def test_20_model_related(self):
model = 'model'
dataset = 'dataset'
labels = 'labels'
git_server = os.path.join(self.tmp_dir, GIT_PATH)
self.assertIn(messages[0], check_output(MLGIT_INIT))
self.assertIn(messages[2] % (git_server, model), check_output(MLGIT_REMOTE_ADD % (model, git_server)))
self.assertIn(messages[7] % (STORE_TYPE, BUCKET_NAME, PROFILE),
check_output(MLGIT_STORE_ADD % (BUCKET_NAME, PROFILE)))
self.assertIn(messages[8] % (git_server, os.path.join(self.tmp_dir, '.ml-git', model, 'metadata')),
check_output(MLGIT_ENTITY_INIT % 'model'))
edit_config_yaml(os.path.join(self.tmp_dir, '.ml-git'))
workspace_model = os.path.join(model, model + '-ex')
os.makedirs(workspace_model)
version = 1
create_spec(self, model, self.tmp_dir, version)
with open(os.path.join(self.tmp_dir, workspace_model, 'file1'), 'wb') as z:
z.write(b'0' * 1024)
self.assertIn(messages[2] % (git_server, dataset), check_output(MLGIT_REMOTE_ADD % (dataset, git_server)))
self.assertIn(messages[7] % (STORE_TYPE, BUCKET_NAME, PROFILE),
check_output(MLGIT_STORE_ADD % (BUCKET_NAME, PROFILE)))
self.assertIn(messages[8] % (git_server, os.path.join(self.tmp_dir, '.ml-git', dataset, 'metadata')),
check_output(MLGIT_ENTITY_INIT % 'dataset'))
edit_config_yaml(os.path.join(self.tmp_dir, '.ml-git'))
workspace_dataset = os.path.join(dataset, dataset + '-ex')
os.makedirs(workspace_dataset)
version = 1
create_spec(self, dataset, self.tmp_dir, version)
with open(os.path.join(self.tmp_dir, workspace_dataset, 'file1'), 'wb') as z:
z.write(b'0' * 1024)
self.assertIn(messages[13] % 'dataset', check_output(MLGIT_ADD % ('dataset', 'dataset-ex', '--bumpversion')))
self.assertIn(messages[17] % (os.path.join(self.tmp_dir, '.ml-git', 'dataset', 'metadata'),
os.path.join('computer-vision', 'images', 'dataset-ex')),
check_output(MLGIT_COMMIT % ('dataset', 'dataset-ex', '')))
self.assertIn(messages[47], check_output(MLGIT_PUSH % ('dataset', 'dataset-ex')))
self.assertIn(messages[2] % (git_server, labels), check_output(MLGIT_REMOTE_ADD % (labels, git_server)))
self.assertIn(messages[7] % (STORE_TYPE, BUCKET_NAME, PROFILE),
check_output(MLGIT_STORE_ADD % (BUCKET_NAME, PROFILE)))
self.assertIn(messages[8] % (git_server, os.path.join(self.tmp_dir, '.ml-git', labels, 'metadata')),
check_output(MLGIT_ENTITY_INIT % labels))
edit_config_yaml(os.path.join(self.tmp_dir, '.ml-git'))
workspace_labels = os.path.join(labels, labels + '-ex')
os.makedirs(workspace_labels)
version = 1
create_spec(self, labels, self.tmp_dir, version)
with open(os.path.join(self.tmp_dir, workspace_labels, 'file1'), 'wb') as z:
z.write(b'0' * 1024)
self.assertIn(messages[15], check_output(MLGIT_ADD % ('labels', 'labels-ex', '--bumpversion')))
self.assertIn(messages[17] % (os.path.join(self.tmp_dir, '.ml-git', 'labels', 'metadata'),
os.path.join('computer-vision', 'images', 'labels-ex')),
check_output(MLGIT_COMMIT % ('labels', 'labels-ex', '')))
self.assertIn(messages[47], check_output(MLGIT_PUSH % ('labels', 'labels-ex')))
self.assertIn(messages[14], check_output(MLGIT_ADD % ('model', 'model-ex', '--bumpversion')))
self.assertIn(messages[17] % (os.path.join(self.tmp_dir, '.ml-git', 'model', 'metadata'),
os.path.join('computer-vision', 'images', 'model-ex')),
check_output(MLGIT_COMMIT % ('model', 'model-ex', '--dataset=dataset-ex') + ' --labels=labels-ex'))
self.assertIn(messages[47], check_output(MLGIT_PUSH % ('model', 'model-ex')))
set_write_read(os.path.join(self.tmp_dir, workspace_model, 'file1'))
set_write_read(os.path.join(self.tmp_dir, workspace_dataset, 'file1'))
set_write_read(os.path.join(self.tmp_dir, workspace_labels, 'file1'))
if not sys.platform.startswith('linux'):
recursive_write_read(os.path.join(self.tmp_dir, '.ml-git'))
clear(os.path.join(self.tmp_dir, model))
clear(os.path.join(self.tmp_dir, dataset))
clear(os.path.join(self.tmp_dir, labels))
clear(os.path.join(self.tmp_dir, '.ml-git', model))
clear(os.path.join(self.tmp_dir, '.ml-git', dataset))
clear(os.path.join(self.tmp_dir, '.ml-git', labels))
self.assertIn(messages[8] % (git_server, os.path.join(self.tmp_dir, '.ml-git', model, 'metadata')),
check_output(MLGIT_ENTITY_INIT % model))
self.assertIn('', check_output(MLGIT_CHECKOUT % ('model', 'computer-vision__images__model-ex__2')
+ ' -d -l'))
self.assertTrue(os.path.exists(os.path.join(self.tmp_dir, model)))
self.assertTrue(os.path.exists(os.path.join(self.tmp_dir, dataset)))
self.assertTrue(os.path.exists(os.path.join(self.tmp_dir, labels)))
