def test_18_recover_corrupted_file_added(self):
entity_init(DATASETS, self)
add_file(self, DATASETS, '--bumpversion', 'new')
corrupted_file = os.path.join(self.tmp_dir, DATASETS, DATASET_NAME,
'newfile0')
with open(corrupted_file) as f:
content = f.read()
os.chmod(corrupted_file, S_IWUSR | S_IREAD)
with open(corrupted_file, 'wb') as z:
z.write(b'0' * 0)
command_output = check_output(
MLGIT_ADD % (DATASETS, DATASET_NAME, '--bumpversion'))
self.assertIn(output_messages['WARN_CORRUPTED_CANNOT_BE_ADD'],
command_output)
self.assertIn('newfile0', command_output)
os.chmod(corrupted_file, S_IWUSR | S_IREAD)
with open(corrupted_file, 'w') as z:
z.write(content)
command_output = check_output(
MLGIT_ADD % (DATASETS, DATASET_NAME, '--bumpversion'))
self.assertNotIn(output_messages['WARN_CORRUPTED_CANNOT_BE_ADD'],
command_output)
self.assertNotIn('newfile0', command_output)
def test_14_add_with_ignore_file(self):
entity_init(DATASETS, self)
workspace = os.path.join(self.tmp_dir, DATASETS, DATASET_NAME)
os.mkdir(os.path.join(workspace, 'data'))
os.mkdir(os.path.join(workspace, 'ignored-folder'))
create_file(workspace, 'image.png', '0')
create_file(workspace, 'image2.jpg', '1', file_path='ignored-folder')
create_file(workspace, 'file1', '0')
create_file(workspace, 'file2', '1')
create_ignore_file(workspace)
output = check_output(MLGIT_ADD % (DATASETS, DATASET_NAME, ''))
self.assertIn(output_messages['INFO_ADDING_PATH'] % DATASETS, output)
self.assertNotIn(ERROR_MESSAGE, output)
metadata = os.path.join(self.tmp_dir, ML_GIT_DIR, DATASETS, 'index',
'metadata', DATASET_NAME)
metadata_file = os.path.join(metadata, 'MANIFEST.yaml')
index_file = os.path.join(metadata, 'INDEX.yaml')
ignore_file = os.path.join(metadata, MLGIT_IGNORE_FILE_NAME)
self.assertTrue(os.path.exists(metadata_file))
self.assertTrue(os.path.exists(ignore_file))
self.assertTrue(os.path.exists(index_file))
self._check_index(index_file, ['data/file1', 'data/file2'],
['data/image.png', 'ignored-folder/image2.jpg'])
def test_06_create_with_wizard_enabled(self):
entity_type = DATASETS
entity_init(entity_type, self)
add_file(self, entity_type, '--bumpversion', 'new')
runner = CliRunner()
result = runner.invoke(entity.datasets, ['create', 'ENTITY-NAME', '--wizard'], input='test\nstrict')
self.assertIn(prompt_msg.CATEGORIES_MESSAGE, result.output)
def _commit_entity(self, entity_type):
entity_init(entity_type, self)
add_file(self, entity_type, '--bumpversion', 'new')
self.assertIn(output_messages['INFO_COMMIT_REPO'] % (os.path.join(self.tmp_dir, ML_GIT_DIR, entity_type, 'metadata'), entity_type + '-ex'),
check_output(MLGIT_COMMIT % (entity_type, entity_type + '-ex', '')))
HEAD = os.path.join(self.tmp_dir, ML_GIT_DIR, entity_type, 'refs', entity_type + '-ex', 'HEAD')
self.assertTrue(os.path.exists(HEAD))
def set_up_test(self, repo_type=MODELS):
self.TAG_TIMES = []
entity_name = '{}-ex'.format(repo_type)
entity_init(repo_type, self)
data_path = os.path.join(self.tmp_dir, repo_type, entity_name)
create_file(data_path, 'file', '0', '')
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 test_05_commit_with_wizard_enabled(self):
entity_type = LABELS
entity_init(entity_type, self)
add_file(self, entity_type, '--bumpversion', 'new')
runner = CliRunner()
result = runner.invoke(entity.labels, ['commit', entity_type + '-ex', '--wizard'], input='\n'.join(['', 'message']))
self.assertIn(prompt_msg.COMMIT_VERSION.format('labels', '1'), result.output)
self.assertIn(prompt_msg.COMMIT_MESSAGE, result.output)
def test_13_commit_with_invalid_related_entity_name(self):
entity_type = MODELS
entity_name = entity_type + '-ex'
entity_init(entity_type, self)
add_file(self, entity_type, '--bumpversion', 'new')
self.assertIn(output_messages['ERROR_ENTITY_NOT_FIND'].format('wrong-entity'),
check_output(MLGIT_COMMIT % (entity_type, entity_name, ' --labels=wrong-entity')))
HEAD = os.path.join(self.tmp_dir, ML_GIT_DIR, entity_type, 'refs', entity_name, 'HEAD')
self.assertFalse(os.path.exists(HEAD))
def test_12_commit_with_empty_related_entity_name(self):
entity_type = MODELS
entity_name = entity_type + '-ex'
entity_init(entity_type, self)
add_file(self, entity_type, '--bumpversion', 'new')
self.assertIn(output_messages['ERROR_INVALID_VALUE_FOR'] % ('--labels', 'Value cannot be empty'),
check_output(MLGIT_COMMIT % (entity_type, entity_name, ' --labels=')))
HEAD = os.path.join(self.tmp_dir, ML_GIT_DIR, entity_type, 'refs', entity_name, 'HEAD')
self.assertFalse(os.path.exists(HEAD))
def test_19_commit_files_to_labels_with_wizard_enabled(self):
entity_type = LABELS
entity_init(entity_type, self)
add_file(self, entity_type, '--bumpversion', 'new')
runner = CliRunner()
result = runner.invoke(entity.labels, ['commit', entity_type + '-ex', '--wizard'], input='\n'.join(['', 'message', 'y', ' ', 'dataset-ex']))
self.assertIn(prompt_msg.COMMIT_VERSION.format('labels', '1'), result.output)
self.assertIn(prompt_msg.COMMIT_MESSAGE, result.output)
self.assertIn(prompt_msg.WANT_LINK_TO_LABEL_ENTITY.format(parse_entity_type_to_singular(DATASETS)), result.output)
self.assertIn(prompt_msg.DEFINE_LINKED_DATASET, result.output)
def test_06_add_command_with_corrupted_file_added(self):
entity_init('dataset', self)
add_file(self, 'dataset', '--bumpversion', 'new')
corrupted_file = os.path.join(self.tmp_dir, 'dataset', 'dataset-ex', 'newfile0')
os.chmod(corrupted_file, S_IWUSR | S_IREAD)
with open(corrupted_file, 'wb') as z:
z.write(b'0' * 0)
self.assertIn(messages[67], check_output(MLGIT_ADD % ('dataset', 'dataset-ex', '--bumpversion')))
def test_16_commit_with_multiple_related_entities(self):
entity_type = MODELS
entity_name = entity_type + '-ex'
entity_init(entity_type, self)
add_file(self, entity_type, '--bumpversion', 'new')
self.assertIn(output_messages['ERROR_OPTION_WITH_MULTIPLE_VALUES'].format('wrong-entity'),
check_output(MLGIT_COMMIT % (entity_type, entity_name, ' --labels=A --labels=B')))
self.assertIn(output_messages['ERROR_OPTION_WITH_MULTIPLE_VALUES'],
check_output(MLGIT_COMMIT % (entity_type, entity_name, ' --labels=A --labels=B')))
self.assertIn(output_messages['ERROR_OPTION_WITH_MULTIPLE_VALUES'],
check_output(MLGIT_COMMIT % (entity_type, entity_name, ' --dataset=A --dataset=B')))
HEAD = os.path.join(self.tmp_dir, ML_GIT_DIR, entity_type, 'refs', entity_name, 'HEAD')
self.assertFalse(os.path.exists(HEAD))
def test_09_commit_with_ignore_file(self):
entity_init(DATASETS, self)
workspace = os.path.join(self.tmp_dir, DATASETS, DATASET_NAME)
os.mkdir(os.path.join(workspace, 'data'))
create_file(workspace, 'image.png', '0')
create_file(workspace, 'file1', '0')
create_ignore_file(workspace)
self.assertNotIn(ERROR_MESSAGE, check_output(MLGIT_ADD % (DATASETS, DATASET_NAME, '')))
self.assertNotIn(ERROR_MESSAGE, check_output(MLGIT_COMMIT % (DATASETS, DATASET_NAME, '')))
metadata = os.path.join(self.tmp_dir, ML_GIT_DIR, DATASETS, 'metadata', DATASET_NAME)
manifest_file = os.path.join(metadata, 'MANIFEST.yaml')
ignore_file = os.path.join(metadata, MLGIT_IGNORE_FILE_NAME)
self.assertTrue(os.path.exists(ignore_file))
self.assertTrue(os.path.exists(manifest_file))
def set_up_test(self, repo_type=DATASETS, with_metrics=False):
entity = '{}-ex'.format(repo_type)
entity_init(repo_type, self)
data_path = os.path.join(self.tmp_dir, repo_type, entity)
create_file(data_path, 'file', '0', '')
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_13_add_entity_with_readme_file_in_data(self):
entity_init(DATASETS, self)
workspace = os.path.join(self.tmp_dir, DATASETS, DATASET_NAME)
create_file(workspace, 'README.md', '0', file_path='')
os.mkdir(os.path.join(workspace, 'data'))
create_file(workspace, 'README.md', '0', file_path='data')
output = check_output(MLGIT_ADD % (DATASETS, DATASET_NAME, ''))
self.assertIn(output_messages['INFO_ADDING_PATH'] % DATASETS, output)
self.assertNotIn(ERROR_MESSAGE, output)
metadata = os.path.join(self.tmp_dir, ML_GIT_DIR, DATASETS, 'index',
'metadata', DATASET_NAME)
metadata_file = os.path.join(metadata, 'MANIFEST.yaml')
index_file = os.path.join(metadata, 'INDEX.yaml')
self.assertTrue(os.path.exists(metadata_file))
self.assertTrue(os.path.exists(index_file))
def test_04_add_files_with_bumpversion(self):
entity_init(DATASETS, self)
add_file(self, DATASETS, '--bumpversion', 'new')
def test_03_add_files_to_labels(self):
entity_init(LABELS, self)
add_file(self, LABELS, '', 'new')
def test_02_add_files_to_model(self):
entity_init(MODELS, self)
add_file(self, MODELS, '', 'new')
def test_01_add_files_to_dataset(self):
entity_init(DATASETS, self)
add_file(self, DATASETS, '', 'new')
def test_14_first_commit_without_add(self):
entity_type = DATASETS
entity_init(entity_type, self)
self.assertIn(output_messages['ERROR_COMMIT_WITHOUT_ADD'].format(DATASETS), check_output(MLGIT_COMMIT % (entity_type, entity_type + '-ex', '')))
HEAD = os.path.join(self.tmp_dir, ML_GIT_DIR, entity_type, 'refs', entity_type + '-ex', 'HEAD')
self.assertFalse(os.path.exists(HEAD))
def test_01_add_files_to_dataset(self):
entity_init('dataset', self)
add_file(self, 'dataset', '', 'new')
def test_02_add_files_to_model(self):
entity_init('model', self)
add_file(self, 'model', '', 'new')
def test_03_add_files_to_labels(self):
entity_init('labels', self)
add_file(self, 'labels', '', 'new')
def test_04_add_files_with_bumpversion(self):
entity_init('dataset', self)
add_file(self, 'dataset', '--bumpversion', 'new')
