def test_diff_refs_modified_file(self):
repo_type = DATASETS
mdpath = os.path.join(self.test_dir, '.ml-git', repo_type, 'metadata')
entity = 'dataset-ex'
specpath = os.path.join('vision-computer', 'images', entity)
config_test = deepcopy(config)
config_test['mlgit_path'] = '.ml-git'
m = Metadata(entity, mdpath, config_test, repo_type)
m.init()
ensure_path_exists(os.path.join(mdpath, specpath, entity))
manifestpath = os.path.join(os.path.join(mdpath, specpath),
'MANIFEST.yaml')
shutil.copy('hdata/dataset-ex.spec',
os.path.join(mdpath, specpath, '{}.spec'.format(entity)))
yaml_save(files_mock, manifestpath)
sha1 = m.commit(manifestpath, 'test')
files_mock_copy = deepcopy(files_mock)
del files_mock_copy[
'zdj7WZzR8Tw87Dx3dm76W5aehnT23GSbXbQ9qo73JgtwREGwB']
files_mock_copy['NewHash'] = {'7.jpg'}
yaml_save(files_mock_copy, manifestpath)
sha2 = m.commit(manifestpath, 'test')
added_files, deleted_files, modified_file = m.diff_refs_with_modified_files(
entity, sha1, sha2)
self.assertTrue(len(added_files) == 0)
self.assertTrue(len(deleted_files) == 0)
self.assertTrue(len(modified_file) == 1)
def test_diff_refs_add_file(self):
repo_type = DATASETS
mdpath = os.path.join(self.test_dir, '.ml-git', repo_type, 'metadata')
entity = 'dataset-ex'
specpath = os.path.join('vision-computer', 'images', entity)
config_test = deepcopy(config)
config_test['mlgit_path'] = '.ml-git'
m = Metadata(entity, mdpath, config_test, repo_type)
m.init()
ensure_path_exists(os.path.join(mdpath, specpath, entity))
manifestpath = os.path.join(os.path.join(mdpath, specpath),
'MANIFEST.yaml')
shutil.copy('hdata/dataset-ex.spec',
os.path.join(mdpath, specpath, '{}.spec'.format(entity)))
yaml_save(files_mock, manifestpath)
sha1 = m.commit(manifestpath, 'test')
files_mock_copy = deepcopy(files_mock)
files_mock_copy[
'zPaksM5tNewHashQ2VABPvvfC3VW6wFRTWKvFhUW5QaDx6JMoma'] = {
'11.jpg'
}
yaml_save(files_mock_copy, manifestpath)
sha2 = m.commit(manifestpath, 'test')
added_files, deleted_files, modified_file = m.diff_refs_with_modified_files(
entity, sha1, sha2)
self.assertTrue(len(added_files) == 1)
self.assertTrue(len(deleted_files) == 0)
self.assertTrue(len(modified_file) == 0)
def test_get_metrics(self):
repo_type = MODELS
mdpath = os.path.join(self.test_dir, 'mdata', repo_type, 'metadata')
specpath = os.path.join('vision-computer', 'images')
entity = 'model-ex'
m = Metadata(entity, self.test_dir, config, repo_type)
m.init()
ensure_path_exists(os.path.join(mdpath, specpath, entity))
spec_metadata_path = os.path.join(mdpath, specpath, entity, 'model-ex.spec')
shutil.copy('hdata/dataset-ex.spec', spec_metadata_path)
spec_file = yaml_load(spec_metadata_path)
spec_file[MODEL_SPEC_KEY] = deepcopy(spec_file[DATASET_SPEC_KEY])
del spec_file[DATASET_SPEC_KEY]
spec_file[MODEL_SPEC_KEY]['metrics'] = {'metric_1': 0, 'metric_2': 1}
yaml_save(spec_file, spec_metadata_path)
tag = 'vision-computer__images__model-ex__1'
sha = m.commit(spec_metadata_path, specpath)
m.tag_add(tag)
metrics = m._get_metrics(entity, sha)
test_table = PrettyTable()
test_table.field_names = ['Name', 'Value']
test_table.align['Name'] = 'l'
test_table.align['Value'] = 'l'
test_table.add_row(['metric_1', 0])
test_table.add_row(['metric_2', 1])
test_metrics = '\nmetrics:\n{}'.format(test_table.get_string())
self.assertEqual(metrics, test_metrics)
def test_get_spec_content_from_ref(self):
mdpath = os.path.join(self.test_dir, 'mdata', DATASETS, 'metadata')
specpath = 'dataset-ex'
m = Metadata(specpath, self.test_dir, config, DATASETS)
m.init()
ensure_path_exists(os.path.join(mdpath, specpath))
spec_metadata_path = os.path.join(mdpath, specpath) + '/dataset-ex.spec'
shutil.copy('hdata/dataset-ex.spec', spec_metadata_path)
sha = m.commit(spec_metadata_path, specpath)
tag = m.tag_add(sha)
path = 'dataset-ex/dataset-ex.spec'
content = yaml_load_str(m._get_spec_content_from_ref(tag.commit, path))
spec_file = yaml_load(spec_metadata_path)
self.assertEqual(content, spec_file)
def test_get_specs_to_compare(self):
mdpath = os.path.join(self.test_dir, 'mdata', DATASETS, 'metadata')
specpath = 'dataset-ex'
m = Metadata(specpath, self.test_dir, config, DATASETS)
m.init()
ensure_path_exists(os.path.join(mdpath, specpath))
spec_metadata_path = os.path.join(mdpath, specpath) + '/dataset-ex.spec'
shutil.copy('hdata/dataset-ex.spec', spec_metadata_path)
sha = m.commit(spec_metadata_path, specpath)
m.tag_add(sha)
specs = m.get_specs_to_compare(specpath)
spec_file = yaml_load(spec_metadata_path)
for c, v in specs:
self.assertEqual(c, spec_file[DATASETS]['manifest'])
self.assertIsNotNone(v, {DATASETS: {'manifest': {}}})
def test_get_metrics_without_metrics(self):
repo_type = MODELS
mdpath = os.path.join(self.test_dir, 'mdata', repo_type, 'metadata')
specpath = os.path.join('vision-computer', 'images')
entity = 'model-ex'
m = Metadata(entity, self.test_dir, config, repo_type)
m.init()
ensure_path_exists(os.path.join(mdpath, specpath, entity))
spec_metadata_path = os.path.join(mdpath, specpath, entity) + '/model-ex.spec'
shutil.copy('hdata/dataset-ex.spec', spec_metadata_path)
spec_file = yaml_load(spec_metadata_path)
spec_file[MODEL_SPEC_KEY] = deepcopy(spec_file[DATASET_SPEC_KEY])
del spec_file[DATASET_SPEC_KEY]
yaml_save(spec_file, spec_metadata_path)
tag = 'vision-computer__images__model-ex__1'
sha = m.commit(spec_metadata_path, specpath)
m.tag_add(tag)
metrics = m._get_metrics(entity, sha)
self.assertEqual(metrics, '')
