def test_columns_without_local_data_and_without_key_argument(
self, repo_20_filled_samples):
repo = repo_20_filled_samples
co = repo.checkout()
from hangar.backends import backend_decoder
# perform a mock for nonlocal data
for k in co._columns._columns['writtenaset']._samples:
co._columns._columns['writtenaset']._samples[k] = backend_decoder(
b'50:daeaaeeaebv')
col = co.columns['writtenaset']
with pytest.raises(RuntimeError):
HangarDataset((col, ))
# perform a mock for nonlocal data
co = repo.checkout()
template = backend_decoder(b'50:daeaaeeaebv')
co._columns._columns['writtenaset']._samples['4'] = template
col = co.columns['writtenaset']
dataset = HangarDataset((col, ))
dataset_available_keys = dataset._keys
assert len(dataset_available_keys) == 19
assert '4' not in dataset_available_keys
column_reported_local_keys = list(col.keys(local=True))
for dset_avail_key in dataset_available_keys:
assert dset_avail_key in column_reported_local_keys
assert len(dataset_available_keys) == len(column_reported_local_keys)
co.close()
def test_keys_not_valid(self, repo_20_filled_samples):
co = repo_20_filled_samples.checkout()
first_col = co.columns['writtenaset']
keys = ['w', 'r', 'o', 'n', 'g']
dataset = HangarDataset((first_col, ), keys=keys)
with pytest.raises(KeyError):
dataset.index_get(1)
co.close()
def test_keys_multiple_column_success(self, repo_20_filled_samples):
co = repo_20_filled_samples.checkout()
first_col = co.columns['writtenaset']
second_col = co.columns['second_aset']
keys = [('1', '2'), ('2', '3'), ('3', '4')]
dataset = HangarDataset((first_col, second_col), keys=keys)
for i, key in enumerate(keys):
data = dataset.index_get(i)
assert np.allclose(data[0], first_col[key[0]])
assert np.allclose(data[1], second_col[key[1]])
co.close()
def test_column_without_wrapping_list(self, repo_20_filled_samples,
array5by7):
co = repo_20_filled_samples.checkout()
first_col = co.columns['writtenaset']
second_col = co.columns['second_aset']
dataset = HangarDataset((first_col, second_col))
key1, key2 = dataset._keys[0]
assert key1 == key2
target = array5by7[:] = int(key1)
assert np.allclose(dataset.index_get(0), target)
co.close()
def test_keys_single_column_success(self, repo_20_filled_samples):
co = repo_20_filled_samples.checkout()
first_col = co.columns['writtenaset']
keys = ['1', '2', '3']
dataset = HangarDataset((first_col, ), keys=keys)
assert dataset._keys == keys
co.close()
def test_fails_on_write_enabled_columns(self, repo_20_filled_samples):
repo = repo_20_filled_samples
co = repo.checkout(write=True)
first_aset = co.columns['writtenaset']
with pytest.raises(PermissionError):
HangarDataset((first_aset, ))
co.close()
def test_keys_non_local(self, repo_20_filled_samples):
repo = repo_20_filled_samples
co = repo.checkout()
# perform a mock for nonlocal data
from hangar.backends import backend_decoder
template = backend_decoder(b'50:daeaaeeaebv')
co._columns._columns['writtenaset']._samples['4'] = template
col = co.columns['writtenaset']
col_reported_remote_keys = col.remote_reference_keys
assert col_reported_remote_keys == ('4', )
assert len(col_reported_remote_keys) == 1
dataset = HangarDataset((col, ), keys=('0', *col_reported_remote_keys))
with pytest.raises(KeyError):
# TODO: hangar internal should raise FileNotFoundError?
dataset.index_get(1)
co.close()
def test_columns_without_common_keys_and_without_key_argument(
self, repo_20_filled_samples):
co = repo_20_filled_samples.checkout(write=True)
first_col = co.columns['writtenaset']
first_col['AnExtraKey'] = first_col['0']
co.commit('added an extra key')
co.close()
co = repo_20_filled_samples.checkout()
first_col = co.columns['writtenaset']
second_col = co.columns['second_aset']
with pytest.raises(KeyError):
HangarDataset((first_col, second_col))
co.close()
def test_keys_nested_column_success(self, repo_20_filled_subsamples):
co = repo_20_filled_subsamples.checkout()
col1 = co['writtenaset']
col2 = co['second_aset']
dataset = HangarDataset([col1, col2])
data = dataset.index_get(1)
assert tuple(data[0].keys()) == tuple(data[1].keys()) == (4, 5, 6)
assert isinstance(data[0], dict)
assert isinstance(data[1], dict)
keys = (((0, ...), (0, 1)), ((1, ...), (1, 4)))
dataset = HangarDataset([col1, col2], keys=keys)
data = dataset.index_get(1)
assert tuple(data[0].keys()) == (4, 5, 6)
assert np.allclose(data[1], col2[1][4])
co.close()
def test_no_column(self):
with pytest.raises(TypeError):
HangarDataset([])