def test_load_dataset_returns_cached_despite_update(upstream, repo, dataset):
ds = repo.load_dataset('dataset')
ds_updated = ds.copy()
ds_updated.df['b'] += 1
upstream.add_dataset(ds_updated, commit_message="Update dataset")
loaded = repo.load_dataset('dataset')
assert loaded.equals(ds)
def test_load_dataset_cached(upstream, repo, dataset):
# Load dataset to get it into cache
repo.load_dataset('dataset')
# Remove dataset from cloud to make sure it is returned from cache
upstream.clear_cloud()
loaded = repo.load_dataset('dataset')
assert loaded.equals(dataset)
def test_pull_dataset_enables_loading(repo, uncached_dataset):
repo.pull_datasets()
loaded = repo.load_dataset(uncached_dataset.identifier)
assert loaded.equals(uncached_dataset)
def test_load_dataset_fail_before_pull(repo, uncached_dataset):
with pytest.raises(NoOutputOrStageError):
repo.load_dataset(uncached_dataset.identifier)
def test_load_dataset_does_not_exist(repo):
with pytest.raises(NoOutputOrStageError):
repo.load_dataset('does-not-exist')
def test_load_dataset_not_cached(repo, cache, dataset):
assert dataset not in cache
loaded = repo.load_dataset(dataset.identifier)
assert dataset in cache
assert loaded.equals(dataset)