def test_save_and_load_catalog(data_set, dummy_dataframe, tmpdir):
catalog = DataCatalogWithDefault(data_sets={"test": data_set},
default=default_csv)
path = str(tmpdir.mkdir("sub").join("test"))
catalog.save(path, dummy_dataframe)
reloaded_df = catalog.load(path)
assert dummy_dataframe.equals(reloaded_df)
def test_load_from_unregistered(dummy_dataframe, tmpdir):
catalog = DataCatalogWithDefault(data_sets={}, default=default_csv)
path = str(tmpdir.mkdir("sub").join("test.csv"))
catalog.save(path, dummy_dataframe)
reloaded_df = catalog.load(path)
assert dummy_dataframe.equals(reloaded_df)
def test_remember_save(tmpdir, dummy_dataframe):
catalog = DataCatalogWithDefault(data_sets={},
default=default_csv,
remember=True)
path = str(tmpdir.mkdir("sub").join("test.csv"))
catalog.save(path, dummy_dataframe)
assert tmpdir.join("sub").join("test.csv") in catalog.list()
def test_default_none():
with pytest.raises(
TypeError,
match="Default must be a callable with a "
"single input string argument: the "
"key of the requested data set.",
):
DataCatalogWithDefault(data_sets={}, default=None)
def test_from_sane_config_default(sane_config, dummy_dataframe, tmpdir):
catalog = DataCatalog.from_config(sane_config["catalog"],
sane_config["credentials"])
catalog_with_default = DataCatalogWithDefault.from_data_catalog(
catalog, default_csv)
path = str(tmpdir.mkdir("sub").join("missing.csv"))
catalog_with_default.save(path, dummy_dataframe)
reloaded_df = catalog_with_default.load(path)
assert dummy_dataframe.equals(reloaded_df)
def test_from_sane_config(sane_config):
with pytest.raises(ValueError,
match="Cannot instantiate a `DataCatalogWithDefault`"):
DataCatalogWithDefault.from_config(sane_config["catalog"],
sane_config["credentials"])
def test_remember_load():
catalog = DataCatalogWithDefault(data_sets={},
default=default_memory,
remember=True)
assert catalog.load("any") == 5
assert "any" in catalog.list()