def test_only_encrypt_data(compute_backend_pii_df):
sensitivewidget = sensitive_data(compute_backend_pii_df,
mode="encrypt",
detect_infotypes=False)
assert isinstance(sensitivewidget, SensitiveDataWidget)
assert _is_dataframe(sensitivewidget.encrypt)
assert isinstance(sensitivewidget.encrypt["name"][1], str)
assert isinstance(sensitivewidget.encrypt["domain"][1], str)
assert isinstance(sensitivewidget.redact, type(None))
assert isinstance(sensitivewidget.infotypes, type(None))
def test_sensitive_data_cols(compute_backend_pii_df):
sensitivewidget = sensitive_data(compute_backend_pii_df,
mode="redact",
columns=["name"],
detect_infotypes=False)
assert isinstance(sensitivewidget, SensitiveDataWidget)
assert _is_dataframe(sensitivewidget.redact)
assert sensitivewidget.redact.shape == (1, 1)
assert isinstance(sensitivewidget.infotypes, type(None))
assert isinstance(sensitivewidget.encrypt, type(None))
def test_only_redact_data(compute_backend_pii_df):
sensitivewidget = sensitive_data(compute_backend_pii_df,
mode="redact",
detect_infotypes=False)
assert isinstance(sensitivewidget, SensitiveDataWidget)
assert sensitivewidget.redact.shape == (1, 2)
assert _is_dataframe(sensitivewidget.redact)
assert sensitivewidget.redact["name"][1] == "<PERSON>"
assert sensitivewidget.redact["domain"][1] == "<DOMAIN_NAME>"
assert isinstance(sensitivewidget.redact["name"][1], str)
assert isinstance(sensitivewidget.redact["domain"][1], str)
assert isinstance(sensitivewidget.encrypt, type(None))
assert isinstance(sensitivewidget.infotypes, type(None))
def test_encrypt_data_and_infotypes(compute_backend_pii_df):
sensitivewidget = sensitive_data(compute_backend_pii_df,
mode="encrypt",
detect_infotypes=True,
sample_size=1)
assert isinstance(sensitivewidget.encrypt["name"][1], str)
assert isinstance(sensitivewidget.encrypt["domain"][1], str)
assert isinstance(sensitivewidget.infotypes, dict)
assert sensitivewidget.infotypes["domain"][0] == "DOMAIN_NAME"
assert sensitivewidget.infotypes["name"][0] == "PERSON"
assert len(sensitivewidget.infotypes) == 2
assert isinstance(sensitivewidget.redact, type(None))
assert _is_dataframe(sensitivewidget.show())
def test_sample_size(compute_backend_pii_df):
with pytest.raises(ValueError):
sensitive_data(compute_backend_pii_df,
mode=None,
detect_infotypes=True,
sample_size=9)
def test_mode_value(compute_backend_pii_df):
with pytest.raises(ValueError):
sensitive_data(compute_backend_pii_df, mode="invalid mode")
def test_column_type(compute_backend_pii_df):
with pytest.raises(TypeError):
sensitive_data(compute_backend_pii_df, columns="this is not a list")
def test_type_df_type(compute_backend_pii_text):
with pytest.raises(ValueError):
sensitive_data(compute_backend_pii_text)