def test_load_from_private_reader_path_connect(self):
priv = PrivateReader.from_connection(pums,
privacy=privacy,
metadata=meta_path)
assert (isinstance(priv.reader, PandasReader))
res = priv.execute("SELECT COUNT(age) FROM PUMS.PUMS GROUP BY sex")
assert (len(res) == 3)
def test_count_query(self):
priv = PrivateReader.from_connection(pums,
privacy=privacy,
metadata=meta_path)
assert (priv.odometer.spent == (0.0, 0.0))
assert (priv.odometer.k == 0)
res = priv.execute("SELECT COUNT(age) FROM PUMS.PUMS GROUP BY educ")
assert (priv.odometer.k == 2)
def test_count_query(self):
priv = PrivateReader.from_connection(pums,
privacy=privacy,
metadata=meta_path)
res = priv._get_mechanism_costs(
"SELECT COUNT(age) FROM PUMS.PUMS GROUP BY educ")
res = [r for r in res if r]
assert (len(res) == 2)
def test_variance(self):
meta_obj['PUMS.PUMS'].row_privacy = True
meta_obj['PUMS.PUMS']['pid'].is_key = False
meta_obj['PUMS.PUMS'].censor_dims = True
priv = PrivateReader.from_connection(pums,
privacy=privacy,
metadata=meta_obj)
res = priv.execute("SELECT VAR(age) FROM PUMS.PUMS GROUP BY educ")
assert (priv.odometer.k == 3)
def test_count_row_privacy(self):
meta_obj['PUMS.PUMS'].row_privacy = True
meta_obj['PUMS.PUMS']['pid'].is_key = False
priv = PrivateReader.from_connection(pums,
privacy=privacy,
metadata=meta_obj)
assert (priv.odometer.spent == (0.0, 0.0))
assert (priv.odometer.k == 0)
res = priv.execute("SELECT COUNT(*) FROM PUMS.PUMS GROUP BY educ")
assert (priv.odometer.k == 1)
def test_count_row_privacy(self):
meta_obj['PUMS.PUMS'].row_privacy = True
meta_obj['PUMS.PUMS']['pid'].is_key = False
priv = PrivateReader.from_connection(pums,
privacy=privacy,
metadata=meta_obj)
res = priv._get_mechanism_costs(
"SELECT COUNT(*) FROM PUMS.PUMS GROUP BY educ")
res = [r for r in res if r]
assert (len(res) == 1)
def test_sum(self):
meta_obj['PUMS.PUMS'].row_privacy = True
meta_obj['PUMS.PUMS']['pid'].is_key = False
meta_obj['PUMS.PUMS'].censor_dims = True
priv = PrivateReader.from_connection(pums,
privacy=privacy,
metadata=meta_obj)
res = priv._get_mechanism_costs(
"SELECT SUM(age) FROM PUMS.PUMS GROUP BY educ")
res = [r for r in res if r]
assert (len(res) == 2)
def test_count_row_privacy_col_quantifier(self):
meta_obj['PUMS.PUMS'].row_privacy = True
meta_obj['PUMS.PUMS']['pid'].is_key = False
meta_obj['PUMS.PUMS'].censor_dims = True
priv = PrivateReader.from_connection(pums,
privacy=privacy,
metadata=meta_obj)
assert (priv.odometer.spent == (0.0, 0.0))
assert (priv.odometer.k == 0)
res = priv.execute(
"SELECT COUNT(DISTINCT age) FROM PUMS.PUMS GROUP BY educ")
assert (priv.odometer.k == 2)
def test_two_var(self):
meta_obj['PUMS.PUMS'].row_privacy = True
meta_obj['PUMS.PUMS']['pid'].is_key = False
meta_obj['PUMS.PUMS'].censor_dims = True
priv = PrivateReader.from_connection(pums,
privacy=privacy,
metadata=meta_obj)
query = "SELECT VAR(age), VAR(income) FROM PUMS.PUMS GROUP BY sex"
res = priv._get_mechanism_costs(query)
res = [r for r in res if r]
assert (len(res) == 5)
eps, _ = priv.get_privacy_cost(query)
assert (eps <= 5.0)
def test_load_from_private_reader_path(self):
reader = PandasReader(pums, meta_path)
priv = PrivateReader(reader, meta_path, privacy=Privacy(epsilon=1.0))
res = priv.execute("SELECT COUNT(age) FROM PUMS.PUMS GROUP BY sex")
assert (len(res) == 3)