def compile(protocol: callable, enable_optimizations: [bool, None] = True):
dag = Dag(protocol())
if enable_optimizations:
compile_dag(dag)
else:
compile_dag_without_optimizations(dag)
return dag
def test_agg_mean(party_data, expected):
cols_in_one = create_cols(party_data[0])
rel_one = create("in1", cols_in_one, party_data[0]["stored_with"])
agg = aggregate(rel_one, "agg", party_data[0]["col_names"][:1],
party_data[0]["col_names"][1], "mean")
div = divide(agg, "div", party_data[0]["col_names"][1], [10])
collect(div, {1, 2})
d = Dag({rel_one})
compile_dag(d)
compare_to_expected(d, expected)
def test_single_dataset(party_data, expected):
cols_in_one = create_cols(party_data[0])
rel_one = create("in1", cols_in_one, party_data[0]["stored_with"])
agg = aggregate(rel_one, "agg", party_data[0]["col_names"][:1],
party_data[0]["col_names"][1], "mean")
div = divide(agg, "div", party_data[0]["col_names"][1], [10])
collect(div, {1, 2})
d = Dag({rel_one})
compile_dag(d)
p = HeuristicPart(d)
parts = p.partition(100)
compare_partition_to_expected(parts, expected)
def test_agg_variance(party_data, expected):
cols_in_one = create_cols(party_data[0])
cols_in_two = create_cols(party_data[1])
rel_one = create("in1", cols_in_one, party_data[0]["stored_with"])
rel_two = create("in2", cols_in_two, party_data[1]["stored_with"])
cc = concat([rel_one, rel_two], "concat", party_data[0]["col_names"])
variance = aggregate(cc, "variance", [party_data[0]["col_names"][0]],
party_data[0]["col_names"][1], "variance")
mult = multiply(variance, "mult", party_data[0]["col_names"][0],
[party_data[0]["col_names"][1], 7])
collect(mult, {1, 2})
d = Dag({rel_one, rel_two})
compile_dag(d)
compare_to_expected(d, expected)
def test_join(party_data, expected):
cols_in_one = create_cols(party_data[0])
cols_in_two = create_cols(party_data[1])
rel_one = create("in1", cols_in_one, party_data[0]["stored_with"])
rel_two = create("in2", cols_in_two, party_data[1]["stored_with"])
j = join(rel_one, rel_two, "join", [party_data[0]["col_names"][0]], [party_data[1]["col_names"][0]])
agg = aggregate(j, "agg", [party_data[0]["col_names"][0]], party_data[0]["col_names"][1], "mean")
div = divide(agg, "div", party_data[0]["col_names"][1], [10])
collect(div, {1, 2})
d = Dag({rel_one, rel_two})
compile_dag(d)
p = HeuristicPart(d)
parts = p.partition(100)
compare_partition_to_expected(parts, expected)
def test_three_datasets_concat(party_data, expected):
cols_in_one = create_cols(party_data[0])
cols_in_two = create_cols(party_data[1])
cols_in_three = create_cols(party_data[2])
rel_one = create("in1", cols_in_one, party_data[0]["stored_with"])
rel_two = create("in2", cols_in_two, party_data[1]["stored_with"])
rel_three = create("in3", cols_in_three, party_data[2]["stored_with"])
cc = concat([rel_one, rel_two, rel_three], "cc",
party_data[0]["col_names"])
agg = aggregate(cc, "agg", party_data[0]["col_names"][:1],
party_data[0]["col_names"][1], "mean")
div = divide(agg, "div", party_data[0]["col_names"][1], [10])
collect(div, {1, 2, 3})
d = Dag({rel_one, rel_two, rel_three})
compile_dag(d)
p = HeuristicPart(d)
parts = p.partition(100)
compare_partition_to_expected(parts, expected)