def test_all_formulas(self):
self.assertEqual(
get_all_formulas(read_sample("x,y\n1,2\n3,4")),
set(
[
frozendict({}),
frozendict({"x": "1"}),
frozendict({"y": "2"}),
frozendict({"x": "3"}),
frozendict({"y": "4"}),
frozendict({"y": "2", "x": "1"}),
frozendict({"y": "4", "x": "3"}),
]
),
)
self.assertEqual(
get_all_formulas(read_sample("x\n{1 2}\n{3 4}")),
set(
[
frozendict({}),
frozendict({"x": frozenset(["1"])}),
frozendict({"x": frozenset(["4"])}),
frozendict({"x": frozenset(["3"])}),
frozendict({"x": frozenset(["2"])}),
frozendict({"x": frozenset(["1", "2"])}),
frozendict({"x": frozenset(["3", "4"])}),
]
),
)
def test_all_formulas(self):
self.assertEqual(
get_all_formulas(read_sample('x,y\n1,2\n3,4')),
{frozenset([Cell(attribute='x', sequence=0, value='1')]),
frozenset([Cell(attribute='x', sequence=0, value='3')]),
frozenset([Cell(attribute='y', sequence=0, value='4')]),
frozenset([Cell(attribute='x', sequence=0, value='1'), Cell(attribute='y', sequence=0, value='2')]),
frozenset([Cell(attribute='y', sequence=0, value='2')]),
frozenset([Cell(attribute='y', sequence=0, value='4'), Cell(attribute='x', sequence=0, value='3')])})
self.assertEqual(
get_all_formulas(read_sample('x\n{1 2}\n{3 4}')),
{frozenset([Cell(attribute='x', sequence=0, value='3'), Cell(attribute='x', sequence=0, value='4')]),
frozenset([Cell(attribute='x', sequence=0, value='4')]),
frozenset([Cell(attribute='x', sequence=0, value='1')]),
frozenset([Cell(attribute='x', sequence=0, value='1'), Cell(attribute='x', sequence=0, value='2')]),
frozenset([Cell(attribute='x', sequence=0, value='3')]),
frozenset([Cell(attribute='x', sequence=0, value='2')])})
def test_relation(self):
self.assertEqual(
read_sample(example),
[
frozendict({"y": "c", "x": frozenset(["a", "b"])}),
frozendict({"y": "c"}),
frozendict({"y": "c", "x": ("a", "d")}),
],
)
def test_relation(self):
self.assertEqual(
read_sample(example),
[frozenset([Cell(attribute='x', sequence=0, value='a'),
Cell(attribute='x', sequence=0, value='b'),
Cell(attribute='y', sequence=0, value='c')]),
frozenset([Cell(attribute='y', sequence=0, value='c')]),
frozenset([Cell(attribute='x', sequence=0, value='a'),
Cell(attribute='y', sequence=0, value='c'),
Cell(attribute='x', sequence=1, value='d')])])
def test_cost(self):
self.assertEqual(cost(formulas=read_sample('x, y\n1\n, 4'),
relation=read_sample('x, y\n1, 2\n1, 4\n1, 4')),
3)
self.assertEqual(cost(formulas=read_sample('x\n{1}\n{4}'),
relation=read_sample('x\n{1 2}\n{1 4}\n{1 4}')),
3)
self.assertEqual(cost(formulas=read_sample('x,y,z\na,b,\n,,c'),
relation=read_sample(three_attr_null)),
6)
def test_repr(self):
self.assertEqual(map(tuple_rep, read_sample(example)),
['x[0]:a x[0]:b y[0]:c', 'y[0]:c', 'x[0]:a x[1]:d y[0]:c'])
self.assertEqual(relation_rep(read_sample(example)),
'x[0]:a x[0]:b y[0]:c\ny[0]:c\nx[0]:a x[1]:d y[0]:c')
def test_repr(self):
self.assertEqual(map(tuple_rep, read_sample(example)), ["x:{a b} y:c", "y:c", "x:[a d] y:c"])
def test_cost(self):
self.assertEqual(cost(formulas=read_sample("x, y\n1\n, 4"), relation=read_sample("x, y\n1, 2\n1, 4\n1, 4")), 3)
self.assertEqual(cost(formulas=read_sample("x\n{1}\n{4}"), relation=read_sample("x\n{1 2}\n{1 4}\n{1 4}")), 3)
def run():
relation = read_sample(larger_example)
return find_incremental(relation, 5)
summary = improved_summary
best_cost = improved_cost
# resort cells
all_cells = [x for x in all_cells if x[0] > 0]
all_cells.sort()
all_cells.reverse()
return best_cost, summary
if __name__ == '__main__':
logger.setLevel(logging.INFO)
relation = read_sample(three_attr_null)
print relation_rep(relation)
print
best_cost, best = find_incremental(relation, 3)
print('Best cost:', best_cost)
print(relation_rep(best))
# speedtest
logger.setLevel(logging.WARN)
def run():
relation = read_sample(larger_example)
return find_incremental(relation, 5)
