def find_rectangle_candidates(
cutout: np.array,
points: np.array,
candidate_limit: int = 20) -> List[Tuple[np.array]]:
"""Returns a list of rectangles with maximum space overlap with original figure."""
heap = Heap()
for quad in tqdm(combinations(points, 4),
total=num_combinations(len(points), 4)):
zeros = np.zeros_like(cutout)
for p in quad:
zeros[p[0], p[1]] = 1
chull = convex_hull_image(zeros)
space_part = np.sum(chull * cutout)
heap.insert((space_part, quad))
best_space, _ = heap.peek()
retlist = []
total = 0
while total < candidate_limit and heap.peek()[0] >= 0.9 * best_space:
space, cur_points = heap.pop()
retlist.append(cur_points)
return retlist
class TestHeap(unittest.TestCase):
def setUp(self) -> None:
self.heap = Heap()
self.heap.heap = [1, 3, 2]
def test_peek(self) -> None:
self.assertEqual(self.heap.peek(), 1)
def test_append(self) -> None:
self.heap.append(2)
self.assertEqual(self.heap.heap, [1, 2, 2, 3])
self.heap.append(1)
self.assertEqual(self.heap.heap, [1, 1, 2, 3, 2])
self.heap.append(3)
self.assertEqual(self.heap.heap, [1, 1, 2, 3, 2, 3])
def test_pop(self) -> None:
self.assertEqual(self.heap.pop(), 1)
self.assertEqual(self.heap.heap, [2, 3])
self.assertEqual(self.heap.pop(), 2)
self.assertEqual(self.heap.heap, [3])
self.assertEqual(self.heap.pop(), 3)
self.assertEqual(self.heap.heap, [])
def test_is_empty(self) -> None:
self.assertFalse(self.heap.is_empty())
self.heap.heap = []
self.assertTrue(self.heap.is_empty())
def test_clear(self) -> None:
self.assertNotEqual(self.heap.heap, [])
self.heap.clear()
self.assertEqual(self.heap.heap, [])
def test_big_test(self) -> None:
self.heap.clear()
test_case = 10_000
for i in reversed(range(test_case)):
self.heap.append(i)
for i in range(test_case):
self.assertEqual(self.heap.pop(), i)
def test_repr(self) -> None:
self.heap.append(4)
self.heap.append(5)
self.heap.append(6)
self.assertEqual(self.heap.__repr__(), "Heap [1, 3, 2, 4, 5, 6]")
self.heap.append(7)
self.assertEqual(self.heap.__repr__(), "Heap [1, 3, 2, 4, 5, 6, ...]")
