def test_shape(self):
bands = [CompressedBand((16, 16), np.float32),
CompressedBand((16, 16), np.float32),
CompressedBand((16, 16), np.float32)]
indexer1 = BandIndexer([bands[0]])
self.assertEqual(indexer1.shape, (16, 16))
indexer3 = BandIndexer(bands)
self.assertEqual(indexer3.shape, (16, 16))
def test_get_masked(self):
values = np.ones([16, 16])
band = CompressedBand((16, 16), np.float32)
band.setblock(0, 0, values)
indexer = BandIndexer([band])
mask = np.zeros([16, 16], dtype=np.bool)
mask[8:, 2:] = True
self.assertEqual(np.sum(indexer[mask]), 112)
def test_set_errors(self):
bands = [CompressedBand((16, 16), np.float32),
CompressedBand((16, 16), np.float32),
CompressedBand((16, 16), np.float32)]
indexer = BandIndexer(bands)
with self.assertRaises(TypeError):
indexer["a", 0, 0] = 1.0
with self.assertRaises(TypeError):
indexer[0, [1, 3], 0] = 1.0
with self.assertRaises(TypeError):
indexer[0, 0, None] = 1.0
def test_get_set_typeerror(self):
bands = [CompressedBand((16, 16), np.float32),
CompressedBand((16, 16), np.float32),
CompressedBand((16, 16), np.float32)]
indexer = BandIndexer(bands)
with self.assertRaises(TypeError):
indexer[None]
with self.assertRaises(TypeError):
indexer[None] = 3.0
def test_get_mask2(self):
values = np.ones([3, 3])
bands = [CompressedBand((3, 3), np.float32),
CompressedBand((3, 3), np.float32),
CompressedBand((3, 3), np.float32)]
bands[0].setblock(0, 0, values)
bands[1].setblock(0, 0, 2*values)
bands[2].setblock(0, 0, 3*values)
indexer = BandIndexer(bands)
mask = np.array([[True, False, False], [False, True, False], [False, False, True]])
result = indexer[mask]
npt.assert_equal(np.tile(np.array([1, 2, 3]), (3, 1)), result)
def test_set_masked_array(self):
bands = [CompressedBand((16, 16), np.float32),
CompressedBand((16, 16), np.float32),
CompressedBand((16, 16), np.float32)]
mask = np.zeros([16, 16], dtype=np.bool)
mask[8:, 2:] = True
indexer = BandIndexer(bands)
indexer[:,:,:] = np.zeros([16, 16])
indexer[mask] = np.ones(8*14)
self.assertEqual(np.sum(indexer[:,:]), 336)
def test_set_multibanded_masked_scalar(self):
values = np.ones([16, 16])
bands = [CompressedBand((16, 16), np.float32),
CompressedBand((16, 16), np.float32),
CompressedBand((16, 16), np.float32)]
mask = np.zeros([16, 16], dtype=np.bool)
mask[8:, 2:] = True
indexer = BandIndexer(bands)
indexer[:,:] = np.zeros([16, 16])
indexer[mask] = 1.0
self.assertEqual(np.sum(indexer[:,:]), 336)
return
def test_set_multibanded(self):
values = np.ones([16, 16])
bands = [
CompressedBand((16, 16), np.float32),
CompressedBand((16, 16), np.float32),
CompressedBand((16, 16), np.float32)
]
indexer = BandIndexer(bands)
indexer[:, :, 0] = values
indexer[:, :, 1:] = 2 * values
self.assertTrue(np.all(bands[0][:, :] == 1.0))
self.assertTrue(np.all(bands[1][:, :] == 2.0))
self.assertTrue(np.all(bands[2][:, :] == 2.0))
return
def test_set_slice(self):
values = np.ones([16, 16])
bands = [CompressedBand((16, 16), np.float32),
CompressedBand((16, 16), np.float32),
CompressedBand((16, 16), np.float32)]
indexer = BandIndexer(bands)
indexer[:,:,0] = values
indexer[:,:,1:] = 2*values
npt.assert_equal(bands[0].getblock(0, 0, 16, 16), 1.0)
npt.assert_equal(bands[1].getblock(0, 0, 16, 16), 2.0)
npt.assert_equal(bands[2].getblock(0, 0, 16, 16), 2.0)
indexer[:,:,1:] = np.dstack([2*values, 3*values])
npt.assert_equal(bands[1].getblock(0, 0, 16, 16), 2.0)
npt.assert_equal(bands[2].getblock(0, 0, 16, 16), 3.0)
def test_set_mask2(self):
values = np.ones([3, 3])
bands = [CompressedBand((3, 3), np.float32),
CompressedBand((3, 3), np.float32),
CompressedBand((3, 3), np.float32)]
bands[0].setblock(0, 0, values)
bands[1].setblock(0, 0, 2*values)
bands[2].setblock(0, 0, 3*values)
indexer = BandIndexer(bands)
mask = np.array([[True, False, False], [False, True, False], [False, False, True]])
indexer[mask] = -1
for band in bands:
self.assertEqual(band.getblock(0, 0, 1, 1)[0], -1.0)
self.assertEqual(band.getblock(1, 1, 1, 1)[0], -1.0)
self.assertEqual(band.getblock(2, 2, 1, 1)[0], -1.0)
def test_set_integer(self):
bands = [CompressedBand((16, 16), np.float32),
CompressedBand((16, 16), np.float32),
CompressedBand((16, 16), np.float32)]
indexer = BandIndexer(bands)
indexer[3,4,0] = -1.0
self.assertEqual(bands[0].getblock(3, 4, 1, 1)[0], -1.0)
indexer[4,5,:] = np.array([1.0, 2.0, 3.0])
self.assertEqual(bands[0].getblock(4, 5, 1, 1)[0], 1.0)
self.assertEqual(bands[1].getblock(4, 5, 1, 1)[0], 2.0)
self.assertEqual(bands[2].getblock(4, 5, 1, 1)[0], 3.0)
indexer[5,6,:] = 5.0
self.assertEqual(bands[0].getblock(5, 6, 1, 1)[0], 5.0)
self.assertEqual(bands[1].getblock(5, 6, 1, 1)[0], 5.0)
self.assertEqual(bands[2].getblock(5, 6, 1, 1)[0], 5.0)
def test_get_multibanded(self):
values = np.ones([16, 16])
bands = [CompressedBand((16, 16), np.float32),
CompressedBand((16, 16), np.float32),
CompressedBand((16, 16), np.float32)]
bands[0][:,:] = values
bands[1][:,:] = 2*values
bands[2][:,:] = 3*values
indexer = BandIndexer(bands)
result = indexer[4:7,2:8,:]
self.assertEqual(result.shape, (3, 6, 3))
self.assertTrue(np.all(result[0,0,:] == np.array([1.0, 2.0, 3.0])))
# make sure it works with a scalar band index
result = indexer[4:7,2:8,1]
self.assertEqual(result.shape, (3, 6))
self.assertTrue(np.all(result == 2.0))
return
def test_get_integer(self):
values = np.ones([16, 16])
bands = [CompressedBand((16, 16), np.float32),
CompressedBand((16, 16), np.float32),
CompressedBand((16, 16), np.float32)]
bands[0].setblock(0, 0, values)
bands[1].setblock(0, 0, 2*values)
bands[2].setblock(0, 0, 3*values)
indexer = BandIndexer(bands)
result = indexer[10,2]
npt.assert_equal(result, np.array([1.0, 2.0, 3.0]))
result = indexer[10,2,:]
npt.assert_equal(result, np.array([1.0, 2.0, 3.0]))
# make sure it works with a scalar band index
result = indexer[10,2,1]
self.assertEqual(result, 2.0)
def test_get_mask3(self):
values = np.ones([3, 3])
bands = [CompressedBand((3, 3), np.float32),
CompressedBand((3, 3), np.float32),
CompressedBand((3, 3), np.float32)]
bands[0].setblock(0, 0, values)
bands[1].setblock(0, 0, 2*values)
bands[2].setblock(0, 0, 3*values)
indexer = BandIndexer(bands)
# Create mask in band, row, column order than use np.moveaxis to put in
# row, column, band order for karta
mask = np.array([
[[True, False, False], [False, True, False], [False, False, True]],
[[False, False, False], [False, False, False], [False, True, False]],
[[True, False, False], [False, False, False], [True, False, True]]])
result = indexer[np.moveaxis(mask, 0, -1)]
expected = np.array([1, 3, 1, 3, 2, 1, 3])
npt.assert_equal(expected, result)
def test_set_masked(self):
values = np.ones([16, 16])
band = CompressedBand((16, 16), np.float32)
band[:, :] = values
indexer = BandIndexer([band])
mask = np.zeros([16, 16], dtype=np.bool)
mask[8:, 2:] = True
indexer[mask] = -1
self.assertEqual(np.sum(indexer[:, :]), 32)
def test_set_mask3(self):
values = np.ones([3, 3])
bands = [CompressedBand((3, 3), np.float32),
CompressedBand((3, 3), np.float32),
CompressedBand((3, 3), np.float32)]
bands[0].setblock(0, 0, values)
bands[1].setblock(0, 0, 2*values)
bands[2].setblock(0, 0, 3*values)
indexer = BandIndexer(bands)
# Create mask in band, row, column order than use np.moveaxis to put in
# row, column, band order for karta
mask = np.array([
[[True, False, False], [False, True, False], [False, False, True]],
[[False, False, False], [False, False, False], [False, True, False]],
[[True, False, False], [False, False, False], [True, False, True]]])
indexer[np.moveaxis(mask, 0, -1)] = -1
self.assertEqual(bands[0].getblock(0, 0, 1, 1)[0], -1)
self.assertEqual(bands[1].getblock(2, 1, 1, 1)[0], -1)
self.assertEqual(bands[2].getblock(0, 0, 1, 1)[0], -1)
self.assertEqual(bands[2].getblock(2, 0, 1, 1)[0], -1)
def test_get_slice(self):
values = np.ones([16, 16])
bands = [CompressedBand((16, 16), np.float32),
CompressedBand((16, 16), np.float32),
CompressedBand((16, 16), np.float32)]
bands[0].setblock(0, 0, values)
bands[1].setblock(0, 0, 2*values)
bands[2].setblock(0, 0, 3*values)
indexer = BandIndexer(bands)
result = indexer[4:7,2:8,:]
self.assertEqual(result.shape, (3, 6, 3))
npt.assert_equal(result[0,0,:], np.array([1.0, 2.0, 3.0]))
result = indexer[12:]
self.assertEqual(result.shape, (4, 16, 3))
npt.assert_equal(result[0,0,:], np.array([1.0, 2.0, 3.0]))
# make sure it works with a scalar band index
result = indexer[4:7,2:8,1]
self.assertEqual(result.shape, (3, 6))
npt.assert_equal(result, 2.0)