moveaxis operator now accepts negative indices and sequence of ints a…

…s well. (apache#14321)

* nd.moveaxis now accepts negative indices and sequence of ints as well.

* Retrigger CI

* retrigger CI

* Retriever

* Retrigger CI

* Retrigger CI

* Retrigger CI, hope for success

* Update test_ndarray.py

* Let’s try to retrigger CI again.

* Retrigger CI

* Disable the new testcase to track bugs

* enable some new testcases

* Update test_ndarray.py

* Update test_ndarray.py

python/mxnet/ndarray/ndarray.py

@@ -2513,10 +2513,10 @@ def moveaxis(tensor, source, destination):
     ----------
     tensor : mx.nd.array
         The array which axes should be reordered
-    source : int
-        Original position of the axes to move.
-    destination : int
-        Destination position for each of the original axes.
+    source : int or sequence of int
+        Original position of the axes to move. Can be negative but must be unique.
+    destination : int or sequence of int
+        Destination position for each of the original axes. Can be negative but must be unique.
 
     Returns
     -------
@@ -2528,19 +2528,32 @@ def moveaxis(tensor, source, destination):
     >>> X = mx.nd.array([[1, 2, 3], [4, 5, 6]])
     >>> mx.nd.moveaxis(X, 0, 1).shape
     (3L, 2L)
+
+    >>> X = mx.nd.zeros((3, 4, 5))
+    >>> mx.nd.moveaxis(X, [0, 1], [-1, -2]).shape
+    (5, 4, 3)
     """
-    axes = list(range(tensor.ndim))
     try:
-        axes.pop(source)
+        source = np.core.numeric.normalize_axis_tuple(source, tensor.ndim)
     except IndexError:
         raise ValueError('Source should verify 0 <= source < tensor.ndim'
                          'Got %d' % source)
     try:
-        axes.insert(destination, source)
+        destination = np.core.numeric.normalize_axis_tuple(destination, tensor.ndim)
     except IndexError:
-        raise ValueError('Destination should verify 0 <= destination < tensor.ndim'
-                         'Got %d' % destination)
-    return op.transpose(tensor, axes)
+        raise ValueError('Destination should verify 0 <= destination < tensor.ndim (%d).'
+                         % tensor.ndim, 'Got %d' % destination)
+
+    if len(source) != len(destination):
+        raise ValueError('`source` and `destination` arguments must have '
+                         'the same number of elements')
+
+    order = [n for n in range(tensor.ndim) if n not in source]
+
+    for dest, src in sorted(zip(destination, source)):
+        order.insert(dest, src)
+
+    return op.transpose(tensor, order)
 
 
 # pylint: disable= no-member, protected-access, too-many-arguments, redefined-outer-name

tests/python/unittest/test_ndarray.py

@@ -32,6 +32,7 @@
 from numpy.testing import assert_allclose
 import mxnet.autograd
 
+
 def check_with_uniform(uf, arg_shapes, dim=None, npuf=None, rmin=-10, type_list=[np.float32]):
     """check function consistency with uniform random numbers"""
     if isinstance(arg_shapes, int):
@@ -60,6 +61,7 @@ def check_with_uniform(uf, arg_shapes, dim=None, npuf=None, rmin=-10, type_list=
         else:
             assert_almost_equal(out1, out2, atol=1e-5)
 
+
 def random_ndarray(dim):
     shape = tuple(np.random.randint(1, int(1000**(1.0/dim)), size=dim))
     data = mx.nd.array(np.random.uniform(-10, 10, shape))
@@ -144,12 +146,14 @@ def test_ndarray_elementwise():
             check_with_uniform(mx.nd.square, 1, dim, np.square, rmin=0)
             check_with_uniform(lambda x: mx.nd.norm(x).asscalar(), 1, dim, np.linalg.norm)
 
+
 @with_seed()
 def test_ndarray_elementwisesum():
     ones = mx.nd.ones((10,), dtype=np.int32)
     res = mx.nd.ElementWiseSum(ones, ones*2, ones*4, ones*8)
     assert same(res.asnumpy(), ones.asnumpy()*15)
 
+
 @with_seed()
 def test_ndarray_negate():
     npy = np.random.uniform(-10, 10, (2,3,4))
@@ -162,6 +166,7 @@ def test_ndarray_negate():
     # we compute (-arr)
     assert_almost_equal(npy, arr.asnumpy())
 
+
 @with_seed()
 def test_ndarray_reshape():
     tensor = (mx.nd.arange(30) + 1).reshape(2, 3, 5)
@@ -360,6 +365,7 @@ def test_buffer_load():
                 # test garbage values
                 assertRaises(mx.base.MXNetError,  mx.nd.load_frombuffer, buf_single_ndarray[:-10])
 
+
 @with_seed()
 def test_ndarray_slice():
     shape = (10,)
@@ -391,6 +397,7 @@ def test_ndarray_slice():
         assert same(A[:, i].asnumpy(), A2[:, i])
         assert same(A[i, :].asnumpy(), A2[i, :])
 
+
 @with_seed()
 def test_ndarray_crop():
     # get crop
@@ -524,6 +531,7 @@ def test_reduce_inner(numpy_reduce_func, nd_reduce_func, multi_axes):
                              keepdims:np_reduce(np.float32(data), axis, keepdims, np.argmin),
                       mx.nd.argmin, False)
 
+
 @with_seed()
 def test_broadcast():
     sample_num = 1000
@@ -626,7 +634,7 @@ def check_broadcast_binary(fn):
 def test_moveaxis():
     X = mx.nd.array([[[1, 2, 3], [4, 5, 6]],
                      [[7, 8, 9], [10, 11, 12]]])
-    res = mx.nd.moveaxis(X, 0, 3).asnumpy()
+    res = mx.nd.moveaxis(X, 0, 2).asnumpy()
     true_res = mx.nd.array([[[  1.,   7.],
                              [  2.,   8.],
                              [  3.,   9.]],
@@ -636,6 +644,66 @@ def test_moveaxis():
     assert same(res, true_res.asnumpy())
     assert mx.nd.moveaxis(X, 2, 0).shape == (3, 2, 2)
 
+    def test_move_to_end():
+        x = mx.nd.random.normal(0, 1, (5, 6, 7))
+        for source, expected in [(0, (6, 7, 5)),
+                                 (1, (5, 7, 6)),
+                                 (2, (5, 6, 7)),
+                                 (-1, (5, 6, 7))]:
+            actual = mx.nd.moveaxis(x, source, -1).shape
+            assert actual == expected
+
+    def test_move_new_position():
+        x = mx.nd.random.normal(0, 1, (1, 2, 3, 4))
+        for source, destination, expected in [
+            (0, 1, (2, 1, 3, 4)),
+            (1, 2, (1, 3, 2, 4)),
+            (1, -1, (1, 3, 4, 2)),
+        ]:
+            actual = mx.nd.moveaxis(x, source, destination).shape
+            assert actual == expected
+
+    def test_preserve_order():
+        x = mx.nd.zeros((1, 2, 3, 4))
+        for source, destination in [
+            (0, 0),
+            (3, -1),
+            (-1, 3),
+            ([0, -1], [0, -1]),
+            ([2, 0], [2, 0]),
+            (range(4), range(4)),
+        ]:
+            actual = mx.nd.moveaxis(x, source, destination).shape
+            assert actual == (1, 2, 3, 4)
+
+    def test_move_multiples():
+        x = mx.nd.zeros((4, 1, 2, 3))
+        for source, destination, expected in [
+            ([0, 1], [2, 3], (2, 3, 4, 1)),
+            ([2, 3], [0, 1], (2, 3, 4, 1)),
+            ([0, 1, 2], [2, 3, 0], (2, 3, 4, 1)),
+            ([3, 0], [1, 0], (4, 3, 1, 2)),
+            ([0, 3], [0, 1], (4, 3, 1, 2)),
+        ]:
+            actual = mx.nd.moveaxis(x, source, destination).shape
+            assert actual == expected
+
+    def test_errors():
+        x = mx.nd.random.normal(0, 1, (1, 2, 3))
+        assert_exception(mx.nd.moveaxis, ValueError, x, 3, 0)
+        assert_exception(mx.nd.moveaxis, ValueError, x, -4, 0)
+        assert_exception(mx.nd.moveaxis, ValueError, x, 0, 5)
+        assert_exception(mx.nd.moveaxis, ValueError, x, [0, 0], [0, 1])
+        assert_exception(mx.nd.moveaxis, ValueError, x, [0, 1], [1, 1])
+        assert_exception(mx.nd.moveaxis, ValueError, x, 0, [0, 1])
+        assert_exception(mx.nd.moveaxis, ValueError, x, [0, 1], [0])
+
+    test_move_to_end()
+    test_move_new_position()
+    test_preserve_order()
+    test_move_multiples()
+    test_errors()
+
 
 @with_seed()
 def test_arange():
@@ -653,6 +721,7 @@ def test_arange():
                         dtype="int32").asnumpy()
     assert_almost_equal(pred, gt)
 
+
 @with_seed()
 def test_order():
     ctx = default_context()
@@ -885,6 +954,7 @@ def get_large_matrix():
                      k=dat_size*dat_size*dat_size*dat_size, is_ascend=False)
         assert_almost_equal(nd_ret_sort, gt)
 
+
 @with_seed()
 def test_ndarray_equal():
     x = mx.nd.zeros((2, 3))