!13745 Enable zero dimension when it comes out of operator.

From: @liangzhibo
Reviewed-by: @ginfung,@zh_qh,@ginfung
Signed-off-by: @zh_qh

mindspore/common/tensor.py

@@ -74,7 +74,7 @@ class Tensor(Tensor_):
         >>> assert t3.dtype == ms.float32
     """
 
-    def __init__(self, input_data=None, dtype=None, shape=None, init=None):
+    def __init__(self, input_data=None, dtype=None, shape=None, init=None, check_zero_dims=True):
         self.init_finished = False
         # If input data is numpy number, convert it to np array
         if isinstance(input_data, np_types):
@@ -92,6 +92,14 @@ class Tensor(Tensor_):
         if isinstance(shape, numbers.Number):
             shape = (shape,)
 
+        if check_zero_dims:
+            if input_data is not None and isinstance(input_data, (tuple, list, np.ndarray)) \
+                    and np.array(input_data).ndim > 1 and np.array(input_data).size == 0:
+                raise ValueError("input_data can not contain zero dimension.")
+            if shape is not None:
+                if 0 in shape:
+                    raise ValueError("Shape can not contain zero value.")
+
         # If input_data is tuple/list/numpy.ndarray, it's support in check_type method.
         if init is None:
             validator.check_value_type('input_data', input_data, (Tensor_, np.ndarray, list, tuple, float, int, bool),

mindspore/core/ir/tensor.cc

@@ -465,9 +465,7 @@ Tensor::Tensor(const Tensor &tensor)
       cache_tensor_ptr_(tensor.cache_tensor_ptr_),
       hashmap_tensor_ptr_(tensor.hashmap_tensor_ptr_),
       padding_type_(tensor.padding_type()),
-      device_event_(tensor.device_event_) {
-  CheckShape(tensor.shape_);
-}
+      device_event_(tensor.device_event_) {}
 
 Tensor::Tensor(const Tensor &tensor, TypeId data_type)
     : MetaTensor(data_type, tensor.shape_),
@@ -481,43 +479,29 @@ Tensor::Tensor(const Tensor &tensor, TypeId data_type)
       cache_tensor_ptr_(tensor.cache_tensor_ptr_),
       hashmap_tensor_ptr_(tensor.hashmap_tensor_ptr_),
       padding_type_(tensor.padding_type()),
-      device_event_(tensor.device_event_) {
-  CheckShape(tensor.shape_);
-}
+      device_event_(tensor.device_event_) {}
 
 Tensor::Tensor(TypeId data_type, const ShapeVector &shape, TensorDataPtr data)
-    : MetaTensor(data_type, shape), data_(std::move(data)), id_(MakeId()) {
-  CheckShape(shape);
-}
+    : MetaTensor(data_type, shape), data_(std::move(data)), id_(MakeId()) {}
 
 Tensor::Tensor(TypeId data_type, const ShapeVector &shape)
-    : Tensor(data_type, shape, MakeTensorData(data_type, shape)) {
-  CheckShape(shape);
-}
+    : Tensor(data_type, shape, MakeTensorData(data_type, shape)) {}
 
 Tensor::Tensor(TypeId data_type, const ShapeVector &shape, void *data, size_t data_len)
-    : Tensor(data_type, shape, MakeTensorData(data_type, shape, data, data_len)) {
-  CheckShape(shape);
-}
+    : Tensor(data_type, shape, MakeTensorData(data_type, shape, data, data_len)) {}
 
 Tensor::Tensor(TypeId data_type, const ShapeVector &shape, void *data, TypeId src_data_type)
-    : Tensor(data_type, shape, MakeTensorData(data_type, shape, data, src_data_type)) {
-  CheckShape(shape);
-}
+    : Tensor(data_type, shape, MakeTensorData(data_type, shape, data, src_data_type)) {}
 
 Tensor::Tensor(const std::vector<int64_t> &input, const TypePtr &data_type)
     : MetaTensor(TypeIdOf(data_type, kNumberTypeInt32), {static_cast<int>(input.size())}),
       data_(MakeTensorData(data_type_, shape_, input.data(), input.size())),
-      id_(MakeId()) {
-  CheckShape(shape_);
-}
+      id_(MakeId()) {}
 
 Tensor::Tensor(const std::vector<double> &input, const TypePtr &data_type)
     : MetaTensor(TypeIdOf(data_type, kNumberTypeFloat32), {static_cast<int>(input.size())}),
       data_(MakeTensorData(data_type_, shape_, input.data(), input.size())),
-      id_(MakeId()) {
-  CheckShape(shape_);
-}
+      id_(MakeId()) {}
 
 Tensor::Tensor(int64_t input, const TypePtr &data_type)
     : MetaTensor(TypeIdOf(data_type, kNumberTypeInt32), {}),
@@ -622,17 +606,6 @@ std::string Tensor::ToStringRepr() const {
   return buf.str();
 }
 
-void Tensor::CheckShape(const ShapeVector &shape) const {
-  // Check tensor's shape, ignore one-dimensional tensor, including empty tensor with shape=(0,).
-  if (shape.size() > 1) {
-    for (const auto &s : shape) {
-      if (s == 0) {
-        MS_EXCEPTION(ValueError) << "Zero is not supported in the shape of Tensor. ";
-      }
-    }
-  }
-}
-
 void Tensor::data_sync(bool need_wait) const {
   if (need_wait) {
     Wait();

mindspore/ops/operations/array_ops.py

@@ -2980,7 +2980,8 @@ class StridedSlice(PrimitiveWithInfer):
 
         ret_shape = self._compute_slicing_shape(x['shape'], begin_v, end_v, strides_v)
 
-        value = None if all(ret_shape) else Tensor(np.array([]).reshape(ret_shape), x['dtype'].element_type())
+        value = None if all(ret_shape) else Tensor(np.array([]).reshape(ret_shape), x['dtype'].element_type(),
+                                                   check_zero_dims=False)
         if "max_value" in x and "min_value" in x:
             validator.check_value_type("min_value", x["min_value"], [tuple, list], self.name)
             validator.check_value_type("max_value", x["max_value"], [tuple, list], self.name)

tests/ut/python/ir/test_tensor.py

@@ -67,17 +67,6 @@ def test_tensor():
     assert isinstance(t4, ms.Tensor)
     assert t4.dtype == ms.int64
 
-def test_tensor_empty():
-    t = ms.Tensor(np.ones(0), ms.float32)
-    assert isinstance(t, ms.Tensor)
-    assert t.shape == (0,)
-
-
-def test_tensor_shape_has_zero():
-    with pytest.raises(ValueError):
-        t = ms.Tensor(np.ones((1, 0)), ms.float32)
-        print(t)
-
 
 def test_tensor_type_float16():
     t_float16 = ms.Tensor(np.array([[1, 2, 3], [4, 5, 6]], dtype=np.float16))

tests/ut/python/pipeline/parse/test_tensor_zero_dimension.py

@@ -0,0 +1,59 @@
+# Copyright 2021 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ============================================================================
+
+import numpy as np
+import pytest
+import mindspore
+
+import mindspore.nn as nn
+from mindspore import Tensor
+from mindspore import context
+from mindspore.ops import operations as P
+from mindspore.common.initializer import One
+
+context.set_context(mode=context.GRAPH_MODE)
+
+def test_zero_dimension_list():
+    Tensor([])
+    with pytest.raises(ValueError) as ex:
+        Tensor([[]])
+    assert "input_data can not contain zero dimension." in str(ex.value)
+
+
+def test_zero_dimension_np_array():
+    with pytest.raises(ValueError) as ex:
+        Tensor(np.ones((1, 0, 3)))
+    assert "input_data can not contain zero dimension." in str(ex.value)
+
+
+def test_zero_dimension_with_zero_shape():
+    with pytest.raises(ValueError) as ex:
+        Tensor(shape=(1, 0, 3), dtype=mindspore.float32, init=One())
+    assert "Shape can not contain zero value." in str(ex.value)
+
+
+def test_zero_dimension_with_operator():
+    class Net(nn.Cell):
+        def __init__(self):
+            super(Net, self).__init__()
+            self.strided_slice = P.StridedSlice()
+
+        def construct(self, x):
+            a = self.strided_slice(x, (2, 4, 4), (-1, 2, 1), (1, 1, 1))
+            return a
+
+    x = Tensor(np.ones((1, 3, 3)))
+    net = Net()
+    net(x)