!419 Tensor assign with bool Tensor

Merge pull request !419 from candanzg/tensor_assign_bool_index

mindspore/_extends/parse/resources.py

@@ -83,6 +83,7 @@ convert_object_map = {
     T.mul:          multitype_ops.mul,
     T.truediv:      multitype_ops.div,
     T.getitem:      multitype_ops.getitem,
+    T.setitem:      multitype_ops.setitem,
     T.floordiv:     multitype_ops.floordiv,
     T.mod:          multitype_ops.mod,
     T.pow:          multitype_ops.pow_,
@@ -118,7 +119,6 @@ convert_object_map = {
     T.iter:         M.ms_iter,
     T.next:         M.ms_next,
     T.hasnext:      M.hasnext,
-    T.setitem:      M.setitem,
 
     T.make_tuple:   F.make_tuple,
     T.make_dict:    F.make_dict,

mindspore/ops/composite/multitype_ops/__init__.py

@@ -23,6 +23,7 @@ from .pow_impl import pow_
 from .floordiv_impl import floordiv
 from .mod_impl import mod
 from .getitem_impl import getitem
+from .setitem_impl import setitem
 from .zeros_like_impl import zeros_like
 from .ones_like_impl import ones_like
 from .equal_impl import equal
@@ -55,6 +56,7 @@ __all__ = [
     'greater_equal',
     'negative',
     'getitem',
+    'setitem',
     'logical_and',
     'logical_or',
     'logical_not'

mindspore/ops/composite/multitype_ops/_multitype_ops_util.py

@@ -0,0 +1,45 @@
+# Copyright 2020 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.
+# ============================================================================
+
+"""constexpr util"""
+
+from ...primitive import constexpr
+
+
+@constexpr
+def is_same_type(inst, type_):
+    """
+    Check whether an object is an instance of a target type.
+
+    Inputs:
+        inst (mindspore.dtype): Inspected type.
+        type_ (mindspore.dtype): Target type.
+
+    Outputs:
+        bool, the check result.
+    """
+    return inst == type_
+
+
+@constexpr
+def error_msg(msg="", format_values=""):
+    """
+    Used to throw exception information.
+
+    Inputs:
+        msg (str): information content.
+    """
+
+    raise ValueError(msg.format(*format_values))

mindspore/ops/composite/multitype_ops/setitem_impl.py

@@ -0,0 +1,194 @@
+# Copyright 2020 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.
+# ============================================================================
+
+"""Implementation for setitem."""
+
+from ...composite import base
+from ....common import dtype as mstype
+from ... import functional as F
+from . import _multitype_ops_util as mult_util
+
+setitem = base.MultitypeFuncGraph('setitem')
+
+@setitem.register("List", "Number", "String")
+def _list_setitem_with_string(data, number_index, value):
+    """
+    Assign value to list.
+
+    Inputs:
+        data (list): Data of type lis.
+        number_index (Number): Index of data.
+        value (String): Value given.
+
+    Outputs:
+        List, type is same as the element type of data.
+    """
+    return F.list_setitem(data, number_index, value)
+
+
+@setitem.register("List", "Number", "Number")
+def _list_setitem_with_number(data, number_index, value):
+    """
+    Assign value to list.
+
+    Inputs:
+        data (list): Data of type lis.
+        number_index (Number): Index of data.
+        value (Number): Value given.
+
+    Outputs:
+        List, type is same as the element type of data.
+    """
+    return F.list_setitem(data, number_index, value)
+
+
+@setitem.register("List", "Number", "Tensor")
+def _list_setitem_with_Tensor(data, number_index, value):
+    """
+    Assign value to list.
+
+    Inputs:
+        data (list): Data of type lis.
+        number_index (Number): Index of data.
+        value (Tensor): Value given.
+
+    Outputs:
+        List, type is same as the element type of data.
+    """
+    return F.list_setitem(data, number_index, value)
+
+
+@setitem.register("List", "Number", "List")
+def _list_setitem_with_List(data, number_index, value):
+    """
+    Assign value to list.
+
+    Inputs:
+        data (list): Data of type lis.
+        number_index (Number): Index of data.
+        value (List): Value given.
+
+    Outputs:
+        List, type is same as the element type of data.
+    """
+    return F.list_setitem(data, number_index, value)
+
+
+@setitem.register("Dictionary", "String", "Tensor")
+def _dict_setitem_with_tensor(data, key, value):
+    """
+    Assign value to dictionary.
+
+    Inputs:
+        data (Dictionary): Data of type dict.
+        key (str): Key of the data.
+        value (Tensor): Value given.
+
+    Outputs:
+        Dict, type is as same as the element type of data.
+    """
+    return F.dict_setitem(data, key, value)
+
+
+@setitem.register("Dictionary", "String", "Number")
+def _dict_setitem_with_number(data, key, value):
+    """
+    Assign value to dictionary.
+
+    Inputs:
+        data (Dictionary): Data of type dict.
+        key (str): Key of the data.
+        value (Number): Value given.
+
+    Outputs:
+        Dict, type is as same as the element type of data.
+    """
+    return F.dict_setitem(data, key, value)
+
+
+@setitem.register("Tensor", "Tensor", "Tensor")
+def _tensor_setitem_by_tensor_v1(data, index, value_tensor):
+    """
+    Tensor assignment.
+
+    Note:
+        Syntax support: A[B] = U and A[A>n] = U.
+        Restraint condition: 1) A, U is a Tensor, and B is a bool Tensor.
+                             2) A.shape == B.shape
+                             3) U.size == 1
+                             4) n is a number
+
+    Inputs:
+        data (Tensor): Assigned tensor.
+        index (Tensor): Tensor of bool type.
+        value_tensor (Tensor): Tensor with size 1.
+
+    Outputs:
+        Tensor, element type and shape is same as data.
+    """
+    index_dtype = F.dtype(index)
+    index_shape = F.shape(index)
+    is_bool = mult_util.is_same_type(index_dtype, mstype.bool_)
+    if not is_bool:
+        return mult_util.error_msg(
+            "The tensor index should be a bool type tensor. {} type tensor is not supported yet.", (index_dtype,))
+    data_shape = F.shape(data)
+    if index_shape != data_shape:
+        return mult_util.error_msg(
+            "The tensor(shape={}) and tensor index(shape={}) should be the same shape.", (data_shape, index_shape))
+    size = F.size(value_tensor)
+    if size != 1:
+        return mult_util.error_msg(
+            "When assign value is a tensor, its size should be 1, but current size is {}.", (size,))
+    dtype = F.dtype(data)
+    u_cast = F.cast(value_tensor, dtype)
+    one_data = F.ones_like(data)
+    u = F.tensor_mul(one_data, u_cast)
+    return F.select(index, u, data)
+
+
+@setitem.register("Tensor", "Tensor", "Number")
+def _tensor_setitem_by_tensor_v2(data, index, value):
+    """
+    Tensor assignment.
+
+    Note:
+        Syntax support: A[B] = u and A[A>n] = u.
+        Restraint condition: 1) A is a Tensor, and B is a bool Tensor.
+                             2) A.shape == B.shape
+                             3) u is a scalar
+                             4) n is a number
+
+    Inputs:
+        data (Tensor): Assigned tensor.
+        index (Tensor): Tensor of bool type.
+        value_tensor (Number): Assignment value.
+
+    Outputs:
+        Tensor, element type and shape is same as data.
+    """
+    index_dtype = F.dtype(index)
+    index_shape = F.shape(index)
+    is_bool = mult_util.is_same_type(index_dtype, mstype.bool_)
+    if not is_bool:
+        return mult_util.error_msg(
+            "The tensor index should be a bool type tensor. {} type tensor is not supported yet.", (index_dtype,))
+    shape = F.shape(data)
+    if index_shape != shape:
+        return mult_util.error_msg(
+            "The tensor(shape={}) and tensor index(shape={}) should be the same shape.", (shape, index_shape))
+    dtype = F.dtype(data)
+    u = F.fill(dtype, shape, value)
+    return F.select(index, u, data)

mindspore/ops/functional.py

@@ -31,6 +31,9 @@ dtype = P.DType()
 issubclass_ = P.IsSubClass()
 isinstance_ = P.IsInstance()
 fill = P.Fill()
+select = P.Select()
+size = P.Size()
+ones_like = P.OnesLike()
 shape = P.Shape()
 rank = P.Rank()
 reshape = P.Reshape()
@@ -68,7 +71,9 @@ scalar_cast = P.ScalarCast()
 tuple_setitem = Primitive('tuple_setitem')
 tuple_getitem = Primitive('tuple_getitem')
 list_getitem = Primitive('list_getitem')
+list_setitem = Primitive('list_setitem')
 dict_getitem = Primitive('dict_getitem')
+dict_setitem = Primitive('dict_setitem')
 tuple_div = Primitive("tuple_div")
 tuple_len = Primitive("tuple_len")
 tuple_reversed = Primitive("tuple_reversed")

tests/ut/python/ops/test_tensor_slice.py

@@ -18,6 +18,7 @@ import pytest
 
 from mindspore import Tensor
 from mindspore import context
+from mindspore import dtype as mstype
 from mindspore.nn import Cell
 
 from ....mindspore_test_framework.mindspore_test import mindspore_test
@@ -79,7 +80,102 @@ class NetWorkReduceToScalar(Cell):
         return ret
 
 
+class TensorAssignWithBoolTensorIndex(Cell):
+    def __init__(self):
+        super(TensorAssignWithBoolTensorIndex, self).__init__()
+        self.t = Tensor(np.arange(6).reshape([2,3]), dtype = mstype.float64)
+
+    def construct(self, a, b, c, u_tensor, _scalar):
+        a[c] = u_scalar
+        a[b] = u_tensor
+        z = a + self.t
+        return z
+
+
+class TensorAssignWithBoolTensorIndexError(Cell):
+    def __init__(self):
+        super(TensorAssignWithBoolTensorIndexError, self).__init__()
+
+    def construct(self, a, b, c, u_tensor):
+        a[b][c] = u_tensor
+        return a
+
+
+class TensorAssignWithBoolTensorIndex2(Cell):
+    def __init__(self):
+        super(TensorAssignWithBoolTensorIndex2, self).__init__()
+        self.t = Tensor(np.arange(6).reshape([2,3]), dtype = mstype.float64)
+
+    def construct(self, a, u_tensor, _scalar):
+        a[a>8] = u_tensor
+        a[a>=6] = u_scalar
+        a[a<3] = u_scalar
+        a[a<=5] = u_tensor
+        a[a==5] = u_scalar
+        z = a + self.t
+        return z
+
+
+class TensorAssignWithBoolTensorIndex2Error(Cell):
+    def __init__(self):
+        super(TensorAssignWithBoolTensorIndex2Error, self).__init__()
+
+    def construct(self, a, u_tensor):
+        a[a>8][a>5] = u_tensor
+        return a
+
+
+a = np.random.uniform(1,10,[2,3])
+b = a > 5
+c = a < 3
+Ta = Tensor(a)
+Tb = Tensor(b)
+Tc = Tensor(c)
+Td = Tensor([True, True])
+u_tensor = Tensor([1])
+u_tensor_error = Tensor([1, 2])
+u_scalar = 5
+
+
+def test_tensor_assign_bool_index():
+    net1 = TensorAssignWithBoolTensorIndex()
+    net2 = TensorAssignWithBoolTensorIndex2()
+
+    net1(Ta, Tb, Tc, u_tensor, u_scalar)
+    with pytest.raises(ValueError):
+        net1(Ta, Td, Tc, u_tensor, u_scalar)
+    with pytest.raises(ValueError):
+        net1(Ta, u_tensor, Tc, u_tensor, u_scalar)
+    with pytest.raises(ValueError):
+        net1(Ta, Tb, Td, u_tensor, u_scalar)
+    with pytest.raises(ValueError):
+        net1(Ta, Tb, Ta, u_tensor, u_scalar)
+    with pytest.raises(ValueError):
+        net1(Ta, Tb, Tc, u_tensor_error, u_scalar)
+    #net1(Ta, u_tensor, Tc, u_tensor_error, u_scalar)
+    with pytest.raises(ValueError):
+        net2(Ta, u_tensor_error, u_scalar)
+    net3 = TensorAssignWithBoolTensorIndexError()
+    with pytest.raises(AttributeError):
+        net3(Ta, Tb, Tc, u_tensor)
+    with pytest.raises(AttributeError):
+        net3(Ta, Tb, Tc, u_scalar)
+    net4 = TensorAssignWithBoolTensorIndex2Error()
+    with pytest.raises(AttributeError):
+        net4(Ta, u_tensor)
+    with pytest.raises(AttributeError):
+        net4(Ta, u_scalar)
+
+
 test_cases = [
+    ('TensorAssignWithBoolTensorIndex', {
+        'block': TensorAssignWithBoolTensorIndex(),
+        'desc_inputs': [Ta, Tb, Tc, u_tensor, u_scalar],
+    }),
+    ('TensorAssignWithBoolTensorIndex2', {
+        'block': TensorAssignWithBoolTensorIndex2(),
+        'desc_inputs': [Ta, u_tensor, u_scalar],
+    }),
     ('SlicePositive', {
         'block': NetWorkSlicePositive(),
         'desc_inputs': [Tensor(np.ones([6, 8, 10], np.int32))],