python - 这是创建 numpy 数组的只读 View 的正确方法吗？

Question

我想创建一个对 NumPy 数组的只读引用。这是制作 b 的正确方法吗？对 a 的只读引用( a 是任何 NumPy 数组)？

def get_readonly_view(a):
    b = a.view()
    b.flags.writeable = False
    return b

具体来说，我想确保上述内容不会“复制” a 的内容。 ? (我尝试用 np.shares_memory 测试它并返回 True 。但我不确定这是否是正确的测试。)

另外我想知道是否 get_readonly_view已经在 NumPy 中实现了吗？

更新。 建议将数组转换为类属性以使其只读。我认为这不起作用:

import numpy as np

class Foo:

    def __init__(self):
        self._a = np.arange(15).reshape((3, 5))


    @property
    def a(self):
        return self._a


    def bar(self):
        print(self._a)

但是客户端可以更改 _a的内容:

>> baz = Foo()
>> baz.bar()
[[ 0  1  2  3  4]
 [ 5  6  7  8  9]
 [10 11 12 13 14]]
>> baz.a[1, 2] = 10
>> baz.bar()
[[ 0  1  2  3  4]
 [ 5  6 10  8  9]
 [10 11 12 13 14]]

而我想要 baz.a[1, 2] = 10引发异常。

Answer 1

您的方法似乎是创建 read-only 的建议方法。 view .

特别是arr.view() (也可以写成切片 arr[:] )将创建对 arr 的引用, 同时修改 writeable flag 是使 NumPy 数组只读的建议方法。

该文档还提供了一些关于 writeable 继承的附加信息。属性(property):

The data area can be written to. Setting this to False locks the data, making it read-only. A view (slice, etc.) inherits WRITEABLE from its base array at creation time, but a view of a writeable array may be subsequently locked while the base array remains writeable. (The opposite is not true, in that a view of a locked array may not be made writeable. However, currently, locking a base object does not lock any views that already reference it, so under that circumstance it is possible to alter the contents of a locked array via a previously created writeable view onto it.) Attempting to change a non-writeable array raises a RuntimeError exception.

只是重申并检查正在发生的事情:

import numpy as np


def get_readonly_view(arr):
    result = arr.view()
    result.flags.writeable = False
    return result


a = np.zeros((2, 2))
b = get_readonly_view(a)

print(a.flags)
#   C_CONTIGUOUS : True
#   F_CONTIGUOUS : False
#   OWNDATA : True
#   WRITEABLE : True
#   ALIGNED : True
#   WRITEBACKIFCOPY : False
#   UPDATEIFCOPY : False
print(b.flags)
#   C_CONTIGUOUS : True
#   F_CONTIGUOUS : False
#   OWNDATA : False
#   WRITEABLE : False
#   ALIGNED : True
#   WRITEBACKIFCOPY : False
#   UPDATEIFCOPY : False

print(a.base)
# None
print(b.base)
# [[0. 0.]
#  [0. 0.]]

a[1, 1] = 1.0
# ...works
b[0, 0] = 1.0
# raises  ValueError: assignment destination is read-only