Python 创建自定义对象的 numpy 数组会出现错误“SystemError：错误返回而未设置异常”
声明：本页面是StackOverFlow热门问题的中英对照翻译，遵循CC BY-SA 4.0协议，如果您需要使用它，必须同样遵循CC BY-SA许可，注明原文地址和作者信息，同时你必须将它归于原作者（不是我）：StackOverFlow
原文地址: http://stackoverflow.com/questions/20482837/
Warning: these are provided under cc-by-sa 4.0 license. You are free to use/share it, But you must attribute it to the original authors (not me):
StackOverFlow
Creating numpy array of custom objects gives error "SystemError: error return without exception set"
提问by Dannnno
I am trying to use numpy to store some custom objects I've made. The following is a simplified version of my program
我正在尝试使用 numpy 来存储我制作的一些自定义对象。以下是我的程序的简化版
import numpy as np
class Element:
def __init__(self): pass
a = Element()
periodicTable = np.array(range(7*32)).reshape((7,32))
periodicTable[0][0] = a
However when I run this I get
但是，当我运行它时，我得到
Traceback (most recent call last):
File "C:/Users/Dan/Desktop/a.py", line 9, in <module>
periodicTable[0][0] = a
SystemError: error return without exception set
I'm not really sure what I'm doing wrong - as far as I can tell everything I've done should be legal. The cryptic error message itself isn't very helpful - I believe that is a numpy issue however I've been unable to identify my problem.
我不确定我做错了什么——据我所知，我所做的一切都应该是合法的。神秘的错误消息本身并不是很有帮助 - 我相信这是一个很棘手的问题，但是我一直无法确定我的问题。
采纳答案by Warren Weckesser
@user2357112 identified the problem: you are assigning an Elementinstance to a numpy array that holds integers. This is what I get when I try something similar:
@user2357112 发现了问题：您正在将一个Element实例分配给一个包含整数的 numpy 数组。当我尝试类似的东西时，这就是我得到的：
>>> import numpy as np
>>> np.__version__
'1.7.1'
>>> p = np.array([1,2,3])
>>> class Foo:
... pass
...
>>> p[0] = Foo()
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
SystemError: error return without exception set
>>>
It is not surprising that this is not allowed. The cryptic error message, however, is almost certainly a numpy bug.
这是不允许的，这并不奇怪。然而，这个神秘的错误信息几乎可以肯定是一个 numpy 错误。
One way to fix the issue is to use an array of type object. Change this line:
解决此问题的一种方法是使用类型为 的数组object。改变这一行：
periodicTable = np.array(range(7*32)).reshape((7,32))
to this:
对此：
periodicTable = np.empty((7,32), dtype=object)
Update
更新
In numpy 1.10.1, the error message is still a bit cryptic:
在 numpy 1.10.1 中，错误信息仍然有点神秘：
>>> import numpy as np
>>> np.__version__
'1.10.1'
>>> p = np.array([1, 2, 3])
>>> class Foo:
... pass
...
>>> p[0] = Foo()
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
AttributeError: Foo instance has no attribute '__trunc__'
Update 2
更新 2
The error message is better is later versions of numpy:
错误信息更好的是 numpy 的更高版本：
In [1]: import numpy as np
In [2]: np.__version__
Out[2]: '1.12.1'
In [3]: class Foo:
...: pass
...:
In [4]: p = np.array([1, 2, 3])
In [5]: p[0] = Foo()
---------------------------------------------------------------------------
TypeError Traceback (most recent call last)
<ipython-input-5-739d5e5f795b> in <module>()
----> 1 p[0] = Foo()
TypeError: int() argument must be a string, a bytes-like object or a number, not 'Foo'
