将大小为 1 的 Python 标量和 numpy 数组转换为标量

Question

I have a Python function that receives Python scalars or NumPy arrays of size 1. I would like to get the exact scalar out of this without too much complication. item() unconditionally converts to native Python types, so that's not good:

a = np.array([1])
b = np.asarray(a).item()
print(a[0], type(a[0]))
print(b, type(b))

1 <class 'numpy.int64'>
1 <class 'int'>

I also need to scratch the asarray since it converts to numpy types right away:

a = 1
b = np.asarray(a).flat[0]
print(a, type(a))
print(b, type(b))

1 <class 'int'>
1 <class 'numpy.int64'>

Any helper functions that I'm missing?

Answer 1

Based on your comment, I think you want (although this is not entirely clear from your post) to get the first item out of an array with length 1, but it should also work if the input is already an integer/float/complex number.

You can use type checking here:

def GetItem(inputobject):
   if type(inputobject) is np.ndarray:
      return inputobject[0]
   else:
      return inputobject

example:

>>> GetItem(np.array([2], dtype = np.complex128))
(2+0j)

>>> type(GetItem(np.array([2], dtype = np.complex128)))
np.complex128

>>> GetItem(42)
42

---

This won't work for many cases, like when the array is of shape (1,1,1,1)

If your objects can have between 0 and some_large_number amount of dimensions but there is still only one item, this works:

def GetItem(inputobject):
    while type(inputobject) is np.ndarray:
        inputobject = inputobject[0]
    return inputobject