在函数内的函数内创建和返回局部变量不起作用 numpy

Question

我正在尝试为一件作品匹配数组尺寸。我正在尝试在函数内创建并返回一个变量，但收到以下错误：

UnboundLocalError: local variable 'first' referenced before assignment

我不明白这一点，因为第一次遇到这个词实际上是我在定义它是什么。

我不能使用global，因为它是函数中的函数。我的代码如下，应该是可重现的。

a= np.array([(1, 2, 3), (4, 5, 6), (7, 8, 9), (10,11,12)])
b = np.array([(10, 20, 30), (40, 50, 60)])
def broadcast(a, b):
    def match_lengths(a_ar, b_ar):
        a_shape = list(np.shape(a_ar))
        b_shape = list(np.shape(b_ar))
        if len(a_shape) == len(b_shape):
            print('Already same dimensions')
            pass
        else:
            if len(a_shape) > len(b_shape):
                extra_dims = len(a_shape) - len(b_shape)
                smaller_arr = b_shape
            else:
                extra_dims = len(b_shape) - len(a_shape)
                smaller_arr = a_shape
            dim = (1,)
            add_dims = dim * extra_dims
            shapes = add_dims + tuple(smaller_arr)


            if smaller_arr == a_shape:
                first = np.reshape(a_ar, shapes)
            else:
                second = np.reshape(b_ar, shapes)
        return first, second

     match_lengths(a, b)
     a_shape = list(np.shape(first))
     b_shape = list(np.shape(second)) 
     return a_shape, b_shape

有人知道为什么会这样吗？

Answer 1

为什么不使用函数返回的值？

first, second = match_lengths(a, b)

这是正常的函数用法。如果函数返回一些值，则将它们分配给变量。

Answer 2

您可以尝试以下方法吗，因为return first, second 正在尝试返回一个未初始化的变量。

a= np.array([(1, 2, 3), (4, 5, 6), (7, 8, 9), (10,11,12)])
b = np.array([(10, 20, 30), (40, 50, 60)])
def broadcast(fun, a, b):

    def match_lengths(a_ar, b_ar):
        first, second = np.array([]), np.array([])
        a_shape = list(np.shape(a_ar))
        b_shape = list(np.shape(b_ar))
        if len(a_shape) == len(b_shape):
            print('Already same dimensions')
            pass
        else:
            if len(a_shape) > len(b_shape):
                extra_dims = len(a_shape) - len(b_shape)
                smaller_arr = b_shape
            else:
                extra_dims = len(b_shape) - len(a_shape)
                smaller_arr = a_shape
            dim = (1,)
            add_dims = dim * extra_dims
            shapes = add_dims + tuple(smaller_arr)


            if smaller_arr == a_shape:
                first = np.reshape(a_ar, shapes)
            else:
                second = np.reshape(b_ar, shapes)
        return first, second

     first, second = match_lengths(a, b)
     a_shape = list(np.shape(first))
     b_shape = list(np.shape(second)) 
     return a_shape, b_shape