有效地将平面列表切成多层嵌套列表

Question

For example, I have a flat list

 [1, 2, 3, 4, 5, 6, 7, 8, 9, 'A', 'B', 'C', 'D', 'E', 'F', 'G']

I want to transform it into 4-deep list

[[[[1, 2], [3, 4]], [[5, 6], [7, 8]]], [[[9, 'A'], ['B', 'C']], [['D', 'E'] ['F', 'G']]]]

Is there a way to do it without creating a separate variable for every level? What is the most memory- and performance-efficient way?

UPDATE: Also, is there a way to do it in a non-symmetrical fashion?

[[[[1, 2, 3], 4], [[5, 6, 7], 8]]], [[[9, 'A', 'B'], 'C']], [['D', 'E', 'F'], 'G']]]]

Answer 1

Note that your first list has 15 elements instead of 16. Also, what should A be? Is it a constant you've defined somewhere else? I'll just assume it's a string : 'A' .

If you work with np.arrays , you could simply reshape your array:

import numpy as np
r = np.array([1, 2, 3, 4, 5, 6, 7, 8, 9, 'A', 'B', 'C', 'D', 'E', 'F', 'G'])
r.reshape(2,2,2,2)

It outputs:

array([[[['1', '2'],
         ['3', '4']],

        [['5', '6'],
         ['7', '8']]]


       [[['9', 'A'],
         ['B', 'C']],

        [['D', 'E'],
         ['F', 'G']]]
      dtype='<U11')

This should be really efficient because numpy doesn't change the underlying data format. It's still a flat array, displayed differently.

Numpy doesn't support irregular shapes. You'll have to work with standard python lists then:

i = iter([1, 2, 3, 4, 5, 6, 7, 8, 9, 'A', 'B', 'C', 'D', 'E', 'F', 'G'])

l1 = []

for _ in range(2):
    l2 = []
    for _ in range(2):
        l3 = []
        l4 = []
        for _ in range(3):
            l4.append(next(i))
        l3.append(l4)
        l3.append(next(i))
        l2.append(l3)
    l1.append(l2)

print(l1)
# [[[[1, 2, 3], 4], [[5, 6, 7], 8]], [[[9, 'A', 'B'], 'C'], [['D', 'E', 'F'], 'G']]]

As you said, you'll have to define a temporary variable for each level. I guess you could use list comprehensions, but they wouldn't be pretty.