Better way to initialize multidimensional array with mutable object

Question

I want to create a multidimensional array and initialize it with copies of a mutable object. This is what I have so far:

import copy

def create_array(dimensions):
    dimensions = copy.deepcopy(dimensions)
    dimensions.reverse()
    a = [0] * dimensions[0]
    del dimensions[0]
    for d in dimensions:
        a = [copy.deepcopy(a) for _ in range(d)]
    return a

def create_array_mutable(dimensions, obj):
    a = create_array(dimensions)
    def set(x):
        if isinstance(x[0], list):
            for e in x:
                set(e)
        else:
            for i in range(len(x)):
                x[i] = copy.deepcopy(obj)
    set(a)
    return a

I wonder if there is a better way to do it (without the copies and the recursion)?

if you can use numpy then the first answer here seems to be close to what you are looking for - http://stackoverflow.com/questions/4877624/numpy-array-of-objects (although that is calling the constructor rather than copy...) — andrew cooke, Aug 12 '11 at 13:04

score 2 · Accepted Answer · answered Sep 19 '11 at 13:59

How about:

import copy

def create_array_mutable(dims, obj):
  if len(dims) == 0:
    return copy.deepcopy(obj)
  else:
    return [create_array_mutable(dims[1:], obj) for i in xrange(dims[0])]

class C(object): pass

print create_array_mutable((2,3,4), C())

This creates a 2x3x4 array of unique instances of C.

The solution is still recursive, but I think recursion is a pretty good fit for this problem.

Better way to initialize multidimensional array with mutable object

1 Answers1