Runtime slicing

Question

can someone explain me how to chop numpy.array at runtime? I do not know the rank (number of measurements) at the "coding time".

Minimal example:

import numpy as np
a = np.arange(16).reshape(4,4) # 2D matrix
targetsize = [2,3] # desired shape

b_correct = dynSlicing(a, targetsize)
b_wrong = np.resize(a, targetsize)

print a
[[ 0  1  2  3]
 [ 4  5  6  7]
 [ 8  9 10 11]
 [12 13 14 15]]
print b_correct
[[0 1 2]
 [4 5 6]]
print b_wrong
[[0 1 2]
 [3 4 5]]

And my ugly dynSlicing ():

def dynSlicing(data, targetsize):
    ndims = len(targetsize)

    if(ndims==1):
        return data[:targetsize[0]],
    elif(ndims==2):
        return data[:targetsize[0], :targetsize[1]]
    elif(ndims==3):
        return data[:targetsize[0], :targetsize[1], :targetsize[2]]
    elif(ndims==4):
        return data[:targetsize[0], :targetsize[1], :targetsize[2], :targetsize[3]]

Resize () will not do the job since it wraps the array before deleting items.

Thanks Tebas

Answer 1

Transferring a tuple of slicer objects performs the task:

def dynSlicing(data, targetsize):
    return data[tuple(slice(x) for x in targetsize)]

Answer 2

A simple solution:

b = a[tuple(map(slice,targetsize))]

Answer 3

"" . , .

: http://www.scipy.org/Tentative_NumPy_Tutorial#head-1529ae93dd5d431ffe3a1001a4ab1a394e70a5f2