python: change numpy.array access method to start at 1 instead of 0

Question

I'm porting some code from R to python (note: in R lists start at the 1st, not 0th element) and instead of changing every place I access an array I want to create an child class of numpy.array so that, for example, the following code

    import numpy
    class array_starting_at_one(numpy.array):
        ???
    def myfunc(A):
        print A[1,1,1]
        print A[1:3,1,:]
    A = array_starting_at_one([[[1, 2, 3], [4, 5, 6]], [[11, 12, 13], [14, 15, 16]]])
    myfunc(A)

outputs

    1
    [[ 1  2  3]
     [11 12 13]]

Anyone know how to fill in the ??? in the above code?

You can probably do this, but I would advise you not to. 1) It'll probably involve more view-casting than you want. 2) In the long run, It'll probably be more confusing than just trying to keep the indexing conventions of the two languages straight in your hea -- especially if you plan on working with other's who use numpy frequently. (Note, I come from a Fortran background where the indices start at 1 as well, and after a short amount of time, the differences are no big deal ...) — mgilson
– mgilson, Commented Oct 6, 2014 at 18:42
P.S. I'm aware R interprets 1:3 as [1,2,3] instead of [1,2] (as python does), but I, rushing ahead somewhat stupidly, already converted all these ranges the painstaking way. — mathandy
– mathandy, Commented Oct 6, 2014 at 18:51

mathandy · Accepted Answer · 2014-10-13 01:28:44Z

Here's the solution I came up with:

import numpy
def adj(attr):
    if attr==None:
        return attr
    else:
        return attr-1
def adjust_slice(x):
    if isinstance(x,int):
        return x-1
    elif isinstance(x,slice):
        return slice(*[adj(attrib) for attrib in (x.start,x.stop,x.step)])
    elif isinstance(x,list):
        return slice(x[0]-1,x[-1]-1,1)
    else:
        raise Exception("Expected slice, list, or int.")
class array_starting_at_one(list):
    def __init__(self,np_array):
        self.np_array = numpy.array(np_array)
    def __getitem__(self,i):
        if isinstance(i,int):
            i=i-1
        elif isinstance(i,tuple):
            i = tuple([adjust_slice(x) for x in i])
        else:
            return array_starting_at_one(self.np_array[adjust_slice(x)])
        return self.np_array[i]
    def __setitem__(self,i,y):
        if isinstance(i,int):
            self.np_array[i-1] = y
        elif isinstance(i,tuple):
            self.np_array[tuple([adjust_slice(x) for x in i])] = y
        else:
            self.np_array[adjust_slice(x)] = y
    def __getslice__(self,i,j):
        return array_starting_at_one(self.np_array[(i-1):(j-1)])
    def __setslice__(self,i,j,y):
        self.np_array[i-1:j-1]=y
    def __repr__(self):
        print self.np_array
    def __str__(self):
        return str(self.np_array)

It worked for what I needed it to, but I didn't take the time to test it too carefully, so please use with caution.

Collectives™ on Stack Overflow

python: change numpy.array access method to start at 1 instead of 0

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related