Fast indexing of a boolean numpy array

[Jake Vanderplas]

Hi all,
I was wondering if I could get some input on this: I have an application where I need to quickly index a numpy boolean array in a cython script.  If it were an integer or float array, I know how I'd do this, using templated ndarray types as in the online tutorials.  It doesn't appear that the same solution is available for boolean types.
Below is a simple example of summation using the solution I came up with.  It works in the few cases I've tried, but I'd appreciate feedback: are there any details I'm missing that could make this sort of solution unstable?  Is there a more elegant way to quickly loop through a boolean array in cython?  My mask array is large enough that converting it to an int type is not an option.
Thanks
   Jake

import numpy as np
cimport numpy as np

# Here's how I would quickly cycle through an integer array
def sum_int(np.ndarray[int, ndim=2, mode='c'] mask):
    cdef unsigned int i, j, N1, N2, tot
    tot = 0
    N1 = mask.shape[0]
    N2 = mask.shape[1]
   
    for i from 0 <= i < N1:
        for j from 0 <= j < N2:
            if mask[i, j] != 0:
                tot += 1
   
    return tot

# cython does not support any bool dtype, so here's how I quickly cycle through a boolean array
def sum_bool(np.ndarray mask):
    assert mask.dtype == bool
    assert mask.flags['C_CONTIGUOUS']

    cdef np.npy_bool* mptr = <np.npy_bool*> mask.data

    cdef unsigned int N = mask.size
    cdef unsigned int i, tot=0
    for i from 0 <= i < N:
        if mptr[i] != 0:
            tot += 1

    return tot