npindex 0.1.1

Provide an intuitive way to make ndarray indexing variable.

Summary

Provide an intuitive way to make ndarray indexing variable like this:

from npindex import npidx

green_idx = npidx[:, :, 1]

# assume that im1 and im2 are numpy ndarray
# that were return values of OpenCV's cv2.imread() function.
green_of_im1 = im1[green_idx]
green_of_im2 = im2[green_idx]

This is all npindex does.

Usage

The signagure of npindex.npidx

npidx(d1, d2=None, d3=None, /)

Only positional argument is permitted. And 1, 2, 3 dimentional ndarrays are only supported.

Each arguments are treated as a slice object when arguments are passed in the form of slicing notation:

npidx[1:2, 3:4, 5:6]
# => (slice(1, 2, None), slice(3, 4, None), slice(5, 6, None))

Or, treated as an integer when arguments are passed just as an integer:

npidx[1, 2, 3] # => (1, 2, 3)

What is the point of that? OK, I’ll show you it in the next section.

How to use effectively

We can access to items of ndarray by passing one or more slice objects inside its square brackets like this:

import numpy as np
from npindex import npidx

a = np.arange(10) # a => array([0, 1, 2, 3, 4, 5, 6, 7, 8, 9])
i = npidx[1:5] # i => slice(1, 5, None)
a[i] # => array([1, 2, 3, 4])

a2 = a.reshape(2, 5) # a => array([[0, 1, 2, 3, 4], [5, 6, 7, 8, 9]])
i2 = npidx[:, 2:4] # i2 => (slice(None, None, None), slice(2, 4, None))
a2[i2] # => array([[2, 3], [7, 8]])

And it is also possible to access just one value from ndarray:

import numpy as np
from npindex import npidx

a = np.arange(10) # a => array([0, 1, 2, 3, 4, 5, 6, 7, 8, 9])
i = npidx[1] # i => 1
a[i] # => 1

a2 = a.reshape(2, 5) # a => array([[0, 1, 2, 3, 4], [5, 6, 7, 8, 9]])
i2 = npidx[0, 4] # (0, 4)
a2[i2] # => 4

When we want to access values from two or more ndarray object, npidx is useful.:

import numpy as np
from npindex import npidx

range = npidx[100:300, 150:250, :]
small_im1 = im1[range]
small_im2 = im2[range]

Install

Linux, Windows, macOS:

pip install npindex