How To Extract Specific Numbers From Numpy Arrays Based On The Repetition Of Numbers

I have two numpy arrays and want to extract some specific parts of them. These arrays are: arr_A=np.arange(17, 29) arr_B=np.arange(17, 27) These arrays are number of some points.

Solution 1:

Based on your comments and clarifications, here is a way you can solve this.

How it works -

0. fetch_sublists() generator breaks the list into sublists and front pads each with 0s.
1. get_stars() calls the above generator to construct the matrix by stacking the arrays, flips to face the zeros upwards based on orientation, and saves it to m
2. Uses this orientation to find the first non zero elements from the top direction and saves it to star
3. Rotates the matrix in the required direction and overwrites it on m
4. Finally returns both matrix as mat and side elements as star.

#Generates the sublists with padding0
def fetch_sublists(arr, rep):
    itr = iter(arr)
    maxlen = np.max(rep)
    for size in rep:
        sublist = []
        for _ in range(size):
            sublist.append(next(itr))
        yield np.pad(sublist, maxlen-size)[:maxlen]
      
#Constructs the matrix based on orientation and returns the side elements
def get_stars(arr, rep, orientation='right'):
    if orientation=='right':
        m = np.flipud(list(fetch_sublists(arr, rep)))
        star_elements = m[(m!=0).argmax(0), np.arange(m.shape[1])]
        m = np.rot90(m,k=-1)
    else:
        m = np.array(list(fetch_sublists(arr, rep)))
        star_elements = np.flip(m[(m!=0).argmax(0), np.arange(m.shape[1])])
        m = m.T
    return m, star_elements


#Case 1 with right side orientation

mat, star = get_stars(arr_A, rep_A, 'right')

print('Matrix -')
print(mat)
print('')
print('Star elements -')
print(star)

Matrix -
[[17 21  0  0  0][18 22  0  0  0][19 23 25  0  0][20 24 26 27 28]]

Star elements -
[21 22 25 28]

#Case 2 with left side orientation

mat, star = get_stars(arr_B, rep_B, 'left')

print('Matrix -')
print(mat)
print('')
print('Star elements -')
print(star)

Matrix -
[[ 0  0 22][ 0 18 23][ 0 19 24][ 0 20 25][17 21 26]]

Star elements -
[17 20 19 18 22]

---

IIUC, you are trying to break a list into uneven chunks as the first part. Note, that you cant use numpy to store the final array since it needs same elements in each object in each dimension, and here each of the sublists contains different number of elements.

You can use generators for this as -

def fetch_sublists(arr, rep):
    itr = iter(arr)
    for size in rep:
        sublist = []
        for _ in range(size):
            sublist.append(next(itr))
        yield sublist

print(list(fetch_sublists(arr_A, rep_A)))
print(list(fetch_sublists(arr_B, rep_B)))

[[17, 18, 19, 20], [21, 22, 23, 24], [25, 26], [27], [28]][[17], [18, 19, 20, 21], [22, 23, 24, 25, 26]]

What is unclear is what you want after that in the second part. Will request you elaborate on that so I can update my answer accordingly.