Python Merge Two Numpy Arrays Based on Condition

Question

How can I merge the following two arrays, by looking up a value from array A in array B?

Array A:

array([['GG', 'AB', IPv4Network('1.2.3.41/26')],
       ['GG', 'AC', IPv4Network('1.2.3.42/25')],
       ['GG', 'AD', IPv4Network('1.2.3.43/24')],
       ['GG', 'AE', IPv4Network('1.2.3.47/23')],
       ['GG', 'AF', IPv4Network('1.2.3.5/24')]],
      dtype=object)

and Array B:

array([['123456', 'A1', IPv4Address('1.2.3.5'), nan],
       ['987654', 'B1', IPv4Address('1.2.3.47'), nan]],
      dtype=object)  

The goal here is to create Array C, by looking up the IPv4Address from Array B in Array A and comparing them, and getting the corresponding array's second value and storing it:

Array C:

array([['123456', 'A1', IPv4Address('1.2.3.5'), nan, 'AF'],
       ['987654', 'B1', IPv4Address('1.2.3.47'), nan, 'AE']],
      dtype=object) 

The ip addresses are of this type: https://docs.python.org/3/library/ipaddress.html#ipaddress.ip_network

How can I achieve this?

edit:

Please note that the merging is conditioned on the IPs matching, so the resulting array C will have the same number of arrays as the Array B, but it will have one more value. The suggested duplicate links are not answering the same question.

Answer 1

This should do what you asked for (at least the output is exactly what you wanted), I made some minor assumptions to deal with your #dummydata, but that should not matter too much.

Code:

import numpy as np
import ipaddress as ip

array_A = np.array([['GG', 'AB', ip.ip_network('192.168.0.0/32')],
                    ['GG', 'AC', ip.ip_network('192.168.0.0/31')],
                    ['GG', 'AD', ip.ip_network('192.168.0.0/30')],
                    ['GG', 'AE', ip.ip_network('192.168.0.0/29')],
                    ['GG', 'AF', ip.ip_network('192.168.0.0/28')]],
                   dtype=object)

array_B = np.array([['123456', 'A1', ip.ip_network('192.168.0.0/28'), np.nan],
                    ['987654', 'B1', ip.ip_network('192.168.0.0/29'), np.nan]],
                   dtype=object)


def merge_by_ip(A, B):
    # initializing an empty array with len(B) rows and 5 columns for the values you want to save in it
    C = np.empty([len(B), 5],dtype=object)
    for n in range(len(B)):
        for a in A:
            # checking condition: if ip address in a is ip address in b
            if a[2] == B[n][2]:
                # add the entry of b with the second value of a to the new Array c
                C[n] = np.append(B[n], a[1])
    return C


print(merge_by_ip(array_A, array_B))

Output:

[['123456' 'A1' IPv4Network('192.168.0.0/28') nan 'AF']
 ['987654' 'B1' IPv4Network('192.168.0.0/29') nan 'AE']]

Note:

This solution has O(m * n) complexity, which isn't necessary, there are many out-of-the box (Pandas) and custom (e.g. using dict) ways to merge with lower complexity.