numpy unique with customized sorting of unique elements of an array

Given:

R=["ip1", "ip7", "ip12", "ip5", "ip2", "ip22", "ip7", "ip1", "ip17", "ip22"]

I would like to get unique values of my list R with their corresponding indices.

Right now, I have name,idx=np.unique(R,return_inverse=True) which returns:

array(['ip1', 'ip12', 'ip17', 'ip2', 'ip22', 'ip5', 'ip7'], dtype='<U4') # name
[0 6 1 5 3 4 6 0 2 4]                                                    # idx

But I would like to use customized sorting with results as follows:

['ip1', 'ip2', 'ip5', 'ip7', 'ip12', 'ip17', 'ip22']
[0 3 4 2 1 6 3 0 5 6]

In list, I can use Rs=sorted(R, key=lambda x: int(x[2:])) with customized key but I can’t get unique values and corresponding indices.

Is there any way to manipulate sorting key np.unique or is there already a better approach handling this?

>Solution :

Run unique after conversion to int:

_, i, idx = np.unique([int(x[2:]) for x in R],
                      return_index=True,
                      return_inverse=True)

names = np.array(R)[i]

output:

# names
array(['ip1', 'ip2', 'ip5', 'ip7', 'ip12', 'ip17', 'ip22'], dtype='<U4')

# idx
array([0, 3, 4, 2, 1, 6, 3, 0, 5, 6])