I have a 3-dimensional array filled with strings, mostly of numbers but also some text. If a string contains only one digit (ie. 1, 5), I want to add a zero to it (01,05,14,etc.). I can't get it to work for my NumPy array though.
I tried (among others):
strlist = ['1','2','3','4','5','6','7','8','9']
arr[np.isin(arr, strlist)] = '0' + arr[np.isin(arr, strlist)]
But this doesn't work. Does anyone have any tips?
NumPy has several useful functions for arrays of strings. See NumPy docs on String operations. The function you are looking for is np.core.defchararray.zfill or its alias np.char.zfill.
Taking an example array from David Buck's answer:
>>> import numpy as np
>>> arr = np.array([[['3', '6', '12'],
['0', '1', '3'],
['5', 'T', '8'],
['19', '15', '11']],
[['6', '3', '1'],
['10', '10', 'QR'],
['7', '11', '9'],
['12', '13', '11']],
[['1', 'G', '3'],
['10', '9', '2'],
['18', '12', '17'],
['6', '1', '10']]])
>>> np.char.zfill(arr, 2)
array([[['03', '06', '12'],
['00', '01', '03'],
['05', '0T', '08'],
['19', '15', '11']],
[['06', '03', '01'],
['10', '10', 'QR'],
['07', '11', '09'],
['12', '13', '11']],
[['01', '0G', '03'],
['10', '09', '02'],
['18', '12', '17'],
['06', '01', '10']]], dtype='<U2')
If you want to avoid adding zeros to elements that are not digits, we can use boolean array indexing and np.core.defchararray.isdigit function or its alias np.char.isdigit:
>>> mask = np.char.isdigit(arr)
>>> mask
array([[[ True, True, True],
[ True, True, True],
[ True, False, True],
[ True, True, True]],
[[ True, True, True],
[ True, True, False],
[ True, True, True],
[ True, True, True]],
[[ True, False, True],
[ True, True, True],
[ True, True, True],
[ True, True, True]]])
>>> arr[mask] = np.char.zfill(arr[mask], 2)
>>> arr
array([[['03', '06', '12'],
['00', '01', '03'],
['05', 'T', '08'],
['19', '15', '11']],
[['06', '03', '01'],
['10', '10', 'QR'],
['07', '11', '09'],
['12', '13', '11']],
[['01', 'G', '03'],
['10', '09', '02'],
['18', '12', '17'],
['06', '01', '10']]], dtype='<U2')
You can define a function that pads integers/passes over non integers and then use vectorize to apply it to the whole array.
import numpy as np
def pad(value):
try:
return '{0:0>2}'.format(int(value))
except:
return value
vfunc = np.vectorize(pad)
arr = vfunc(arr)
print(arr)
Applying that to an input of:
arr = np.array([[['3', '6', '12'],
['0', '1', '3'],
['5', 'T', '8'],
['19', '15', '11']],
[['6', '3', '1'],
['10', '10', 'QR'],
['7', '11', '9'],
['12', '13', '11']],
[['1', 'G', '3'],
['10', '9', '2'],
['18', '12', '17'],
['6', '1', '10']],])
returns
[[['03' '06' '12']
['00' '01' '03']
['05' 'T' '08']
['19' '15' '11']]
[['06' '03' '01']
['10' '10' 'QR']
['07' '11' '09']
['12' '13' '11']]
[['01' 'G' '03']
['10' '09' '02']
['18' '12' '17']
['06' '01' '10']]]
