Convert string to 2 dimensional numpy array

Question

I have a string which is loaded from a .mat file and is in the following format :

(array([172.169, 73.2]), array([128.83, 102.31]), array([143.49, 124.43]), array([186.83, 95.69]), 'R', array([], dtype=float64))

Is there any way I can convert this directly to a numpy 2D array without parsing over the entire string and removing the "arrays" manually? Neither numpy.fromstring nor numpy.frombuffer will work here.

If it's a string, why have you not surrounded it with quotes? — Saimon
– Saimon, Commented Apr 1, 2020 at 9:15
@DavidBuck It's similar, but not exactly, since to convert it to a list I have to parse the string anyways. — Rijul Ganguly
– Rijul Ganguly, Commented Apr 1, 2020 at 9:18
@Back2Basics No, R is not a typo. This is an annotation file for hands. R indicates the file is for the right hand. It's not to be there in the final array, I can manually remove that — Rijul Ganguly
– Rijul Ganguly, Commented Apr 1, 2020 at 9:19

kederrac · Accepted Answer · 2020-04-01 09:25:04Z

2

you can use np.fromstring with a regular expression:

import re

# s is your string
np.fromstring(', '.join(re.findall(r'\[(.+?)\]', s)), sep =', ').reshape((2, 2, 2))

output:

array([[[172.169,  73.2  ],
        [128.83 , 102.31 ]],

       [[143.49 , 124.43 ],
        [186.83 ,  95.69 ]]])

edited Apr 1, 2020 at 9:25

answered Apr 1, 2020 at 9:21

kederrac

17.4k6 gold badges36 silver badges58 bronze badges

Sign up to request clarification or add additional context in comments.

Collectives™ on Stack Overflow

Convert string to 2 dimensional numpy array

1 Answer 1

Comments

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Related