将字符串更改为逗号分隔的 numpy int 数组

Question

I have a string that comes in bytes that has commas inside of it.

ex. b'-8 ,0 ,54 ,-30 ,28'

I first change it into a string using

msg = str(msg, 'utf-8')

This part works. However I need to make this string into a numpy int array. I have tried splitting at the commas, but I wind up just getting a 1 dimensional numpy array. I would like for each value in the array to be split by a comma.

msg = str(msg, 'utf-8')

z = [x.strip() for x in msg.split(',')]

x = np.array(z)
y = x.astype(np.int)

The error I get is

ValueError: Error when checking input: expected dense_1_input to have shape (5,) but got array with shape (1,)

Thank for the help!

Answer 1

All you are missing is a conversion from a string to an int inside of your list comprehension:

msg = str(msg, 'utf-8')
z = [int(x.strip()) for x in msg.split(',')]
x = np.array(z)

split() returns an array of strings, so you were getting z as an array of strings that look like numbers. int() is able to convert such strings into their numerical representations.

Answer 2

In [213]: b'-8 ,0 ,54 ,-30 ,28'.decode()                                                                     
Out[213]: '-8 ,0 ,54 ,-30 ,28'
In [214]: b'-8 ,0 ,54 ,-30 ,28'.decode().split(',')                                                          
Out[214]: ['-8 ', '0 ', '54 ', '-30 ', '28']
In [215]: np.array(b'-8 ,0 ,54 ,-30 ,28'.decode().split(','), dtype=int)                                     
Out[215]: array([ -8,   0,  54, -30,  28])
In [216]: np.array(b'-8 ,0 ,54 ,-30 ,28'.decode().split(','), dtype=int).reshape(-1,1)                       
Out[216]: 
array([[ -8],
       [  0],
       [ 54],
       [-30],
       [ 28]])