Numpy Ndarray has no convenience method to set encoding (expect some thing like this np.astype(str, encoding))

we need to set the encoding during type conversion np dtype object to np dtype string with a encoding “utf-8” , but what i see is only ascii or unicode np.astype(np.unicode_)

unicode_np_array = np.ndarray(shape=(1,1,1), dtype=object)
for index, x in np.ndenumerate(unicode_np_array):
   unicode_np_array.itemset(index,b'search_key_\xc3\x84')
np.astype(np.unicode_)
yields this error 
UnicodeDecodeError: 'ascii' codec can't decode
byte 0xc3 in position 12: ordinal not in range(128)

because of that we do ndenumerate and apply decoding for each element which has a performance impact . It would be nice if there is any convenience method to do this which internally parallelize the encoding.

import numpy as np
import time

unicode_np_array = np.ndarray(shape=(1,1,1), dtype=object)
for index, x in np.ndenumerate(unicode_np_array):
   unicode_np_array.itemset(index,b'search_key_\xc3\x84')

print ("The unicode numpy array :"+ str(unicode_np_array))

ascii_np_array = np.ndarray(shape=(1,1,1), dtype=object)
for index, x in np.ndenumerate(ascii_np_array):
   ascii_np_array.itemset(index,b'search_key')
print ("The ascii numpy array :"+ str(ascii_np_array))

unicode_conversion_start_time = time.time()
for index, x in np.ndenumerate(unicode_np_array):
   unicode_np_array.itemset(index,x.decode('utf-8'))
unicode_conversion_end_time = time.time()

unicode_conversion_time = (unicode_conversion_end_time - unicode_conversion_start_time)*1000
print ("The unicode numpy array after conversion :"+ str(unicode_np_array))
print ("The unicode numpy conversion time :"+ str(unicode_conversion_time))

ascii_conversion_start_time = time.time()
ascii_np_array = ascii_np_array.astype(str)
ascii_conversion_end_time = time.time()

ascii_conversion_time = (ascii_conversion_end_time - ascii_conversion_start_time)*1000
print ("The ascii numpy array after conversion :"+ str(ascii_np_array))
print ("The ascii numpy conversion time :"+ str(ascii_conversion_time))

[MartinNowak]

np.char.decode(unicode_np_array.astype(np.bytes_), 'UTF-8')

[ScriptPup]

After hours of banging by head against this, turns out to be incredibly simple. Thanks @mattip, that does it.