Env: Python 3.6, Numpy 1.14.0
a = {ndarray}['中','中']
I need to convert str type to numpy.string_.
A single str can be converted like this: np.string_('中', encoding='utf-8')]
But, how to convert a ndarray with a single statement not loop?
a.astype(np.string_) will raise a error:
{SystemError} returned a result with an error set
np.char has functions that apply str methods to elements of an array:
In [367]: a = np.array(['中','中'])
In [368]: np.string_(a[0], encoding='utf-8')
Out[368]: b'\xe4\xb8\xad'
In [369]: np.char.encode(a, encoding='utf-8')
Out[369]: array([b'\xe4\xb8\xad', b'\xe4\xb8\xad'], dtype='|S3')
np.string_ is a bytestring, while np.str is unicode (in py3). So this translates a 'U1' array into a 'S3' array.
In [370]: a
Out[370]: array(['中', '中'], dtype='<U1')
