python - str object and string dtypes mismatch in Cython -

i'm trying run simple cython code ipython notebook. have following code snippet:

%load_ext cythonmagic %%cython cimport cython import numpy np cimport numpy np  cdef int test(np.ndarray[np.str, ndim = 1] a):     return 6  print test(np.array(['gona','haraka'])) 

what want pass numpy 1d array of strings function. when function test executed, returns 0 , following error:

exception valueerror: "buffer dtype mismatch, expected 'str object' got string" in '_cython_magic_505ff8c1b7497cde585006f723e794bd.test' ignored

what can reason this?
how pass numpy array of strings?
if use str instead of np.str or give dtype=np.str in numpy array when function test called, same error.

i tested cython 0.20.1 , can use general np.ndarray definition, not specifying data type or number of dimensions:

cimport cython import numpy np cimport numpy np  cdef int test(np.ndarray a):     return 6  print test(np.array(['gona', 'haraka'])) 

if want better performance can pass numpy string array using pointer , passing array around through char * buffer. following example shows how can achieved. increases character code adding 1 each non-zero value:

import numpy np cimport numpy np  cdef int f(char *a, int size):     cdef int     in range(size):         if a[i]!=0:             a[i] += 1  def main():     cdef char *inp     cdef np.ndarray     = np.array(['aaaa', 'bbbbbb'])     inp = a.data     print     f(inp, a.itemsize*a.shape[0])     print 

when run main() get:

['aaaa' 'bbbbbb'] ['bbbb' 'cccccc']