python - Numpy array slicing transposes internal data -

i writing c extension shall deal numpy arrays. have written function read , output numpy array. using it, noticed weird behavior appears when use slicing in input array.

the c function read (boolan) array:

char **pymatrix_to_carrayptrschar(pyarrayobject *arrayin) {     char **result, *array;     int i, n, m, j;      n = arrayin->dimensions[0];     m = arrayin->dimensions[1];     result = ptrvectorchar(n, m);      array = (char *) arrayin->data; /* pointer arrayin data int */     (i = 0; < n; i++) {       result[i] = &array[i * m];     }     printarrchar(result, n, m);     return result; } 

ptrvectorchar function memory allocation:

char **ptrvectorchar(long dim1) {     char **v;     if (!(v = malloc(dim1 * sizeof(char*)))) {         pyerr_setstring(pyexc_memoryerror,               "in **ptrvectorchar. allocation of memory character array failed.");         exit(0);     }     return v; } 

and printing done with:

void printarrchar(char **arr, int dim1, int dim2) {     int i, j;     (i = 0; < dim1; i++) {         (j = 0; j < dim2; j++) {             printf("%i ", arr[i][j]);         }         printf("\n");     } } 

my python script reproducing error is:

import numpy np import myextension np.random.seed(1)  x = np.array((1,1,1,1,1,1)).astype(bool) = np.round(np.random.rand(trialnr, lakenr)).astype(bool) aslicing = a[:, x]  print("a:") print(a + 0)  print("aslicing:") print(aslicing + 0)  print("c output a:") myextension.myfunction(a)  print("c output aslicing:") myextension.myfunction(aslicing) 

output is:

a: [[0 1 0 0 0 0]  [0 0 0 1 0 1]  [0 1 0 1 0 1]  [0 0 1 1 0 1]  [1 1 0 0 0 1]  [0 0 1 1 1 0]  [1 1 0 1 1 1]  [0 1 0 0 1 0]  [0 0 0 1 0 0]  [0 0 1 0 1 1]]  aslicing: [[0 1 0 0 0 0]  [0 0 0 1 0 1]  [0 1 0 1 0 1]  [0 0 1 1 0 1]  [1 1 0 0 0 1]  [0 0 1 1 1 0]  [1 1 0 1 1 1]  [0 1 0 0 1 0]  [0 0 0 1 0 0]  [0 0 1 0 1 1]]  c output a: 0 1 0 0 0 0  0 0 0 1 0 1  0 1 0 1 0 1  0 0 1 1 0 1  1 1 0 0 0 1  0 0 1 1 1 0  1 1 0 1 1 1  0 1 0 0 1 0  0 0 0 1 0 0  0 0 1 0 1 1   c output aslicing: 0 0 0 0 1 0  1 0 0 0 1 0  1 0 1 0 1 1  0 0 0 0 0 1  0 1 0 0 0 1  0 1 1 1 0 1  1 0 1 0 0 0  0 0 0 1 1 1  0 1 0 1 1 1  1 0 1 0 0 1  

as can seen easily, a , aslicing same arrays python. however, c function reads data in sees data kind of transposed. c regards asliced if

a.t.reshape((10,6)) 

does know why error appears , how circumvent it? of course, transposing within c code easy. however, want program able deal both types of arrays.

i prefer solution within c extension, i.e. useres of extension shall not have care whether input "sliced" or not. nevertheless, tried put deep copy of asliced extension - had same wrong result asliced had.

i working python 3.4 64bit, numpy 1.9.1, win8 64bit , visual studio 10 64bit c compiler.

as hpaulj pointed out, possible figure out memory structure using flag f_contiguous. spent lot of time trying find way read flag within c. far found out, can done evaluating value of trials_array->flags % 2. however, did not find reference clear explenation issue.

trials_array->flags integer number. numpy flag constants npy_c_contiguous, npy_f_contiguous, etc. integers powers of two. seems true flag set if respective position in binary representation of trials_array->flags 1.

even knowing memory structure not trivial thought read array in. found much easier way convert numpy arrays c arrays: use

char *myarray; pyarrayobject *myarray_numpy;  pyarray_ascarray(&myarray_numpy, (void *) &myarray, myarray_numpy->dimensions, 2, pyarray_descrfromtype(npy_bool));  //do array  pyarray_free(myarray_numpy, myarray); 

i found example using these functions here.