fun

static int32_t *get_token(int32_t **buf, int *toklen) {
    int i = 0, datalen = 16;
    int32_t *b = *buf;
    int32_t *data;
    if ((data = (int32_t *) malloc(datalen * sizeof(int32_t))) == NULL)
        return NULL;
    for (;;) {
        int32_t uc = *b++;
        utf8proc_propval_t cat = utf8proc_get_property(uc).category;
        if (!uc) {
            *toklen = i;
            *buf = b;
            return data = realloc(data, i);
        }
        if (cat != UTF8PROC_CATEGORY_ZS) {
            data[i++] = uc;
            if (i >= datalen) {
                int32_t *t;
                if ((t = (int32_t *) realloc(data, datalen <<= 1)) == NULL) {
                    free(data);
                    return NULL;
                }
                data = t;
            }
        } else { // end of token; find start of next
            *toklen = i;
            data = realloc(data, i);
            for (;;) {
                uc = *b;
                if (!uc)
                    return data;
                if (utf8proc_get_property(uc).category != UTF8PROC_CATEGORY_ZS) {
                    *buf = b;
                    return data;
                }
                b++;
            }
        }
    }
    return NULL; // unreachable
}