src/functional/mbc.c

   1 #include <stdio.h>
   2 #include <string.h>
   3 #include <wchar.h>
   4 #include <stdlib.h>
   5 #include <locale.h>
   6 #include <langinfo.h>
   7 #include "test.h"
   8
   9 /*
  10  * f = function call to test (or any expression)
  11  * x = expected result
  12  * m = message to print on failure
  13  */
  14 #define T(f, x, m) (void)( \
  15         memset(&st, 0, sizeof st), \
  16         (i = (f)) == (x) || \
  17                 t_error("%s failed (%s) got %d want %d\n", #f, m, i, x) )
  18 #define TCHAR(f, x, m) (void)( \
  19         memset(&st, 0, sizeof st), \
  20         (i = (f)) == (x) || \
  21                 t_error("%s failed (%s) got 0x%04x want 0x%04x\n", #f, m, i, x) )
  22
  23 int main(void)
  24 {
  25         const char *cs;
  26         int i;
  27         mbstate_t st, st2;
  28         wchar_t wc, wcs[32];
  29
  30         (void)(
  31         setlocale(LC_CTYPE, "en_US.UTF-8") ||
  32         setlocale(LC_CTYPE, "en_GB.UTF-8") ||
  33         setlocale(LC_CTYPE, "en.UTF-8") ||
  34         setlocale(LC_CTYPE, "POSIX.UTF-8") ||
  35         setlocale(LC_CTYPE, "C.UTF-8") ||
  36         setlocale(LC_CTYPE, "UTF-8") ||
  37         setlocale(LC_CTYPE, "") );
  38
  39         T(mbsrtowcs(wcs, (cs="abcdef",&cs), 3, &st), 3, "wrong semantics for wcs buf len");
  40         T(mbsrtowcs(wcs, (cs="abcdef",&cs), 8, &st), 6, "wrong semantics for wcs buf len");
  41         T(mbsrtowcs(NULL, (cs="abcdef",&cs), 2, &st), 6, "wrong semantics for NULL wcs");
  42
  43         if (strcmp(nl_langinfo(CODESET), "UTF-8"))
  44                 return t_error("cannot set UTF-8 locale for test (codeset=%s)\n", nl_langinfo(CODESET));
  45
  46         T(mbrtowc(&wc, "\x80", 1, &st), -1, "failed to catch error");
  47         T(mbrtowc(&wc, "\xc0", 1, &st), -1, "failed to catch illegal initial");
  48
  49         T(mbrtowc(&wc, "\xc0\x80", 2, &st), -1, "aliasing nul");
  50         T(mbrtowc(&wc, "\xc0\xaf", 2, &st), -1, "aliasing slash");
  51         T(mbrtowc(&wc, "\xe0\x80\xaf", 3, &st), -1, "aliasing slash");
  52         T(mbrtowc(&wc, "\xf0\x80\x80\xaf", 4, &st), -1, "aliasing slash");
  53         T(mbrtowc(&wc, "\xf8\x80\x80\x80\xaf", 5, &st), -1, "aliasing slash");
  54         T(mbrtowc(&wc, "\xfc\x80\x80\x80\x80\xaf", 6, &st), -1, "aliasing slash");
  55         T(mbrtowc(&wc, "\xe0\x82\x80", 3, &st), -1, "aliasing U+0080");
  56         T(mbrtowc(&wc, "\xe0\x9f\xbf", 3, &st), -1, "aliasing U+07FF");
  57         T(mbrtowc(&wc, "\xf0\x80\xa0\x80", 4, &st), -1, "aliasing U+0800");
  58         T(mbrtowc(&wc, "\xf0\x8f\xbf\xbd", 4, &st), -1, "aliasing U+FFFD");
  59
  60         T(mbrtowc(&wc, "\xed\xa0\x80", 3, &st), -1, "failed to catch surrogate");
  61         T(mbrtowc(&wc, "\xef\xbf\xbe", 3, &st), 3, "failed to accept U+FFFE");
  62         T(mbrtowc(&wc, "\xef\xbf\xbf", 3, &st), 3, "failed to accept U+FFFF");
  63         T(mbrtowc(&wc, "\xf4\x8f\xbf\xbe", 4, &st), 4, "failed to accept U+10FFFE");
  64         T(mbrtowc(&wc, "\xf4\x8f\xbf\xbf", 4, &st), 4, "failed to accept U+10FFFF");
  65
  66         T(mbrtowc(&wc, "\xc2\x80", 2, &st), 2, "wrong length");
  67         TCHAR((mbrtowc(&wc, "\xc2\x80", 2, &st),wc), 0x80, "wrong char");
  68         T(mbrtowc(&wc, "\xe0\xa0\x80", 3, &st), 3, "wrong length");
  69         TCHAR((mbrtowc(&wc, "\xe0\xa0\x80", 3, &st),wc), 0x800, "wrong char");
  70         T(mbrtowc(&wc, "\xf0\x90\x80\x80", 4, &st), 4, "wrong length");
  71         TCHAR((mbrtowc(&wc, "\xf0\x90\x80\x80", 4, &st),wc), 0x10000, "wrong char");
  72
  73         memset(&st2, 0, sizeof st2);
  74         T(mbrtowc(&wc, "\xc2", 1, &st2), -2, "failed to accept initial byte");
  75         T(mbrtowc(&wc, "\x80", 1, &st2), 1, "failed to resume");
  76         TCHAR(wc, 0x80, "wrong char");
  77
  78         memset(&st2, 0, sizeof st2);
  79         T(mbrtowc(&wc, "\xc2", 1, &st2), -2, "failed to accept initial byte");
  80         T(mbsrtowcs(wcs, (cs="\xa0""abc",&cs), 32, &st2), 4, "failed to resume");
  81         TCHAR(wcs[0], 0xa0, "wrong char");
  82         TCHAR(wcs[1], 'a', "wrong char");
  83         T(!cs, 1, "wrong final position");
  84         return t_status;
  85 }