mingw-w64-crt/testcases/t_mbrtowc.c

#ifdef NDEBUG
#undef NDEBUG
#endif

#include <assert.h>
#include <errno.h>
#include <locale.h>
#include <stdio.h>
#include <stdlib.h>
#include <wchar.h>

/* mingw-w64 headers */
#include "libtest.h"

/**
 * This test is for mingw-w64's implementation of mbrtowc function.
 *
 * This implementation is used with all msvcr*.dll CRTs, but not UCRT.
 * This test is skipped for UCRT.
 *
 * Also note that mingw-w64's implementation only works with SBCS and DBCS
 * code pages (MB_CUR_MAX == 1 || MB_CUR_MAX == 2).
 */

static void set_conversion_state (mbstate_t *state, int bytes) {
#ifdef _UCRT
  state->_Wchar = bytes;
#else
  *state = bytes;
#endif
}

char Ascii[] = {'a'};
char NonAscii[] = {(char) 0x80};
char Multibyte[] = {(char) 0x81, (char) 0x81};
char InvalidMultibyte[] = {(char) 0x81, 0};

int main (void) {
  mingw_test_init ();

#ifdef _UCRT
  return 77;
#endif

  mbstate_t state = {0};
  wchar_t   wc = WEOF;

  /**
   * Test "C" locale
   */
  assert (setlocale (LC_ALL, "C") != NULL);
  assert (MB_CUR_MAX == 1);

  /**
   * All bytes in range [0,255] are valid and must convert to themselves
   */
  for (unsigned char c = 0;; ++c) {
    assert (mbrtowc (&wc, (char *) &c, MB_CUR_MAX, &state) == !!c);
    assert (wc == c);
    assert (mbsinit (&state));
    assert (errno == 0);

    if (c == 0xFF) {
      break;
    }
  }

  /**
   * Detect invalid conversion state
   *
   * NOTE: this is optional error condition specified in POSIX.
   * This check fails with CRT's mbrtowc.
   */
  set_conversion_state (&state, Ascii[0]);
  wc = WEOF;

  assert (mbrtowc (&wc, (char *) &Ascii, MB_CUR_MAX, &state) == (size_t) -1);
  assert (wc == WEOF);
  assert (!mbsinit (&state));
  assert (errno == EINVAL);

  // reset errno
  _set_errno (0);

  // reset `state`
  set_conversion_state (&state, 0);

  /**
   * Test SBCS code page
   */
  if (setlocale (LC_ALL, ".1252") == NULL) {
    wprintf (L"Skipping tests for SBCS code page 1252\n");
    goto skip_sbcs;
  }

  assert (MB_CUR_MAX == 1);

  /**
   * All bytes in range [0,127] are valid ASCII characters
   */
  for (unsigned char c = 0; c < 0x80; ++c) {
    wc = WEOF;

    assert (mbrtowc (&wc, (char *) &c, MB_CUR_MAX, &state) == !!c);
    assert (wc == c);
    assert (mbsinit (&state));
    assert (errno == 0);
  }

  /**
   * All bytes in range [128,159] are valid characters
   */
  for (unsigned char c = 0x80; c < 0xA0; ++c) {
    wc = WEOF;

    assert (mbrtowc (&wc, (char *) &c, MB_CUR_MAX, &state) == 1);
    if (c == 0x81 || c == 0x8D || c == 0x8F || c == 0x90 || c == 0x9D) {
      assert (wc == c);
    } else {
      assert (wc != WEOF);
      assert (wc != c);
    }
    assert (mbsinit (&state));
    assert (errno == 0);
  }

  /**
   * All bytes in range [160,255] are valid and must convert to themselves
   */
  for (unsigned char c = 0xA0;; ++c) {
    wc = WEOF;

    assert (mbrtowc (&wc, (char *) &c, MB_CUR_MAX, &state) == 1);
    assert (wc == c);
    assert (mbsinit (&state));
    assert (errno == 0);

    if (c == 0xFF) {
      break;
    }
  }

skip_sbcs:
  /**
   * Disable tests for DBCS code pages with msvcrt10.dll since it does not
   * support multibyte characters.
   *
   * Calling setlocale with locale string which requests DBCS code page
   * result in runtime error.
   */
#if __MSVCRT_VERSION__ != 0x0100
  /**
   * Test DBCS code page
   */
  if (setlocale (LC_ALL, ".932") == NULL) {
    wprintf (L"Skipping tests for DBCS code page 932\n");
    goto skip_dbcs;
  }

  assert (MB_CUR_MAX == 2);

  /**
   * Make sure ASCII characters are handled correctly
   */
  for (char c = 0;; ++c) {
    wc = WEOF;

    assert (mbrtowc (&wc, &c, 1, &state) == !!c);
    assert (wc == c);
    assert (mbsinit (&state));
    assert (errno == 0);

    if (c == 0x7F) {
      break;
    }
  }

  /**
   * Try convert incomplete multibyte character
   */
  wc = WEOF;

  assert (mbrtowc (&wc, (char *) Multibyte, 1, &state) == (size_t) -2);
  /* This assertion fails with CRT's version */
  assert (wc == WEOF);
  assert (!mbsinit (&state));
  assert (errno == 0);

  /**
   * Complete multibyte character
   */
  wc = WEOF;

  assert (mbrtowc (&wc, (char *) Multibyte + 1, 1, &state) == 1);
  assert (wc != WEOF);
  assert (mbsinit (&state));
  assert (errno == 0);

  /**
   * Convert complete multibyte character
   */
  wc = WEOF;

  assert (mbrtowc (&wc, (char *) Multibyte, MB_CUR_MAX, &state) == 2);
  assert (wc != WEOF);
  assert (mbsinit (&state));
  assert (errno == 0);

  /**
   * Try convert invalid multibyte character
   */
  wc = WEOF;

  assert (mbrtowc (&wc, (char *) InvalidMultibyte, MB_CUR_MAX, &state) == (size_t) -1);
  /* This assertion fails with CRT's version */
  assert (wc == WEOF);
  assert (mbsinit (&state));
  assert (errno == EILSEQ);

skip_dbcs:
#endif
  return 0;
}