rostests/tests/wcstombs-tests/wcstombs-tests.c

   1 /*
   2  * PROJECT:    ReactOS wcstombs Test Suite
   3  * LICENSE:    GPL v2 or any later version
   4  * FILE:       tests/wcstombs-tests/wcstombs-tests.c
   5  * PURPOSE:    Application for testing the CRT API's (wcstombs and wctomb) and the Win32 API WideCharToMultiByte for the Unicode to MultiByte string conversion
   6  * COPYRIGHT:  Copyright 2008 Colin Finck <colin@reactos.org>
   7  */
   8
   9 #include <windows.h>
  10 #include <stdio.h>
  11 #include <stdlib.h>
  12 #include <locale.h>
  13 #include <errno.h>
  14
  15 /* Macros for simplification */
  16 #define SETLOCALE(locale) \
  17     loc = setlocale(LC_ALL, locale); \
  18     if(!loc) \
  19     { \
  20         puts("setlocale failed for " locale ", this locale is probably not installed on your system"); \
  21         return; \
  22     }
  23
  24 #define OK(condition, fail_message, ...) \
  25     if(!(condition)) \
  26         printf("%d: " fail_message "\n", __LINE__, ##__VA_ARGS__);
  27
  28 /* Global variables for easier handling */
  29 char mbc;
  30 char mbs[5];
  31 int ret;
  32 wchar_t wc1 = 228;                                  /* Western Windows-1252 character */
  33 wchar_t wc2 = 1088;                                 /* Russian Windows-1251 character not displayable for Windows-1252 */
  34 wchar_t wcs[5] = {'T', 'h', 1088, 'i', 0};          /* String with ASCII characters and a Russian character */
  35 wchar_t dbwcs[3] = {28953, 25152, 0};               /* String with Chinese (codepage 950) characters */
  36
  37
  38 void CRT_Tests()
  39 {
  40     char* loc;
  41
  42     puts("CRT-Tests");
  43     puts("---------");
  44
  45     /* Current locale is "C", wcstombs should return the length of the input buffer without the terminating null character */
  46     ret = wcstombs(NULL, dbwcs, 0);
  47     OK(ret == 2, "ret is %d", ret);
  48
  49     ret = wcstombs(mbs, dbwcs, ret);
  50     OK(ret == -1, "ret is %d", ret);
  51     OK(mbs[0] == 0, "mbs[0] is %d", mbs[0]);
  52     OK(errno == EILSEQ, "errno is %d", errno);
  53
  54     ret = wcstombs(NULL, wcs, 0);
  55     OK(ret == 4, "ret is %d", ret);
  56
  57     ret = wcstombs(mbs, wcs, ret);
  58     OK(ret == -1, "ret is %d", ret);
  59     OK(!strcmp(mbs, "Th"), "mbs is %s", mbs);
  60     OK(errno == EILSEQ, "errno is %d", errno);
  61
  62     ret = wctomb(&mbc, wcs[0]);
  63     OK(ret == 1, "ret is %d", ret);
  64     OK(mbc == 84, "mbc is %d", mbc);
  65
  66     mbc = 84;
  67     ret = wcstombs(&mbc, &dbwcs[0], 1);
  68     OK(ret == -1, "ret is %d", ret);
  69     OK(mbc == 84, "mbc is %d", mbc);
  70
  71     ret = wcstombs(mbs, wcs, 0);
  72     OK(ret == 0, "ret is %d", ret);
  73
  74     /* The length for the null character (in any locale) is 0, but if you pass a variable, it will be set to 0 and wctomb returns 1 */
  75     ret = wctomb(NULL, 0);
  76     OK(ret == 0, "ret is %d", ret);
  77
  78     ret = wctomb(&mbc, 0);
  79     OK(ret == 1, "ret is %d", ret);
  80     OK(mbc == 0, "mbc is %d", mbc);
  81
  82     /* msvcr80.dll changes mbc in the following call back to 0, msvcrt.dll from WinXP SP2 leaves it untouched */
  83     mbc = 84;
  84     ret = wctomb(&mbc, dbwcs[0]);
  85     OK(ret == -1, "ret is %d", ret);
  86     OK(errno == EILSEQ, "errno is %d", errno);
  87     OK(mbc == 84, "mbc is %d", mbc);
  88
  89     /* With a real locale, -1 also becomes a possible return value in case of an invalid character */
  90     SETLOCALE("German");
  91     ret = wcstombs(NULL, dbwcs, 0);
  92     OK(ret == -1, "ret is %d", ret);
  93     OK(errno == EILSEQ, "errno is %d", errno);
  94
  95     ret = wcstombs(NULL, wcs, 2);
  96     OK(ret == -1, "ret is %d", ret);
  97     OK(errno == EILSEQ, "errno is %d", errno);
  98
  99     /* Test if explicitly setting the locale back to "C" also leads to the same results as above */
 100     SETLOCALE("C");
 101
 102     ret = wcstombs(NULL, dbwcs, 0);
 103     OK(ret == 2, "ret is %d", ret);
 104
 105     ret = wcstombs(NULL, wcs, 0);
 106     OK(ret == 4, "ret is %d", ret);
 107
 108     /* Test wctomb() as well */
 109     SETLOCALE("English");
 110
 111     ret = wctomb(&mbc, wc1);
 112     OK(ret == 1, "ret is %d", ret);
 113     OK(mbc == -28, "mbc is %d", mbc);
 114
 115     ret = wctomb(&mbc, wc2);
 116     OK(ret == -1, "ret is %d", ret);
 117     OK(errno == EILSEQ, "errno is %d", errno);
 118     OK(mbc == 63, "mbc is %d", mbc);
 119
 120     SETLOCALE("Russian");
 121
 122     ret = wcstombs(mbs, wcs, sizeof(mbs));
 123     OK(ret == 4, "ret is %d", ret);
 124     OK(!strcmp(mbs, "Thði"), "mbs is %s", mbs);
 125
 126     ret = wctomb(&mbc, wc2);
 127     OK(ret == 1, "ret is %d", ret);
 128     OK(mbc == -16, "mbc is %d", mbc);
 129
 130     ret = wctomb(&mbc, wc1);
 131     OK(ret == 1, "ret is %d", ret);
 132     OK(mbc == 97, "mbc is %d", mbc);
 133
 134     SETLOCALE("English");
 135
 136     ret = wcstombs(&mbc, wcs, 1);
 137     OK(ret == 1, "ret is %d", ret);
 138     OK(mbc == 84, "mbc is %d", mbc);
 139
 140     ZeroMemory(mbs, sizeof(mbs));
 141     ret = wcstombs(mbs, wcs, sizeof(mbs));
 142     OK(ret == -1, "ret is %d", ret);
 143     OK(errno == EILSEQ, "errno is %d", errno);
 144     OK(!strcmp(mbs, "Th?i"), "mbs is %s", mbs);
 145     mbs[0] = 0;
 146
 147     /* wcstombs mustn't add any null character automatically.
 148        So in this case, we should get the same string again, even if we only copied the first three bytes. */
 149     ret = wcstombs(mbs, wcs, 3);
 150     OK(ret == -1, "ret is %d", ret);
 151     OK(errno == EILSEQ, "errno is %d", errno);
 152     OK(!strcmp(mbs, "Th?i"), "mbs is %s", mbs);
 153     ZeroMemory(mbs, 5);
 154
 155     /* Now this shouldn't be the case like above as we zeroed the complete string buffer. */
 156     ret = wcstombs(mbs, wcs, 3);
 157     OK(ret == -1, "ret is %d", ret);
 158     OK(errno == EILSEQ, "errno is %d", errno);
 159     OK(!strcmp(mbs, "Th?"), "mbs is %s", mbs);
 160
 161     /* Double-byte tests */
 162     SETLOCALE("Chinese");
 163     ret = wcstombs(mbs, dbwcs, sizeof(mbs));
 164     OK(ret == 4, "ret is %d", ret);
 165     OK(!strcmp(mbs, "µH©Ò"), "mbs is %s", mbs);
 166     ZeroMemory(mbs, 5);
 167
 168     /* Length-only tests */
 169     SETLOCALE("English");
 170     ret = wcstombs(NULL, wcs, 0);
 171     OK(ret == -1, "ret is %d", ret);
 172     OK(errno == EILSEQ, "errno is %d", errno);
 173
 174     SETLOCALE("Chinese");
 175     ret = wcstombs(NULL, dbwcs, 0);
 176     OK(ret == 4, "ret is %d", ret);
 177
 178     /* This call causes an ERROR_INSUFFICIENT_BUFFER in the called WideCharToMultiByte function.
 179        For some reason, wcstombs under Windows doesn't reset the last error to the previous value here, so we can check for ERROR_INSUFFICIENT_BUFFER with GetLastError().
 180        This could also be seen as an indication that Windows uses WideCharToMultiByte internally for wcstombs. */
 181     ret = wcstombs(mbs, dbwcs, 1);
 182     OK(ret == 0, "ret is %d", ret);
 183     OK(mbs[0] == 0, "mbs[0] is %d", mbs[0]);
 184
 185     /* ERROR_INSUFFICIENT_BUFFER is also the result of this call with SBCS characters. WTF?!
 186        Anyway this is a Win32 error not related to the CRT, so we leave out this criteria. */
 187     ret = wcstombs(mbs, wcs, 1);
 188     OK(ret == 1, "ret is %d", ret);
 189     OK(mbs[0] == 84, "mbs[0] is %d", mbs[0]);
 190
 191     putchar('\n');
 192 }
 193
 194 void Win32_Tests(LPBOOL bUsedDefaultChar)
 195 {
 196     /*int i;*/
 197
 198     SetLastError(0xdeadbeef);
 199
 200     puts("Win32-Tests");
 201     puts("-----------");
 202
 203     ret = WideCharToMultiByte(1252, WC_NO_BEST_FIT_CHARS, &wc1, 1, &mbc, 1, NULL, bUsedDefaultChar);
 204     OK(ret == 1, "ret is %d", ret);
 205     OK(mbc == -28, "mbc is %d", mbc);
 206     if(bUsedDefaultChar) OK(*bUsedDefaultChar == FALSE, "bUsedDefaultChar is %d", *bUsedDefaultChar);
 207     OK(GetLastError() == 0xdeadbeef, "GetLastError() is %lu", GetLastError());
 208
 209     ret = WideCharToMultiByte(1252, WC_NO_BEST_FIT_CHARS, &wc2, 1, &mbc, 1, NULL, bUsedDefaultChar);
 210     OK(ret == 1, "ret is %d", ret);
 211     OK(mbc == 63, "mbc is %d", mbc);
 212     if(bUsedDefaultChar) OK(*bUsedDefaultChar == TRUE, "bUsedDefaultChar is %d", *bUsedDefaultChar);
 213     OK(GetLastError() == 0xdeadbeef, "GetLastError() is %lu", GetLastError());
 214
 215     ret = WideCharToMultiByte(1251, WC_NO_BEST_FIT_CHARS, &wc2, 1, &mbc, 1, NULL, bUsedDefaultChar);
 216     OK(ret == 1, "ret is %d", ret);
 217     OK(mbc == -16, "mbc is %d", mbc);
 218     if(bUsedDefaultChar) OK(*bUsedDefaultChar == FALSE, "bUsedDefaultChar is %d", *bUsedDefaultChar);
 219     OK(GetLastError() == 0xdeadbeef, "GetLastError() is %lu", GetLastError());
 220
 221     ret = WideCharToMultiByte(1251, WC_NO_BEST_FIT_CHARS, &wc1, 1, &mbc, 1, NULL, bUsedDefaultChar);
 222     OK(ret == 1, "ret is %d", ret);
 223     OK(mbc == 97, "mbc is %d", mbc);
 224     if(bUsedDefaultChar) OK(*bUsedDefaultChar == FALSE, "bUsedDefaultChar is %d", *bUsedDefaultChar);
 225     OK(GetLastError() == 0xdeadbeef, "GetLastError() is %lu", GetLastError());
 226
 227     /* This call triggers the last Win32 error */
 228     ret = WideCharToMultiByte(1252, WC_NO_BEST_FIT_CHARS, wcs, -1, &mbc, 1, NULL, bUsedDefaultChar);
 229     OK(ret == 0, "ret is %d", ret);
 230     OK(mbc == 84, "mbc is %d", mbc);
 231     if(bUsedDefaultChar) OK(*bUsedDefaultChar == FALSE, "bUsedDefaultChar is %d", *bUsedDefaultChar);
 232     OK(GetLastError() == ERROR_INSUFFICIENT_BUFFER, "GetLastError() is %lu", GetLastError());
 233     SetLastError(0xdeadbeef);
 234
 235     ret = WideCharToMultiByte(1252, WC_NO_BEST_FIT_CHARS, wcs, -1, mbs, sizeof(mbs), NULL, bUsedDefaultChar);
 236     OK(ret == 5, "ret is %d", ret);
 237     OK(!strcmp(mbs, "Th?i"), "mbs is %s", mbs);
 238     if(bUsedDefaultChar) OK(*bUsedDefaultChar == TRUE, "bUsedDefaultChar is %d", *bUsedDefaultChar);
 239     OK(GetLastError() == 0xdeadbeef, "GetLastError() is %lu", GetLastError());
 240     mbs[0] = 0;
 241
 242     /* WideCharToMultiByte mustn't add any null character automatically.
 243        So in this case, we should get the same string again, even if we only copied the first three bytes. */
 244     ret = WideCharToMultiByte(1252, WC_NO_BEST_FIT_CHARS, wcs, 3, mbs, sizeof(mbs), NULL, bUsedDefaultChar);
 245     OK(ret == 3, "ret is %d", ret);
 246     OK(!strcmp(mbs, "Th?i"), "mbs is %s", mbs);
 247     if(bUsedDefaultChar) OK(*bUsedDefaultChar == TRUE, "bUsedDefaultChar is %d", *bUsedDefaultChar);
 248     OK(GetLastError() == 0xdeadbeef, "GetLastError() is %lu", GetLastError());
 249     ZeroMemory(mbs, 5);
 250
 251     /* Now this shouldn't be the case like above as we zeroed the complete string buffer. */
 252     ret = WideCharToMultiByte(1252, WC_NO_BEST_FIT_CHARS, wcs, 3, mbs, sizeof(mbs), NULL, bUsedDefaultChar);
 253     OK(ret == 3, "ret is %d", ret);
 254     OK(!strcmp(mbs, "Th?"), "mbs is %s", mbs);
 255     if(bUsedDefaultChar) OK(*bUsedDefaultChar == TRUE, "bUsedDefaultChar is %d", *bUsedDefaultChar);
 256     OK(GetLastError() == 0xdeadbeef, "GetLastError() is %lu", GetLastError());
 257
 258     /* Double-byte tests */
 259     ret = WideCharToMultiByte(950, WC_NO_BEST_FIT_CHARS, dbwcs, -1, mbs, sizeof(mbs), NULL, bUsedDefaultChar);
 260     OK(ret == 5, "ret is %d", ret);
 261     OK(!strcmp(mbs, "µH©Ò"), "mbs is %s", mbs);
 262     if(bUsedDefaultChar) OK(*bUsedDefaultChar == FALSE, "bUsedDefaultChar is %d", *bUsedDefaultChar);
 263     OK(GetLastError() == 0xdeadbeef, "GetLastError() is %lu", GetLastError());
 264
 265     ret = WideCharToMultiByte(950, WC_NO_BEST_FIT_CHARS, dbwcs, 1, &mbc, 1, NULL, bUsedDefaultChar);
 266     OK(ret == 0, "ret is %d", ret);
 267     if(bUsedDefaultChar) OK(*bUsedDefaultChar == FALSE, "bUsedDefaultChar == FALSE");
 268     OK(GetLastError() == ERROR_INSUFFICIENT_BUFFER, "GetLastError() is %lu", GetLastError());
 269     SetLastError(0xdeadbeef);
 270     ZeroMemory(mbs, 5);
 271
 272     ret = WideCharToMultiByte(950, WC_NO_BEST_FIT_CHARS, dbwcs, 1, mbs, sizeof(mbs), NULL, bUsedDefaultChar);
 273     OK(ret == 2, "ret is %d", ret);
 274     OK(!strcmp(mbs, "µH"), "mbs is %s", mbs);
 275     if(bUsedDefaultChar) OK(*bUsedDefaultChar == FALSE, "bUsedDefaultChar is %d", *bUsedDefaultChar);
 276     OK(GetLastError() == 0xdeadbeef, "GetLastError() is %lu", GetLastError());
 277
 278     /* Length-only tests */
 279     ret = WideCharToMultiByte(1252, WC_NO_BEST_FIT_CHARS, &wc2, 1, NULL, 0, NULL, bUsedDefaultChar);
 280     OK(ret == 1, "ret is %d", ret);
 281     if(bUsedDefaultChar) OK(*bUsedDefaultChar == TRUE, "bUsedDefaultChar is %d", *bUsedDefaultChar);
 282     OK(GetLastError() == 0xdeadbeef, "GetLastError() is %lu", GetLastError());
 283
 284     ret = WideCharToMultiByte(1252, WC_NO_BEST_FIT_CHARS, wcs, -1, NULL, 0, NULL, bUsedDefaultChar);
 285     OK(ret == 5, "ret is %d", ret);
 286     if(bUsedDefaultChar) OK(*bUsedDefaultChar == TRUE, "bUsedDefaultChar is %d", *bUsedDefaultChar);
 287     OK(GetLastError() == 0xdeadbeef, "GetLastError() is %lu", GetLastError());
 288
 289     ret = WideCharToMultiByte(950, WC_NO_BEST_FIT_CHARS, dbwcs, 1, NULL, 0, NULL, bUsedDefaultChar);
 290     OK(ret == 2, "ret is %d", ret);
 291     if(bUsedDefaultChar) OK(*bUsedDefaultChar == FALSE, "bUsedDefaultChar is %d", *bUsedDefaultChar);
 292     OK(GetLastError() == 0xdeadbeef, "GetLastError() is %lu", GetLastError());
 293
 294     ret = WideCharToMultiByte(950, WC_NO_BEST_FIT_CHARS, dbwcs, -1, NULL, 0, NULL, bUsedDefaultChar);
 295     OK(ret == 5, "ret is %d", ret);
 296     if(bUsedDefaultChar) OK(*bUsedDefaultChar == FALSE, "bUsedDefaultChar is %d", *bUsedDefaultChar);
 297     OK(GetLastError() == 0xdeadbeef, "GetLastError() is %lu", GetLastError());
 298
 299     /* Abnormal uses of WideCharToMultiByte */
 300     ret = WideCharToMultiByte(1252, WC_NO_BEST_FIT_CHARS, NULL, 5, mbs, sizeof(mbs), NULL, bUsedDefaultChar);
 301     OK(ret == 0, "ret is %d", ret);
 302     if(bUsedDefaultChar) OK(*bUsedDefaultChar == FALSE, "bUsedDefaultChar is %d", *bUsedDefaultChar);
 303     OK(GetLastError() == ERROR_INVALID_PARAMETER, "GetLastError() is %lu", GetLastError());
 304     SetLastError(0xdeadbeef);
 305
 306     ret = WideCharToMultiByte(0, WC_NO_BEST_FIT_CHARS, dbwcs, 5, mbs, sizeof(mbs), NULL, bUsedDefaultChar);
 307     OK(ret == 5, "ret is %d", ret);
 308     OK(!strcmp(mbs, "??"), "mbs is %s", mbs);
 309     if(bUsedDefaultChar) OK(*bUsedDefaultChar == TRUE, "bUsedDefaultChar is %d", *bUsedDefaultChar);
 310
 311     ret = WideCharToMultiByte(1252, WC_NO_BEST_FIT_CHARS, wcs, -1, (LPSTR)wcs, 5, NULL, bUsedDefaultChar);
 312     OK(ret == 0, "ret is %d", ret);
 313     OK(GetLastError() == ERROR_INVALID_PARAMETER, "GetLastError() is %lu", GetLastError());
 314     SetLastError(0xdeadbeef);
 315
 316     ret = WideCharToMultiByte(1252, WC_NO_BEST_FIT_CHARS, wcs, -1, mbs, -1, NULL, bUsedDefaultChar);
 317     OK(ret == 0, "ret is %d", ret);
 318     OK(GetLastError() == ERROR_INVALID_PARAMETER, "GetLastError() is %lu", GetLastError());
 319     SetLastError(0xdeadbeef);
 320
 321     putchar('\n');
 322 }
 323
 324 int main()
 325 {
 326     BOOL UsedDefaultChar;
 327
 328     CRT_Tests();
 329
 330     /* There are two code pathes in WideCharToMultiByte, one when Flags || DefaultChar || UsedDefaultChar is set and one when it's not.
 331        Test both here. */
 332     Win32_Tests(NULL);
 333     Win32_Tests(&UsedDefaultChar);
 334
 335     return 0;
 336 }