| 1 | /* Test of conversion of multibyte character to wide character.
|
|---|
| 2 | Copyright (C) 2008-2022 Free Software Foundation, Inc.
|
|---|
| 3 |
|
|---|
| 4 | This program is free software: you can redistribute it and/or modify
|
|---|
| 5 | it under the terms of the GNU General Public License as published by
|
|---|
| 6 | the Free Software Foundation, either version 3 of the License, or
|
|---|
| 7 | (at your option) any later version.
|
|---|
| 8 |
|
|---|
| 9 | This program is distributed in the hope that it will be useful,
|
|---|
| 10 | but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|---|
| 11 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|---|
| 12 | GNU General Public License for more details.
|
|---|
| 13 |
|
|---|
| 14 | You should have received a copy of the GNU General Public License
|
|---|
| 15 | along with this program. If not, see <https://www.gnu.org/licenses/>. */
|
|---|
| 16 |
|
|---|
| 17 | #include <config.h>
|
|---|
| 18 |
|
|---|
| 19 | #include <wchar.h>
|
|---|
| 20 |
|
|---|
| 21 | #include <errno.h>
|
|---|
| 22 | #include <locale.h>
|
|---|
| 23 | #include <stdio.h>
|
|---|
| 24 | #include <stdlib.h>
|
|---|
| 25 | #include <string.h>
|
|---|
| 26 |
|
|---|
| 27 | #include "localcharset.h"
|
|---|
| 28 | #include "macros.h"
|
|---|
| 29 |
|
|---|
| 30 | #if defined _WIN32 && !defined __CYGWIN__
|
|---|
| 31 |
|
|---|
| 32 | static int
|
|---|
| 33 | test_one_locale (const char *name, int codepage)
|
|---|
| 34 | {
|
|---|
| 35 | mbstate_t state;
|
|---|
| 36 | wchar_t wc;
|
|---|
| 37 | size_t ret;
|
|---|
| 38 |
|
|---|
| 39 | # if 1
|
|---|
| 40 | /* Portable code to set the locale. */
|
|---|
| 41 | {
|
|---|
| 42 | char name_with_codepage[1024];
|
|---|
| 43 |
|
|---|
| 44 | sprintf (name_with_codepage, "%s.%d", name, codepage);
|
|---|
| 45 |
|
|---|
| 46 | /* Set the locale. */
|
|---|
| 47 | if (setlocale (LC_ALL, name_with_codepage) == NULL)
|
|---|
| 48 | return 77;
|
|---|
| 49 | }
|
|---|
| 50 | # else
|
|---|
| 51 | /* Hacky way to set a locale.codepage combination that setlocale() refuses
|
|---|
| 52 | to set. */
|
|---|
| 53 | {
|
|---|
| 54 | /* Codepage of the current locale, set with setlocale().
|
|---|
| 55 | Not necessarily the same as GetACP(). */
|
|---|
| 56 | extern __declspec(dllimport) unsigned int __lc_codepage;
|
|---|
| 57 |
|
|---|
| 58 | /* Set the locale. */
|
|---|
| 59 | if (setlocale (LC_ALL, name) == NULL)
|
|---|
| 60 | return 77;
|
|---|
| 61 |
|
|---|
| 62 | /* Clobber the codepage and MB_CUR_MAX, both set by setlocale(). */
|
|---|
| 63 | __lc_codepage = codepage;
|
|---|
| 64 | switch (codepage)
|
|---|
| 65 | {
|
|---|
| 66 | case 1252:
|
|---|
| 67 | case 1256:
|
|---|
| 68 | MB_CUR_MAX = 1;
|
|---|
| 69 | break;
|
|---|
| 70 | case 932:
|
|---|
| 71 | case 950:
|
|---|
| 72 | case 936:
|
|---|
| 73 | MB_CUR_MAX = 2;
|
|---|
| 74 | break;
|
|---|
| 75 | case 54936:
|
|---|
| 76 | case 65001:
|
|---|
| 77 | MB_CUR_MAX = 4;
|
|---|
| 78 | break;
|
|---|
| 79 | }
|
|---|
| 80 |
|
|---|
| 81 | /* Test whether the codepage is really available. */
|
|---|
| 82 | memset (&state, '\0', sizeof (mbstate_t));
|
|---|
| 83 | if (mbrtowc (&wc, " ", 1, &state) == (size_t)(-1))
|
|---|
| 84 | return 77;
|
|---|
| 85 | }
|
|---|
| 86 | # endif
|
|---|
| 87 |
|
|---|
| 88 | /* Test zero-length input. */
|
|---|
| 89 | {
|
|---|
| 90 | memset (&state, '\0', sizeof (mbstate_t));
|
|---|
| 91 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 92 | ret = mbrtowc (&wc, "x", 0, &state);
|
|---|
| 93 | /* gnulib's implementation returns (size_t)(-2).
|
|---|
| 94 | The AIX 5.1 implementation returns (size_t)(-1).
|
|---|
| 95 | glibc's implementation returns 0. */
|
|---|
| 96 | ASSERT (ret == (size_t)(-2) || ret == (size_t)(-1) || ret == 0);
|
|---|
| 97 | ASSERT (mbsinit (&state));
|
|---|
| 98 | }
|
|---|
| 99 |
|
|---|
| 100 | /* Test NUL byte input. */
|
|---|
| 101 | {
|
|---|
| 102 | memset (&state, '\0', sizeof (mbstate_t));
|
|---|
| 103 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 104 | ret = mbrtowc (&wc, "", 1, &state);
|
|---|
| 105 | ASSERT (ret == 0);
|
|---|
| 106 | ASSERT (wc == 0);
|
|---|
| 107 | ASSERT (mbsinit (&state));
|
|---|
| 108 | ret = mbrtowc (NULL, "", 1, &state);
|
|---|
| 109 | ASSERT (ret == 0);
|
|---|
| 110 | ASSERT (mbsinit (&state));
|
|---|
| 111 | }
|
|---|
| 112 |
|
|---|
| 113 | /* Test single-byte input. */
|
|---|
| 114 | {
|
|---|
| 115 | int c;
|
|---|
| 116 | char buf[1];
|
|---|
| 117 |
|
|---|
| 118 | memset (&state, '\0', sizeof (mbstate_t));
|
|---|
| 119 | for (c = 0; c < 0x100; c++)
|
|---|
| 120 | switch (c)
|
|---|
| 121 | {
|
|---|
| 122 | case '\t': case '\v': case '\f':
|
|---|
| 123 | case ' ': case '!': case '"': case '#': case '%':
|
|---|
| 124 | case '&': case '\'': case '(': case ')': case '*':
|
|---|
| 125 | case '+': case ',': case '-': case '.': case '/':
|
|---|
| 126 | case '0': case '1': case '2': case '3': case '4':
|
|---|
| 127 | case '5': case '6': case '7': case '8': case '9':
|
|---|
| 128 | case ':': case ';': case '<': case '=': case '>':
|
|---|
| 129 | case '?':
|
|---|
| 130 | case 'A': case 'B': case 'C': case 'D': case 'E':
|
|---|
| 131 | case 'F': case 'G': case 'H': case 'I': case 'J':
|
|---|
| 132 | case 'K': case 'L': case 'M': case 'N': case 'O':
|
|---|
| 133 | case 'P': case 'Q': case 'R': case 'S': case 'T':
|
|---|
| 134 | case 'U': case 'V': case 'W': case 'X': case 'Y':
|
|---|
| 135 | case 'Z':
|
|---|
| 136 | case '[': case '\\': case ']': case '^': case '_':
|
|---|
| 137 | case 'a': case 'b': case 'c': case 'd': case 'e':
|
|---|
| 138 | case 'f': case 'g': case 'h': case 'i': case 'j':
|
|---|
| 139 | case 'k': case 'l': case 'm': case 'n': case 'o':
|
|---|
| 140 | case 'p': case 'q': case 'r': case 's': case 't':
|
|---|
| 141 | case 'u': case 'v': case 'w': case 'x': case 'y':
|
|---|
| 142 | case 'z': case '{': case '|': case '}': case '~':
|
|---|
| 143 | /* c is in the ISO C "basic character set". */
|
|---|
| 144 | buf[0] = c;
|
|---|
| 145 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 146 | ret = mbrtowc (&wc, buf, 1, &state);
|
|---|
| 147 | ASSERT (ret == 1);
|
|---|
| 148 | ASSERT (wc == c);
|
|---|
| 149 | ASSERT (mbsinit (&state));
|
|---|
| 150 | ret = mbrtowc (NULL, buf, 1, &state);
|
|---|
| 151 | ASSERT (ret == 1);
|
|---|
| 152 | ASSERT (mbsinit (&state));
|
|---|
| 153 | break;
|
|---|
| 154 | }
|
|---|
| 155 | }
|
|---|
| 156 |
|
|---|
| 157 | /* Test special calling convention, passing a NULL pointer. */
|
|---|
| 158 | {
|
|---|
| 159 | memset (&state, '\0', sizeof (mbstate_t));
|
|---|
| 160 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 161 | ret = mbrtowc (&wc, NULL, 5, &state);
|
|---|
| 162 | ASSERT (ret == 0);
|
|---|
| 163 | ASSERT (wc == (wchar_t) 0xBADFACE);
|
|---|
| 164 | ASSERT (mbsinit (&state));
|
|---|
| 165 | }
|
|---|
| 166 |
|
|---|
| 167 | switch (codepage)
|
|---|
| 168 | {
|
|---|
| 169 | case 1252:
|
|---|
| 170 | /* Locale encoding is CP1252, an extension of ISO-8859-1. */
|
|---|
| 171 | {
|
|---|
| 172 | char input[] = "B\374\337er"; /* "BÃŒÃer" */
|
|---|
| 173 | memset (&state, '\0', sizeof (mbstate_t));
|
|---|
| 174 |
|
|---|
| 175 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 176 | ret = mbrtowc (&wc, input, 1, &state);
|
|---|
| 177 | ASSERT (ret == 1);
|
|---|
| 178 | ASSERT (wc == 'B');
|
|---|
| 179 | ASSERT (mbsinit (&state));
|
|---|
| 180 | input[0] = '\0';
|
|---|
| 181 |
|
|---|
| 182 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 183 | ret = mbrtowc (&wc, input + 1, 1, &state);
|
|---|
| 184 | ASSERT (ret == 1);
|
|---|
| 185 | ASSERT (wctob (wc) == (unsigned char) '\374');
|
|---|
| 186 | ASSERT (wc == 0x00FC);
|
|---|
| 187 | ASSERT (mbsinit (&state));
|
|---|
| 188 | input[1] = '\0';
|
|---|
| 189 |
|
|---|
| 190 | /* Test support of NULL first argument. */
|
|---|
| 191 | ret = mbrtowc (NULL, input + 2, 3, &state);
|
|---|
| 192 | ASSERT (ret == 1);
|
|---|
| 193 | ASSERT (mbsinit (&state));
|
|---|
| 194 |
|
|---|
| 195 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 196 | ret = mbrtowc (&wc, input + 2, 3, &state);
|
|---|
| 197 | ASSERT (ret == 1);
|
|---|
| 198 | ASSERT (wctob (wc) == (unsigned char) '\337');
|
|---|
| 199 | ASSERT (wc == 0x00DF);
|
|---|
| 200 | ASSERT (mbsinit (&state));
|
|---|
| 201 | input[2] = '\0';
|
|---|
| 202 |
|
|---|
| 203 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 204 | ret = mbrtowc (&wc, input + 3, 2, &state);
|
|---|
| 205 | ASSERT (ret == 1);
|
|---|
| 206 | ASSERT (wc == 'e');
|
|---|
| 207 | ASSERT (mbsinit (&state));
|
|---|
| 208 | input[3] = '\0';
|
|---|
| 209 |
|
|---|
| 210 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 211 | ret = mbrtowc (&wc, input + 4, 1, &state);
|
|---|
| 212 | ASSERT (ret == 1);
|
|---|
| 213 | ASSERT (wc == 'r');
|
|---|
| 214 | ASSERT (mbsinit (&state));
|
|---|
| 215 | }
|
|---|
| 216 | return 0;
|
|---|
| 217 |
|
|---|
| 218 | case 1256:
|
|---|
| 219 | /* Locale encoding is CP1256, not the same as ISO-8859-6. */
|
|---|
| 220 | {
|
|---|
| 221 | char input[] = "x\302\341\346y"; /* "xØ¢ÙÙy" */
|
|---|
| 222 | memset (&state, '\0', sizeof (mbstate_t));
|
|---|
| 223 |
|
|---|
| 224 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 225 | ret = mbrtowc (&wc, input, 1, &state);
|
|---|
| 226 | ASSERT (ret == 1);
|
|---|
| 227 | ASSERT (wc == 'x');
|
|---|
| 228 | ASSERT (mbsinit (&state));
|
|---|
| 229 | input[0] = '\0';
|
|---|
| 230 |
|
|---|
| 231 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 232 | ret = mbrtowc (&wc, input + 1, 1, &state);
|
|---|
| 233 | ASSERT (ret == 1);
|
|---|
| 234 | ASSERT (wctob (wc) == (unsigned char) '\302');
|
|---|
| 235 | ASSERT (wc == 0x0622);
|
|---|
| 236 | ASSERT (mbsinit (&state));
|
|---|
| 237 | input[1] = '\0';
|
|---|
| 238 |
|
|---|
| 239 | /* Test support of NULL first argument. */
|
|---|
| 240 | ret = mbrtowc (NULL, input + 2, 3, &state);
|
|---|
| 241 | ASSERT (ret == 1);
|
|---|
| 242 | ASSERT (mbsinit (&state));
|
|---|
| 243 |
|
|---|
| 244 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 245 | ret = mbrtowc (&wc, input + 2, 3, &state);
|
|---|
| 246 | ASSERT (ret == 1);
|
|---|
| 247 | ASSERT (wctob (wc) == (unsigned char) '\341');
|
|---|
| 248 | ASSERT (wc == 0x0644);
|
|---|
| 249 | ASSERT (mbsinit (&state));
|
|---|
| 250 | input[2] = '\0';
|
|---|
| 251 |
|
|---|
| 252 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 253 | ret = mbrtowc (&wc, input + 3, 2, &state);
|
|---|
| 254 | ASSERT (ret == 1);
|
|---|
| 255 | ASSERT (wctob (wc) == (unsigned char) '\346');
|
|---|
| 256 | ASSERT (wc == 0x0648);
|
|---|
| 257 | ASSERT (mbsinit (&state));
|
|---|
| 258 | input[3] = '\0';
|
|---|
| 259 |
|
|---|
| 260 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 261 | ret = mbrtowc (&wc, input + 4, 1, &state);
|
|---|
| 262 | ASSERT (ret == 1);
|
|---|
| 263 | ASSERT (wc == 'y');
|
|---|
| 264 | ASSERT (mbsinit (&state));
|
|---|
| 265 | }
|
|---|
| 266 | return 0;
|
|---|
| 267 |
|
|---|
| 268 | case 932:
|
|---|
| 269 | /* Locale encoding is CP932, similar to Shift_JIS. */
|
|---|
| 270 | {
|
|---|
| 271 | char input[] = "<\223\372\226\173\214\352>"; /* "<æ¥æ¬èª>" */
|
|---|
| 272 | memset (&state, '\0', sizeof (mbstate_t));
|
|---|
| 273 |
|
|---|
| 274 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 275 | ret = mbrtowc (&wc, input, 1, &state);
|
|---|
| 276 | ASSERT (ret == 1);
|
|---|
| 277 | ASSERT (wc == '<');
|
|---|
| 278 | ASSERT (mbsinit (&state));
|
|---|
| 279 | input[0] = '\0';
|
|---|
| 280 |
|
|---|
| 281 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 282 | ret = mbrtowc (&wc, input + 1, 2, &state);
|
|---|
| 283 | ASSERT (ret == 2);
|
|---|
| 284 | ASSERT (wctob (wc) == EOF);
|
|---|
| 285 | ASSERT (wc == 0x65E5);
|
|---|
| 286 | ASSERT (mbsinit (&state));
|
|---|
| 287 | input[1] = '\0';
|
|---|
| 288 | input[2] = '\0';
|
|---|
| 289 |
|
|---|
| 290 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 291 | ret = mbrtowc (&wc, input + 3, 1, &state);
|
|---|
| 292 | ASSERT (ret == (size_t)(-2));
|
|---|
| 293 | ASSERT (wc == (wchar_t) 0xBADFACE);
|
|---|
| 294 | ASSERT (!mbsinit (&state));
|
|---|
| 295 | input[3] = '\0';
|
|---|
| 296 |
|
|---|
| 297 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 298 | ret = mbrtowc (&wc, input + 4, 4, &state);
|
|---|
| 299 | ASSERT (ret == 1);
|
|---|
| 300 | ASSERT (wctob (wc) == EOF);
|
|---|
| 301 | ASSERT (wc == 0x672C);
|
|---|
| 302 | ASSERT (mbsinit (&state));
|
|---|
| 303 | input[4] = '\0';
|
|---|
| 304 |
|
|---|
| 305 | /* Test support of NULL first argument. */
|
|---|
| 306 | ret = mbrtowc (NULL, input + 5, 3, &state);
|
|---|
| 307 | ASSERT (ret == 2);
|
|---|
| 308 | ASSERT (mbsinit (&state));
|
|---|
| 309 |
|
|---|
| 310 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 311 | ret = mbrtowc (&wc, input + 5, 3, &state);
|
|---|
| 312 | ASSERT (ret == 2);
|
|---|
| 313 | ASSERT (wctob (wc) == EOF);
|
|---|
| 314 | ASSERT (wc == 0x8A9E);
|
|---|
| 315 | ASSERT (mbsinit (&state));
|
|---|
| 316 | input[5] = '\0';
|
|---|
| 317 | input[6] = '\0';
|
|---|
| 318 |
|
|---|
| 319 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 320 | ret = mbrtowc (&wc, input + 7, 1, &state);
|
|---|
| 321 | ASSERT (ret == 1);
|
|---|
| 322 | ASSERT (wc == '>');
|
|---|
| 323 | ASSERT (mbsinit (&state));
|
|---|
| 324 |
|
|---|
| 325 | /* Test some invalid input. */
|
|---|
| 326 | memset (&state, '\0', sizeof (mbstate_t));
|
|---|
| 327 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 328 | ret = mbrtowc (&wc, "\377", 1, &state); /* 0xFF */
|
|---|
| 329 | ASSERT ((ret == (size_t)-1 && errno == EILSEQ) || ret == (size_t)-2);
|
|---|
| 330 |
|
|---|
| 331 | memset (&state, '\0', sizeof (mbstate_t));
|
|---|
| 332 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 333 | ret = mbrtowc (&wc, "\225\377", 2, &state); /* 0x95 0xFF */
|
|---|
| 334 | ASSERT ((ret == (size_t)-1 && errno == EILSEQ) || (ret == 2 && wc == 0x30FB));
|
|---|
| 335 | }
|
|---|
| 336 | return 0;
|
|---|
| 337 |
|
|---|
| 338 | case 950:
|
|---|
| 339 | /* Locale encoding is CP950, similar to Big5. */
|
|---|
| 340 | {
|
|---|
| 341 | char input[] = "<\244\351\245\273\273\171>"; /* "<æ¥æ¬èª>" */
|
|---|
| 342 | memset (&state, '\0', sizeof (mbstate_t));
|
|---|
| 343 |
|
|---|
| 344 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 345 | ret = mbrtowc (&wc, input, 1, &state);
|
|---|
| 346 | ASSERT (ret == 1);
|
|---|
| 347 | ASSERT (wc == '<');
|
|---|
| 348 | ASSERT (mbsinit (&state));
|
|---|
| 349 | input[0] = '\0';
|
|---|
| 350 |
|
|---|
| 351 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 352 | ret = mbrtowc (&wc, input + 1, 2, &state);
|
|---|
| 353 | ASSERT (ret == 2);
|
|---|
| 354 | ASSERT (wctob (wc) == EOF);
|
|---|
| 355 | ASSERT (wc == 0x65E5);
|
|---|
| 356 | ASSERT (mbsinit (&state));
|
|---|
| 357 | input[1] = '\0';
|
|---|
| 358 | input[2] = '\0';
|
|---|
| 359 |
|
|---|
| 360 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 361 | ret = mbrtowc (&wc, input + 3, 1, &state);
|
|---|
| 362 | ASSERT (ret == (size_t)(-2));
|
|---|
| 363 | ASSERT (wc == (wchar_t) 0xBADFACE);
|
|---|
| 364 | ASSERT (!mbsinit (&state));
|
|---|
| 365 | input[3] = '\0';
|
|---|
| 366 |
|
|---|
| 367 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 368 | ret = mbrtowc (&wc, input + 4, 4, &state);
|
|---|
| 369 | ASSERT (ret == 1);
|
|---|
| 370 | ASSERT (wctob (wc) == EOF);
|
|---|
| 371 | ASSERT (wc == 0x672C);
|
|---|
| 372 | ASSERT (mbsinit (&state));
|
|---|
| 373 | input[4] = '\0';
|
|---|
| 374 |
|
|---|
| 375 | /* Test support of NULL first argument. */
|
|---|
| 376 | ret = mbrtowc (NULL, input + 5, 3, &state);
|
|---|
| 377 | ASSERT (ret == 2);
|
|---|
| 378 | ASSERT (mbsinit (&state));
|
|---|
| 379 |
|
|---|
| 380 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 381 | ret = mbrtowc (&wc, input + 5, 3, &state);
|
|---|
| 382 | ASSERT (ret == 2);
|
|---|
| 383 | ASSERT (wctob (wc) == EOF);
|
|---|
| 384 | ASSERT (wc == 0x8A9E);
|
|---|
| 385 | ASSERT (mbsinit (&state));
|
|---|
| 386 | input[5] = '\0';
|
|---|
| 387 | input[6] = '\0';
|
|---|
| 388 |
|
|---|
| 389 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 390 | ret = mbrtowc (&wc, input + 7, 1, &state);
|
|---|
| 391 | ASSERT (ret == 1);
|
|---|
| 392 | ASSERT (wc == '>');
|
|---|
| 393 | ASSERT (mbsinit (&state));
|
|---|
| 394 |
|
|---|
| 395 | /* Test some invalid input. */
|
|---|
| 396 | memset (&state, '\0', sizeof (mbstate_t));
|
|---|
| 397 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 398 | ret = mbrtowc (&wc, "\377", 1, &state); /* 0xFF */
|
|---|
| 399 | ASSERT ((ret == (size_t)-1 && errno == EILSEQ) || ret == (size_t)-2);
|
|---|
| 400 |
|
|---|
| 401 | memset (&state, '\0', sizeof (mbstate_t));
|
|---|
| 402 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 403 | ret = mbrtowc (&wc, "\225\377", 2, &state); /* 0x95 0xFF */
|
|---|
| 404 | ASSERT ((ret == (size_t)-1 && errno == EILSEQ) || (ret == 2 && wc == '?'));
|
|---|
| 405 | }
|
|---|
| 406 | return 0;
|
|---|
| 407 |
|
|---|
| 408 | case 936:
|
|---|
| 409 | /* Locale encoding is CP936 = GBK, an extension of GB2312. */
|
|---|
| 410 | {
|
|---|
| 411 | char input[] = "<\310\325\261\276\325\132>"; /* "<æ¥æ¬èª>" */
|
|---|
| 412 | memset (&state, '\0', sizeof (mbstate_t));
|
|---|
| 413 |
|
|---|
| 414 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 415 | ret = mbrtowc (&wc, input, 1, &state);
|
|---|
| 416 | ASSERT (ret == 1);
|
|---|
| 417 | ASSERT (wc == '<');
|
|---|
| 418 | ASSERT (mbsinit (&state));
|
|---|
| 419 | input[0] = '\0';
|
|---|
| 420 |
|
|---|
| 421 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 422 | ret = mbrtowc (&wc, input + 1, 2, &state);
|
|---|
| 423 | ASSERT (ret == 2);
|
|---|
| 424 | ASSERT (wctob (wc) == EOF);
|
|---|
| 425 | ASSERT (wc == 0x65E5);
|
|---|
| 426 | ASSERT (mbsinit (&state));
|
|---|
| 427 | input[1] = '\0';
|
|---|
| 428 | input[2] = '\0';
|
|---|
| 429 |
|
|---|
| 430 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 431 | ret = mbrtowc (&wc, input + 3, 1, &state);
|
|---|
| 432 | ASSERT (ret == (size_t)(-2));
|
|---|
| 433 | ASSERT (wc == (wchar_t) 0xBADFACE);
|
|---|
| 434 | ASSERT (!mbsinit (&state));
|
|---|
| 435 | input[3] = '\0';
|
|---|
| 436 |
|
|---|
| 437 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 438 | ret = mbrtowc (&wc, input + 4, 4, &state);
|
|---|
| 439 | ASSERT (ret == 1);
|
|---|
| 440 | ASSERT (wctob (wc) == EOF);
|
|---|
| 441 | ASSERT (wc == 0x672C);
|
|---|
| 442 | ASSERT (mbsinit (&state));
|
|---|
| 443 | input[4] = '\0';
|
|---|
| 444 |
|
|---|
| 445 | /* Test support of NULL first argument. */
|
|---|
| 446 | ret = mbrtowc (NULL, input + 5, 3, &state);
|
|---|
| 447 | ASSERT (ret == 2);
|
|---|
| 448 | ASSERT (mbsinit (&state));
|
|---|
| 449 |
|
|---|
| 450 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 451 | ret = mbrtowc (&wc, input + 5, 3, &state);
|
|---|
| 452 | ASSERT (ret == 2);
|
|---|
| 453 | ASSERT (wctob (wc) == EOF);
|
|---|
| 454 | ASSERT (wc == 0x8A9E);
|
|---|
| 455 | ASSERT (mbsinit (&state));
|
|---|
| 456 | input[5] = '\0';
|
|---|
| 457 | input[6] = '\0';
|
|---|
| 458 |
|
|---|
| 459 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 460 | ret = mbrtowc (&wc, input + 7, 1, &state);
|
|---|
| 461 | ASSERT (ret == 1);
|
|---|
| 462 | ASSERT (wc == '>');
|
|---|
| 463 | ASSERT (mbsinit (&state));
|
|---|
| 464 |
|
|---|
| 465 | /* Test some invalid input. */
|
|---|
| 466 | memset (&state, '\0', sizeof (mbstate_t));
|
|---|
| 467 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 468 | ret = mbrtowc (&wc, "\377", 1, &state); /* 0xFF */
|
|---|
| 469 | ASSERT ((ret == (size_t)-1 && errno == EILSEQ) || ret == (size_t)-2);
|
|---|
| 470 |
|
|---|
| 471 | memset (&state, '\0', sizeof (mbstate_t));
|
|---|
| 472 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 473 | ret = mbrtowc (&wc, "\225\377", 2, &state); /* 0x95 0xFF */
|
|---|
| 474 | ASSERT ((ret == (size_t)-1 && errno == EILSEQ) || (ret == 2 && wc == '?'));
|
|---|
| 475 | }
|
|---|
| 476 | return 0;
|
|---|
| 477 |
|
|---|
| 478 | case 54936:
|
|---|
| 479 | /* Locale encoding is CP54936 = GB18030. */
|
|---|
| 480 | if (strcmp (locale_charset (), "GB18030") != 0)
|
|---|
| 481 | return 77;
|
|---|
| 482 | {
|
|---|
| 483 | char input[] = "B\250\271\201\060\211\070er"; /* "BÃŒÃer" */
|
|---|
| 484 | memset (&state, '\0', sizeof (mbstate_t));
|
|---|
| 485 |
|
|---|
| 486 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 487 | ret = mbrtowc (&wc, input, 1, &state);
|
|---|
| 488 | ASSERT (ret == 1);
|
|---|
| 489 | ASSERT (wc == 'B');
|
|---|
| 490 | ASSERT (mbsinit (&state));
|
|---|
| 491 | input[0] = '\0';
|
|---|
| 492 |
|
|---|
| 493 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 494 | ret = mbrtowc (&wc, input + 1, 1, &state);
|
|---|
| 495 | ASSERT (ret == (size_t)(-2));
|
|---|
| 496 | ASSERT (wc == (wchar_t) 0xBADFACE);
|
|---|
| 497 | ASSERT (!mbsinit (&state));
|
|---|
| 498 | input[1] = '\0';
|
|---|
| 499 |
|
|---|
| 500 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 501 | ret = mbrtowc (&wc, input + 2, 7, &state);
|
|---|
| 502 | ASSERT (ret == 1);
|
|---|
| 503 | ASSERT (wctob (wc) == EOF);
|
|---|
| 504 | ASSERT (wc == 0x00FC);
|
|---|
| 505 | ASSERT (mbsinit (&state));
|
|---|
| 506 | input[2] = '\0';
|
|---|
| 507 |
|
|---|
| 508 | /* Test support of NULL first argument. */
|
|---|
| 509 | ret = mbrtowc (NULL, input + 3, 6, &state);
|
|---|
| 510 | ASSERT (ret == 4);
|
|---|
| 511 | ASSERT (mbsinit (&state));
|
|---|
| 512 |
|
|---|
| 513 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 514 | ret = mbrtowc (&wc, input + 3, 6, &state);
|
|---|
| 515 | ASSERT (ret == 4);
|
|---|
| 516 | ASSERT (wctob (wc) == EOF);
|
|---|
| 517 | ASSERT (wc == 0x00DF);
|
|---|
| 518 | ASSERT (mbsinit (&state));
|
|---|
| 519 | input[3] = '\0';
|
|---|
| 520 | input[4] = '\0';
|
|---|
| 521 | input[5] = '\0';
|
|---|
| 522 | input[6] = '\0';
|
|---|
| 523 |
|
|---|
| 524 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 525 | ret = mbrtowc (&wc, input + 7, 2, &state);
|
|---|
| 526 | ASSERT (ret == 1);
|
|---|
| 527 | ASSERT (wc == 'e');
|
|---|
| 528 | ASSERT (mbsinit (&state));
|
|---|
| 529 | input[7] = '\0';
|
|---|
| 530 |
|
|---|
| 531 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 532 | ret = mbrtowc (&wc, input + 8, 1, &state);
|
|---|
| 533 | ASSERT (ret == 1);
|
|---|
| 534 | ASSERT (wc == 'r');
|
|---|
| 535 | ASSERT (mbsinit (&state));
|
|---|
| 536 |
|
|---|
| 537 | /* Test some invalid input. */
|
|---|
| 538 | memset (&state, '\0', sizeof (mbstate_t));
|
|---|
| 539 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 540 | ret = mbrtowc (&wc, "\377", 1, &state); /* 0xFF */
|
|---|
| 541 | ASSERT (ret == (size_t)-1);
|
|---|
| 542 | ASSERT (errno == EILSEQ);
|
|---|
| 543 |
|
|---|
| 544 | memset (&state, '\0', sizeof (mbstate_t));
|
|---|
| 545 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 546 | ret = mbrtowc (&wc, "\225\377", 2, &state); /* 0x95 0xFF */
|
|---|
| 547 | ASSERT (ret == (size_t)-1);
|
|---|
| 548 | ASSERT (errno == EILSEQ);
|
|---|
| 549 |
|
|---|
| 550 | memset (&state, '\0', sizeof (mbstate_t));
|
|---|
| 551 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 552 | ret = mbrtowc (&wc, "\201\045", 2, &state); /* 0x81 0x25 */
|
|---|
| 553 | ASSERT (ret == (size_t)-1);
|
|---|
| 554 | ASSERT (errno == EILSEQ);
|
|---|
| 555 |
|
|---|
| 556 | memset (&state, '\0', sizeof (mbstate_t));
|
|---|
| 557 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 558 | ret = mbrtowc (&wc, "\201\060\377", 3, &state); /* 0x81 0x30 0xFF */
|
|---|
| 559 | ASSERT (ret == (size_t)-1);
|
|---|
| 560 | ASSERT (errno == EILSEQ);
|
|---|
| 561 |
|
|---|
| 562 | memset (&state, '\0', sizeof (mbstate_t));
|
|---|
| 563 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 564 | ret = mbrtowc (&wc, "\201\060\377\064", 4, &state); /* 0x81 0x30 0xFF 0x34 */
|
|---|
| 565 | ASSERT (ret == (size_t)-1);
|
|---|
| 566 | ASSERT (errno == EILSEQ);
|
|---|
| 567 |
|
|---|
| 568 | memset (&state, '\0', sizeof (mbstate_t));
|
|---|
| 569 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 570 | ret = mbrtowc (&wc, "\201\060\211\072", 4, &state); /* 0x81 0x30 0x89 0x3A */
|
|---|
| 571 | ASSERT (ret == (size_t)-1);
|
|---|
| 572 | ASSERT (errno == EILSEQ);
|
|---|
| 573 | }
|
|---|
| 574 | return 0;
|
|---|
| 575 |
|
|---|
| 576 | case 65001:
|
|---|
| 577 | /* Locale encoding is CP65001 = UTF-8. */
|
|---|
| 578 | if (strcmp (locale_charset (), "UTF-8") != 0)
|
|---|
| 579 | return 77;
|
|---|
| 580 | {
|
|---|
| 581 | char input[] = "B\303\274\303\237er"; /* "BÃŒÃer" */
|
|---|
| 582 | memset (&state, '\0', sizeof (mbstate_t));
|
|---|
| 583 |
|
|---|
| 584 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 585 | ret = mbrtowc (&wc, input, 1, &state);
|
|---|
| 586 | ASSERT (ret == 1);
|
|---|
| 587 | ASSERT (wc == 'B');
|
|---|
| 588 | ASSERT (mbsinit (&state));
|
|---|
| 589 | input[0] = '\0';
|
|---|
| 590 |
|
|---|
| 591 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 592 | ret = mbrtowc (&wc, input + 1, 1, &state);
|
|---|
| 593 | ASSERT (ret == (size_t)(-2));
|
|---|
| 594 | ASSERT (wc == (wchar_t) 0xBADFACE);
|
|---|
| 595 | ASSERT (!mbsinit (&state));
|
|---|
| 596 | input[1] = '\0';
|
|---|
| 597 |
|
|---|
| 598 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 599 | ret = mbrtowc (&wc, input + 2, 5, &state);
|
|---|
| 600 | ASSERT (ret == 1);
|
|---|
| 601 | ASSERT (wctob (wc) == EOF);
|
|---|
| 602 | ASSERT (wc == 0x00FC);
|
|---|
| 603 | ASSERT (mbsinit (&state));
|
|---|
| 604 | input[2] = '\0';
|
|---|
| 605 |
|
|---|
| 606 | /* Test support of NULL first argument. */
|
|---|
| 607 | ret = mbrtowc (NULL, input + 3, 4, &state);
|
|---|
| 608 | ASSERT (ret == 2);
|
|---|
| 609 | ASSERT (mbsinit (&state));
|
|---|
| 610 |
|
|---|
| 611 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 612 | ret = mbrtowc (&wc, input + 3, 4, &state);
|
|---|
| 613 | ASSERT (ret == 2);
|
|---|
| 614 | ASSERT (wctob (wc) == EOF);
|
|---|
| 615 | ASSERT (wc == 0x00DF);
|
|---|
| 616 | ASSERT (mbsinit (&state));
|
|---|
| 617 | input[3] = '\0';
|
|---|
| 618 | input[4] = '\0';
|
|---|
| 619 |
|
|---|
| 620 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 621 | ret = mbrtowc (&wc, input + 5, 2, &state);
|
|---|
| 622 | ASSERT (ret == 1);
|
|---|
| 623 | ASSERT (wc == 'e');
|
|---|
| 624 | ASSERT (mbsinit (&state));
|
|---|
| 625 | input[5] = '\0';
|
|---|
| 626 |
|
|---|
| 627 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 628 | ret = mbrtowc (&wc, input + 6, 1, &state);
|
|---|
| 629 | ASSERT (ret == 1);
|
|---|
| 630 | ASSERT (wc == 'r');
|
|---|
| 631 | ASSERT (mbsinit (&state));
|
|---|
| 632 |
|
|---|
| 633 | /* Test some invalid input. */
|
|---|
| 634 | memset (&state, '\0', sizeof (mbstate_t));
|
|---|
| 635 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 636 | ret = mbrtowc (&wc, "\377", 1, &state); /* 0xFF */
|
|---|
| 637 | ASSERT (ret == (size_t)-1);
|
|---|
| 638 | ASSERT (errno == EILSEQ);
|
|---|
| 639 |
|
|---|
| 640 | memset (&state, '\0', sizeof (mbstate_t));
|
|---|
| 641 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 642 | ret = mbrtowc (&wc, "\303\300", 2, &state); /* 0xC3 0xC0 */
|
|---|
| 643 | ASSERT (ret == (size_t)-1);
|
|---|
| 644 | ASSERT (errno == EILSEQ);
|
|---|
| 645 |
|
|---|
| 646 | memset (&state, '\0', sizeof (mbstate_t));
|
|---|
| 647 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 648 | ret = mbrtowc (&wc, "\343\300", 2, &state); /* 0xE3 0xC0 */
|
|---|
| 649 | ASSERT (ret == (size_t)-1);
|
|---|
| 650 | ASSERT (errno == EILSEQ);
|
|---|
| 651 |
|
|---|
| 652 | memset (&state, '\0', sizeof (mbstate_t));
|
|---|
| 653 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 654 | ret = mbrtowc (&wc, "\343\300\200", 3, &state); /* 0xE3 0xC0 0x80 */
|
|---|
| 655 | ASSERT (ret == (size_t)-1);
|
|---|
| 656 | ASSERT (errno == EILSEQ);
|
|---|
| 657 |
|
|---|
| 658 | memset (&state, '\0', sizeof (mbstate_t));
|
|---|
| 659 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 660 | ret = mbrtowc (&wc, "\343\200\300", 3, &state); /* 0xE3 0x80 0xC0 */
|
|---|
| 661 | ASSERT (ret == (size_t)-1);
|
|---|
| 662 | ASSERT (errno == EILSEQ);
|
|---|
| 663 |
|
|---|
| 664 | memset (&state, '\0', sizeof (mbstate_t));
|
|---|
| 665 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 666 | ret = mbrtowc (&wc, "\363\300", 2, &state); /* 0xF3 0xC0 */
|
|---|
| 667 | ASSERT (ret == (size_t)-1);
|
|---|
| 668 | ASSERT (errno == EILSEQ);
|
|---|
| 669 |
|
|---|
| 670 | memset (&state, '\0', sizeof (mbstate_t));
|
|---|
| 671 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 672 | ret = mbrtowc (&wc, "\363\300\200\200", 4, &state); /* 0xF3 0xC0 0x80 0x80 */
|
|---|
| 673 | ASSERT (ret == (size_t)-1);
|
|---|
| 674 | ASSERT (errno == EILSEQ);
|
|---|
| 675 |
|
|---|
| 676 | memset (&state, '\0', sizeof (mbstate_t));
|
|---|
| 677 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 678 | ret = mbrtowc (&wc, "\363\200\300", 3, &state); /* 0xF3 0x80 0xC0 */
|
|---|
| 679 | ASSERT (ret == (size_t)-1);
|
|---|
| 680 | ASSERT (errno == EILSEQ);
|
|---|
| 681 |
|
|---|
| 682 | memset (&state, '\0', sizeof (mbstate_t));
|
|---|
| 683 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 684 | ret = mbrtowc (&wc, "\363\200\300\200", 4, &state); /* 0xF3 0x80 0xC0 0x80 */
|
|---|
| 685 | ASSERT (ret == (size_t)-1);
|
|---|
| 686 | ASSERT (errno == EILSEQ);
|
|---|
| 687 |
|
|---|
| 688 | memset (&state, '\0', sizeof (mbstate_t));
|
|---|
| 689 | wc = (wchar_t) 0xBADFACE;
|
|---|
| 690 | ret = mbrtowc (&wc, "\363\200\200\300", 4, &state); /* 0xF3 0x80 0x80 0xC0 */
|
|---|
| 691 | ASSERT (ret == (size_t)-1);
|
|---|
| 692 | ASSERT (errno == EILSEQ);
|
|---|
| 693 | }
|
|---|
| 694 | return 0;
|
|---|
| 695 |
|
|---|
| 696 | default:
|
|---|
| 697 | return 1;
|
|---|
| 698 | }
|
|---|
| 699 | }
|
|---|
| 700 |
|
|---|
| 701 | int
|
|---|
| 702 | main (int argc, char *argv[])
|
|---|
| 703 | {
|
|---|
| 704 | int codepage = atoi (argv[argc - 1]);
|
|---|
| 705 | int result;
|
|---|
| 706 | int i;
|
|---|
| 707 |
|
|---|
| 708 | result = 77;
|
|---|
| 709 | for (i = 1; i < argc - 1; i++)
|
|---|
| 710 | {
|
|---|
| 711 | int ret = test_one_locale (argv[i], codepage);
|
|---|
| 712 |
|
|---|
| 713 | if (ret != 77)
|
|---|
| 714 | result = ret;
|
|---|
| 715 | }
|
|---|
| 716 |
|
|---|
| 717 | if (result == 77)
|
|---|
| 718 | {
|
|---|
| 719 | fprintf (stderr, "Skipping test: found no locale with codepage %d\n",
|
|---|
| 720 | codepage);
|
|---|
| 721 | }
|
|---|
| 722 | return result;
|
|---|
| 723 | }
|
|---|
| 724 |
|
|---|
| 725 | #else
|
|---|
| 726 |
|
|---|
| 727 | int
|
|---|
| 728 | main (int argc, char *argv[])
|
|---|
| 729 | {
|
|---|
| 730 | fputs ("Skipping test: not a native Windows system\n", stderr);
|
|---|
| 731 | return 77;
|
|---|
| 732 | }
|
|---|
| 733 |
|
|---|
| 734 | #endif
|
|---|