OLD | NEW |
(Empty) | |
| 1 /* Convert multibyte character to wide character. |
| 2 Copyright (C) 1999-2002, 2005-2012 Free Software Foundation, Inc. |
| 3 Written by Bruno Haible <bruno@clisp.org>, 2008. |
| 4 |
| 5 This program is free software: you can redistribute it and/or modify |
| 6 it under the terms of the GNU General Public License as published by |
| 7 the Free Software Foundation; either version 3 of the License, or |
| 8 (at your option) any later version. |
| 9 |
| 10 This program is distributed in the hope that it will be useful, |
| 11 but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 13 GNU General Public License for more details. |
| 14 |
| 15 You should have received a copy of the GNU General Public License |
| 16 along with this program. If not, see <http://www.gnu.org/licenses/>. */ |
| 17 |
| 18 #include <config.h> |
| 19 |
| 20 /* Specification. */ |
| 21 #include <wchar.h> |
| 22 |
| 23 #if GNULIB_defined_mbstate_t |
| 24 /* Implement mbrtowc() on top of mbtowc(). */ |
| 25 |
| 26 # include <errno.h> |
| 27 # include <stdlib.h> |
| 28 |
| 29 # include "localcharset.h" |
| 30 # include "streq.h" |
| 31 # include "verify.h" |
| 32 |
| 33 |
| 34 verify (sizeof (mbstate_t) >= 4); |
| 35 |
| 36 static char internal_state[4]; |
| 37 |
| 38 size_t |
| 39 mbrtowc (wchar_t *pwc, const char *s, size_t n, mbstate_t *ps) |
| 40 { |
| 41 char *pstate = (char *)ps; |
| 42 |
| 43 if (s == NULL) |
| 44 { |
| 45 pwc = NULL; |
| 46 s = ""; |
| 47 n = 1; |
| 48 } |
| 49 |
| 50 if (n == 0) |
| 51 return (size_t)(-2); |
| 52 |
| 53 /* Here n > 0. */ |
| 54 |
| 55 if (pstate == NULL) |
| 56 pstate = internal_state; |
| 57 |
| 58 { |
| 59 size_t nstate = pstate[0]; |
| 60 char buf[4]; |
| 61 const char *p; |
| 62 size_t m; |
| 63 |
| 64 switch (nstate) |
| 65 { |
| 66 case 0: |
| 67 p = s; |
| 68 m = n; |
| 69 break; |
| 70 case 3: |
| 71 buf[2] = pstate[3]; |
| 72 /*FALLTHROUGH*/ |
| 73 case 2: |
| 74 buf[1] = pstate[2]; |
| 75 /*FALLTHROUGH*/ |
| 76 case 1: |
| 77 buf[0] = pstate[1]; |
| 78 p = buf; |
| 79 m = nstate; |
| 80 buf[m++] = s[0]; |
| 81 if (n >= 2 && m < 4) |
| 82 { |
| 83 buf[m++] = s[1]; |
| 84 if (n >= 3 && m < 4) |
| 85 buf[m++] = s[2]; |
| 86 } |
| 87 break; |
| 88 default: |
| 89 errno = EINVAL; |
| 90 return (size_t)(-1); |
| 91 } |
| 92 |
| 93 /* Here m > 0. */ |
| 94 |
| 95 # if __GLIBC__ || defined __UCLIBC__ |
| 96 /* Work around bug <http://sourceware.org/bugzilla/show_bug.cgi?id=9674> */ |
| 97 mbtowc (NULL, NULL, 0); |
| 98 # endif |
| 99 { |
| 100 int res = mbtowc (pwc, p, m); |
| 101 |
| 102 if (res >= 0) |
| 103 { |
| 104 if (pwc != NULL && ((*pwc == 0) != (res == 0))) |
| 105 abort (); |
| 106 if (nstate >= (res > 0 ? res : 1)) |
| 107 abort (); |
| 108 res -= nstate; |
| 109 pstate[0] = 0; |
| 110 return res; |
| 111 } |
| 112 |
| 113 /* mbtowc does not distinguish between invalid and incomplete multibyte |
| 114 sequences. But mbrtowc needs to make this distinction. |
| 115 There are two possible approaches: |
| 116 - Use iconv() and its return value. |
| 117 - Use built-in knowledge about the possible encodings. |
| 118 Given the low quality of implementation of iconv() on the systems that |
| 119 lack mbrtowc(), we use the second approach. |
| 120 The possible encodings are: |
| 121 - 8-bit encodings, |
| 122 - EUC-JP, EUC-KR, GB2312, EUC-TW, BIG5, GB18030, SJIS, |
| 123 - UTF-8. |
| 124 Use specialized code for each. */ |
| 125 if (m >= 4 || m >= MB_CUR_MAX) |
| 126 goto invalid; |
| 127 /* Here MB_CUR_MAX > 1 and 0 < m < 4. */ |
| 128 { |
| 129 const char *encoding = locale_charset (); |
| 130 |
| 131 if (STREQ_OPT (encoding, "UTF-8", 'U', 'T', 'F', '-', '8', 0, 0, 0, 0)) |
| 132 { |
| 133 /* Cf. unistr/u8-mblen.c. */ |
| 134 unsigned char c = (unsigned char) p[0]; |
| 135 |
| 136 if (c >= 0xc2) |
| 137 { |
| 138 if (c < 0xe0) |
| 139 { |
| 140 if (m == 1) |
| 141 goto incomplete; |
| 142 } |
| 143 else if (c < 0xf0) |
| 144 { |
| 145 if (m == 1) |
| 146 goto incomplete; |
| 147 if (m == 2) |
| 148 { |
| 149 unsigned char c2 = (unsigned char) p[1]; |
| 150 |
| 151 if ((c2 ^ 0x80) < 0x40 |
| 152 && (c >= 0xe1 || c2 >= 0xa0) |
| 153 && (c != 0xed || c2 < 0xa0)) |
| 154 goto incomplete; |
| 155 } |
| 156 } |
| 157 else if (c <= 0xf4) |
| 158 { |
| 159 if (m == 1) |
| 160 goto incomplete; |
| 161 else /* m == 2 || m == 3 */ |
| 162 { |
| 163 unsigned char c2 = (unsigned char) p[1]; |
| 164 |
| 165 if ((c2 ^ 0x80) < 0x40 |
| 166 && (c >= 0xf1 || c2 >= 0x90) |
| 167 && (c < 0xf4 || (c == 0xf4 && c2 < 0x90))) |
| 168 { |
| 169 if (m == 2) |
| 170 goto incomplete; |
| 171 else /* m == 3 */ |
| 172 { |
| 173 unsigned char c3 = (unsigned char) p[2]; |
| 174 |
| 175 if ((c3 ^ 0x80) < 0x40) |
| 176 goto incomplete; |
| 177 } |
| 178 } |
| 179 } |
| 180 } |
| 181 } |
| 182 goto invalid; |
| 183 } |
| 184 |
| 185 /* As a reference for this code, you can use the GNU libiconv |
| 186 implementation. Look for uses of the RET_TOOFEW macro. */ |
| 187 |
| 188 if (STREQ_OPT (encoding, |
| 189 "EUC-JP", 'E', 'U', 'C', '-', 'J', 'P', 0, 0, 0)) |
| 190 { |
| 191 if (m == 1) |
| 192 { |
| 193 unsigned char c = (unsigned char) p[0]; |
| 194 |
| 195 if ((c >= 0xa1 && c < 0xff) || c == 0x8e || c == 0x8f) |
| 196 goto incomplete; |
| 197 } |
| 198 if (m == 2) |
| 199 { |
| 200 unsigned char c = (unsigned char) p[0]; |
| 201 |
| 202 if (c == 0x8f) |
| 203 { |
| 204 unsigned char c2 = (unsigned char) p[1]; |
| 205 |
| 206 if (c2 >= 0xa1 && c2 < 0xff) |
| 207 goto incomplete; |
| 208 } |
| 209 } |
| 210 goto invalid; |
| 211 } |
| 212 if (STREQ_OPT (encoding, |
| 213 "EUC-KR", 'E', 'U', 'C', '-', 'K', 'R', 0, 0, 0) |
| 214 || STREQ_OPT (encoding, |
| 215 "GB2312", 'G', 'B', '2', '3', '1', '2', 0, 0, 0) |
| 216 || STREQ_OPT (encoding, |
| 217 "BIG5", 'B', 'I', 'G', '5', 0, 0, 0, 0, 0)) |
| 218 { |
| 219 if (m == 1) |
| 220 { |
| 221 unsigned char c = (unsigned char) p[0]; |
| 222 |
| 223 if (c >= 0xa1 && c < 0xff) |
| 224 goto incomplete; |
| 225 } |
| 226 goto invalid; |
| 227 } |
| 228 if (STREQ_OPT (encoding, |
| 229 "EUC-TW", 'E', 'U', 'C', '-', 'T', 'W', 0, 0, 0)) |
| 230 { |
| 231 if (m == 1) |
| 232 { |
| 233 unsigned char c = (unsigned char) p[0]; |
| 234 |
| 235 if ((c >= 0xa1 && c < 0xff) || c == 0x8e) |
| 236 goto incomplete; |
| 237 } |
| 238 else /* m == 2 || m == 3 */ |
| 239 { |
| 240 unsigned char c = (unsigned char) p[0]; |
| 241 |
| 242 if (c == 0x8e) |
| 243 goto incomplete; |
| 244 } |
| 245 goto invalid; |
| 246 } |
| 247 if (STREQ_OPT (encoding, |
| 248 "GB18030", 'G', 'B', '1', '8', '0', '3', '0', 0, 0)) |
| 249 { |
| 250 if (m == 1) |
| 251 { |
| 252 unsigned char c = (unsigned char) p[0]; |
| 253 |
| 254 if ((c >= 0x90 && c <= 0xe3) || (c >= 0xf8 && c <= 0xfe)) |
| 255 goto incomplete; |
| 256 } |
| 257 else /* m == 2 || m == 3 */ |
| 258 { |
| 259 unsigned char c = (unsigned char) p[0]; |
| 260 |
| 261 if (c >= 0x90 && c <= 0xe3) |
| 262 { |
| 263 unsigned char c2 = (unsigned char) p[1]; |
| 264 |
| 265 if (c2 >= 0x30 && c2 <= 0x39) |
| 266 { |
| 267 if (m == 2) |
| 268 goto incomplete; |
| 269 else /* m == 3 */ |
| 270 { |
| 271 unsigned char c3 = (unsigned char) p[2]; |
| 272 |
| 273 if (c3 >= 0x81 && c3 <= 0xfe) |
| 274 goto incomplete; |
| 275 } |
| 276 } |
| 277 } |
| 278 } |
| 279 goto invalid; |
| 280 } |
| 281 if (STREQ_OPT (encoding, "SJIS", 'S', 'J', 'I', 'S', 0, 0, 0, 0, 0)) |
| 282 { |
| 283 if (m == 1) |
| 284 { |
| 285 unsigned char c = (unsigned char) p[0]; |
| 286 |
| 287 if ((c >= 0x81 && c <= 0x9f) || (c >= 0xe0 && c <= 0xea) |
| 288 || (c >= 0xf0 && c <= 0xf9)) |
| 289 goto incomplete; |
| 290 } |
| 291 goto invalid; |
| 292 } |
| 293 |
| 294 /* An unknown multibyte encoding. */ |
| 295 goto incomplete; |
| 296 } |
| 297 |
| 298 incomplete: |
| 299 { |
| 300 size_t k = nstate; |
| 301 /* Here 0 <= k < m < 4. */ |
| 302 pstate[++k] = s[0]; |
| 303 if (k < m) |
| 304 { |
| 305 pstate[++k] = s[1]; |
| 306 if (k < m) |
| 307 pstate[++k] = s[2]; |
| 308 } |
| 309 if (k != m) |
| 310 abort (); |
| 311 } |
| 312 pstate[0] = m; |
| 313 return (size_t)(-2); |
| 314 |
| 315 invalid: |
| 316 errno = EILSEQ; |
| 317 /* The conversion state is undefined, says POSIX. */ |
| 318 return (size_t)(-1); |
| 319 } |
| 320 } |
| 321 } |
| 322 |
| 323 #else |
| 324 /* Override the system's mbrtowc() function. */ |
| 325 |
| 326 # undef mbrtowc |
| 327 |
| 328 size_t |
| 329 rpl_mbrtowc (wchar_t *pwc, const char *s, size_t n, mbstate_t *ps) |
| 330 { |
| 331 # if MBRTOWC_NULL_ARG2_BUG || MBRTOWC_RETVAL_BUG |
| 332 if (s == NULL) |
| 333 { |
| 334 pwc = NULL; |
| 335 s = ""; |
| 336 n = 1; |
| 337 } |
| 338 # endif |
| 339 |
| 340 # if MBRTOWC_RETVAL_BUG |
| 341 { |
| 342 static mbstate_t internal_state; |
| 343 |
| 344 /* Override mbrtowc's internal state. We cannot call mbsinit() on the |
| 345 hidden internal state, but we can call it on our variable. */ |
| 346 if (ps == NULL) |
| 347 ps = &internal_state; |
| 348 |
| 349 if (!mbsinit (ps)) |
| 350 { |
| 351 /* Parse the rest of the multibyte character byte for byte. */ |
| 352 size_t count = 0; |
| 353 for (; n > 0; s++, n--) |
| 354 { |
| 355 wchar_t wc; |
| 356 size_t ret = mbrtowc (&wc, s, 1, ps); |
| 357 |
| 358 if (ret == (size_t)(-1)) |
| 359 return (size_t)(-1); |
| 360 count++; |
| 361 if (ret != (size_t)(-2)) |
| 362 { |
| 363 /* The multibyte character has been completed. */ |
| 364 if (pwc != NULL) |
| 365 *pwc = wc; |
| 366 return (wc == 0 ? 0 : count); |
| 367 } |
| 368 } |
| 369 return (size_t)(-2); |
| 370 } |
| 371 } |
| 372 # endif |
| 373 |
| 374 # if MBRTOWC_NUL_RETVAL_BUG |
| 375 { |
| 376 wchar_t wc; |
| 377 size_t ret = mbrtowc (&wc, s, n, ps); |
| 378 |
| 379 if (ret != (size_t)(-1) && ret != (size_t)(-2)) |
| 380 { |
| 381 if (pwc != NULL) |
| 382 *pwc = wc; |
| 383 if (wc == 0) |
| 384 ret = 0; |
| 385 } |
| 386 return ret; |
| 387 } |
| 388 # else |
| 389 { |
| 390 # if MBRTOWC_NULL_ARG1_BUG |
| 391 wchar_t dummy; |
| 392 |
| 393 if (pwc == NULL) |
| 394 pwc = &dummy; |
| 395 # endif |
| 396 |
| 397 return mbrtowc (pwc, s, n, ps); |
| 398 } |
| 399 # endif |
| 400 } |
| 401 |
| 402 #endif |
OLD | NEW |