| 1 |
/* |
| 2 |
* Copyright (C) 2000-2002, 2005-2006, 2008-2009, 2011 Free Software Foundation, Inc. |
| 3 |
* This file is part of the GNU LIBICONV Library. |
| 4 |
* |
| 5 |
* The GNU LIBICONV Library is free software; you can redistribute it |
| 6 |
* and/or modify it under the terms of the GNU Library General Public |
| 7 |
* License as published by the Free Software Foundation; either version 2 |
| 8 |
* of the License, or (at your option) any later version. |
| 9 |
* |
| 10 |
* The GNU LIBICONV Library is distributed in the hope that it will be |
| 11 |
* useful, but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 12 |
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
| 13 |
* Library General Public License for more details. |
| 14 |
* |
| 15 |
* You should have received a copy of the GNU Library General Public |
| 16 |
* License along with the GNU LIBICONV Library; see the file COPYING.LIB. |
| 17 |
* If not, see <http://www.gnu.org/licenses/>. |
| 18 |
*/ |
| 19 |
|
| 20 |
/* This file defines three conversion loops: |
| 21 |
- from wchar_t to anything else, |
| 22 |
- from anything else to wchar_t, |
| 23 |
- from wchar_t to wchar_t. |
| 24 |
*/ |
| 25 |
|
| 26 |
#if HAVE_WCRTOMB || HAVE_MBRTOWC |
| 27 |
/* Tru64 with Desktop Toolkit C has a bug: <stdio.h> must be included before |
| 28 |
<wchar.h>. |
| 29 |
BSD/OS 4.0.1 has a bug: <stddef.h>, <stdio.h> and <time.h> must be |
| 30 |
included before <wchar.h>. |
| 31 |
In some builds of uClibc, <wchar.h> is nonexistent and wchar_t is defined |
| 32 |
by <stddef.h>. */ |
| 33 |
# include <stddef.h> |
| 34 |
# include <stdio.h> |
| 35 |
# include <time.h> |
| 36 |
# include <wchar.h> |
| 37 |
# define BUF_SIZE 64 /* assume MB_LEN_MAX <= 64 */ |
| 38 |
/* Some systems, like BeOS, have multibyte encodings but lack mbstate_t. */ |
| 39 |
extern size_t mbrtowc (); |
| 40 |
# ifdef mbstate_t |
| 41 |
# define mbrtowc(pwc, s, n, ps) (mbrtowc)(pwc, s, n, 0) |
| 42 |
# define mbsinit(ps) 1 |
| 43 |
# endif |
| 44 |
# ifndef mbsinit |
| 45 |
# if !HAVE_MBSINIT |
| 46 |
# define mbsinit(ps) 1 |
| 47 |
# endif |
| 48 |
# endif |
| 49 |
#endif |
| 50 |
|
| 51 |
/* |
| 52 |
* The first two conversion loops have an extended conversion descriptor. |
| 53 |
*/ |
| 54 |
struct wchar_conv_struct { |
| 55 |
struct conv_struct parent; |
| 56 |
#if HAVE_WCRTOMB || HAVE_MBRTOWC |
| 57 |
mbstate_t state; |
| 58 |
#endif |
| 59 |
}; |
| 60 |
|
| 61 |
|
| 62 |
#if HAVE_WCRTOMB |
| 63 |
|
| 64 |
/* From wchar_t to anything else. */ |
| 65 |
|
| 66 |
#ifndef LIBICONV_PLUG |
| 67 |
|
| 68 |
#if 0 |
| 69 |
|
| 70 |
struct wc_to_mb_fallback_locals { |
| 71 |
struct wchar_conv_struct * l_wcd; |
| 72 |
char* l_outbuf; |
| 73 |
size_t l_outbytesleft; |
| 74 |
int l_errno; |
| 75 |
}; |
| 76 |
|
| 77 |
/* A callback that writes a string given in the locale encoding. */ |
| 78 |
static void wc_to_mb_write_replacement (const char *buf, size_t buflen, |
| 79 |
void* callback_arg) |
| 80 |
{ |
| 81 |
struct wc_to_mb_fallback_locals * plocals = |
| 82 |
(struct wc_to_mb_fallback_locals *) callback_arg; |
| 83 |
/* Do nothing if already encountered an error in a previous call. */ |
| 84 |
if (plocals->l_errno == 0) { |
| 85 |
/* Attempt to convert the passed buffer to the target encoding. |
| 86 |
Here we don't support characters split across multiple calls. */ |
| 87 |
const char* bufptr = buf; |
| 88 |
size_t bufleft = buflen; |
| 89 |
size_t res = unicode_loop_convert(&plocals->l_wcd->parent, |
| 90 |
&bufptr,&bufleft, |
| 91 |
&plocals->l_outbuf,&plocals->l_outbytesleft); |
| 92 |
if (res == (size_t)(-1)) { |
| 93 |
if (errno == EILSEQ || errno == EINVAL) |
| 94 |
/* Invalid buf contents. */ |
| 95 |
plocals->l_errno = EILSEQ; |
| 96 |
else if (errno == E2BIG) |
| 97 |
/* Output buffer too small. */ |
| 98 |
plocals->l_errno = E2BIG; |
| 99 |
else |
| 100 |
abort(); |
| 101 |
} else { |
| 102 |
/* Successful conversion. */ |
| 103 |
if (bufleft > 0) |
| 104 |
abort(); |
| 105 |
} |
| 106 |
} |
| 107 |
} |
| 108 |
|
| 109 |
#else |
| 110 |
|
| 111 |
struct wc_to_mb_fallback_locals { |
| 112 |
char* l_outbuf; |
| 113 |
size_t l_outbytesleft; |
| 114 |
int l_errno; |
| 115 |
}; |
| 116 |
|
| 117 |
/* A callback that writes a string given in the target encoding. */ |
| 118 |
static void wc_to_mb_write_replacement (const char *buf, size_t buflen, |
| 119 |
void* callback_arg) |
| 120 |
{ |
| 121 |
struct wc_to_mb_fallback_locals * plocals = |
| 122 |
(struct wc_to_mb_fallback_locals *) callback_arg; |
| 123 |
/* Do nothing if already encountered an error in a previous call. */ |
| 124 |
if (plocals->l_errno == 0) { |
| 125 |
/* Attempt to copy the passed buffer to the output buffer. */ |
| 126 |
if (plocals->l_outbytesleft < buflen) |
| 127 |
plocals->l_errno = E2BIG; |
| 128 |
else { |
| 129 |
memcpy(plocals->l_outbuf, buf, buflen); |
| 130 |
plocals->l_outbuf += buflen; |
| 131 |
plocals->l_outbytesleft -= buflen; |
| 132 |
} |
| 133 |
} |
| 134 |
} |
| 135 |
|
| 136 |
#endif |
| 137 |
|
| 138 |
#endif /* !LIBICONV_PLUG */ |
| 139 |
|
| 140 |
static size_t wchar_from_loop_convert (iconv_t icd, |
| 141 |
const char* * inbuf, size_t *inbytesleft, |
| 142 |
char* * outbuf, size_t *outbytesleft) |
| 143 |
{ |
| 144 |
struct wchar_conv_struct * wcd = (struct wchar_conv_struct *) icd; |
| 145 |
size_t result = 0; |
| 146 |
while (*inbytesleft >= sizeof(wchar_t)) { |
| 147 |
const wchar_t * inptr = (const wchar_t *) *inbuf; |
| 148 |
size_t inleft = *inbytesleft; |
| 149 |
char buf[BUF_SIZE]; |
| 150 |
mbstate_t state = wcd->state; |
| 151 |
size_t bufcount = 0; |
| 152 |
while (inleft >= sizeof(wchar_t)) { |
| 153 |
/* Convert one wchar_t to multibyte representation. */ |
| 154 |
size_t count = wcrtomb(buf+bufcount,*inptr,&state); |
| 155 |
if (count == (size_t)(-1)) { |
| 156 |
/* Invalid input. */ |
| 157 |
if (wcd->parent.discard_ilseq) { |
| 158 |
count = 0; |
| 159 |
} |
| 160 |
#ifndef LIBICONV_PLUG |
| 161 |
else if (wcd->parent.fallbacks.wc_to_mb_fallback != NULL) { |
| 162 |
/* Drop the contents of buf[] accumulated so far, and instead |
| 163 |
pass all queued wide characters to the fallback handler. */ |
| 164 |
struct wc_to_mb_fallback_locals locals; |
| 165 |
const wchar_t * fallback_inptr; |
| 166 |
#if 0 |
| 167 |
locals.l_wcd = wcd; |
| 168 |
#endif |
| 169 |
locals.l_outbuf = *outbuf; |
| 170 |
locals.l_outbytesleft = *outbytesleft; |
| 171 |
locals.l_errno = 0; |
| 172 |
for (fallback_inptr = (const wchar_t *) *inbuf; |
| 173 |
fallback_inptr <= inptr; |
| 174 |
fallback_inptr++) |
| 175 |
wcd->parent.fallbacks.wc_to_mb_fallback(*fallback_inptr, |
| 176 |
wc_to_mb_write_replacement, |
| 177 |
&locals, |
| 178 |
wcd->parent.fallbacks.data); |
| 179 |
if (locals.l_errno != 0) { |
| 180 |
errno = locals.l_errno; |
| 181 |
return -1; |
| 182 |
} |
| 183 |
wcd->state = state; |
| 184 |
*inbuf = (const char *) (inptr + 1); |
| 185 |
*inbytesleft = inleft - sizeof(wchar_t); |
| 186 |
*outbuf = locals.l_outbuf; |
| 187 |
*outbytesleft = locals.l_outbytesleft; |
| 188 |
result += 1; |
| 189 |
break; |
| 190 |
} |
| 191 |
#endif |
| 192 |
else { |
| 193 |
errno = EILSEQ; |
| 194 |
return -1; |
| 195 |
} |
| 196 |
} |
| 197 |
inptr++; |
| 198 |
inleft -= sizeof(wchar_t); |
| 199 |
bufcount += count; |
| 200 |
if (count == 0) { |
| 201 |
/* Continue, append next wchar_t. */ |
| 202 |
} else { |
| 203 |
/* Attempt to convert the accumulated multibyte representations |
| 204 |
to the target encoding. */ |
| 205 |
const char* bufptr = buf; |
| 206 |
size_t bufleft = bufcount; |
| 207 |
char* outptr = *outbuf; |
| 208 |
size_t outleft = *outbytesleft; |
| 209 |
size_t res = unicode_loop_convert(&wcd->parent, |
| 210 |
&bufptr,&bufleft, |
| 211 |
&outptr,&outleft); |
| 212 |
if (res == (size_t)(-1)) { |
| 213 |
if (errno == EILSEQ) |
| 214 |
/* Invalid input. */ |
| 215 |
return -1; |
| 216 |
else if (errno == E2BIG) |
| 217 |
/* Output buffer too small. */ |
| 218 |
return -1; |
| 219 |
else if (errno == EINVAL) { |
| 220 |
/* Continue, append next wchar_t, but avoid buffer overrun. */ |
| 221 |
if (bufcount + MB_CUR_MAX > BUF_SIZE) |
| 222 |
abort(); |
| 223 |
} else |
| 224 |
abort(); |
| 225 |
} else { |
| 226 |
/* Successful conversion. */ |
| 227 |
wcd->state = state; |
| 228 |
*inbuf = (const char *) inptr; |
| 229 |
*inbytesleft = inleft; |
| 230 |
*outbuf = outptr; |
| 231 |
*outbytesleft = outleft; |
| 232 |
result += res; |
| 233 |
break; |
| 234 |
} |
| 235 |
} |
| 236 |
} |
| 237 |
} |
| 238 |
return result; |
| 239 |
} |
| 240 |
|
| 241 |
static size_t wchar_from_loop_reset (iconv_t icd, |
| 242 |
char* * outbuf, size_t *outbytesleft) |
| 243 |
{ |
| 244 |
struct wchar_conv_struct * wcd = (struct wchar_conv_struct *) icd; |
| 245 |
if (outbuf == NULL || *outbuf == NULL) { |
| 246 |
/* Reset the states. */ |
| 247 |
memset(&wcd->state,'\0',sizeof(mbstate_t)); |
| 248 |
return unicode_loop_reset(&wcd->parent,NULL,NULL); |
| 249 |
} else { |
| 250 |
if (!mbsinit(&wcd->state)) { |
| 251 |
mbstate_t state = wcd->state; |
| 252 |
char buf[BUF_SIZE]; |
| 253 |
size_t bufcount = wcrtomb(buf,(wchar_t)0,&state); |
| 254 |
if (bufcount == (size_t)(-1) || bufcount == 0 || buf[bufcount-1] != '\0') |
| 255 |
abort(); |
| 256 |
else { |
| 257 |
const char* bufptr = buf; |
| 258 |
size_t bufleft = bufcount-1; |
| 259 |
char* outptr = *outbuf; |
| 260 |
size_t outleft = *outbytesleft; |
| 261 |
size_t res = unicode_loop_convert(&wcd->parent, |
| 262 |
&bufptr,&bufleft, |
| 263 |
&outptr,&outleft); |
| 264 |
if (res == (size_t)(-1)) { |
| 265 |
if (errno == E2BIG) |
| 266 |
return -1; |
| 267 |
else |
| 268 |
abort(); |
| 269 |
} else { |
| 270 |
res = unicode_loop_reset(&wcd->parent,&outptr,&outleft); |
| 271 |
if (res == (size_t)(-1)) |
| 272 |
return res; |
| 273 |
else { |
| 274 |
/* Successful. */ |
| 275 |
wcd->state = state; |
| 276 |
*outbuf = outptr; |
| 277 |
*outbytesleft = outleft; |
| 278 |
return 0; |
| 279 |
} |
| 280 |
} |
| 281 |
} |
| 282 |
} else |
| 283 |
return unicode_loop_reset(&wcd->parent,outbuf,outbytesleft); |
| 284 |
} |
| 285 |
} |
| 286 |
|
| 287 |
#endif |
| 288 |
|
| 289 |
|
| 290 |
#if HAVE_MBRTOWC |
| 291 |
|
| 292 |
/* From anything else to wchar_t. */ |
| 293 |
|
| 294 |
#ifndef LIBICONV_PLUG |
| 295 |
|
| 296 |
struct mb_to_wc_fallback_locals { |
| 297 |
char* l_outbuf; |
| 298 |
size_t l_outbytesleft; |
| 299 |
int l_errno; |
| 300 |
}; |
| 301 |
|
| 302 |
static void mb_to_wc_write_replacement (const wchar_t *buf, size_t buflen, |
| 303 |
void* callback_arg) |
| 304 |
{ |
| 305 |
struct mb_to_wc_fallback_locals * plocals = |
| 306 |
(struct mb_to_wc_fallback_locals *) callback_arg; |
| 307 |
/* Do nothing if already encountered an error in a previous call. */ |
| 308 |
if (plocals->l_errno == 0) { |
| 309 |
/* Attempt to copy the passed buffer to the output buffer. */ |
| 310 |
if (plocals->l_outbytesleft < sizeof(wchar_t)*buflen) |
| 311 |
plocals->l_errno = E2BIG; |
| 312 |
else { |
| 313 |
for (; buflen > 0; buf++, buflen--) { |
| 314 |
*(wchar_t*) plocals->l_outbuf = *buf; |
| 315 |
plocals->l_outbuf += sizeof(wchar_t); |
| 316 |
plocals->l_outbytesleft -= sizeof(wchar_t); |
| 317 |
} |
| 318 |
} |
| 319 |
} |
| 320 |
} |
| 321 |
|
| 322 |
#endif /* !LIBICONV_PLUG */ |
| 323 |
|
| 324 |
static size_t wchar_to_loop_convert (iconv_t icd, |
| 325 |
const char* * inbuf, size_t *inbytesleft, |
| 326 |
char* * outbuf, size_t *outbytesleft) |
| 327 |
{ |
| 328 |
struct wchar_conv_struct * wcd = (struct wchar_conv_struct *) icd; |
| 329 |
size_t result = 0; |
| 330 |
while (*inbytesleft > 0) { |
| 331 |
size_t incount; |
| 332 |
for (incount = 1; ; ) { |
| 333 |
/* Here incount <= *inbytesleft. */ |
| 334 |
char buf[BUF_SIZE]; |
| 335 |
const char* inptr = *inbuf; |
| 336 |
size_t inleft = incount; |
| 337 |
char* bufptr = buf; |
| 338 |
size_t bufleft = BUF_SIZE; |
| 339 |
size_t res = unicode_loop_convert(&wcd->parent, |
| 340 |
&inptr,&inleft, |
| 341 |
&bufptr,&bufleft); |
| 342 |
if (res == (size_t)(-1)) { |
| 343 |
if (errno == EILSEQ) |
| 344 |
/* Invalid input. */ |
| 345 |
return -1; |
| 346 |
else if (errno == EINVAL) { |
| 347 |
/* Incomplete input. Next try with one more input byte. */ |
| 348 |
} else |
| 349 |
/* E2BIG shouldn't occur. */ |
| 350 |
abort(); |
| 351 |
} else { |
| 352 |
/* Successful conversion. */ |
| 353 |
size_t bufcount = bufptr-buf; /* = BUF_SIZE-bufleft */ |
| 354 |
mbstate_t state = wcd->state; |
| 355 |
wchar_t wc; |
| 356 |
res = mbrtowc(&wc,buf,bufcount,&state); |
| 357 |
if (res == (size_t)(-2)) { |
| 358 |
/* Next try with one more input byte. */ |
| 359 |
} else { |
| 360 |
if (res == (size_t)(-1)) { |
| 361 |
/* Invalid input. */ |
| 362 |
if (wcd->parent.discard_ilseq) { |
| 363 |
} |
| 364 |
#ifndef LIBICONV_PLUG |
| 365 |
else if (wcd->parent.fallbacks.mb_to_wc_fallback != NULL) { |
| 366 |
/* Drop the contents of buf[] accumulated so far, and instead |
| 367 |
pass all queued chars to the fallback handler. */ |
| 368 |
struct mb_to_wc_fallback_locals locals; |
| 369 |
locals.l_outbuf = *outbuf; |
| 370 |
locals.l_outbytesleft = *outbytesleft; |
| 371 |
locals.l_errno = 0; |
| 372 |
wcd->parent.fallbacks.mb_to_wc_fallback(*inbuf, incount, |
| 373 |
mb_to_wc_write_replacement, |
| 374 |
&locals, |
| 375 |
wcd->parent.fallbacks.data); |
| 376 |
if (locals.l_errno != 0) { |
| 377 |
errno = locals.l_errno; |
| 378 |
return -1; |
| 379 |
} |
| 380 |
/* Restoring the state is not needed because it is the initial |
| 381 |
state anyway: For all known locale encodings, the multibyte |
| 382 |
to wchar_t conversion doesn't have shift state, and we have |
| 383 |
excluded partial accumulated characters. */ |
| 384 |
/* wcd->state = state; */ |
| 385 |
*inbuf += incount; |
| 386 |
*inbytesleft -= incount; |
| 387 |
*outbuf = locals.l_outbuf; |
| 388 |
*outbytesleft = locals.l_outbytesleft; |
| 389 |
result += 1; |
| 390 |
break; |
| 391 |
} |
| 392 |
#endif |
| 393 |
else |
| 394 |
return -1; |
| 395 |
} else { |
| 396 |
if (*outbytesleft < sizeof(wchar_t)) { |
| 397 |
errno = E2BIG; |
| 398 |
return -1; |
| 399 |
} |
| 400 |
*(wchar_t*) *outbuf = wc; |
| 401 |
/* Restoring the state is not needed because it is the initial |
| 402 |
state anyway: For all known locale encodings, the multibyte |
| 403 |
to wchar_t conversion doesn't have shift state, and we have |
| 404 |
excluded partial accumulated characters. */ |
| 405 |
/* wcd->state = state; */ |
| 406 |
*outbuf += sizeof(wchar_t); |
| 407 |
*outbytesleft -= sizeof(wchar_t); |
| 408 |
} |
| 409 |
*inbuf += incount; |
| 410 |
*inbytesleft -= incount; |
| 411 |
result += res; |
| 412 |
break; |
| 413 |
} |
| 414 |
} |
| 415 |
incount++; |
| 416 |
if (incount > *inbytesleft) { |
| 417 |
/* Incomplete input. */ |
| 418 |
errno = EINVAL; |
| 419 |
return -1; |
| 420 |
} |
| 421 |
} |
| 422 |
} |
| 423 |
return result; |
| 424 |
} |
| 425 |
|
| 426 |
static size_t wchar_to_loop_reset (iconv_t icd, |
| 427 |
char* * outbuf, size_t *outbytesleft) |
| 428 |
{ |
| 429 |
struct wchar_conv_struct * wcd = (struct wchar_conv_struct *) icd; |
| 430 |
size_t res = unicode_loop_reset(&wcd->parent,outbuf,outbytesleft); |
| 431 |
if (res == (size_t)(-1)) |
| 432 |
return res; |
| 433 |
memset(&wcd->state,0,sizeof(mbstate_t)); |
| 434 |
return 0; |
| 435 |
} |
| 436 |
|
| 437 |
#endif |
| 438 |
|
| 439 |
|
| 440 |
/* From wchar_t to wchar_t. */ |
| 441 |
|
| 442 |
static size_t wchar_id_loop_convert (iconv_t icd, |
| 443 |
const char* * inbuf, size_t *inbytesleft, |
| 444 |
char* * outbuf, size_t *outbytesleft) |
| 445 |
{ |
| 446 |
struct conv_struct * cd = (struct conv_struct *) icd; |
| 447 |
const wchar_t* inptr = (const wchar_t*) *inbuf; |
| 448 |
size_t inleft = *inbytesleft / sizeof(wchar_t); |
| 449 |
wchar_t* outptr = (wchar_t*) *outbuf; |
| 450 |
size_t outleft = *outbytesleft / sizeof(wchar_t); |
| 451 |
size_t count = (inleft <= outleft ? inleft : outleft); |
| 452 |
if (count > 0) { |
| 453 |
*inbytesleft -= count * sizeof(wchar_t); |
| 454 |
*outbytesleft -= count * sizeof(wchar_t); |
| 455 |
do { |
| 456 |
wchar_t wc = *inptr++; |
| 457 |
*outptr++ = wc; |
| 458 |
#ifndef LIBICONV_PLUG |
| 459 |
if (cd->hooks.wc_hook) |
| 460 |
(*cd->hooks.wc_hook)(wc, cd->hooks.data); |
| 461 |
#endif |
| 462 |
} while (--count > 0); |
| 463 |
*inbuf = (const char*) inptr; |
| 464 |
*outbuf = (char*) outptr; |
| 465 |
} |
| 466 |
return 0; |
| 467 |
} |
| 468 |
|
| 469 |
static size_t wchar_id_loop_reset (iconv_t icd, |
| 470 |
char* * outbuf, size_t *outbytesleft) |
| 471 |
{ |
| 472 |
return 0; |
| 473 |
} |