| 1 |
207 |
jeremybenn |
/* Simple transformations functions.
|
| 2 |
|
|
Copyright (C) 1997, 1998, 1999, 2000, 2001 Free Software Foundation, Inc.
|
| 3 |
|
|
This file is part of the GNU C Library.
|
| 4 |
|
|
Contributed by Ulrich Drepper <drepper@cygnus.com>, 1997.
|
| 5 |
|
|
|
| 6 |
|
|
The GNU C Library is free software; you can redistribute it and/or
|
| 7 |
|
|
modify it under the terms of the GNU Lesser General Public
|
| 8 |
|
|
License as published by the Free Software Foundation; either
|
| 9 |
|
|
version 2.1 of the License, or (at your option) any later version.
|
| 10 |
|
|
|
| 11 |
|
|
The GNU C Library is distributed in the hope that it will be useful,
|
| 12 |
|
|
but WITHOUT ANY WARRANTY; without even the implied warranty of
|
| 13 |
|
|
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
| 14 |
|
|
Lesser General Public License for more details.
|
| 15 |
|
|
|
| 16 |
|
|
You should have received a copy of the GNU Lesser General Public
|
| 17 |
|
|
License along with the GNU C Library; if not, write to the Free
|
| 18 |
|
|
Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
|
| 19 |
|
|
02111-1307 USA. */
|
| 20 |
|
|
|
| 21 |
|
|
#include <byteswap.h>
|
| 22 |
|
|
#include <dlfcn.h>
|
| 23 |
|
|
#include <endian.h>
|
| 24 |
|
|
#include <errno.h>
|
| 25 |
|
|
#include <gconv.h>
|
| 26 |
|
|
#include <stdint.h>
|
| 27 |
|
|
#include <stdlib.h>
|
| 28 |
|
|
#include <string.h>
|
| 29 |
|
|
#include <wchar.h>
|
| 30 |
|
|
#include <sys/param.h>
|
| 31 |
|
|
|
| 32 |
|
|
#define BUILTIN_ALIAS(s1, s2) /* nothing */
|
| 33 |
|
|
#define BUILTIN_TRANSFORMATION(From, To, Cost, Name, Fct, MinF, MaxF, \
|
| 34 |
|
|
MinT, MaxT) \
|
| 35 |
|
|
extern int Fct (struct __gconv_step *, struct __gconv_step_data *, \
|
| 36 |
|
|
__const unsigned char **, __const unsigned char *, \
|
| 37 |
|
|
unsigned char **, size_t *, int, int);
|
| 38 |
|
|
#include "gconv_builtin.h"
|
| 39 |
|
|
|
| 40 |
|
|
|
| 41 |
|
|
#ifndef EILSEQ
|
| 42 |
|
|
# define EILSEQ EINVAL
|
| 43 |
|
|
#endif
|
| 44 |
|
|
|
| 45 |
|
|
|
| 46 |
|
|
/* Transform from the internal, UCS4-like format, to UCS4. The
|
| 47 |
|
|
difference between the internal ucs4 format and the real UCS4
|
| 48 |
|
|
format is, if any, the endianess. The Unicode/ISO 10646 says that
|
| 49 |
|
|
unless some higher protocol specifies it differently, the byte
|
| 50 |
|
|
order is big endian.*/
|
| 51 |
|
|
#define DEFINE_INIT 0
|
| 52 |
|
|
#define DEFINE_FINI 0
|
| 53 |
|
|
#define MIN_NEEDED_FROM 4
|
| 54 |
|
|
#define MIN_NEEDED_TO 4
|
| 55 |
|
|
#define FROM_DIRECTION 1
|
| 56 |
|
|
#define FROM_LOOP internal_ucs4_loop
|
| 57 |
|
|
#define TO_LOOP internal_ucs4_loop /* This is not used. */
|
| 58 |
|
|
#define FUNCTION_NAME __gconv_transform_internal_ucs4
|
| 59 |
|
|
|
| 60 |
|
|
|
| 61 |
|
|
static inline int
|
| 62 |
|
|
internal_ucs4_loop (struct __gconv_step *step,
|
| 63 |
|
|
struct __gconv_step_data *step_data,
|
| 64 |
|
|
const unsigned char **inptrp, const unsigned char *inend,
|
| 65 |
|
|
unsigned char **outptrp, unsigned char *outend,
|
| 66 |
|
|
size_t *irreversible)
|
| 67 |
|
|
{
|
| 68 |
|
|
const unsigned char *inptr = *inptrp;
|
| 69 |
|
|
unsigned char *outptr = *outptrp;
|
| 70 |
|
|
size_t n_convert = MIN (inend - inptr, outend - outptr) / 4;
|
| 71 |
|
|
int result;
|
| 72 |
|
|
|
| 73 |
|
|
#if __BYTE_ORDER == __LITTLE_ENDIAN
|
| 74 |
|
|
/* Sigh, we have to do some real work. */
|
| 75 |
|
|
size_t cnt;
|
| 76 |
|
|
|
| 77 |
|
|
for (cnt = 0; cnt < n_convert; ++cnt, inptr += 4)
|
| 78 |
|
|
{
|
| 79 |
|
|
*((uint32_t *) outptr) = bswap_32 (*(const uint32_t *) inptr);
|
| 80 |
|
|
outptr = (unsigned char *)((uint32_t *) outptr + 1);
|
| 81 |
|
|
}
|
| 82 |
|
|
|
| 83 |
|
|
*inptrp = inptr;
|
| 84 |
|
|
*outptrp = outptr;
|
| 85 |
|
|
#elif __BYTE_ORDER == __BIG_ENDIAN
|
| 86 |
|
|
/* Simply copy the data. */
|
| 87 |
|
|
*inptrp = inptr + n_convert * 4;
|
| 88 |
|
|
*outptrp = memcpy (outptr, inptr, n_convert * 4);
|
| 89 |
|
|
*outptrp += n_convert * 4;
|
| 90 |
|
|
#else
|
| 91 |
|
|
# error "This endianess is not supported."
|
| 92 |
|
|
#endif
|
| 93 |
|
|
|
| 94 |
|
|
/* Determine the status. */
|
| 95 |
|
|
if (*inptrp == inend)
|
| 96 |
|
|
result = __GCONV_EMPTY_INPUT;
|
| 97 |
|
|
else if (*outptrp + 4 > outend)
|
| 98 |
|
|
result = __GCONV_FULL_OUTPUT;
|
| 99 |
|
|
else
|
| 100 |
|
|
result = __GCONV_INCOMPLETE_INPUT;
|
| 101 |
|
|
|
| 102 |
|
|
return result;
|
| 103 |
|
|
}
|
| 104 |
|
|
|
| 105 |
|
|
#ifndef _STRING_ARCH_unaligned
|
| 106 |
|
|
static inline int
|
| 107 |
|
|
internal_ucs4_loop_unaligned (struct __gconv_step *step,
|
| 108 |
|
|
struct __gconv_step_data *step_data,
|
| 109 |
|
|
const unsigned char **inptrp,
|
| 110 |
|
|
const unsigned char *inend,
|
| 111 |
|
|
unsigned char **outptrp, unsigned char *outend,
|
| 112 |
|
|
size_t *irreversible)
|
| 113 |
|
|
{
|
| 114 |
|
|
const unsigned char *inptr = *inptrp;
|
| 115 |
|
|
unsigned char *outptr = *outptrp;
|
| 116 |
|
|
size_t n_convert = MIN (inend - inptr, outend - outptr) / 4;
|
| 117 |
|
|
int result;
|
| 118 |
|
|
|
| 119 |
|
|
# if __BYTE_ORDER == __LITTLE_ENDIAN
|
| 120 |
|
|
/* Sigh, we have to do some real work. */
|
| 121 |
|
|
size_t cnt;
|
| 122 |
|
|
|
| 123 |
|
|
for (cnt = 0; cnt < n_convert; ++cnt, inptr += 4, outptr += 4)
|
| 124 |
|
|
{
|
| 125 |
|
|
outptr[0] = inptr[3];
|
| 126 |
|
|
outptr[1] = inptr[2];
|
| 127 |
|
|
outptr[2] = inptr[1];
|
| 128 |
|
|
outptr[3] = inptr[0];
|
| 129 |
|
|
}
|
| 130 |
|
|
|
| 131 |
|
|
*inptrp = inptr;
|
| 132 |
|
|
*outptrp = outptr;
|
| 133 |
|
|
# elif __BYTE_ORDER == __BIG_ENDIAN
|
| 134 |
|
|
/* Simply copy the data. */
|
| 135 |
|
|
*inptrp = inptr + n_convert * 4;
|
| 136 |
|
|
*outptrp = memcpy (outptr, inptr, n_convert * 4);
|
| 137 |
|
|
*outptrp += n_convert * 4;
|
| 138 |
|
|
# else
|
| 139 |
|
|
# error "This endianess is not supported."
|
| 140 |
|
|
# endif
|
| 141 |
|
|
|
| 142 |
|
|
/* Determine the status. */
|
| 143 |
|
|
if (*inptrp == inend)
|
| 144 |
|
|
result = __GCONV_EMPTY_INPUT;
|
| 145 |
|
|
else if (*outptrp + 4 > outend)
|
| 146 |
|
|
result = __GCONV_FULL_OUTPUT;
|
| 147 |
|
|
else
|
| 148 |
|
|
result = __GCONV_INCOMPLETE_INPUT;
|
| 149 |
|
|
|
| 150 |
|
|
return result;
|
| 151 |
|
|
}
|
| 152 |
|
|
#endif
|
| 153 |
|
|
|
| 154 |
|
|
|
| 155 |
|
|
static inline int
|
| 156 |
|
|
internal_ucs4_loop_single (struct __gconv_step *step,
|
| 157 |
|
|
struct __gconv_step_data *step_data,
|
| 158 |
|
|
const unsigned char **inptrp,
|
| 159 |
|
|
const unsigned char *inend,
|
| 160 |
|
|
unsigned char **outptrp, unsigned char *outend,
|
| 161 |
|
|
size_t *irreversible)
|
| 162 |
|
|
{
|
| 163 |
|
|
mbstate_t *state = step_data->__statep;
|
| 164 |
|
|
size_t cnt = state->__count & 7;
|
| 165 |
|
|
|
| 166 |
|
|
while (*inptrp < inend && cnt < 4)
|
| 167 |
|
|
state->__value.__wchb[cnt++] = *(*inptrp)++;
|
| 168 |
|
|
|
| 169 |
|
|
if (__builtin_expect (cnt < 4, 0))
|
| 170 |
|
|
{
|
| 171 |
|
|
/* Still not enough bytes. Store the ones in the input buffer. */
|
| 172 |
|
|
state->__count &= ~7;
|
| 173 |
|
|
state->__count |= cnt;
|
| 174 |
|
|
|
| 175 |
|
|
return __GCONV_INCOMPLETE_INPUT;
|
| 176 |
|
|
}
|
| 177 |
|
|
|
| 178 |
|
|
#if __BYTE_ORDER == __LITTLE_ENDIAN
|
| 179 |
|
|
(*outptrp)[0] = state->__value.__wchb[3];
|
| 180 |
|
|
(*outptrp)[1] = state->__value.__wchb[2];
|
| 181 |
|
|
(*outptrp)[2] = state->__value.__wchb[1];
|
| 182 |
|
|
(*outptrp)[3] = state->__value.__wchb[0];
|
| 183 |
|
|
|
| 184 |
|
|
*outptrp += 4;
|
| 185 |
|
|
#elif __BYTE_ORDER == __BIG_ENDIAN
|
| 186 |
|
|
{
|
| 187 |
|
|
uint32_t **p = (uint32_t **)outptrp;
|
| 188 |
|
|
uint32_t *q = *p;
|
| 189 |
|
|
/* XXX unaligned */
|
| 190 |
|
|
*q = state->__value.__wch;
|
| 191 |
|
|
outptrp = (unsigned char *)(p + 1);
|
| 192 |
|
|
}
|
| 193 |
|
|
#else
|
| 194 |
|
|
# error "This endianess is not supported."
|
| 195 |
|
|
#endif
|
| 196 |
|
|
|
| 197 |
|
|
/* Clear the state buffer. */
|
| 198 |
|
|
state->__count &= ~7;
|
| 199 |
|
|
|
| 200 |
|
|
return __GCONV_OK;
|
| 201 |
|
|
}
|
| 202 |
|
|
|
| 203 |
|
|
#include <iconv/skeleton.c>
|
| 204 |
|
|
|
| 205 |
|
|
|
| 206 |
|
|
/* Transform from UCS4 to the internal, UCS4-like format. Unlike
|
| 207 |
|
|
for the other direction we have to check for correct values here. */
|
| 208 |
|
|
#define DEFINE_INIT 0
|
| 209 |
|
|
#define DEFINE_FINI 0
|
| 210 |
|
|
#define MIN_NEEDED_FROM 4
|
| 211 |
|
|
#define MIN_NEEDED_TO 4
|
| 212 |
|
|
#define FROM_DIRECTION 1
|
| 213 |
|
|
#define FROM_LOOP ucs4_internal_loop
|
| 214 |
|
|
#define TO_LOOP ucs4_internal_loop /* This is not used. */
|
| 215 |
|
|
#define FUNCTION_NAME __gconv_transform_ucs4_internal
|
| 216 |
|
|
|
| 217 |
|
|
|
| 218 |
|
|
static inline int
|
| 219 |
|
|
ucs4_internal_loop (struct __gconv_step *step,
|
| 220 |
|
|
struct __gconv_step_data *step_data,
|
| 221 |
|
|
const unsigned char **inptrp, const unsigned char *inend,
|
| 222 |
|
|
unsigned char **outptrp, unsigned char *outend,
|
| 223 |
|
|
size_t *irreversible)
|
| 224 |
|
|
{
|
| 225 |
|
|
int flags = step_data->__flags;
|
| 226 |
|
|
const unsigned char *inptr = *inptrp;
|
| 227 |
|
|
unsigned char *outptr = *outptrp;
|
| 228 |
|
|
size_t n_convert = MIN (inend - inptr, outend - outptr) / 4;
|
| 229 |
|
|
int result;
|
| 230 |
|
|
size_t cnt;
|
| 231 |
|
|
|
| 232 |
|
|
for (cnt = 0; cnt < n_convert; ++cnt, inptr += 4)
|
| 233 |
|
|
{
|
| 234 |
|
|
uint32_t inval;
|
| 235 |
|
|
|
| 236 |
|
|
#if __BYTE_ORDER == __LITTLE_ENDIAN
|
| 237 |
|
|
inval = bswap_32 (*(const uint32_t *) inptr);
|
| 238 |
|
|
#else
|
| 239 |
|
|
inval = *(const uint32_t *) inptr;
|
| 240 |
|
|
#endif
|
| 241 |
|
|
|
| 242 |
|
|
if (__builtin_expect (inval > 0x7fffffff, 0))
|
| 243 |
|
|
{
|
| 244 |
|
|
/* The value is too large. We don't try transliteration here since
|
| 245 |
|
|
this is not an error because of the lack of possibilities to
|
| 246 |
|
|
represent the result. This is a genuine bug in the input since
|
| 247 |
|
|
UCS4 does not allow such values. */
|
| 248 |
|
|
if (irreversible == NULL)
|
| 249 |
|
|
/* We are transliterating, don't try to correct anything. */
|
| 250 |
|
|
return __GCONV_ILLEGAL_INPUT;
|
| 251 |
|
|
|
| 252 |
|
|
if (flags & __GCONV_IGNORE_ERRORS)
|
| 253 |
|
|
{
|
| 254 |
|
|
/* Just ignore this character. */
|
| 255 |
|
|
*irreversible = *irreversible + 1;
|
| 256 |
|
|
continue;
|
| 257 |
|
|
}
|
| 258 |
|
|
|
| 259 |
|
|
*inptrp = inptr;
|
| 260 |
|
|
*outptrp = outptr;
|
| 261 |
|
|
return __GCONV_ILLEGAL_INPUT;
|
| 262 |
|
|
}
|
| 263 |
|
|
|
| 264 |
|
|
*((uint32_t *) outptr) = inval;
|
| 265 |
|
|
outptr = (unsigned char *) ((uint32_t *) outptr + 1);
|
| 266 |
|
|
}
|
| 267 |
|
|
|
| 268 |
|
|
*inptrp = inptr;
|
| 269 |
|
|
*outptrp = outptr;
|
| 270 |
|
|
|
| 271 |
|
|
/* Determine the status. */
|
| 272 |
|
|
if (*inptrp == inend)
|
| 273 |
|
|
result = __GCONV_EMPTY_INPUT;
|
| 274 |
|
|
else if (*outptrp + 4 > outend)
|
| 275 |
|
|
result = __GCONV_FULL_OUTPUT;
|
| 276 |
|
|
else
|
| 277 |
|
|
result = __GCONV_INCOMPLETE_INPUT;
|
| 278 |
|
|
|
| 279 |
|
|
return result;
|
| 280 |
|
|
}
|
| 281 |
|
|
|
| 282 |
|
|
#ifndef _STRING_ARCH_unaligned
|
| 283 |
|
|
static inline int
|
| 284 |
|
|
ucs4_internal_loop_unaligned (struct __gconv_step *step,
|
| 285 |
|
|
struct __gconv_step_data *step_data,
|
| 286 |
|
|
const unsigned char **inptrp,
|
| 287 |
|
|
const unsigned char *inend,
|
| 288 |
|
|
unsigned char **outptrp, unsigned char *outend,
|
| 289 |
|
|
size_t *irreversible)
|
| 290 |
|
|
{
|
| 291 |
|
|
int flags = step_data->__flags;
|
| 292 |
|
|
const unsigned char *inptr = *inptrp;
|
| 293 |
|
|
unsigned char *outptr = *outptrp;
|
| 294 |
|
|
size_t n_convert = MIN (inend - inptr, outend - outptr) / 4;
|
| 295 |
|
|
int result;
|
| 296 |
|
|
size_t cnt;
|
| 297 |
|
|
|
| 298 |
|
|
for (cnt = 0; cnt < n_convert; ++cnt, inptr += 4)
|
| 299 |
|
|
{
|
| 300 |
|
|
if (__builtin_expect (inptr[0] > 0x80, 0))
|
| 301 |
|
|
{
|
| 302 |
|
|
/* The value is too large. We don't try transliteration here since
|
| 303 |
|
|
this is not an error because of the lack of possibilities to
|
| 304 |
|
|
represent the result. This is a genuine bug in the input since
|
| 305 |
|
|
UCS4 does not allow such values. */
|
| 306 |
|
|
if (irreversible == NULL)
|
| 307 |
|
|
/* We are transliterating, don't try to correct anything. */
|
| 308 |
|
|
return __GCONV_ILLEGAL_INPUT;
|
| 309 |
|
|
|
| 310 |
|
|
if (flags & __GCONV_IGNORE_ERRORS)
|
| 311 |
|
|
{
|
| 312 |
|
|
/* Just ignore this character. */
|
| 313 |
|
|
*irreversible = *irreversible + 1;
|
| 314 |
|
|
continue;
|
| 315 |
|
|
}
|
| 316 |
|
|
|
| 317 |
|
|
*inptrp = inptr;
|
| 318 |
|
|
*outptrp = outptr;
|
| 319 |
|
|
return __GCONV_ILLEGAL_INPUT;
|
| 320 |
|
|
}
|
| 321 |
|
|
|
| 322 |
|
|
# if __BYTE_ORDER == __LITTLE_ENDIAN
|
| 323 |
|
|
outptr[3] = inptr[0];
|
| 324 |
|
|
outptr[2] = inptr[1];
|
| 325 |
|
|
outptr[1] = inptr[2];
|
| 326 |
|
|
outptr[0] = inptr[3];
|
| 327 |
|
|
# else
|
| 328 |
|
|
outptr[0] = inptr[0];
|
| 329 |
|
|
outptr[1] = inptr[1];
|
| 330 |
|
|
outptr[2] = inptr[2];
|
| 331 |
|
|
outptr[3] = inptr[3];
|
| 332 |
|
|
# endif
|
| 333 |
|
|
outptr += 4;
|
| 334 |
|
|
}
|
| 335 |
|
|
|
| 336 |
|
|
*inptrp = inptr;
|
| 337 |
|
|
*outptrp = outptr;
|
| 338 |
|
|
|
| 339 |
|
|
/* Determine the status. */
|
| 340 |
|
|
if (*inptrp == inend)
|
| 341 |
|
|
result = __GCONV_EMPTY_INPUT;
|
| 342 |
|
|
else if (*outptrp + 4 > outend)
|
| 343 |
|
|
result = __GCONV_FULL_OUTPUT;
|
| 344 |
|
|
else
|
| 345 |
|
|
result = __GCONV_INCOMPLETE_INPUT;
|
| 346 |
|
|
|
| 347 |
|
|
return result;
|
| 348 |
|
|
}
|
| 349 |
|
|
#endif
|
| 350 |
|
|
|
| 351 |
|
|
|
| 352 |
|
|
static inline int
|
| 353 |
|
|
ucs4_internal_loop_single (struct __gconv_step *step,
|
| 354 |
|
|
struct __gconv_step_data *step_data,
|
| 355 |
|
|
const unsigned char **inptrp,
|
| 356 |
|
|
const unsigned char *inend,
|
| 357 |
|
|
unsigned char **outptrp, unsigned char *outend,
|
| 358 |
|
|
size_t *irreversible)
|
| 359 |
|
|
{
|
| 360 |
|
|
mbstate_t *state = step_data->__statep;
|
| 361 |
|
|
int flags = step_data->__flags;
|
| 362 |
|
|
size_t cnt = state->__count & 7;
|
| 363 |
|
|
|
| 364 |
|
|
while (*inptrp < inend && cnt < 4)
|
| 365 |
|
|
state->__value.__wchb[cnt++] = *(*inptrp)++;
|
| 366 |
|
|
|
| 367 |
|
|
if (__builtin_expect (cnt < 4, 0))
|
| 368 |
|
|
{
|
| 369 |
|
|
/* Still not enough bytes. Store the ones in the input buffer. */
|
| 370 |
|
|
state->__count &= ~7;
|
| 371 |
|
|
state->__count |= cnt;
|
| 372 |
|
|
|
| 373 |
|
|
return __GCONV_INCOMPLETE_INPUT;
|
| 374 |
|
|
}
|
| 375 |
|
|
|
| 376 |
|
|
if (__builtin_expect (((unsigned char *) state->__value.__wchb)[0] > 0x80,
|
| 377 |
|
|
0))
|
| 378 |
|
|
{
|
| 379 |
|
|
/* The value is too large. We don't try transliteration here since
|
| 380 |
|
|
this is not an error because of the lack of possibilities to
|
| 381 |
|
|
represent the result. This is a genuine bug in the input since
|
| 382 |
|
|
UCS4 does not allow such values. */
|
| 383 |
|
|
if (!(flags & __GCONV_IGNORE_ERRORS))
|
| 384 |
|
|
{
|
| 385 |
|
|
*inptrp -= cnt - (state->__count & 7);
|
| 386 |
|
|
return __GCONV_ILLEGAL_INPUT;
|
| 387 |
|
|
}
|
| 388 |
|
|
}
|
| 389 |
|
|
else
|
| 390 |
|
|
{
|
| 391 |
|
|
#if __BYTE_ORDER == __LITTLE_ENDIAN
|
| 392 |
|
|
(*outptrp)[0] = state->__value.__wchb[3];
|
| 393 |
|
|
(*outptrp)[1] = state->__value.__wchb[2];
|
| 394 |
|
|
(*outptrp)[2] = state->__value.__wchb[1];
|
| 395 |
|
|
(*outptrp)[3] = state->__value.__wchb[0];
|
| 396 |
|
|
#elif __BYTE_ORDER == __BIG_ENDIAN
|
| 397 |
|
|
(*outptrp)[0] = state->__value.__wchb[0];
|
| 398 |
|
|
(*outptrp)[1] = state->__value.__wchb[1];
|
| 399 |
|
|
(*outptrp)[2] = state->__value.__wchb[2];
|
| 400 |
|
|
(*outptrp)[3] = state->__value.__wchb[3];
|
| 401 |
|
|
#endif
|
| 402 |
|
|
|
| 403 |
|
|
*outptrp += 4;
|
| 404 |
|
|
}
|
| 405 |
|
|
|
| 406 |
|
|
/* Clear the state buffer. */
|
| 407 |
|
|
state->__count &= ~7;
|
| 408 |
|
|
|
| 409 |
|
|
return __GCONV_OK;
|
| 410 |
|
|
}
|
| 411 |
|
|
|
| 412 |
|
|
#include <iconv/skeleton.c>
|
| 413 |
|
|
|
| 414 |
|
|
|
| 415 |
|
|
/* Similarly for the little endian form. */
|
| 416 |
|
|
#define DEFINE_INIT 0
|
| 417 |
|
|
#define DEFINE_FINI 0
|
| 418 |
|
|
#define MIN_NEEDED_FROM 4
|
| 419 |
|
|
#define MIN_NEEDED_TO 4
|
| 420 |
|
|
#define FROM_DIRECTION 1
|
| 421 |
|
|
#define FROM_LOOP internal_ucs4le_loop
|
| 422 |
|
|
#define TO_LOOP internal_ucs4le_loop /* This is not used. */
|
| 423 |
|
|
#define FUNCTION_NAME __gconv_transform_internal_ucs4le
|
| 424 |
|
|
|
| 425 |
|
|
|
| 426 |
|
|
static inline int
|
| 427 |
|
|
internal_ucs4le_loop (struct __gconv_step *step,
|
| 428 |
|
|
struct __gconv_step_data *step_data,
|
| 429 |
|
|
const unsigned char **inptrp, const unsigned char *inend,
|
| 430 |
|
|
unsigned char **outptrp, unsigned char *outend,
|
| 431 |
|
|
size_t *irreversible)
|
| 432 |
|
|
{
|
| 433 |
|
|
const unsigned char *inptr = *inptrp;
|
| 434 |
|
|
unsigned char *outptr = *outptrp;
|
| 435 |
|
|
size_t n_convert = MIN (inend - inptr, outend - outptr) / 4;
|
| 436 |
|
|
int result;
|
| 437 |
|
|
|
| 438 |
|
|
#if __BYTE_ORDER == __BIG_ENDIAN
|
| 439 |
|
|
/* Sigh, we have to do some real work. */
|
| 440 |
|
|
size_t cnt;
|
| 441 |
|
|
|
| 442 |
|
|
for (cnt = 0; cnt < n_convert; ++cnt, inptr += 4)
|
| 443 |
|
|
*((uint32_t *) outptr)++ = bswap_32 (*(const uint32_t *) inptr);
|
| 444 |
|
|
|
| 445 |
|
|
*inptrp = inptr;
|
| 446 |
|
|
*outptrp = outptr;
|
| 447 |
|
|
#elif __BYTE_ORDER == __LITTLE_ENDIAN
|
| 448 |
|
|
/* Simply copy the data. */
|
| 449 |
|
|
*inptrp = inptr + n_convert * 4;
|
| 450 |
|
|
*outptrp = memcpy (outptr, inptr, n_convert * 4);
|
| 451 |
|
|
*outptrp += n_convert * 4;
|
| 452 |
|
|
#else
|
| 453 |
|
|
# error "This endianess is not supported."
|
| 454 |
|
|
#endif
|
| 455 |
|
|
|
| 456 |
|
|
/* Determine the status. */
|
| 457 |
|
|
if (*inptrp == inend)
|
| 458 |
|
|
result = __GCONV_EMPTY_INPUT;
|
| 459 |
|
|
else if (*outptrp + 4 > outend)
|
| 460 |
|
|
result = __GCONV_FULL_OUTPUT;
|
| 461 |
|
|
else
|
| 462 |
|
|
result = __GCONV_INCOMPLETE_INPUT;
|
| 463 |
|
|
|
| 464 |
|
|
return result;
|
| 465 |
|
|
}
|
| 466 |
|
|
|
| 467 |
|
|
#ifndef _STRING_ARCH_unaligned
|
| 468 |
|
|
static inline int
|
| 469 |
|
|
internal_ucs4le_loop_unaligned (struct __gconv_step *step,
|
| 470 |
|
|
struct __gconv_step_data *step_data,
|
| 471 |
|
|
const unsigned char **inptrp,
|
| 472 |
|
|
const unsigned char *inend,
|
| 473 |
|
|
unsigned char **outptrp, unsigned char *outend,
|
| 474 |
|
|
size_t *irreversible)
|
| 475 |
|
|
{
|
| 476 |
|
|
const unsigned char *inptr = *inptrp;
|
| 477 |
|
|
unsigned char *outptr = *outptrp;
|
| 478 |
|
|
size_t n_convert = MIN (inend - inptr, outend - outptr) / 4;
|
| 479 |
|
|
int result;
|
| 480 |
|
|
|
| 481 |
|
|
# if __BYTE_ORDER == __BIG_ENDIAN
|
| 482 |
|
|
/* Sigh, we have to do some real work. */
|
| 483 |
|
|
size_t cnt;
|
| 484 |
|
|
|
| 485 |
|
|
for (cnt = 0; cnt < n_convert; ++cnt, inptr += 4, outptr += 4)
|
| 486 |
|
|
{
|
| 487 |
|
|
outptr[0] = inptr[3];
|
| 488 |
|
|
outptr[1] = inptr[2];
|
| 489 |
|
|
outptr[2] = inptr[1];
|
| 490 |
|
|
outptr[3] = inptr[0];
|
| 491 |
|
|
}
|
| 492 |
|
|
|
| 493 |
|
|
*inptrp = inptr;
|
| 494 |
|
|
*outptrp = outptr;
|
| 495 |
|
|
# elif __BYTE_ORDER == __LITTLE_ENDIAN
|
| 496 |
|
|
/* Simply copy the data. */
|
| 497 |
|
|
*inptrp = inptr + n_convert * 4;
|
| 498 |
|
|
*outptrp = memcpy (outptr, inptr, n_convert * 4);
|
| 499 |
|
|
*outptrp += n_convert * 4;
|
| 500 |
|
|
# else
|
| 501 |
|
|
# error "This endianess is not supported."
|
| 502 |
|
|
# endif
|
| 503 |
|
|
|
| 504 |
|
|
/* Determine the status. */
|
| 505 |
|
|
if (*inptrp + 4 > inend)
|
| 506 |
|
|
result = __GCONV_EMPTY_INPUT;
|
| 507 |
|
|
else if (*outptrp + 4 > outend)
|
| 508 |
|
|
result = __GCONV_FULL_OUTPUT;
|
| 509 |
|
|
else
|
| 510 |
|
|
result = __GCONV_INCOMPLETE_INPUT;
|
| 511 |
|
|
|
| 512 |
|
|
return result;
|
| 513 |
|
|
}
|
| 514 |
|
|
#endif
|
| 515 |
|
|
|
| 516 |
|
|
|
| 517 |
|
|
static inline int
|
| 518 |
|
|
internal_ucs4le_loop_single (struct __gconv_step *step,
|
| 519 |
|
|
struct __gconv_step_data *step_data,
|
| 520 |
|
|
const unsigned char **inptrp,
|
| 521 |
|
|
const unsigned char *inend,
|
| 522 |
|
|
unsigned char **outptrp, unsigned char *outend,
|
| 523 |
|
|
size_t *irreversible)
|
| 524 |
|
|
{
|
| 525 |
|
|
mbstate_t *state = step_data->__statep;
|
| 526 |
|
|
size_t cnt = state->__count & 7;
|
| 527 |
|
|
|
| 528 |
|
|
while (*inptrp < inend && cnt < 4)
|
| 529 |
|
|
state->__value.__wchb[cnt++] = *(*inptrp)++;
|
| 530 |
|
|
|
| 531 |
|
|
if (__builtin_expect (cnt < 4, 0))
|
| 532 |
|
|
{
|
| 533 |
|
|
/* Still not enough bytes. Store the ones in the input buffer. */
|
| 534 |
|
|
state->__count &= ~7;
|
| 535 |
|
|
state->__count |= cnt;
|
| 536 |
|
|
|
| 537 |
|
|
return __GCONV_INCOMPLETE_INPUT;
|
| 538 |
|
|
}
|
| 539 |
|
|
|
| 540 |
|
|
#if __BYTE_ORDER == __BIG_ENDIAN
|
| 541 |
|
|
(*outptrp)[0] = state->__value.__wchb[3];
|
| 542 |
|
|
(*outptrp)[1] = state->__value.__wchb[2];
|
| 543 |
|
|
(*outptrp)[2] = state->__value.__wchb[1];
|
| 544 |
|
|
(*outptrp)[3] = state->__value.__wchb[0];
|
| 545 |
|
|
|
| 546 |
|
|
*outptrp += 4;
|
| 547 |
|
|
#else
|
| 548 |
|
|
{
|
| 549 |
|
|
/* XXX unaligned */
|
| 550 |
|
|
uint32_t **p = (uint32_t **)outptrp;
|
| 551 |
|
|
uint32_t *q = *p;
|
| 552 |
|
|
*q = state->__value.__wch;
|
| 553 |
|
|
outptrp = (unsigned char **)(p + 1);
|
| 554 |
|
|
}
|
| 555 |
|
|
#endif
|
| 556 |
|
|
|
| 557 |
|
|
/* Clear the state buffer. */
|
| 558 |
|
|
state->__count &= ~7;
|
| 559 |
|
|
|
| 560 |
|
|
return __GCONV_OK;
|
| 561 |
|
|
}
|
| 562 |
|
|
|
| 563 |
|
|
#include <iconv/skeleton.c>
|
| 564 |
|
|
|
| 565 |
|
|
|
| 566 |
|
|
/* And finally from UCS4-LE to the internal encoding. */
|
| 567 |
|
|
#define DEFINE_INIT 0
|
| 568 |
|
|
#define DEFINE_FINI 0
|
| 569 |
|
|
#define MIN_NEEDED_FROM 4
|
| 570 |
|
|
#define MIN_NEEDED_TO 4
|
| 571 |
|
|
#define FROM_DIRECTION 1
|
| 572 |
|
|
#define FROM_LOOP ucs4le_internal_loop
|
| 573 |
|
|
#define TO_LOOP ucs4le_internal_loop /* This is not used. */
|
| 574 |
|
|
#define FUNCTION_NAME __gconv_transform_ucs4le_internal
|
| 575 |
|
|
|
| 576 |
|
|
|
| 577 |
|
|
static inline int
|
| 578 |
|
|
ucs4le_internal_loop (struct __gconv_step *step,
|
| 579 |
|
|
struct __gconv_step_data *step_data,
|
| 580 |
|
|
const unsigned char **inptrp, const unsigned char *inend,
|
| 581 |
|
|
unsigned char **outptrp, unsigned char *outend,
|
| 582 |
|
|
size_t *irreversible)
|
| 583 |
|
|
{
|
| 584 |
|
|
int flags = step_data->__flags;
|
| 585 |
|
|
const unsigned char *inptr = *inptrp;
|
| 586 |
|
|
unsigned char *outptr = *outptrp;
|
| 587 |
|
|
size_t n_convert = MIN (inend - inptr, outend - outptr) / 4;
|
| 588 |
|
|
int result;
|
| 589 |
|
|
size_t cnt;
|
| 590 |
|
|
|
| 591 |
|
|
for (cnt = 0; cnt < n_convert; ++cnt, inptr += 4)
|
| 592 |
|
|
{
|
| 593 |
|
|
uint32_t inval;
|
| 594 |
|
|
|
| 595 |
|
|
#if __BYTE_ORDER == __BIG_ENDIAN
|
| 596 |
|
|
inval = bswap_32 (*(const uint32_t *) inptr);
|
| 597 |
|
|
#else
|
| 598 |
|
|
inval = *(const uint32_t *) inptr;
|
| 599 |
|
|
#endif
|
| 600 |
|
|
|
| 601 |
|
|
if (__builtin_expect (inval > 0x7fffffff, 0))
|
| 602 |
|
|
{
|
| 603 |
|
|
/* The value is too large. We don't try transliteration here since
|
| 604 |
|
|
this is not an error because of the lack of possibilities to
|
| 605 |
|
|
represent the result. This is a genuine bug in the input since
|
| 606 |
|
|
UCS4 does not allow such values. */
|
| 607 |
|
|
if (irreversible == NULL)
|
| 608 |
|
|
/* We are transliterating, don't try to correct anything. */
|
| 609 |
|
|
return __GCONV_ILLEGAL_INPUT;
|
| 610 |
|
|
|
| 611 |
|
|
if (flags & __GCONV_IGNORE_ERRORS)
|
| 612 |
|
|
{
|
| 613 |
|
|
/* Just ignore this character. */
|
| 614 |
|
|
++*irreversible;
|
| 615 |
|
|
continue;
|
| 616 |
|
|
}
|
| 617 |
|
|
|
| 618 |
|
|
return __GCONV_ILLEGAL_INPUT;
|
| 619 |
|
|
}
|
| 620 |
|
|
|
| 621 |
|
|
*((uint32_t *) outptr) = inval;
|
| 622 |
|
|
outptr = (unsigned char *)((uint32_t *) outptr + 1);
|
| 623 |
|
|
}
|
| 624 |
|
|
|
| 625 |
|
|
*inptrp = inptr;
|
| 626 |
|
|
*outptrp = outptr;
|
| 627 |
|
|
|
| 628 |
|
|
/* Determine the status. */
|
| 629 |
|
|
if (*inptrp == inend)
|
| 630 |
|
|
result = __GCONV_EMPTY_INPUT;
|
| 631 |
|
|
else if (*outptrp + 4 > outend)
|
| 632 |
|
|
result = __GCONV_FULL_OUTPUT;
|
| 633 |
|
|
else
|
| 634 |
|
|
result = __GCONV_INCOMPLETE_INPUT;
|
| 635 |
|
|
|
| 636 |
|
|
return result;
|
| 637 |
|
|
}
|
| 638 |
|
|
|
| 639 |
|
|
#ifndef _STRING_ARCH_unaligned
|
| 640 |
|
|
static inline int
|
| 641 |
|
|
ucs4le_internal_loop_unaligned (struct __gconv_step *step,
|
| 642 |
|
|
struct __gconv_step_data *step_data,
|
| 643 |
|
|
const unsigned char **inptrp,
|
| 644 |
|
|
const unsigned char *inend,
|
| 645 |
|
|
unsigned char **outptrp, unsigned char *outend,
|
| 646 |
|
|
size_t *irreversible)
|
| 647 |
|
|
{
|
| 648 |
|
|
int flags = step_data->__flags;
|
| 649 |
|
|
const unsigned char *inptr = *inptrp;
|
| 650 |
|
|
unsigned char *outptr = *outptrp;
|
| 651 |
|
|
size_t n_convert = MIN (inend - inptr, outend - outptr) / 4;
|
| 652 |
|
|
int result;
|
| 653 |
|
|
size_t cnt;
|
| 654 |
|
|
|
| 655 |
|
|
for (cnt = 0; cnt < n_convert; ++cnt, inptr += 4)
|
| 656 |
|
|
{
|
| 657 |
|
|
if (__builtin_expect (inptr[3] > 0x80, 0))
|
| 658 |
|
|
{
|
| 659 |
|
|
/* The value is too large. We don't try transliteration here since
|
| 660 |
|
|
this is not an error because of the lack of possibilities to
|
| 661 |
|
|
represent the result. This is a genuine bug in the input since
|
| 662 |
|
|
UCS4 does not allow such values. */
|
| 663 |
|
|
if (irreversible == NULL)
|
| 664 |
|
|
/* We are transliterating, don't try to correct anything. */
|
| 665 |
|
|
return __GCONV_ILLEGAL_INPUT;
|
| 666 |
|
|
|
| 667 |
|
|
if (flags & __GCONV_IGNORE_ERRORS)
|
| 668 |
|
|
{
|
| 669 |
|
|
/* Just ignore this character. */
|
| 670 |
|
|
++*irreversible;
|
| 671 |
|
|
continue;
|
| 672 |
|
|
}
|
| 673 |
|
|
|
| 674 |
|
|
*inptrp = inptr;
|
| 675 |
|
|
*outptrp = outptr;
|
| 676 |
|
|
return __GCONV_ILLEGAL_INPUT;
|
| 677 |
|
|
}
|
| 678 |
|
|
|
| 679 |
|
|
# if __BYTE_ORDER == __BIG_ENDIAN
|
| 680 |
|
|
outptr[3] = inptr[0];
|
| 681 |
|
|
outptr[2] = inptr[1];
|
| 682 |
|
|
outptr[1] = inptr[2];
|
| 683 |
|
|
outptr[0] = inptr[3];
|
| 684 |
|
|
# else
|
| 685 |
|
|
outptr[0] = inptr[0];
|
| 686 |
|
|
outptr[1] = inptr[1];
|
| 687 |
|
|
outptr[2] = inptr[2];
|
| 688 |
|
|
outptr[3] = inptr[3];
|
| 689 |
|
|
# endif
|
| 690 |
|
|
|
| 691 |
|
|
outptr += 4;
|
| 692 |
|
|
}
|
| 693 |
|
|
|
| 694 |
|
|
*inptrp = inptr;
|
| 695 |
|
|
*outptrp = outptr;
|
| 696 |
|
|
|
| 697 |
|
|
/* Determine the status. */
|
| 698 |
|
|
if (*inptrp == inend)
|
| 699 |
|
|
result = __GCONV_EMPTY_INPUT;
|
| 700 |
|
|
else if (*outptrp + 4 > outend)
|
| 701 |
|
|
result = __GCONV_FULL_OUTPUT;
|
| 702 |
|
|
else
|
| 703 |
|
|
result = __GCONV_INCOMPLETE_INPUT;
|
| 704 |
|
|
|
| 705 |
|
|
return result;
|
| 706 |
|
|
}
|
| 707 |
|
|
#endif
|
| 708 |
|
|
|
| 709 |
|
|
|
| 710 |
|
|
static inline int
|
| 711 |
|
|
ucs4le_internal_loop_single (struct __gconv_step *step,
|
| 712 |
|
|
struct __gconv_step_data *step_data,
|
| 713 |
|
|
const unsigned char **inptrp,
|
| 714 |
|
|
const unsigned char *inend,
|
| 715 |
|
|
unsigned char **outptrp, unsigned char *outend,
|
| 716 |
|
|
size_t *irreversible)
|
| 717 |
|
|
{
|
| 718 |
|
|
mbstate_t *state = step_data->__statep;
|
| 719 |
|
|
int flags = step_data->__flags;
|
| 720 |
|
|
size_t cnt = state->__count & 7;
|
| 721 |
|
|
|
| 722 |
|
|
while (*inptrp < inend && cnt < 4)
|
| 723 |
|
|
state->__value.__wchb[cnt++] = *(*inptrp)++;
|
| 724 |
|
|
|
| 725 |
|
|
if (__builtin_expect (cnt < 4, 0))
|
| 726 |
|
|
{
|
| 727 |
|
|
/* Still not enough bytes. Store the ones in the input buffer. */
|
| 728 |
|
|
state->__count &= ~7;
|
| 729 |
|
|
state->__count |= cnt;
|
| 730 |
|
|
|
| 731 |
|
|
return __GCONV_INCOMPLETE_INPUT;
|
| 732 |
|
|
}
|
| 733 |
|
|
|
| 734 |
|
|
if (__builtin_expect (((unsigned char *) state->__value.__wchb)[3] > 0x80,
|
| 735 |
|
|
0))
|
| 736 |
|
|
{
|
| 737 |
|
|
/* The value is too large. We don't try transliteration here since
|
| 738 |
|
|
this is not an error because of the lack of possibilities to
|
| 739 |
|
|
represent the result. This is a genuine bug in the input since
|
| 740 |
|
|
UCS4 does not allow such values. */
|
| 741 |
|
|
if (!(flags & __GCONV_IGNORE_ERRORS))
|
| 742 |
|
|
return __GCONV_ILLEGAL_INPUT;
|
| 743 |
|
|
}
|
| 744 |
|
|
else
|
| 745 |
|
|
{
|
| 746 |
|
|
#if __BYTE_ORDER == __BIG_ENDIAN
|
| 747 |
|
|
(*outptrp)[0] = state->__value.__wchb[3];
|
| 748 |
|
|
(*outptrp)[1] = state->__value.__wchb[2];
|
| 749 |
|
|
(*outptrp)[2] = state->__value.__wchb[1];
|
| 750 |
|
|
(*outptrp)[3] = state->__value.__wchb[0];
|
| 751 |
|
|
#elif __BYTE_ORDER == __BIG_ENDIAN
|
| 752 |
|
|
(*outptrp)[0] = state->__value.__wchb[0];
|
| 753 |
|
|
(*outptrp)[1] = state->__value.__wchb[1];
|
| 754 |
|
|
(*outptrp)[2] = state->__value.__wchb[2];
|
| 755 |
|
|
(*outptrp)[3] = state->__value.__wchb[3];
|
| 756 |
|
|
#endif
|
| 757 |
|
|
|
| 758 |
|
|
*outptrp += 4;
|
| 759 |
|
|
}
|
| 760 |
|
|
|
| 761 |
|
|
/* Clear the state buffer. */
|
| 762 |
|
|
state->__count &= ~7;
|
| 763 |
|
|
|
| 764 |
|
|
return __GCONV_OK;
|
| 765 |
|
|
}
|
| 766 |
|
|
|
| 767 |
|
|
#include <iconv/skeleton.c>
|
| 768 |
|
|
|
| 769 |
|
|
|
| 770 |
|
|
/* Convert from ISO 646-IRV to the internal (UCS4-like) format. */
|
| 771 |
|
|
#define DEFINE_INIT 0
|
| 772 |
|
|
#define DEFINE_FINI 0
|
| 773 |
|
|
#define MIN_NEEDED_FROM 1
|
| 774 |
|
|
#define MIN_NEEDED_TO 4
|
| 775 |
|
|
#define FROM_DIRECTION 1
|
| 776 |
|
|
#define FROM_LOOP ascii_internal_loop
|
| 777 |
|
|
#define TO_LOOP ascii_internal_loop /* This is not used. */
|
| 778 |
|
|
#define FUNCTION_NAME __gconv_transform_ascii_internal
|
| 779 |
|
|
#define ONE_DIRECTION 1
|
| 780 |
|
|
|
| 781 |
|
|
#define MIN_NEEDED_INPUT MIN_NEEDED_FROM
|
| 782 |
|
|
#define MIN_NEEDED_OUTPUT MIN_NEEDED_TO
|
| 783 |
|
|
#define LOOPFCT FROM_LOOP
|
| 784 |
|
|
#define BODY \
|
| 785 |
|
|
{ \
|
| 786 |
|
|
if (__builtin_expect (*inptr > '\x7f', 0)) \
|
| 787 |
|
|
{ \
|
| 788 |
|
|
/* The value is too large. We don't try transliteration here since \
|
| 789 |
|
|
this is not an error because of the lack of possibilities to \
|
| 790 |
|
|
represent the result. This is a genuine bug in the input since \
|
| 791 |
|
|
ASCII does not allow such values. */ \
|
| 792 |
|
|
if (! ignore_errors_p ()) \
|
| 793 |
|
|
{ \
|
| 794 |
|
|
/* This is no correct ANSI_X3.4-1968 character. */ \
|
| 795 |
|
|
result = __GCONV_ILLEGAL_INPUT; \
|
| 796 |
|
|
break; \
|
| 797 |
|
|
} \
|
| 798 |
|
|
\
|
| 799 |
|
|
*irreversible = *irreversible + 1; \
|
| 800 |
|
|
++inptr; \
|
| 801 |
|
|
} \
|
| 802 |
|
|
else \
|
| 803 |
|
|
{ \
|
| 804 |
|
|
/* It's an one byte sequence. */ \
|
| 805 |
|
|
*((uint32_t *) outptr) = *inptr; \
|
| 806 |
|
|
++inptr; \
|
| 807 |
|
|
outptr = (unsigned char *)((uint32_t *) outptr + 1); \
|
| 808 |
|
|
} \
|
| 809 |
|
|
}
|
| 810 |
|
|
#define LOOP_NEED_FLAGS
|
| 811 |
|
|
#include <iconv/loop.c>
|
| 812 |
|
|
#include <iconv/skeleton.c>
|
| 813 |
|
|
|
| 814 |
|
|
|
| 815 |
|
|
/* Convert from the internal (UCS4-like) format to ISO 646-IRV. */
|
| 816 |
|
|
#define DEFINE_INIT 0
|
| 817 |
|
|
#define DEFINE_FINI 0
|
| 818 |
|
|
#define MIN_NEEDED_FROM 4
|
| 819 |
|
|
#define MIN_NEEDED_TO 1
|
| 820 |
|
|
#define FROM_DIRECTION 1
|
| 821 |
|
|
#define FROM_LOOP internal_ascii_loop
|
| 822 |
|
|
#define TO_LOOP internal_ascii_loop /* This is not used. */
|
| 823 |
|
|
#define FUNCTION_NAME __gconv_transform_internal_ascii
|
| 824 |
|
|
#define ONE_DIRECTION 1
|
| 825 |
|
|
|
| 826 |
|
|
#define MIN_NEEDED_INPUT MIN_NEEDED_FROM
|
| 827 |
|
|
#define MIN_NEEDED_OUTPUT MIN_NEEDED_TO
|
| 828 |
|
|
#define LOOPFCT FROM_LOOP
|
| 829 |
|
|
#define BODY \
|
| 830 |
|
|
{ \
|
| 831 |
|
|
if (__builtin_expect (*((const uint32_t *) inptr) > 0x7f, 0)) \
|
| 832 |
|
|
{ \
|
| 833 |
|
|
UNICODE_TAG_HANDLER (*((const uint32_t *) inptr), 4); \
|
| 834 |
|
|
STANDARD_ERR_HANDLER (4); \
|
| 835 |
|
|
} \
|
| 836 |
|
|
else \
|
| 837 |
|
|
{ \
|
| 838 |
|
|
/* It's an one byte sequence. */ \
|
| 839 |
|
|
*outptr++ = *((const uint32_t *) inptr); \
|
| 840 |
|
|
inptr = ((const uint32_t *) inptr + 1); \
|
| 841 |
|
|
} \
|
| 842 |
|
|
}
|
| 843 |
|
|
#define LOOP_NEED_FLAGS
|
| 844 |
|
|
#include <iconv/loop.c>
|
| 845 |
|
|
#include <iconv/skeleton.c>
|
| 846 |
|
|
|
| 847 |
|
|
|
| 848 |
|
|
/* Convert from the internal (UCS4-like) format to UTF-8. */
|
| 849 |
|
|
#define DEFINE_INIT 0
|
| 850 |
|
|
#define DEFINE_FINI 0
|
| 851 |
|
|
#define MIN_NEEDED_FROM 4
|
| 852 |
|
|
#define MIN_NEEDED_TO 1
|
| 853 |
|
|
#define MAX_NEEDED_TO 6
|
| 854 |
|
|
#define FROM_DIRECTION 1
|
| 855 |
|
|
#define FROM_LOOP internal_utf8_loop
|
| 856 |
|
|
#define TO_LOOP internal_utf8_loop /* This is not used. */
|
| 857 |
|
|
#define FUNCTION_NAME __gconv_transform_internal_utf8
|
| 858 |
|
|
#define ONE_DIRECTION 1
|
| 859 |
|
|
|
| 860 |
|
|
#define MIN_NEEDED_INPUT MIN_NEEDED_FROM
|
| 861 |
|
|
#define MIN_NEEDED_OUTPUT MIN_NEEDED_TO
|
| 862 |
|
|
#define MAX_NEEDED_OUTPUT MAX_NEEDED_TO
|
| 863 |
|
|
#define LOOPFCT FROM_LOOP
|
| 864 |
|
|
#define BODY \
|
| 865 |
|
|
{ \
|
| 866 |
|
|
uint32_t wc = *((const uint32_t *) inptr); \
|
| 867 |
|
|
\
|
| 868 |
|
|
if (wc < 0x80) \
|
| 869 |
|
|
/* It's an one byte sequence. */ \
|
| 870 |
|
|
*outptr++ = (unsigned char) wc; \
|
| 871 |
|
|
else if (__builtin_expect (wc <= 0x7fffffff, 1)) \
|
| 872 |
|
|
{ \
|
| 873 |
|
|
size_t step; \
|
| 874 |
|
|
char *start; \
|
| 875 |
|
|
\
|
| 876 |
|
|
for (step = 2; step < 6; ++step) \
|
| 877 |
|
|
if ((wc & (~(uint32_t)0 << (5 * step + 1))) == 0) \
|
| 878 |
|
|
break; \
|
| 879 |
|
|
\
|
| 880 |
|
|
if (__builtin_expect (outptr + step > outend, 0)) \
|
| 881 |
|
|
{ \
|
| 882 |
|
|
/* Too long. */ \
|
| 883 |
|
|
result = __GCONV_FULL_OUTPUT; \
|
| 884 |
|
|
break; \
|
| 885 |
|
|
} \
|
| 886 |
|
|
\
|
| 887 |
|
|
start = outptr; \
|
| 888 |
|
|
*outptr = (unsigned char) (~0xff >> step); \
|
| 889 |
|
|
outptr += step; \
|
| 890 |
|
|
--step; \
|
| 891 |
|
|
do \
|
| 892 |
|
|
{ \
|
| 893 |
|
|
start[step] = 0x80 | (wc & 0x3f); \
|
| 894 |
|
|
wc >>= 6; \
|
| 895 |
|
|
} \
|
| 896 |
|
|
while (--step > 0); \
|
| 897 |
|
|
start[0] |= wc; \
|
| 898 |
|
|
} \
|
| 899 |
|
|
else \
|
| 900 |
|
|
{ \
|
| 901 |
|
|
STANDARD_ERR_HANDLER (4); \
|
| 902 |
|
|
} \
|
| 903 |
|
|
\
|
| 904 |
|
|
inptr += 4; \
|
| 905 |
|
|
}
|
| 906 |
|
|
#define LOOP_NEED_FLAGS
|
| 907 |
|
|
#include <iconv/loop.c>
|
| 908 |
|
|
#include <iconv/skeleton.c>
|
| 909 |
|
|
|
| 910 |
|
|
|
| 911 |
|
|
/* Convert from UTF-8 to the internal (UCS4-like) format. */
|
| 912 |
|
|
#define DEFINE_INIT 0
|
| 913 |
|
|
#define DEFINE_FINI 0
|
| 914 |
|
|
#define MIN_NEEDED_FROM 1
|
| 915 |
|
|
#define MAX_NEEDED_FROM 6
|
| 916 |
|
|
#define MIN_NEEDED_TO 4
|
| 917 |
|
|
#define FROM_DIRECTION 1
|
| 918 |
|
|
#define FROM_LOOP utf8_internal_loop
|
| 919 |
|
|
#define TO_LOOP utf8_internal_loop /* This is not used. */
|
| 920 |
|
|
#define FUNCTION_NAME __gconv_transform_utf8_internal
|
| 921 |
|
|
#define ONE_DIRECTION 1
|
| 922 |
|
|
|
| 923 |
|
|
#define MIN_NEEDED_INPUT MIN_NEEDED_FROM
|
| 924 |
|
|
#define MAX_NEEDED_INPUT MAX_NEEDED_FROM
|
| 925 |
|
|
#define MIN_NEEDED_OUTPUT MIN_NEEDED_TO
|
| 926 |
|
|
#define LOOPFCT FROM_LOOP
|
| 927 |
|
|
#define BODY \
|
| 928 |
|
|
{ \
|
| 929 |
|
|
uint32_t ch; \
|
| 930 |
|
|
uint32_t cnt; \
|
| 931 |
|
|
uint32_t i; \
|
| 932 |
|
|
\
|
| 933 |
|
|
/* Next input byte. */ \
|
| 934 |
|
|
ch = *inptr; \
|
| 935 |
|
|
\
|
| 936 |
|
|
if (ch < 0x80) \
|
| 937 |
|
|
{ \
|
| 938 |
|
|
/* One byte sequence. */ \
|
| 939 |
|
|
cnt = 1; \
|
| 940 |
|
|
++inptr; \
|
| 941 |
|
|
} \
|
| 942 |
|
|
else \
|
| 943 |
|
|
{ \
|
| 944 |
|
|
if (ch >= 0xc2 && ch < 0xe0) \
|
| 945 |
|
|
{ \
|
| 946 |
|
|
/* We expect two bytes. The first byte cannot be 0xc0 or 0xc1, \
|
| 947 |
|
|
otherwise the wide character could have been represented \
|
| 948 |
|
|
using a single byte. */ \
|
| 949 |
|
|
cnt = 2; \
|
| 950 |
|
|
ch &= 0x1f; \
|
| 951 |
|
|
} \
|
| 952 |
|
|
else if (__builtin_expect ((ch & 0xf0) == 0xe0, 1)) \
|
| 953 |
|
|
{ \
|
| 954 |
|
|
/* We expect three bytes. */ \
|
| 955 |
|
|
cnt = 3; \
|
| 956 |
|
|
ch &= 0x0f; \
|
| 957 |
|
|
} \
|
| 958 |
|
|
else if (__builtin_expect ((ch & 0xf8) == 0xf0, 1)) \
|
| 959 |
|
|
{ \
|
| 960 |
|
|
/* We expect four bytes. */ \
|
| 961 |
|
|
cnt = 4; \
|
| 962 |
|
|
ch &= 0x07; \
|
| 963 |
|
|
} \
|
| 964 |
|
|
else if (__builtin_expect ((ch & 0xfc) == 0xf8, 1)) \
|
| 965 |
|
|
{ \
|
| 966 |
|
|
/* We expect five bytes. */ \
|
| 967 |
|
|
cnt = 5; \
|
| 968 |
|
|
ch &= 0x03; \
|
| 969 |
|
|
} \
|
| 970 |
|
|
else if (__builtin_expect ((ch & 0xfe) == 0xfc, 1)) \
|
| 971 |
|
|
{ \
|
| 972 |
|
|
/* We expect six bytes. */ \
|
| 973 |
|
|
cnt = 6; \
|
| 974 |
|
|
ch &= 0x01; \
|
| 975 |
|
|
} \
|
| 976 |
|
|
else \
|
| 977 |
|
|
{ \
|
| 978 |
|
|
int skipped; \
|
| 979 |
|
|
\
|
| 980 |
|
|
if (! ignore_errors_p ()) \
|
| 981 |
|
|
{ \
|
| 982 |
|
|
/* This is an illegal encoding. */ \
|
| 983 |
|
|
result = __GCONV_ILLEGAL_INPUT; \
|
| 984 |
|
|
break; \
|
| 985 |
|
|
} \
|
| 986 |
|
|
\
|
| 987 |
|
|
/* Search the end of this ill-formed UTF-8 character. This \
|
| 988 |
|
|
is the next byte with (x & 0xc0) != 0x80. */ \
|
| 989 |
|
|
skipped = 0; \
|
| 990 |
|
|
do \
|
| 991 |
|
|
{ \
|
| 992 |
|
|
++inptr; \
|
| 993 |
|
|
++skipped; \
|
| 994 |
|
|
} \
|
| 995 |
|
|
while (inptr < inend && (*inptr & 0xc0) == 0x80 && skipped < 5); \
|
| 996 |
|
|
\
|
| 997 |
|
|
continue; \
|
| 998 |
|
|
} \
|
| 999 |
|
|
\
|
| 1000 |
|
|
if (__builtin_expect (inptr + cnt > inend, 0)) \
|
| 1001 |
|
|
{ \
|
| 1002 |
|
|
/* We don't have enough input. But before we report that check \
|
| 1003 |
|
|
that all the bytes are correct. */ \
|
| 1004 |
|
|
for (i = 1; inptr + i < inend; ++i) \
|
| 1005 |
|
|
if ((inptr[i] & 0xc0) != 0x80) \
|
| 1006 |
|
|
break; \
|
| 1007 |
|
|
\
|
| 1008 |
|
|
if (__builtin_expect (inptr + i == inend, 1)) \
|
| 1009 |
|
|
{ \
|
| 1010 |
|
|
result = __GCONV_INCOMPLETE_INPUT; \
|
| 1011 |
|
|
break; \
|
| 1012 |
|
|
} \
|
| 1013 |
|
|
\
|
| 1014 |
|
|
if (ignore_errors_p ()) \
|
| 1015 |
|
|
{ \
|
| 1016 |
|
|
/* Ignore it. */ \
|
| 1017 |
|
|
inptr += i; \
|
| 1018 |
|
|
*irreversible = *irreversible + 1; \
|
| 1019 |
|
|
continue; \
|
| 1020 |
|
|
} \
|
| 1021 |
|
|
\
|
| 1022 |
|
|
result = __GCONV_ILLEGAL_INPUT; \
|
| 1023 |
|
|
break; \
|
| 1024 |
|
|
} \
|
| 1025 |
|
|
\
|
| 1026 |
|
|
/* Read the possible remaining bytes. */ \
|
| 1027 |
|
|
for (i = 1; i < cnt; ++i) \
|
| 1028 |
|
|
{ \
|
| 1029 |
|
|
uint32_t byte = inptr[i]; \
|
| 1030 |
|
|
\
|
| 1031 |
|
|
if ((byte & 0xc0) != 0x80) \
|
| 1032 |
|
|
/* This is an illegal encoding. */ \
|
| 1033 |
|
|
break; \
|
| 1034 |
|
|
\
|
| 1035 |
|
|
ch <<= 6; \
|
| 1036 |
|
|
ch |= byte & 0x3f; \
|
| 1037 |
|
|
} \
|
| 1038 |
|
|
\
|
| 1039 |
|
|
/* If i < cnt, some trail byte was not >= 0x80, < 0xc0. \
|
| 1040 |
|
|
If cnt > 2 and ch < 2^(5*cnt-4), the wide character ch could \
|
| 1041 |
|
|
have been represented with fewer than cnt bytes. */ \
|
| 1042 |
|
|
if (i < cnt || (cnt > 2 && (ch >> (5 * cnt - 4)) == 0)) \
|
| 1043 |
|
|
{ \
|
| 1044 |
|
|
/* This is an illegal encoding. */ \
|
| 1045 |
|
|
if (ignore_errors_p ()) \
|
| 1046 |
|
|
{ \
|
| 1047 |
|
|
inptr += i; \
|
| 1048 |
|
|
*irreversible = *irreversible + 1; \
|
| 1049 |
|
|
continue; \
|
| 1050 |
|
|
} \
|
| 1051 |
|
|
\
|
| 1052 |
|
|
result = __GCONV_ILLEGAL_INPUT; \
|
| 1053 |
|
|
break; \
|
| 1054 |
|
|
} \
|
| 1055 |
|
|
\
|
| 1056 |
|
|
inptr += cnt; \
|
| 1057 |
|
|
} \
|
| 1058 |
|
|
\
|
| 1059 |
|
|
/* Now adjust the pointers and store the result. */ \
|
| 1060 |
|
|
*((uint32_t *) outptr) = ch; \
|
| 1061 |
|
|
outptr = (unsigned char *)((uint32_t *) outptr + 1); \
|
| 1062 |
|
|
}
|
| 1063 |
|
|
#define LOOP_NEED_FLAGS
|
| 1064 |
|
|
|
| 1065 |
|
|
#define STORE_REST \
|
| 1066 |
|
|
{ \
|
| 1067 |
|
|
/* We store the remaining bytes while converting them into the UCS4 \
|
| 1068 |
|
|
format. We can assume that the first byte in the buffer is \
|
| 1069 |
|
|
correct and that it requires a larger number of bytes than there \
|
| 1070 |
|
|
are in the input buffer. */ \
|
| 1071 |
|
|
wint_t ch = **inptrp; \
|
| 1072 |
|
|
size_t cnt; \
|
| 1073 |
|
|
\
|
| 1074 |
|
|
state->__count = inend - *inptrp; \
|
| 1075 |
|
|
\
|
| 1076 |
|
|
if (ch >= 0xc2 && ch < 0xe0) \
|
| 1077 |
|
|
{ \
|
| 1078 |
|
|
/* We expect two bytes. The first byte cannot be 0xc0 or \
|
| 1079 |
|
|
0xc1, otherwise the wide character could have been \
|
| 1080 |
|
|
represented using a single byte. */ \
|
| 1081 |
|
|
cnt = 2; \
|
| 1082 |
|
|
ch &= 0x1f; \
|
| 1083 |
|
|
} \
|
| 1084 |
|
|
else if (__builtin_expect ((ch & 0xf0) == 0xe0, 1)) \
|
| 1085 |
|
|
{ \
|
| 1086 |
|
|
/* We expect three bytes. */ \
|
| 1087 |
|
|
cnt = 3; \
|
| 1088 |
|
|
ch &= 0x0f; \
|
| 1089 |
|
|
} \
|
| 1090 |
|
|
else if (__builtin_expect ((ch & 0xf8) == 0xf0, 1)) \
|
| 1091 |
|
|
{ \
|
| 1092 |
|
|
/* We expect four bytes. */ \
|
| 1093 |
|
|
cnt = 4; \
|
| 1094 |
|
|
ch &= 0x07; \
|
| 1095 |
|
|
} \
|
| 1096 |
|
|
else if (__builtin_expect ((ch & 0xfc) == 0xf8, 1)) \
|
| 1097 |
|
|
{ \
|
| 1098 |
|
|
/* We expect five bytes. */ \
|
| 1099 |
|
|
cnt = 5; \
|
| 1100 |
|
|
ch &= 0x03; \
|
| 1101 |
|
|
} \
|
| 1102 |
|
|
else \
|
| 1103 |
|
|
{ \
|
| 1104 |
|
|
/* We expect six bytes. */ \
|
| 1105 |
|
|
cnt = 6; \
|
| 1106 |
|
|
ch &= 0x01; \
|
| 1107 |
|
|
} \
|
| 1108 |
|
|
\
|
| 1109 |
|
|
/* The first byte is already consumed. */ \
|
| 1110 |
|
|
--cnt; \
|
| 1111 |
|
|
while (++(*inptrp) < inend) \
|
| 1112 |
|
|
{ \
|
| 1113 |
|
|
ch <<= 6; \
|
| 1114 |
|
|
ch |= **inptrp & 0x3f; \
|
| 1115 |
|
|
--cnt; \
|
| 1116 |
|
|
} \
|
| 1117 |
|
|
\
|
| 1118 |
|
|
/* Shift for the so far missing bytes. */ \
|
| 1119 |
|
|
ch <<= cnt * 6; \
|
| 1120 |
|
|
\
|
| 1121 |
|
|
/* Store the value. */ \
|
| 1122 |
|
|
state->__value.__wch = ch; \
|
| 1123 |
|
|
}
|
| 1124 |
|
|
|
| 1125 |
|
|
#define UNPACK_BYTES \
|
| 1126 |
|
|
{ \
|
| 1127 |
|
|
wint_t wch = state->__value.__wch; \
|
| 1128 |
|
|
size_t ntotal; \
|
| 1129 |
|
|
inlen = state->__count; \
|
| 1130 |
|
|
\
|
| 1131 |
|
|
if (state->__value.__wch <= 0x7ff) \
|
| 1132 |
|
|
{ \
|
| 1133 |
|
|
bytebuf[0] = 0xc0; \
|
| 1134 |
|
|
ntotal = 2; \
|
| 1135 |
|
|
} \
|
| 1136 |
|
|
else if (__builtin_expect (state->__value.__wch <= 0xffff, 1)) \
|
| 1137 |
|
|
{ \
|
| 1138 |
|
|
bytebuf[0] = 0xe0; \
|
| 1139 |
|
|
ntotal = 3; \
|
| 1140 |
|
|
} \
|
| 1141 |
|
|
else if (__builtin_expect (state->__value.__wch < 0x1fffff, 1)) \
|
| 1142 |
|
|
{ \
|
| 1143 |
|
|
bytebuf[0] = 0xf0; \
|
| 1144 |
|
|
ntotal = 4; \
|
| 1145 |
|
|
} \
|
| 1146 |
|
|
else if (__builtin_expect (state->__value.__wch < 0x3ffffff, 1)) \
|
| 1147 |
|
|
{ \
|
| 1148 |
|
|
bytebuf[0] = 0xf8; \
|
| 1149 |
|
|
ntotal = 5; \
|
| 1150 |
|
|
} \
|
| 1151 |
|
|
else \
|
| 1152 |
|
|
{ \
|
| 1153 |
|
|
bytebuf[0] = 0xfc; \
|
| 1154 |
|
|
ntotal = 6; \
|
| 1155 |
|
|
} \
|
| 1156 |
|
|
\
|
| 1157 |
|
|
do \
|
| 1158 |
|
|
{ \
|
| 1159 |
|
|
if (--ntotal < inlen) \
|
| 1160 |
|
|
bytebuf[ntotal] = 0x80 | (wch & 0x3f); \
|
| 1161 |
|
|
wch >>= 6; \
|
| 1162 |
|
|
} \
|
| 1163 |
|
|
while (ntotal > 1); \
|
| 1164 |
|
|
\
|
| 1165 |
|
|
bytebuf[0] |= wch; \
|
| 1166 |
|
|
}
|
| 1167 |
|
|
|
| 1168 |
|
|
#include <iconv/loop.c>
|
| 1169 |
|
|
#include <iconv/skeleton.c>
|
| 1170 |
|
|
|
| 1171 |
|
|
|
| 1172 |
|
|
/* Convert from UCS2 to the internal (UCS4-like) format. */
|
| 1173 |
|
|
#define DEFINE_INIT 0
|
| 1174 |
|
|
#define DEFINE_FINI 0
|
| 1175 |
|
|
#define MIN_NEEDED_FROM 2
|
| 1176 |
|
|
#define MIN_NEEDED_TO 4
|
| 1177 |
|
|
#define FROM_DIRECTION 1
|
| 1178 |
|
|
#define FROM_LOOP ucs2_internal_loop
|
| 1179 |
|
|
#define TO_LOOP ucs2_internal_loop /* This is not used. */
|
| 1180 |
|
|
#define FUNCTION_NAME __gconv_transform_ucs2_internal
|
| 1181 |
|
|
#define ONE_DIRECTION 1
|
| 1182 |
|
|
|
| 1183 |
|
|
#define MIN_NEEDED_INPUT MIN_NEEDED_FROM
|
| 1184 |
|
|
#define MIN_NEEDED_OUTPUT MIN_NEEDED_TO
|
| 1185 |
|
|
#define LOOPFCT FROM_LOOP
|
| 1186 |
|
|
#define BODY \
|
| 1187 |
|
|
{ \
|
| 1188 |
|
|
uint16_t u1 = *((const uint16_t *) inptr); \
|
| 1189 |
|
|
\
|
| 1190 |
|
|
if (__builtin_expect (u1 >= 0xd800 && u1 < 0xe000, 0)) \
|
| 1191 |
|
|
{ \
|
| 1192 |
|
|
/* Surrogate characters in UCS-2 input are not valid. Reject \
|
| 1193 |
|
|
them. (Catching this here is not security relevant.) */ \
|
| 1194 |
|
|
if (! ignore_errors_p ()) \
|
| 1195 |
|
|
{ \
|
| 1196 |
|
|
result = __GCONV_ILLEGAL_INPUT; \
|
| 1197 |
|
|
break; \
|
| 1198 |
|
|
} \
|
| 1199 |
|
|
inptr += 2; \
|
| 1200 |
|
|
*irreversible = *irreversible + 1; \
|
| 1201 |
|
|
continue; \
|
| 1202 |
|
|
} \
|
| 1203 |
|
|
\
|
| 1204 |
|
|
*((uint32_t *) outptr) = u1; \
|
| 1205 |
|
|
outptr = (unsigned char *)((uint32_t *) outptr + 1); \
|
| 1206 |
|
|
inptr += 2; \
|
| 1207 |
|
|
}
|
| 1208 |
|
|
#define LOOP_NEED_FLAGS
|
| 1209 |
|
|
#include <iconv/loop.c>
|
| 1210 |
|
|
#include <iconv/skeleton.c>
|
| 1211 |
|
|
|
| 1212 |
|
|
|
| 1213 |
|
|
/* Convert from the internal (UCS4-like) format to UCS2. */
|
| 1214 |
|
|
#define DEFINE_INIT 0
|
| 1215 |
|
|
#define DEFINE_FINI 0
|
| 1216 |
|
|
#define MIN_NEEDED_FROM 4
|
| 1217 |
|
|
#define MIN_NEEDED_TO 2
|
| 1218 |
|
|
#define FROM_DIRECTION 1
|
| 1219 |
|
|
#define FROM_LOOP internal_ucs2_loop
|
| 1220 |
|
|
#define TO_LOOP internal_ucs2_loop /* This is not used. */
|
| 1221 |
|
|
#define FUNCTION_NAME __gconv_transform_internal_ucs2
|
| 1222 |
|
|
#define ONE_DIRECTION 1
|
| 1223 |
|
|
|
| 1224 |
|
|
#define MIN_NEEDED_INPUT MIN_NEEDED_FROM
|
| 1225 |
|
|
#define MIN_NEEDED_OUTPUT MIN_NEEDED_TO
|
| 1226 |
|
|
#define LOOPFCT FROM_LOOP
|
| 1227 |
|
|
#define BODY \
|
| 1228 |
|
|
{ \
|
| 1229 |
|
|
uint32_t val = *((const uint32_t *) inptr); \
|
| 1230 |
|
|
\
|
| 1231 |
|
|
if (__builtin_expect (val >= 0x10000, 0)) \
|
| 1232 |
|
|
{ \
|
| 1233 |
|
|
UNICODE_TAG_HANDLER (val, 4); \
|
| 1234 |
|
|
STANDARD_ERR_HANDLER (4); \
|
| 1235 |
|
|
} \
|
| 1236 |
|
|
else if (__builtin_expect (val >= 0xd800 && val < 0xe000, 0)) \
|
| 1237 |
|
|
{ \
|
| 1238 |
|
|
/* Surrogate characters in UCS-4 input are not valid. \
|
| 1239 |
|
|
We must catch this, because the UCS-2 output might be \
|
| 1240 |
|
|
interpreted as UTF-16 by other programs. If we let \
|
| 1241 |
|
|
surrogates pass through, attackers could make a security \
|
| 1242 |
|
|
hole exploit by synthesizing any desired plane 1-16 \
|
| 1243 |
|
|
character. */ \
|
| 1244 |
|
|
if (! ignore_errors_p ()) \
|
| 1245 |
|
|
{ \
|
| 1246 |
|
|
result = __GCONV_ILLEGAL_INPUT; \
|
| 1247 |
|
|
break; \
|
| 1248 |
|
|
} \
|
| 1249 |
|
|
inptr += 4; \
|
| 1250 |
|
|
*irreversible = *irreversible + 1; \
|
| 1251 |
|
|
continue; \
|
| 1252 |
|
|
} \
|
| 1253 |
|
|
else \
|
| 1254 |
|
|
{ \
|
| 1255 |
|
|
*((uint16_t *) outptr) = val; \
|
| 1256 |
|
|
outptr = (unsigned char *)((uint16_t *) outptr + 1); \
|
| 1257 |
|
|
inptr += 4; \
|
| 1258 |
|
|
} \
|
| 1259 |
|
|
}
|
| 1260 |
|
|
#define LOOP_NEED_FLAGS
|
| 1261 |
|
|
#include <iconv/loop.c>
|
| 1262 |
|
|
#include <iconv/skeleton.c>
|
| 1263 |
|
|
|
| 1264 |
|
|
|
| 1265 |
|
|
/* Convert from UCS2 in other endianness to the internal (UCS4-like) format. */
|
| 1266 |
|
|
#define DEFINE_INIT 0
|
| 1267 |
|
|
#define DEFINE_FINI 0
|
| 1268 |
|
|
#define MIN_NEEDED_FROM 2
|
| 1269 |
|
|
#define MIN_NEEDED_TO 4
|
| 1270 |
|
|
#define FROM_DIRECTION 1
|
| 1271 |
|
|
#define FROM_LOOP ucs2reverse_internal_loop
|
| 1272 |
|
|
#define TO_LOOP ucs2reverse_internal_loop/* This is not used.*/
|
| 1273 |
|
|
#define FUNCTION_NAME __gconv_transform_ucs2reverse_internal
|
| 1274 |
|
|
#define ONE_DIRECTION 1
|
| 1275 |
|
|
|
| 1276 |
|
|
#define MIN_NEEDED_INPUT MIN_NEEDED_FROM
|
| 1277 |
|
|
#define MIN_NEEDED_OUTPUT MIN_NEEDED_TO
|
| 1278 |
|
|
#define LOOPFCT FROM_LOOP
|
| 1279 |
|
|
#define BODY \
|
| 1280 |
|
|
{ \
|
| 1281 |
|
|
uint16_t u1 = bswap_16 (*((const uint16_t *) inptr)); \
|
| 1282 |
|
|
\
|
| 1283 |
|
|
if (__builtin_expect (u1 >= 0xd800 && u1 < 0xe000, 0)) \
|
| 1284 |
|
|
{ \
|
| 1285 |
|
|
/* Surrogate characters in UCS-2 input are not valid. Reject \
|
| 1286 |
|
|
them. (Catching this here is not security relevant.) */ \
|
| 1287 |
|
|
if (! ignore_errors_p ()) \
|
| 1288 |
|
|
{ \
|
| 1289 |
|
|
result = __GCONV_ILLEGAL_INPUT; \
|
| 1290 |
|
|
break; \
|
| 1291 |
|
|
} \
|
| 1292 |
|
|
inptr += 2; \
|
| 1293 |
|
|
*irreversible = *irreversible + 1; \
|
| 1294 |
|
|
continue; \
|
| 1295 |
|
|
} \
|
| 1296 |
|
|
\
|
| 1297 |
|
|
*((uint32_t *) outptr) = u1; \
|
| 1298 |
|
|
outptr = (unsigned char *)((uint32_t *) outptr + 1); \
|
| 1299 |
|
|
inptr += 2; \
|
| 1300 |
|
|
}
|
| 1301 |
|
|
#define LOOP_NEED_FLAGS
|
| 1302 |
|
|
#include <iconv/loop.c>
|
| 1303 |
|
|
#include <iconv/skeleton.c>
|
| 1304 |
|
|
|
| 1305 |
|
|
|
| 1306 |
|
|
/* Convert from the internal (UCS4-like) format to UCS2 in other endianness. */
|
| 1307 |
|
|
#define DEFINE_INIT 0
|
| 1308 |
|
|
#define DEFINE_FINI 0
|
| 1309 |
|
|
#define MIN_NEEDED_FROM 4
|
| 1310 |
|
|
#define MIN_NEEDED_TO 2
|
| 1311 |
|
|
#define FROM_DIRECTION 1
|
| 1312 |
|
|
#define FROM_LOOP internal_ucs2reverse_loop
|
| 1313 |
|
|
#define TO_LOOP internal_ucs2reverse_loop/* This is not used.*/
|
| 1314 |
|
|
#define FUNCTION_NAME __gconv_transform_internal_ucs2reverse
|
| 1315 |
|
|
#define ONE_DIRECTION 1
|
| 1316 |
|
|
|
| 1317 |
|
|
#define MIN_NEEDED_INPUT MIN_NEEDED_FROM
|
| 1318 |
|
|
#define MIN_NEEDED_OUTPUT MIN_NEEDED_TO
|
| 1319 |
|
|
#define LOOPFCT FROM_LOOP
|
| 1320 |
|
|
#define BODY \
|
| 1321 |
|
|
{ \
|
| 1322 |
|
|
uint32_t val = *((const uint32_t *) inptr); \
|
| 1323 |
|
|
if (__builtin_expect (val >= 0x10000, 0)) \
|
| 1324 |
|
|
{ \
|
| 1325 |
|
|
UNICODE_TAG_HANDLER (val, 4); \
|
| 1326 |
|
|
STANDARD_ERR_HANDLER (4); \
|
| 1327 |
|
|
} \
|
| 1328 |
|
|
else if (__builtin_expect (val >= 0xd800 && val < 0xe000, 0)) \
|
| 1329 |
|
|
{ \
|
| 1330 |
|
|
/* Surrogate characters in UCS-4 input are not valid. \
|
| 1331 |
|
|
We must catch this, because the UCS-2 output might be \
|
| 1332 |
|
|
interpreted as UTF-16 by other programs. If we let \
|
| 1333 |
|
|
surrogates pass through, attackers could make a security \
|
| 1334 |
|
|
hole exploit by synthesizing any desired plane 1-16 \
|
| 1335 |
|
|
character. */ \
|
| 1336 |
|
|
if (! ignore_errors_p ()) \
|
| 1337 |
|
|
{ \
|
| 1338 |
|
|
result = __GCONV_ILLEGAL_INPUT; \
|
| 1339 |
|
|
break; \
|
| 1340 |
|
|
} \
|
| 1341 |
|
|
inptr += 4; \
|
| 1342 |
|
|
*irreversible = *irreversible + 1; \
|
| 1343 |
|
|
continue; \
|
| 1344 |
|
|
} \
|
| 1345 |
|
|
else \
|
| 1346 |
|
|
{ \
|
| 1347 |
|
|
*((uint16_t *) outptr) = bswap_16 (val); \
|
| 1348 |
|
|
outptr = (unsigned char *)((uint16_t *) outptr + 1); \
|
| 1349 |
|
|
inptr += 4; \
|
| 1350 |
|
|
} \
|
| 1351 |
|
|
}
|
| 1352 |
|
|
#define LOOP_NEED_FLAGS
|
| 1353 |
|
|
#include <iconv/loop.c>
|
| 1354 |
|
|
#include <iconv/skeleton.c>
|
