lib/libc/citrus/modules/citrus_utf1632.c

   1 /*      $NetBSD: src/lib/libc/citrus/modules/citrus_utf1632.c,v 1.3 2003/06/27 12:55:13 yamt Exp $      */
   2 /*      $DragonFly: src/lib/libc/citrus/modules/citrus_utf1632.c,v 1.1 2005/03/11 23:33:53 joerg Exp $ */
   3
   4 /*-
   5  * Copyright (c)2003 Citrus Project,
   6  * All rights reserved.
   7  *
   8  * Redistribution and use in source and binary forms, with or without
   9  * modification, are permitted provided that the following conditions
  10  * are met:
  11  * 1. Redistributions of source code must retain the above copyright
  12  *    notice, this list of conditions and the following disclaimer.
  13  * 2. Redistributions in binary form must reproduce the above copyright
  14  *    notice, this list of conditions and the following disclaimer in the
  15  *    documentation and/or other materials provided with the distribution.
  16  *
  17  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
  18  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  19  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  20  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
  21  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  22  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  23  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  24  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  25  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  26  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  27  * SUCH DAMAGE.
  28  */
  29
  30 #include <sys/types.h>
  31 #include <sys/endian.h>
  32 #include <assert.h>
  33 #include <errno.h>
  34 #include <limits.h>
  35 #include <locale.h>
  36 #include <stddef.h>
  37 #include <stdio.h>
  38 #include <stdlib.h>
  39 #include <string.h>
  40 #include <wchar.h>
  41
  42 #include "citrus_namespace.h"
  43 #include "citrus_types.h"
  44 #include "citrus_module.h"
  45 #include "citrus_stdenc.h"
  46 #include "citrus_bcs.h"
  47
  48 #include "citrus_utf1632.h"
  49
  50
  51 /* ----------------------------------------------------------------------
  52  * private stuffs used by templates
  53  */
  54
  55 typedef struct {
  56         u_int8_t                ch[4];
  57         int                     chlen;
  58         int                     current_endian;
  59 } _UTF1632State;
  60
  61 typedef struct {
  62         int             preffered_endian;
  63         unsigned int    cur_max;
  64 #define _ENDIAN_UNKNOWN 0
  65 #define _ENDIAN_BIG     1
  66 #define _ENDIAN_LITTLE  2
  67         u_int32_t       mode;
  68 #define _MODE_UTF32             0x00000001U
  69 #define _MODE_FORCE_ENDIAN      0x00000002U
  70 } _UTF1632EncodingInfo;
  71
  72 #define _FUNCNAME(m)                    _citrus_UTF1632_##m
  73 #define _ENCODING_INFO                  _UTF1632EncodingInfo
  74 #define _ENCODING_STATE                 _UTF1632State
  75 #define _ENCODING_MB_CUR_MAX(_ei_)      ((_ei_)->cur_max)
  76 #define _ENCODING_IS_STATE_DEPENDENT    0
  77 #define _STATE_NEEDS_EXPLICIT_INIT(_ps_)        0
  78
  79
  80 static __inline void
  81 /*ARGSUSED*/
  82 _citrus_UTF1632_init_state(_UTF1632EncodingInfo *ei, _UTF1632State *s)
  83 {
  84         memset(s, 0, sizeof(*s));
  85 }
  86
  87 static int
  88 _citrus_UTF1632_mbrtowc_priv(_UTF1632EncodingInfo *ei, wchar_t *pwc,
  89                              const char **s, size_t n, _UTF1632State *psenc,
  90                              size_t *nresult)
  91 {
  92         int chlenbak, endian, needlen;
  93         wchar_t wc;
  94         size_t result;
  95         const char *s0;
  96
  97         _DIAGASSERT(nresult != 0);
  98         _DIAGASSERT(ei != NULL);
  99         _DIAGASSERT(s != NULL);
 100         _DIAGASSERT(psenc != NULL);
 101
 102         s0 = *s;
 103
 104         if (s0 == NULL) {
 105                 _citrus_UTF1632_init_state(ei, psenc);
 106                 *nresult = 0; /* state independent */
 107                 return (0);
 108         }
 109
 110         result = 0;
 111         chlenbak = psenc->chlen;
 112
 113 refetch:
 114         if ((ei->mode & _MODE_UTF32) != 0 || chlenbak>=2)
 115                 needlen = 4;
 116         else
 117                 needlen = 2;
 118
 119         while (chlenbak < needlen) {
 120                 if (n==0)
 121                         goto restart;
 122                 psenc->ch[chlenbak++] = *s0++;
 123                 n--;
 124                 result++;
 125         }
 126
 127         /* judge endian marker */
 128         if ((ei->mode & _MODE_UTF32) == 0) {
 129                 /* UTF16 */
 130                 if (psenc->ch[0]==0xFE && psenc->ch[1]==0xFF) {
 131                         psenc->current_endian = _ENDIAN_BIG;
 132                         chlenbak = 0;
 133                         goto refetch;
 134                 } else if (psenc->ch[0]==0xFF && psenc->ch[1]==0xFE) {
 135                         psenc->current_endian = _ENDIAN_LITTLE;
 136                         chlenbak = 0;
 137                         goto refetch;
 138                 }
 139         } else {
 140                 /* UTF32 */
 141                 if (psenc->ch[0]==0x00 && psenc->ch[1]==0x00 &&
 142                     psenc->ch[2]==0xFE && psenc->ch[3]==0xFF) {
 143                         psenc->current_endian = _ENDIAN_BIG;
 144                         chlenbak = 0;
 145                         goto refetch;
 146                 } else if (psenc->ch[0]==0xFF && psenc->ch[1]==0xFE &&
 147                            psenc->ch[2]==0x00 && psenc->ch[3]==0x00) {
 148                         psenc->current_endian = _ENDIAN_LITTLE;
 149                         chlenbak = 0;
 150                         goto refetch;
 151                 }
 152         }
 153         if ((ei->mode & _MODE_FORCE_ENDIAN) != 0 ||
 154             psenc->current_endian == _ENDIAN_UNKNOWN)
 155                 endian = ei->preffered_endian;
 156         else
 157                 endian = psenc->current_endian;
 158
 159         /* get wc */
 160         if ((ei->mode & _MODE_UTF32) == 0) {
 161                 /* UTF16 */
 162                 if (needlen==2) {
 163                         switch (endian) {
 164                         case _ENDIAN_LITTLE:
 165                                 wc = (psenc->ch[0] |
 166                                       ((wchar_t)psenc->ch[1] << 8));
 167                                 break;
 168                         case _ENDIAN_BIG:
 169                                 wc = (psenc->ch[1] |
 170                                       ((wchar_t)psenc->ch[0] << 8));
 171                                 break;
 172                         }
 173                         if (wc >= 0xD800 && wc <= 0xDBFF) {
 174                                 /* surrogate high */
 175                                 needlen=4;
 176                                 goto refetch;
 177                         }
 178                 } else {
 179                         /* surrogate low */
 180                         wc -= 0xD800; /* wc : surrogate high (see above) */
 181                         wc <<= 10;
 182                         switch (endian) {
 183                         case _ENDIAN_LITTLE:
 184                                 if (psenc->ch[2]<0xDC || psenc->ch[2]>0xDF)
 185                                         goto ilseq;
 186                                 wc |= psenc->ch[2];
 187                                 wc |= (wchar_t)(psenc->ch[3] & 3) << 8;
 188                                 break;
 189                         case _ENDIAN_BIG:
 190                                 if (psenc->ch[3]<0xDC || psenc->ch[3]>0xDF)
 191                                         goto ilseq;
 192                                 wc |= psenc->ch[3];
 193                                 wc |= (wchar_t)(psenc->ch[2] & 3) << 8;
 194                                 break;
 195                         }
 196                         wc += 0x10000;
 197                 }
 198         } else {
 199                 /* UTF32 */
 200                 switch (endian) {
 201                 case _ENDIAN_LITTLE:
 202                         wc = (psenc->ch[0] |
 203                               ((wchar_t)psenc->ch[1] << 8) |
 204                               ((wchar_t)psenc->ch[2] << 16) |
 205                               ((wchar_t)psenc->ch[3] << 24));
 206                         break;
 207                 case _ENDIAN_BIG:
 208                         wc = (psenc->ch[3] |
 209                               ((wchar_t)psenc->ch[2] << 8) |
 210                               ((wchar_t)psenc->ch[1] << 16) |
 211                               ((wchar_t)psenc->ch[0] << 24));
 212                         break;
 213                 }
 214         }
 215
 216
 217         *pwc = wc;
 218         psenc->chlen = 0;
 219         *nresult = result;
 220         *s = s0;
 221
 222         return (0);
 223
 224 ilseq:
 225         *nresult = (size_t)-1;
 226         psenc->chlen = 0;
 227         return (EILSEQ);
 228
 229 restart:
 230         *nresult = (size_t)-2;
 231         psenc->chlen = chlenbak;
 232         *s = s0;
 233         return (0);
 234 }
 235
 236 static int
 237 _citrus_UTF1632_wcrtomb_priv(_UTF1632EncodingInfo *ei, char *s, size_t n,
 238                              wchar_t wc, _UTF1632State *psenc,
 239                              size_t *nresult)
 240 {
 241         int ret;
 242         wchar_t wc2;
 243
 244         _DIAGASSERT(ei != NULL);
 245         _DIAGASSERT(nresult != 0);
 246         _DIAGASSERT(s != NULL);
 247
 248         wc2 = 0;
 249         if ((ei->mode & _MODE_UTF32)==0) {
 250                 /* UTF16 */
 251                 if (wc>0xFFFF) {
 252                         /* surrogate */
 253                         if (wc>0x10FFFF) {
 254                                 ret = EILSEQ;
 255                                 goto err;
 256                         }
 257                         if (n < 4) {
 258                                 ret = E2BIG;
 259                                 goto err;
 260                         }
 261                         wc -= 0x10000;
 262                         wc2 = (wc & 0x3FF) | 0xDC00;
 263                         wc = (wc>>10) | 0xD800;
 264                         *nresult = (size_t)4;
 265                 } else {
 266                         if (n < 2) {
 267                                 ret = E2BIG;
 268                                 goto err;
 269                         }
 270                         *nresult = (size_t)2;
 271                 }
 272
 273 surrogate:
 274                 switch (ei->preffered_endian) {
 275                 case _ENDIAN_BIG:
 276                         s[1] = wc;
 277                         s[0] = (wc >>= 8);
 278                         break;
 279                 case _ENDIAN_LITTLE:
 280                         s[0] = wc;
 281                         s[1] = (wc >>= 8);
 282                         break;
 283                 }
 284                 if (wc2!=0) {
 285                         wc = wc2;
 286                         wc2 = 0;
 287                         s += 2;
 288                         goto surrogate;
 289                 }
 290         } else {
 291                 /* UTF32 */
 292                 if (n < 4) {
 293                         ret = E2BIG;
 294                         goto err;
 295                 }
 296                 switch (ei->preffered_endian) {
 297                 case _ENDIAN_BIG:
 298                         s[3] = wc;
 299                         s[2] = (wc >>= 8);
 300                         s[1] = (wc >>= 8);
 301                         s[0] = (wc >>= 8);
 302                         break;
 303                 case _ENDIAN_LITTLE:
 304                         s[0] = wc;
 305                         s[1] = (wc >>= 8);
 306                         s[2] = (wc >>= 8);
 307                         s[3] = (wc >>= 8);
 308                         break;
 309                 }
 310                 *nresult = (size_t)4;
 311         }
 312
 313         return 0;
 314
 315 err:
 316         *nresult = (size_t)-1;
 317         return ret;
 318 }
 319
 320 static void
 321 parse_variable(_UTF1632EncodingInfo * __restrict ei,
 322                const void * __restrict var, size_t lenvar)
 323 {
 324 #define MATCH(x, act)                                           \
 325 do {                                                            \
 326         if (lenvar >= (sizeof(#x)-1) &&                         \
 327             _bcs_strncasecmp(p, #x, sizeof(#x)-1) == 0) {       \
 328                 act;                                            \
 329                 lenvar -= sizeof(#x)-1;                         \
 330                 p += sizeof(#x)-1;                              \
 331         }                                                       \
 332 } while (/*CONSTCOND*/0)
 333         const char *p;
 334         p = var;
 335         while (lenvar>0) {
 336                 switch (*p) {
 337                 case 'B':
 338                 case 'b':
 339                         MATCH(big, ei->preffered_endian = _ENDIAN_BIG);
 340                         break;
 341                 case 'L':
 342                 case 'l':
 343                         MATCH(little, ei->preffered_endian = _ENDIAN_LITTLE);
 344                         break;
 345                 case 'F':
 346                 case 'f':
 347                         MATCH(force, ei->mode |= _MODE_FORCE_ENDIAN);
 348                         break;
 349                 case 'U':
 350                 case 'u':
 351                         MATCH(utf32, ei->mode |= _MODE_UTF32);
 352                         break;
 353                 }
 354                 p++;
 355                 lenvar--;
 356         }
 357 }
 358
 359 static int
 360 /*ARGSUSED*/
 361 _citrus_UTF1632_encoding_module_init(_UTF1632EncodingInfo * __restrict ei,
 362                                      const void * __restrict var,
 363                                      size_t lenvar)
 364 {
 365         _DIAGASSERT(ei != NULL);
 366
 367         memset((void *)ei, 0, sizeof(*ei));
 368
 369         parse_variable(ei, var, lenvar);
 370
 371         if ((ei->mode&_MODE_UTF32)==0)
 372                 ei->cur_max = 6; /* endian + surrogate */
 373         else
 374                 ei->cur_max = 8; /* endian + normal */
 375
 376         if (ei->preffered_endian == _ENDIAN_UNKNOWN) {
 377 #if BYTE_ORDER == BIG_ENDIAN
 378                 ei->preffered_endian = _ENDIAN_BIG;
 379 #else
 380                 ei->preffered_endian = _ENDIAN_LITTLE;
 381 #endif
 382         }
 383
 384         return (0);
 385 }
 386
 387 static void
 388 /*ARGSUSED*/
 389 _citrus_UTF1632_encoding_module_uninit(_UTF1632EncodingInfo *ei)
 390 {
 391 }
 392
 393 static __inline int
 394 /*ARGSUSED*/
 395 _citrus_UTF1632_stdenc_wctocs(_UTF1632EncodingInfo * __restrict ei,
 396                               _csid_t * __restrict csid,
 397                               _index_t * __restrict idx,
 398                               _wc_t wc)
 399 {
 400
 401         _DIAGASSERT(csid != NULL && idx != NULL);
 402
 403         *csid = 0;
 404         *idx = (_index_t)wc;
 405
 406         return (0);
 407 }
 408
 409 static __inline int
 410 /*ARGSUSED*/
 411 _citrus_UTF1632_stdenc_cstowc(_UTF1632EncodingInfo * __restrict ei,
 412                               _wc_t * __restrict wc,
 413                               _csid_t csid, _index_t idx)
 414 {
 415
 416         _DIAGASSERT(wc != NULL);
 417
 418         if (csid != 0)
 419                 return (EILSEQ);
 420
 421         *wc = (_wc_t)idx;
 422
 423         return (0);
 424 }
 425
 426
 427 /* ----------------------------------------------------------------------
 428  * public interface for stdenc
 429  */
 430
 431 _CITRUS_STDENC_DECLS(UTF1632);
 432 _CITRUS_STDENC_DEF_OPS(UTF1632);
 433
 434 #include "citrus_stdenc_template.h"