lib/libc/citrus/modules/citrus_utf1632.c

   1 /* $NetBSD: citrus_utf1632.c,v 1.8 2008/03/20 11:47:45 tnozaki Exp $ */
   2
   3 /*-
   4  * Copyright (c)2003 Citrus Project,
   5  * All rights reserved.
   6  *
   7  * Redistribution and use in source and binary forms, with or without
   8  * modification, are permitted provided that the following conditions
   9  * are met:
  10  * 1. Redistributions of source code must retain the above copyright
  11  *    notice, this list of conditions and the following disclaimer.
  12  * 2. Redistributions in binary form must reproduce the above copyright
  13  *    notice, this list of conditions and the following disclaimer in the
  14  *    documentation and/or other materials provided with the distribution.
  15  *
  16  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
  17  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  18  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  19  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
  20  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  21  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  22  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  23  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  24  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  25  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  26  * SUCH DAMAGE.
  27  */
  28
  29 #include <sys/types.h>
  30 #include <sys/endian.h>
  31 #include <assert.h>
  32 #include <errno.h>
  33 #include <limits.h>
  34 #include <locale.h>
  35 #include <stddef.h>
  36 #include <stdio.h>
  37 #include <stdlib.h>
  38 #include <string.h>
  39 #include <wchar.h>
  40
  41 #include "citrus_namespace.h"
  42 #include "citrus_types.h"
  43 #include "citrus_module.h"
  44 #include "citrus_stdenc.h"
  45 #include "citrus_bcs.h"
  46
  47 #include "citrus_utf1632.h"
  48
  49
  50 /* ----------------------------------------------------------------------
  51  * private stuffs used by templates
  52  */
  53
  54 typedef struct {
  55         u_int8_t                ch[4];
  56         int                     chlen;
  57         int                     current_endian;
  58 } _UTF1632State;
  59
  60 typedef struct {
  61         int             preffered_endian;
  62         unsigned int    cur_max;
  63 #define _ENDIAN_UNKNOWN 0
  64 #define _ENDIAN_BIG     1
  65 #define _ENDIAN_LITTLE  2
  66         u_int32_t       mode;
  67 #define _MODE_UTF32             0x00000001U
  68 #define _MODE_FORCE_ENDIAN      0x00000002U
  69 } _UTF1632EncodingInfo;
  70
  71 #define _FUNCNAME(m)                    _citrus_UTF1632_##m
  72 #define _ENCODING_INFO                  _UTF1632EncodingInfo
  73 #define _ENCODING_STATE                 _UTF1632State
  74 #define _ENCODING_MB_CUR_MAX(_ei_)      ((_ei_)->cur_max)
  75 #define _ENCODING_IS_STATE_DEPENDENT    0
  76 #define _STATE_NEEDS_EXPLICIT_INIT(_ps_)        0
  77
  78
  79 static __inline void
  80 /*ARGSUSED*/
  81 _citrus_UTF1632_init_state(_UTF1632EncodingInfo *ei __unused, _UTF1632State *s)
  82 {
  83         memset(s, 0, sizeof(*s));
  84 }
  85
  86 static int
  87 _citrus_UTF1632_mbrtowc_priv(_UTF1632EncodingInfo *ei, wchar_t *pwc,
  88                              const char **s, size_t n, _UTF1632State *psenc,
  89                              size_t *nresult)
  90 {
  91         int chlenbak, endian, needlen;
  92         wchar_t wc;
  93         size_t result;
  94         const char *s0;
  95
  96         _DIAGASSERT(nresult != NULL);
  97         _DIAGASSERT(ei != NULL);
  98         _DIAGASSERT(s != NULL);
  99         _DIAGASSERT(psenc != NULL);
 100
 101         s0 = *s;
 102
 103         if (s0 == NULL) {
 104                 _citrus_UTF1632_init_state(ei, psenc);
 105                 *nresult = 0; /* state independent */
 106                 return (0);
 107         }
 108
 109         result = 0;
 110         chlenbak = psenc->chlen;
 111
 112 refetch:
 113         if ((ei->mode & _MODE_UTF32) != 0 || chlenbak>=2)
 114                 needlen = 4;
 115         else
 116                 needlen = 2;
 117
 118         while (chlenbak < needlen) {
 119                 if (n==0)
 120                         goto restart;
 121                 psenc->ch[chlenbak++] = *s0++;
 122                 n--;
 123                 result++;
 124         }
 125
 126         /* judge endian marker */
 127         if ((ei->mode & _MODE_UTF32) == 0) {
 128                 /* UTF16 */
 129                 if (psenc->ch[0]==0xFE && psenc->ch[1]==0xFF) {
 130                         psenc->current_endian = _ENDIAN_BIG;
 131                         chlenbak = 0;
 132                         goto refetch;
 133                 } else if (psenc->ch[0]==0xFF && psenc->ch[1]==0xFE) {
 134                         psenc->current_endian = _ENDIAN_LITTLE;
 135                         chlenbak = 0;
 136                         goto refetch;
 137                 }
 138         } else {
 139                 /* UTF32 */
 140                 if (psenc->ch[0]==0x00 && psenc->ch[1]==0x00 &&
 141                     psenc->ch[2]==0xFE && psenc->ch[3]==0xFF) {
 142                         psenc->current_endian = _ENDIAN_BIG;
 143                         chlenbak = 0;
 144                         goto refetch;
 145                 } else if (psenc->ch[0]==0xFF && psenc->ch[1]==0xFE &&
 146                            psenc->ch[2]==0x00 && psenc->ch[3]==0x00) {
 147                         psenc->current_endian = _ENDIAN_LITTLE;
 148                         chlenbak = 0;
 149                         goto refetch;
 150                 }
 151         }
 152         if ((ei->mode & _MODE_FORCE_ENDIAN) != 0 ||
 153             psenc->current_endian == _ENDIAN_UNKNOWN)
 154                 endian = ei->preffered_endian;
 155         else
 156                 endian = psenc->current_endian;
 157
 158         /* get wc */
 159         if ((ei->mode & _MODE_UTF32) == 0) {
 160                 /* UTF16 */
 161                 if (needlen==2) {
 162                         switch (endian) {
 163                         case _ENDIAN_LITTLE:
 164                                 wc = (psenc->ch[0] |
 165                                       ((wchar_t)psenc->ch[1] << 8));
 166                                 break;
 167                         case _ENDIAN_BIG:
 168                                 wc = (psenc->ch[1] |
 169                                       ((wchar_t)psenc->ch[0] << 8));
 170                                 break;
 171                         default:
 172                                 goto ilseq;
 173                         }
 174                         if (wc >= 0xD800 && wc <= 0xDBFF) {
 175                                 /* surrogate high */
 176                                 needlen=4;
 177                                 goto refetch;
 178                         }
 179                 } else {
 180                         /* surrogate low */
 181                         wc -= 0xD800; /* wc : surrogate high (see above) */
 182                         wc <<= 10;
 183                         switch (endian) {
 184                         case _ENDIAN_LITTLE:
 185                                 if (psenc->ch[2]<0xDC || psenc->ch[2]>0xDF)
 186                                         goto ilseq;
 187                                 wc |= psenc->ch[2];
 188                                 wc |= (wchar_t)(psenc->ch[3] & 3) << 8;
 189                                 break;
 190                         case _ENDIAN_BIG:
 191                                 if (psenc->ch[3]<0xDC || psenc->ch[3]>0xDF)
 192                                         goto ilseq;
 193                                 wc |= psenc->ch[3];
 194                                 wc |= (wchar_t)(psenc->ch[2] & 3) << 8;
 195                                 break;
 196                         default:
 197                                 goto ilseq;
 198                         }
 199                         wc += 0x10000;
 200                 }
 201         } else {
 202                 /* UTF32 */
 203                 switch (endian) {
 204                 case _ENDIAN_LITTLE:
 205                         wc = (psenc->ch[0] |
 206                               ((wchar_t)psenc->ch[1] << 8) |
 207                               ((wchar_t)psenc->ch[2] << 16) |
 208                               ((wchar_t)psenc->ch[3] << 24));
 209                         break;
 210                 case _ENDIAN_BIG:
 211                         wc = (psenc->ch[3] |
 212                               ((wchar_t)psenc->ch[2] << 8) |
 213                               ((wchar_t)psenc->ch[1] << 16) |
 214                               ((wchar_t)psenc->ch[0] << 24));
 215                         break;
 216                 default:
 217                         goto ilseq;
 218                 }
 219                 if (wc >= 0xD800 && wc <= 0xDFFF)
 220                         goto ilseq;
 221         }
 222
 223
 224         *pwc = wc;
 225         psenc->chlen = 0;
 226         *nresult = result;
 227         *s = s0;
 228
 229         return (0);
 230
 231 ilseq:
 232         *nresult = (size_t)-1;
 233         psenc->chlen = 0;
 234         return (EILSEQ);
 235
 236 restart:
 237         *nresult = (size_t)-2;
 238         psenc->chlen = chlenbak;
 239         *s = s0;
 240         return (0);
 241 }
 242
 243 static int
 244 _citrus_UTF1632_wcrtomb_priv(_UTF1632EncodingInfo *ei, char *s, size_t n,
 245                              wchar_t wc, _UTF1632State *psenc,
 246                              size_t *nresult)
 247 {
 248         wchar_t wc2;
 249         static const char _bom[4] = {
 250 #if BYTE_ORDER == BIG_ENDIAN
 251             0x00, 0x00, 0xFE, 0xFF,
 252 #else
 253             0xFF, 0xFE, 0x00, 0x00,
 254 #endif
 255         };
 256         const char *bom = &_bom[0];
 257         size_t cnt;
 258
 259         _DIAGASSERT(ei != NULL);
 260         _DIAGASSERT(nresult != NULL);
 261         _DIAGASSERT(s != NULL);
 262
 263         cnt = (size_t)0;
 264         if (psenc->current_endian == _ENDIAN_UNKNOWN) {
 265                 if ((ei->mode & _MODE_FORCE_ENDIAN) == 0) {
 266                         if (ei->mode & _MODE_UTF32) {
 267                                 cnt = 4;
 268                         } else {
 269                                 cnt = 2;
 270 #if BYTE_ORDER == BIG_ENDIAN
 271                                 bom += 2;
 272 #endif
 273                         }
 274                         if (n < cnt)
 275                                 goto e2big;
 276                         memcpy(s, bom, cnt);
 277                         s += cnt, n -= cnt;
 278                 }
 279                 psenc->current_endian = ei->preffered_endian;
 280         }
 281
 282         wc2 = 0;
 283         if ((ei->mode & _MODE_UTF32)==0) {
 284                 /* UTF16 */
 285                 if (wc>0xFFFF) {
 286                         /* surrogate */
 287                         if (wc>0x10FFFF)
 288                                 goto ilseq;
 289                         if (n < 4)
 290                                 goto e2big;
 291                         cnt += 4;
 292                         wc -= 0x10000;
 293                         wc2 = (wc & 0x3FF) | 0xDC00;
 294                         wc = (wc>>10) | 0xD800;
 295                 } else {
 296                         if (n < 2)
 297                                 goto e2big;
 298                         cnt += 2;
 299                 }
 300
 301 surrogate:
 302                 switch (psenc->current_endian) {
 303                 case _ENDIAN_BIG:
 304                         s[1] = wc;
 305                         s[0] = (wc >>= 8);
 306                         break;
 307                 case _ENDIAN_LITTLE:
 308                         s[0] = wc;
 309                         s[1] = (wc >>= 8);
 310                         break;
 311                 }
 312                 if (wc2!=0) {
 313                         wc = wc2;
 314                         wc2 = 0;
 315                         s += 2;
 316                         goto surrogate;
 317                 }
 318         } else {
 319                 /* UTF32 */
 320                 if (wc >= 0xD800 && wc <= 0xDFFF)
 321                         goto ilseq;
 322                 if (n < 4)
 323                         goto e2big;
 324                 cnt += 4;
 325                 switch (psenc->current_endian) {
 326                 case _ENDIAN_BIG:
 327                         s[3] = wc;
 328                         s[2] = (wc >>= 8);
 329                         s[1] = (wc >>= 8);
 330                         s[0] = (wc >>= 8);
 331                         break;
 332                 case _ENDIAN_LITTLE:
 333                         s[0] = wc;
 334                         s[1] = (wc >>= 8);
 335                         s[2] = (wc >>= 8);
 336                         s[3] = (wc >>= 8);
 337                         break;
 338                 }
 339         }
 340         *nresult = cnt;
 341
 342         return 0;
 343
 344 ilseq:
 345         *nresult = (size_t)-1;
 346         return EILSEQ;
 347 e2big:
 348         *nresult = (size_t)-1;
 349         return E2BIG;
 350 }
 351
 352 static void
 353 parse_variable(_UTF1632EncodingInfo * __restrict ei,
 354                const void * __restrict var, size_t lenvar)
 355 {
 356 #define MATCH(x, act)                                           \
 357 do {                                                            \
 358         if (lenvar >= (sizeof(#x)-1) &&                         \
 359             _bcs_strncasecmp(p, #x, sizeof(#x)-1) == 0) {       \
 360                 act;                                            \
 361                 lenvar -= sizeof(#x)-1;                         \
 362                 p += sizeof(#x)-1;                              \
 363         }                                                       \
 364 } while (/*CONSTCOND*/0)
 365         const char *p;
 366         p = var;
 367         while (lenvar>0) {
 368                 switch (*p) {
 369                 case 'B':
 370                 case 'b':
 371                         MATCH(big, ei->preffered_endian = _ENDIAN_BIG);
 372                         break;
 373                 case 'L':
 374                 case 'l':
 375                         MATCH(little, ei->preffered_endian = _ENDIAN_LITTLE);
 376                         break;
 377                 case 'F':
 378                 case 'f':
 379                         MATCH(force, ei->mode |= _MODE_FORCE_ENDIAN);
 380                         break;
 381                 case 'U':
 382                 case 'u':
 383                         MATCH(utf32, ei->mode |= _MODE_UTF32);
 384                         break;
 385                 }
 386                 p++;
 387                 lenvar--;
 388         }
 389 }
 390
 391 static int
 392 /*ARGSUSED*/
 393 _citrus_UTF1632_encoding_module_init(_UTF1632EncodingInfo * __restrict ei,
 394                                      const void * __restrict var,
 395                                      size_t lenvar)
 396 {
 397         _DIAGASSERT(ei != NULL);
 398
 399         memset((void *)ei, 0, sizeof(*ei));
 400
 401         parse_variable(ei, var, lenvar);
 402
 403         if ((ei->mode&_MODE_UTF32)==0)
 404                 ei->cur_max = 6; /* endian + surrogate */
 405         else
 406                 ei->cur_max = 8; /* endian + normal */
 407
 408         if (ei->preffered_endian == _ENDIAN_UNKNOWN) {
 409 #if BYTE_ORDER == BIG_ENDIAN
 410                 ei->preffered_endian = _ENDIAN_BIG;
 411 #else
 412                 ei->preffered_endian = _ENDIAN_LITTLE;
 413 #endif
 414         }
 415
 416         return (0);
 417 }
 418
 419 static void
 420 /*ARGSUSED*/
 421 _citrus_UTF1632_encoding_module_uninit(_UTF1632EncodingInfo *ei __unused)
 422 {
 423 }
 424
 425 static __inline int
 426 /*ARGSUSED*/
 427 _citrus_UTF1632_stdenc_wctocs(_UTF1632EncodingInfo * __restrict ei __unused,
 428                               _csid_t * __restrict csid,
 429                               _index_t * __restrict idx,
 430                               _wc_t wc)
 431 {
 432
 433         _DIAGASSERT(csid != NULL && idx != NULL);
 434
 435         *csid = 0;
 436         *idx = (_index_t)wc;
 437
 438         return (0);
 439 }
 440
 441 static __inline int
 442 /*ARGSUSED*/
 443 _citrus_UTF1632_stdenc_cstowc(_UTF1632EncodingInfo * __restrict ei __unused,
 444                               _wc_t * __restrict wc,
 445                               _csid_t csid, _index_t idx)
 446 {
 447
 448         _DIAGASSERT(wc != NULL);
 449
 450         if (csid != 0)
 451                 return (EILSEQ);
 452
 453         *wc = (_wc_t)idx;
 454
 455         return (0);
 456 }
 457
 458 static __inline int
 459 /*ARGSUSED*/
 460 _citrus_UTF1632_stdenc_get_state_desc_generic(_UTF1632EncodingInfo * __restrict ei __unused,
 461                                               _UTF1632State * __restrict psenc,
 462                                               int * __restrict rstate)
 463 {
 464
 465         if (psenc->chlen == 0)
 466                 *rstate = _STDENC_SDGEN_INITIAL;
 467         else
 468                 *rstate = _STDENC_SDGEN_INCOMPLETE_CHAR;
 469
 470         return 0;
 471 }
 472
 473 /* ----------------------------------------------------------------------
 474  * public interface for stdenc
 475  */
 476
 477 _CITRUS_STDENC_DECLS(UTF1632);
 478 _CITRUS_STDENC_DEF_OPS(UTF1632);
 479
 480 #include "citrus_stdenc_template.h"