iconvdata/iso-2022-kr.c

   1 /* Conversion module for ISO-2022-KR.
   2    Copyright (C) 1998, 1999, 2000 Free Software Foundation, Inc.
   3    This file is part of the GNU C Library.
   4    Contributed by Ulrich Drepper <drepper@cygnus.com>, 1998.
   5
   6    The GNU C Library is free software; you can redistribute it and/or
   7    modify it under the terms of the GNU Library General Public License as
   8    published by the Free Software Foundation; either version 2 of the
   9    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    Library General Public License for more details.
  15
  16    You should have received a copy of the GNU Library General Public
  17    License along with the GNU C Library; see the file COPYING.LIB.  If not,
  18    write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  19    Boston, MA 02111-1307, USA.  */
  20
  21 #include <dlfcn.h>
  22 #include <gconv.h>
  23 #include <stdint.h>
  24 #include <string.h>
  25 #include "ksc5601.h"
  26
  27 #include <assert.h>
  28
  29 /* This makes obvious what everybody knows: 0x1b is the Esc character.  */
  30 #define ESC     0x1b
  31
  32 /* The shift sequences for this charset (it does not use ESC).  */
  33 #define SI      0x0f
  34 #define SO      0x0e
  35
  36 /* Definitions used in the body of the `gconv' function.  */
  37 #define CHARSET_NAME            "ISO-2022-KR//"
  38 #define DEFINE_INIT             1
  39 #define DEFINE_FINI             1
  40 #define FROM_LOOP               from_iso2022kr_loop
  41 #define TO_LOOP                 to_iso2022kr_loop
  42 #define MIN_NEEDED_FROM         1
  43 #define MAX_NEEDED_FROM         3
  44 #define MIN_NEEDED_TO           4
  45 #define MAX_NEEDED_TO           4
  46 #define PREPARE_LOOP \
  47   int save_set;                                                               \
  48   int *setp = &data->__statep->__count;                                       \
  49   if (!FROM_DIRECTION && !data->__internal_use                                \
  50       && data->__invocation_counter == 0)                                     \
  51     {                                                                         \
  52       /* Emit the designator sequence.  */                                    \
  53       if (outbuf + 4 > outend)                                                \
  54         return __GCONV_FULL_OUTPUT;                                           \
  55                                                                               \
  56       *outbuf++ = ESC;                                                        \
  57       *outbuf++ = '$';                                                        \
  58       *outbuf++ = ')';                                                        \
  59       *outbuf++ = 'C';                                                        \
  60     }
  61 #define EXTRA_LOOP_ARGS         , setp
  62
  63
  64 /* The COUNT element of the state keeps track of the currently selected
  65    character set.  The possible values are:  */
  66 enum
  67 {
  68   ASCII_set = 0,
  69   KSC5601_set = 8
  70 };
  71
  72
  73 /* Since this is a stateful encoding we have to provide code which resets
  74    the output state to the initial state.  This has to be done during the
  75    flushing.  */
  76 #define EMIT_SHIFT_TO_INIT \
  77   if (data->__statep->__count != ASCII_set)                                   \
  78     {                                                                         \
  79       if (FROM_DIRECTION)                                                     \
  80         {                                                                     \
  81           /* It's easy, we don't have to emit anything, we just reset the     \
  82              state for the input.  */                                         \
  83           data->__statep->__count &= 7;                                       \
  84           data->__statep->__count |= ASCII_set;                               \
  85         }                                                                     \
  86       else                                                                    \
  87         {                                                                     \
  88           unsigned char *outbuf = data->__outbuf;                             \
  89                                                                               \
  90           /* We are not in the initial state.  To switch back we have         \
  91              to emit `SI'.  */                                                \
  92           if (__builtin_expect (outbuf == data->__outbufend, 0))              \
  93             /* We don't have enough room in the output buffer.  */            \
  94             status = __GCONV_FULL_OUTPUT;                                     \
  95           else                                                                \
  96             {                                                                 \
  97               /* Write out the shift sequence.  */                            \
  98               *outbuf++ = SI;                                                 \
  99               data->__outbuf = outbuf;                                        \
 100               data->__statep->__count = ASCII_set;                            \
 101             }                                                                 \
 102         }                                                                     \
 103     }
 104
 105
 106 /* Since we might have to reset input pointer we must be able to save
 107    and retore the state.  */
 108 #define SAVE_RESET_STATE(Save) \
 109   if (Save)                                                                   \
 110     save_set = *setp;                                                         \
 111   else                                                                        \
 112     *setp = save_set
 113
 114
 115 /* First define the conversion function from ISO-2022-KR to UCS4.  */
 116 #define MIN_NEEDED_INPUT        MIN_NEEDED_FROM
 117 #define MAX_NEEDED_INPUT        MAX_NEEDED_FROM
 118 #define MIN_NEEDED_OUTPUT       MIN_NEEDED_TO
 119 #define LOOPFCT                 FROM_LOOP
 120 #define BODY \
 121   {                                                                           \
 122     uint32_t ch = *inptr;                                                     \
 123                                                                               \
 124     /* This is a 7bit character set, disallow all 8bit characters.  */        \
 125     if (__builtin_expect (ch, 0) > 0x7f)                                      \
 126       {                                                                       \
 127         if (! ignore_errors_p ())                                             \
 128           {                                                                   \
 129             result = __GCONV_ILLEGAL_INPUT;                                   \
 130             break;                                                            \
 131           }                                                                   \
 132                                                                               \
 133         ++inptr;                                                              \
 134         ++*irreversible;                                                      \
 135         continue;                                                             \
 136       }                                                                       \
 137                                                                               \
 138     /* Recognize escape sequences.  */                                        \
 139     if (__builtin_expect (ch, 0) == ESC)                                      \
 140       {                                                                       \
 141         /* We don't really have to handle escape sequences since all the      \
 142            switching is done using the SI and SO bytes.  But we have to       \
 143            recognize `Esc $ ) C' since this is a kind of flag for this        \
 144            encoding.  We simply ignore it.  */                                \
 145         if (__builtin_expect (inptr + 1 > inend, 0)                           \
 146             || (inptr[1] == '$'                                               \
 147                 && (__builtin_expect (inptr + 2 > inend, 0)                   \
 148                     || (inptr[2] == ')'                                       \
 149                         && __builtin_expect (inptr + 3 > inend, 0)))))        \
 150                                                                               \
 151           {                                                                   \
 152             result = __GCONV_EMPTY_INPUT;                                     \
 153             break;                                                            \
 154           }                                                                   \
 155         if (inptr[1] == '$' && inptr[2] == ')' && inptr[3] == 'C')            \
 156           {                                                                   \
 157             /* Yeah, yeah, we know this is ISO 2022-KR.  */                   \
 158             inptr += 4;                                                       \
 159             continue;                                                         \
 160           }                                                                   \
 161       }                                                                       \
 162     else if (__builtin_expect (ch, 0) == SO)                                  \
 163       {                                                                       \
 164         /* Switch to use KSC.  */                                             \
 165         ++inptr;                                                              \
 166         set = KSC5601_set;                                                    \
 167         continue;                                                             \
 168       }                                                                       \
 169     else if (__builtin_expect (ch, 0) == SI)                                  \
 170       {                                                                       \
 171         /* Switch to use ASCII.  */                                           \
 172         ++inptr;                                                              \
 173         set = ASCII_set;                                                      \
 174         continue;                                                             \
 175       }                                                                       \
 176                                                                               \
 177     if (set == ASCII_set)                                                     \
 178       {                                                                       \
 179         /* Almost done, just advance the input pointer.  */                   \
 180         ++inptr;                                                              \
 181       }                                                                       \
 182     else                                                                      \
 183       {                                                                       \
 184         assert (set == KSC5601_set);                                          \
 185                                                                               \
 186         /* Use the KSC 5601 table.  */                                        \
 187         ch = ksc5601_to_ucs4 (&inptr, inend - inptr, 0);                      \
 188                                                                               \
 189         if (__builtin_expect (ch, 1) == 0)                                    \
 190           {                                                                   \
 191             result = __GCONV_EMPTY_INPUT;                                     \
 192             break;                                                            \
 193           }                                                                   \
 194         else if (__builtin_expect (ch, 0) == __UNKNOWN_10646_CHAR)            \
 195           {                                                                   \
 196             if (! ignore_errors_p ())                                         \
 197               {                                                               \
 198                 /* This is an illegal character.  */                          \
 199                 result = __GCONV_ILLEGAL_INPUT;                               \
 200                 break;                                                        \
 201               }                                                               \
 202                                                                               \
 203             ++*irreversible;                                                  \
 204             ++inptr;                                                          \
 205             continue;                                                         \
 206           }                                                                   \
 207       }                                                                       \
 208                                                                               \
 209     put32 (outptr, ch);                                                       \
 210     outptr += 4;                                                              \
 211   }
 212 #define LOOP_NEED_FLAGS
 213 #define EXTRA_LOOP_DECLS        , int *setp
 214 #define INIT_PARAMS             int set = *setp
 215 #define UPDATE_PARAMS           *setp = set
 216 #include <iconv/loop.c>
 217
 218
 219 /* Next, define the other direction.  */
 220 #define MIN_NEEDED_INPUT        MIN_NEEDED_TO
 221 #define MIN_NEEDED_OUTPUT       MIN_NEEDED_FROM
 222 #define MAX_NEEDED_OUTPUT       MAX_NEEDED_FROM
 223 #define LOOPFCT                 TO_LOOP
 224 #define BODY \
 225   {                                                                           \
 226     uint32_t ch;                                                              \
 227     size_t written = 0;                                                       \
 228                                                                               \
 229     ch = get32 (inptr);                                                       \
 230                                                                               \
 231     /* First see whether we can write the character using the currently       \
 232        selected character set.  */                                            \
 233     if (ch < 0x80)                                                            \
 234       {                                                                       \
 235         if (set != ASCII_set)                                                 \
 236           {                                                                   \
 237             *outptr++ = SI;                                                   \
 238             set = ASCII_set;                                                  \
 239             if (__builtin_expect (outptr == outend, 0))                       \
 240               {                                                               \
 241                 result = __GCONV_FULL_OUTPUT;                                 \
 242                 break;                                                        \
 243               }                                                               \
 244           }                                                                   \
 245                                                                               \
 246         *outptr++ = ch;                                                       \
 247         written = 1;                                                          \
 248       }                                                                       \
 249     else                                                                      \
 250       {                                                                       \
 251         char buf[2];                                                          \
 252                                                                               \
 253         written = ucs4_to_ksc5601 (ch, buf, 2);                               \
 254                                                                               \
 255         if (__builtin_expect (written, 0) == __UNKNOWN_10646_CHAR)            \
 256           {                                                                   \
 257             /* Illegal character.  */                                         \
 258             if (step_data->__trans.__trans_fct != NULL)                       \
 259               {                                                               \
 260                 result = DL_CALL_FCT (step_data->__trans.__trans_fct,         \
 261                                       (step, step_data, *inptrp, &inptr,      \
 262                                        inend, *outptrp, &outptr, outend,      \
 263                                        irreversible));                        \
 264                 if (result != __GCONV_OK)                                     \
 265                   break;                                                      \
 266               }                                                               \
 267             else if (! ignore_errors_p ())                                    \
 268               {                                                               \
 269                 result = __GCONV_ILLEGAL_INPUT;                               \
 270                 break;                                                        \
 271               }                                                               \
 272             else                                                              \
 273               {                                                               \
 274                 ++*irreversible;                                              \
 275                 inptr += 4;                                                   \
 276               }                                                               \
 277             continue;                                                         \
 278           }                                                                   \
 279         else                                                                  \
 280           {                                                                   \
 281             assert (written == 2);                                            \
 282                                                                               \
 283             /* We use KSC 5601.  */                                           \
 284             if (set != KSC5601_set)                                           \
 285               {                                                               \
 286                 *outptr++ = SO;                                               \
 287                 set = KSC5601_set;                                            \
 288               }                                                               \
 289                                                                               \
 290             if (__builtin_expect (outptr + 2 > outend, 0))                    \
 291               {                                                               \
 292                 result = __GCONV_FULL_OUTPUT;                                 \
 293                 break;                                                        \
 294               }                                                               \
 295                                                                               \
 296             *outptr++ = buf[0];                                               \
 297             *outptr++ = buf[1];                                               \
 298           }                                                                   \
 299       }                                                                       \
 300                                                                               \
 301     /* Now that we wrote the output increment the input pointer.  */          \
 302     inptr += 4;                                                               \
 303   }
 304 #define LOOP_NEED_FLAGS
 305 #define EXTRA_LOOP_DECLS        , int *setp
 306 #define INIT_PARAMS             int set = *setp
 307 #define UPDATE_PARAMS           *setp = set
 308 #include <iconv/loop.c>
 309
 310
 311 /* Now define the toplevel functions.  */
 312 #include <iconv/skeleton.c>