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 <gconv.h>
  22 #include <stdint.h>
  23 #include <string.h>
  24 #include "ksc5601.h"
  25
  26 #include <assert.h>
  27
  28 /* This makes obvious what everybody knows: 0x1b is the Esc character.  */
  29 #define ESC     0x1b
  30
  31 /* The shift sequences for this charset (it does not use ESC).  */
  32 #define SI      0x0f
  33 #define SO      0x0e
  34
  35 /* Definitions used in the body of the `gconv' function.  */
  36 #define CHARSET_NAME            "ISO-2022-KR//"
  37 #define DEFINE_INIT             1
  38 #define DEFINE_FINI             1
  39 #define FROM_LOOP               from_iso2022kr_loop
  40 #define TO_LOOP                 to_iso2022kr_loop
  41 #define MIN_NEEDED_FROM         1
  42 #define MAX_NEEDED_FROM         3
  43 #define MIN_NEEDED_TO           4
  44 #define MAX_NEEDED_TO           4
  45 #define PREPARE_LOOP \
  46   int save_set;                                                               \
  47   int *setp = &data->__statep->__count;                                       \
  48   if (!FROM_DIRECTION && !data->__internal_use                                \
  49       && data->__invocation_counter == 0)                                     \
  50     {                                                                         \
  51       /* Emit the designator sequence.  */                                    \
  52       if (outbuf + 4 > outend)                                                \
  53         return __GCONV_FULL_OUTPUT;                                           \
  54                                                                               \
  55       *outbuf++ = ESC;                                                        \
  56       *outbuf++ = '$';                                                        \
  57       *outbuf++ = ')';                                                        \
  58       *outbuf++ = 'C';                                                        \
  59     }
  60 #define EXTRA_LOOP_ARGS         , setp
  61
  62
  63 /* The COUNT element of the state keeps track of the currently selected
  64    character set.  The possible values are:  */
  65 enum
  66 {
  67   ASCII_set = 0,
  68   KSC5601_set = 8
  69 };
  70
  71
  72 /* Since this is a stateful encoding we have to provide code which resets
  73    the output state to the initial state.  This has to be done during the
  74    flushing.  */
  75 #define EMIT_SHIFT_TO_INIT \
  76   if (data->__statep->__count != ASCII_set)                                   \
  77     {                                                                         \
  78       if (FROM_DIRECTION)                                                     \
  79         {                                                                     \
  80           /* It's easy, we don't have to emit anything, we just reset the     \
  81              state for the input.  */                                         \
  82           data->__statep->__count &= 7;                                       \
  83           data->__statep->__count |= ASCII_set;                               \
  84         }                                                                     \
  85       else                                                                    \
  86         {                                                                     \
  87           unsigned char *outbuf = data->__outbuf;                             \
  88                                                                               \
  89           /* We are not in the initial state.  To switch back we have         \
  90              to emit `SI'.  */                                                \
  91           if (__builtin_expect (outbuf == data->__outbufend, 0))              \
  92             /* We don't have enough room in the output buffer.  */            \
  93             status = __GCONV_FULL_OUTPUT;                                     \
  94           else                                                                \
  95             {                                                                 \
  96               /* Write out the shift sequence.  */                            \
  97               *outbuf++ = SI;                                                 \
  98               data->__outbuf = outbuf;                                        \
  99               data->__statep->__count = ASCII_set;                            \
 100             }                                                                 \
 101         }                                                                     \
 102     }
 103
 104
 105 /* Since we might have to reset input pointer we must be able to save
 106    and retore the state.  */
 107 #define SAVE_RESET_STATE(Save) \
 108   if (Save)                                                                   \
 109     save_set = *setp;                                                         \
 110   else                                                                        \
 111     *setp = save_set
 112
 113
 114 /* First define the conversion function from ISO-2022-KR to UCS4.  */
 115 #define MIN_NEEDED_INPUT        MIN_NEEDED_FROM
 116 #define MAX_NEEDED_INPUT        MAX_NEEDED_FROM
 117 #define MIN_NEEDED_OUTPUT       MIN_NEEDED_TO
 118 #define LOOPFCT                 FROM_LOOP
 119 #define BODY \
 120   {                                                                           \
 121     uint32_t ch = *inptr;                                                     \
 122                                                                               \
 123     /* This is a 7bit character set, disallow all 8bit characters.  */        \
 124     if (__builtin_expect (ch, 0) > 0x7f)                                      \
 125       {                                                                       \
 126         if (! ignore_errors_p ())                                             \
 127           {                                                                   \
 128             result = __GCONV_ILLEGAL_INPUT;                                   \
 129             break;                                                            \
 130           }                                                                   \
 131                                                                               \
 132         ++inptr;                                                              \
 133         ++*irreversible;                                                      \
 134         continue;                                                             \
 135       }                                                                       \
 136                                                                               \
 137     /* Recognize escape sequences.  */                                        \
 138     if (__builtin_expect (ch, 0) == ESC)                                      \
 139       {                                                                       \
 140         /* We don't really have to handle escape sequences since all the      \
 141            switching is done using the SI and SO bytes.  But we have to       \
 142            recognize `Esc $ ) C' since this is a kind of flag for this        \
 143            encoding.  We simply ignore it.  */                                \
 144         if ((NEED_LENGTH_TEST && __builtin_expect (inptr + 1 > inend, 0))     \
 145             || (inptr[1] == '$'                                               \
 146                 && ((NEED_LENGTH_TEST                                         \
 147                      && __builtin_expect (inptr + 2 > inend, 0))              \
 148                     || (inptr[2] == ')'                                       \
 149                         && __builtin_expect (inptr + 3 > inend, 0)))))        \
 150                                                                               \
 151           {                                                                   \
 152             result = __GCONV_EMPTY_INPUT;                                     \
 153             break;                                                            \
 154           }                                                                   \
 155         if (inptr[1] == '$' && 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,                                         \
 188                               NEED_LENGTH_TEST ? inend - inptr : 2, 0);       \
 189                                                                               \
 190         if (NEED_LENGTH_TEST && __builtin_expect (ch, 1) == 0)                \
 191           {                                                                   \
 192             result = __GCONV_EMPTY_INPUT;                                     \
 193             break;                                                            \
 194           }                                                                   \
 195         else if (__builtin_expect (ch, 0) == __UNKNOWN_10646_CHAR)            \
 196           {                                                                   \
 197             if (! ignore_errors_p ())                                         \
 198               {                                                               \
 199                 /* This is an illegal character.  */                          \
 200                 result = __GCONV_ILLEGAL_INPUT;                               \
 201                 break;                                                        \
 202               }                                                               \
 203                                                                               \
 204             ++*irreversible;                                                  \
 205             ++inptr;                                                          \
 206             continue;                                                         \
 207           }                                                                   \
 208       }                                                                       \
 209                                                                               \
 210     put32 (outptr, ch);                                                       \
 211     outptr += 4;                                                              \
 212   }
 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 (NEED_LENGTH_TEST && outptr == outend)                         \
 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 (! ignore_errors_p ())                                         \
 259               {                                                               \
 260                 result = __GCONV_ILLEGAL_INPUT;                               \
 261                 break;                                                        \
 262               }                                                               \
 263                                                                               \
 264             ++*irreversible;                                                  \
 265           }                                                                   \
 266         else                                                                  \
 267           {                                                                   \
 268             assert (written == 2);                                            \
 269                                                                               \
 270             /* We use KSC 5601.  */                                           \
 271             if (set != KSC5601_set)                                           \
 272               {                                                               \
 273                 *outptr++ = SO;                                               \
 274                 set = KSC5601_set;                                            \
 275               }                                                               \
 276                                                                               \
 277             if (NEED_LENGTH_TEST && __builtin_expect (outptr + 2 > outend, 0))\
 278               {                                                               \
 279                 result = __GCONV_FULL_OUTPUT;                                 \
 280                 break;                                                        \
 281               }                                                               \
 282                                                                               \
 283             *outptr++ = buf[0];                                               \
 284             *outptr++ = buf[1];                                               \
 285           }                                                                   \
 286       }                                                                       \
 287                                                                               \
 288     /* Now that we wrote the output increment the input pointer.  */          \
 289     inptr += 4;                                                               \
 290   }
 291 #define EXTRA_LOOP_DECLS        , int *setp
 292 #define INIT_PARAMS             int set = *setp
 293 #define UPDATE_PARAMS           *setp = set
 294 #include <iconv/loop.c>
 295
 296
 297 /* Now define the toplevel functions.  */
 298 #include <iconv/skeleton.c>