iconvdata/iso-2022-kr.c

   1 /* Conversion module for ISO-2022-KR.
   2    Copyright (C) 1998, 1999, 2000-2002, 2007, 2008
   3    Free Software Foundation, Inc.
   4    This file is part of the GNU C Library.
   5    Contributed by Ulrich Drepper <drepper@cygnus.com>, 1998.
   6
   7    The GNU C Library is free software; you can redistribute it and/or
   8    modify it under the terms of the GNU Lesser General Public
   9    License as published by the Free Software Foundation; either
  10    version 2.1 of the License, or (at your option) any later version.
  11
  12    The GNU C Library is distributed in the hope that it will be useful,
  13    but WITHOUT ANY WARRANTY; without even the implied warranty of
  14    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  15    Lesser General Public License for more details.
  16
  17    You should have received a copy of the GNU Lesser General Public
  18    License along with the GNU C Library; if not, write to the Free
  19    Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
  20    02111-1307 USA.  */
  21
  22 #include <dlfcn.h>
  23 #include <gconv.h>
  24 #include <stdint.h>
  25 #include <string.h>
  26 #include "ksc5601.h"
  27
  28 #include <assert.h>
  29
  30 /* This makes obvious what everybody knows: 0x1b is the Esc character.  */
  31 #define ESC     0x1b
  32
  33 /* The shift sequences for this charset (it does not use ESC).  */
  34 #define SI      0x0f
  35 #define SO      0x0e
  36
  37 /* Definitions used in the body of the `gconv' function.  */
  38 #define CHARSET_NAME            "ISO-2022-KR//"
  39 #define DEFINE_INIT             1
  40 #define DEFINE_FINI             1
  41 #define FROM_LOOP               from_iso2022kr_loop
  42 #define TO_LOOP                 to_iso2022kr_loop
  43 #define MIN_NEEDED_FROM         1
  44 #define MAX_NEEDED_FROM         4
  45 #define MIN_NEEDED_TO           4
  46 #define MAX_NEEDED_TO           4
  47 #define PREPARE_LOOP \
  48   int save_set;                                                               \
  49   int *setp = &data->__statep->__count;                                       \
  50   if (!FROM_DIRECTION && !data->__internal_use                                \
  51       && data->__invocation_counter == 0)                                     \
  52     {                                                                         \
  53       /* Emit the designator sequence.  */                                    \
  54       if (outbuf + 4 > outend)                                                \
  55         return __GCONV_FULL_OUTPUT;                                           \
  56                                                                               \
  57       *outbuf++ = ESC;                                                        \
  58       *outbuf++ = '$';                                                        \
  59       *outbuf++ = ')';                                                        \
  60       *outbuf++ = 'C';                                                        \
  61     }
  62 #define EXTRA_LOOP_ARGS         , setp
  63
  64
  65 /* The COUNT element of the state keeps track of the currently selected
  66    character set.  The possible values are:  */
  67 enum
  68 {
  69   ASCII_set = 0,
  70   KSC5601_set = 8
  71 };
  72
  73
  74 /* Since this is a stateful encoding we have to provide code which resets
  75    the output state to the initial state.  This has to be done during the
  76    flushing.  */
  77 #define EMIT_SHIFT_TO_INIT \
  78   if (data->__statep->__count != ASCII_set)                                   \
  79     {                                                                         \
  80       if (FROM_DIRECTION)                                                     \
  81         {                                                                     \
  82           /* It's easy, we don't have to emit anything, we just reset the     \
  83              state for the input.  */                                         \
  84           data->__statep->__count &= 7;                                       \
  85           data->__statep->__count |= ASCII_set;                               \
  86         }                                                                     \
  87       else                                                                    \
  88         {                                                                     \
  89           /* We are not in the initial state.  To switch back we have         \
  90              to emit `SI'.  */                                                \
  91           if (__builtin_expect (outbuf == outend, 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->__statep->__count = ASCII_set;                            \
  99             }                                                                 \
 100         }                                                                     \
 101     }
 102
 103
 104 /* Since we might have to reset input pointer we must be able to save
 105    and retore the state.  */
 106 #define SAVE_RESET_STATE(Save) \
 107   if (Save)                                                                   \
 108     save_set = *setp;                                                         \
 109   else                                                                        \
 110     *setp = save_set
 111
 112
 113 /* First define the conversion function from ISO-2022-KR to UCS4.  */
 114 #define MIN_NEEDED_INPUT        MIN_NEEDED_FROM
 115 #define MAX_NEEDED_INPUT        MAX_NEEDED_FROM
 116 #define MIN_NEEDED_OUTPUT       MIN_NEEDED_TO
 117 #define LOOPFCT                 FROM_LOOP
 118 #define BODY \
 119   {                                                                           \
 120     uint32_t ch = *inptr;                                                     \
 121                                                                               \
 122     /* This is a 7bit character set, disallow all 8bit characters.  */        \
 123     if (__builtin_expect (ch > 0x7f, 0))                                      \
 124       STANDARD_FROM_LOOP_ERR_HANDLER (1);                                     \
 125                                                                               \
 126     /* Recognize escape sequences.  */                                        \
 127     if (__builtin_expect (ch, 0) == ESC)                                      \
 128       {                                                                       \
 129         /* We don't really have to handle escape sequences since all the      \
 130            switching is done using the SI and SO bytes.  But we have to       \
 131            recognize `Esc $ ) C' since this is a kind of flag for this        \
 132            encoding.  We simply ignore it.  */                                \
 133         if (__builtin_expect (inptr + 2 > inend, 0)                           \
 134             || (inptr[1] == '$'                                               \
 135                 && (__builtin_expect (inptr + 3 > inend, 0)                   \
 136                     || (inptr[2] == ')'                                       \
 137                         && __builtin_expect (inptr + 4 > inend, 0)))))        \
 138           {                                                                   \
 139             result = __GCONV_INCOMPLETE_INPUT;                                \
 140             break;                                                            \
 141           }                                                                   \
 142         if (inptr[1] == '$' && inptr[2] == ')' && inptr[3] == 'C')            \
 143           {                                                                   \
 144             /* Yeah, yeah, we know this is ISO 2022-KR.  */                   \
 145             inptr += 4;                                                       \
 146             continue;                                                         \
 147           }                                                                   \
 148       }                                                                       \
 149     else if (__builtin_expect (ch, 0) == SO)                                  \
 150       {                                                                       \
 151         /* Switch to use KSC.  */                                             \
 152         ++inptr;                                                              \
 153         set = KSC5601_set;                                                    \
 154         continue;                                                             \
 155       }                                                                       \
 156     else if (__builtin_expect (ch, 0) == SI)                                  \
 157       {                                                                       \
 158         /* Switch to use ASCII.  */                                           \
 159         ++inptr;                                                              \
 160         set = ASCII_set;                                                      \
 161         continue;                                                             \
 162       }                                                                       \
 163                                                                               \
 164     if (set == ASCII_set)                                                     \
 165       {                                                                       \
 166         /* Almost done, just advance the input pointer.  */                   \
 167         ++inptr;                                                              \
 168       }                                                                       \
 169     else                                                                      \
 170       {                                                                       \
 171         assert (set == KSC5601_set);                                          \
 172                                                                               \
 173         /* Use the KSC 5601 table.  */                                        \
 174         ch = ksc5601_to_ucs4 (&inptr, inend - inptr, 0);                      \
 175                                                                               \
 176         if (__builtin_expect (ch == 0, 0))                                    \
 177           {                                                                   \
 178             result = __GCONV_INCOMPLETE_INPUT;                                \
 179             break;                                                            \
 180           }                                                                   \
 181         else if (__builtin_expect (ch == __UNKNOWN_10646_CHAR, 0))            \
 182           {                                                                   \
 183             STANDARD_FROM_LOOP_ERR_HANDLER (1);                               \
 184           }                                                                   \
 185       }                                                                       \
 186                                                                               \
 187     put32 (outptr, ch);                                                       \
 188     outptr += 4;                                                              \
 189   }
 190 #define LOOP_NEED_FLAGS
 191 #define EXTRA_LOOP_DECLS        , int *setp
 192 #define INIT_PARAMS             int set = *setp
 193 #define UPDATE_PARAMS           *setp = set
 194 #include <iconv/loop.c>
 195
 196
 197 /* Next, define the other direction.  */
 198 #define MIN_NEEDED_INPUT        MIN_NEEDED_TO
 199 #define MIN_NEEDED_OUTPUT       MIN_NEEDED_FROM
 200 #define MAX_NEEDED_OUTPUT       MAX_NEEDED_FROM
 201 #define LOOPFCT                 TO_LOOP
 202 #define BODY \
 203   {                                                                           \
 204     uint32_t ch = get32 (inptr);                                              \
 205                                                                               \
 206     /* First see whether we can write the character using the currently       \
 207        selected character set.  */                                            \
 208     if (ch < 0x80)                                                            \
 209       {                                                                       \
 210         if (set != ASCII_set)                                                 \
 211           {                                                                   \
 212             *outptr++ = SI;                                                   \
 213             set = ASCII_set;                                                  \
 214             if (__builtin_expect (outptr == outend, 0))                       \
 215               {                                                               \
 216                 result = __GCONV_FULL_OUTPUT;                                 \
 217                 break;                                                        \
 218               }                                                               \
 219           }                                                                   \
 220                                                                               \
 221         *outptr++ = ch;                                                       \
 222       }                                                                       \
 223     else                                                                      \
 224       {                                                                       \
 225         unsigned char buf[2];                                                 \
 226         /* Fake initialization to keep gcc quiet.  */                         \
 227         asm ("" : "=m" (buf));                                                \
 228                                                                               \
 229         size_t written = ucs4_to_ksc5601 (ch, buf, 2);                        \
 230         if (__builtin_expect (written, 0) == __UNKNOWN_10646_CHAR)            \
 231           {                                                                   \
 232             UNICODE_TAG_HANDLER (ch, 4);                                      \
 233                                                                               \
 234             /* Illegal character.  */                                         \
 235             STANDARD_TO_LOOP_ERR_HANDLER (4);                                 \
 236           }                                                                   \
 237         else                                                                  \
 238           {                                                                   \
 239             assert (written == 2);                                            \
 240                                                                               \
 241             /* We use KSC 5601.  */                                           \
 242             if (set != KSC5601_set)                                           \
 243               {                                                               \
 244                 *outptr++ = SO;                                               \
 245                 set = KSC5601_set;                                            \
 246               }                                                               \
 247                                                                               \
 248             if (__builtin_expect (outptr + 2 > outend, 0))                    \
 249               {                                                               \
 250                 result = __GCONV_FULL_OUTPUT;                                 \
 251                 break;                                                        \
 252               }                                                               \
 253                                                                               \
 254             *outptr++ = buf[0];                                               \
 255             *outptr++ = buf[1];                                               \
 256           }                                                                   \
 257       }                                                                       \
 258                                                                               \
 259     /* Now that we wrote the output increment the input pointer.  */          \
 260     inptr += 4;                                                               \
 261   }
 262 #define LOOP_NEED_FLAGS
 263 #define EXTRA_LOOP_DECLS        , int *setp
 264 #define INIT_PARAMS             int set = *setp
 265 #define REINIT_PARAMS           set = *setp
 266 #define UPDATE_PARAMS           *setp = set
 267 #include <iconv/loop.c>
 268
 269
 270 /* Now define the toplevel functions.  */
 271 #include <iconv/skeleton.c>