iconvdata/iso-2022-kr.c

   1 /* Conversion module for ISO-2022-KR.
   2    Copyright (C) 1998 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 (we 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 set = data->statep->count;                                              \
  48   if (!FROM_DIRECTION && !data->internal_use && data->invocation_counter == 0)\
  49     {                                                                         \
  50       /* Emit the designator sequence.  */                                    \
  51       if (outptr + 4 > outend)                                                \
  52         return GCONV_FULL_OUTPUT;                                             \
  53                                                                               \
  54       *outptr++ = '\x1b';                                                     \
  55       *outptr++ = '\x24';                                                     \
  56       *outptr++ = '\x29';                                                     \
  57       *outptr++ = '\x43';                                                     \
  58     }
  59 #define EXTRA_LOOP_ARGS         , set
  60
  61
  62 /* The COUNT element of the state keeps track of the currently selected
  63    character set.  The possible values are:  */
  64 enum
  65 {
  66   ASCII_set = 0,
  67   KSC5601_set
  68 };
  69
  70
  71 /* Since this is a stateful encoding we have to provide code which resets
  72    the output state to the initial state.  This has to be done during the
  73    flushing.  */
  74 #define EMIT_SHIFT_TO_INIT \
  75   if (data->statep->count != 0)                                               \
  76     {                                                                         \
  77       if (step->data == &from_object)                                         \
  78         /* It's easy, we don't have to emit anything, we just reset the       \
  79            state for the input.  */                                           \
  80         data->statep->count = 0;                                              \
  81       else                                                                    \
  82         {                                                                     \
  83           char *outbuf = data->outbuf;                                        \
  84                                                                               \
  85           /* We are not in the initial state.  To switch back we have         \
  86              to emit `SO'.  */                                                \
  87           if (outbuf == data->outbufend)                                      \
  88             /* We don't have enough room in the output buffer.  */            \
  89             status = GCONV_FULL_OUTPUT;                                       \
  90           else                                                                \
  91             {                                                                 \
  92               /* Write out the shift sequence.  */                            \
  93               *outbuf++ = SO;                                                 \
  94               data->outbuf = outbuf;                                          \
  95               data->statep->count = 0;                                        \
  96             }                                                                 \
  97         }                                                                     \
  98     }
  99
 100
 101 /* Since we might have to reset input pointer we must be able to save
 102    and retore the state.  */
 103 #define SAVE_RESET_STATE(Save) \
 104   if (Save)                                                                   \
 105     save_set = set;                                                           \
 106   else                                                                        \
 107     set = save_set
 108
 109
 110 /* First define the conversion function from ISO-2022-JP to UCS4.  */
 111 #define MIN_NEEDED_INPUT        MIN_NEEDED_FROM
 112 #define MAX_NEEDED_INPUT        MAX_NEEDED_FROM
 113 #define MIN_NEEDED_OUTPUT       MIN_NEEDED_TO
 114 #define LOOPFCT                 FROM_LOOP
 115 #define BODY \
 116   {                                                                           \
 117     uint32_t ch = *inptr;                                                     \
 118                                                                               \
 119     /* This is a 7bit character set, disallow all 8bit characters.  */        \
 120     if (ch > 0x7f)                                                            \
 121       {                                                                       \
 122         result = GCONV_ILLEGAL_INPUT;                                         \
 123         break;                                                                \
 124       }                                                                       \
 125                                                                               \
 126     /* Recognize escape sequences.  */                                        \
 127     if (ch == 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.  Butwe have to        \
 131            recognize `Esc $ ) C' since this is a kind of flag for this        \
 132            encoding.  We simply ignore it.  */                                \
 133         if (inptr + 1 > inend                                                 \
 134             || (inptr[1] == '$'                                               \
 135                 && (inptr + 2 > inend                                         \
 136                     || (inptr[2] == ')' && inptr + 3 > inend))))              \
 137                                                                               \
 138           {                                                                   \
 139             result = GCONV_EMPTY_INPUT;                                       \
 140             break;                                                            \
 141           }                                                                   \
 142         if (inptr[1] == '$' && 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 (ch == SI)                                                        \
 150       {                                                                       \
 151         /* Switch to use KSC.  */                                             \
 152         ++inptr;                                                              \
 153         set = KSC5601_set;                                                    \
 154         continue;                                                             \
 155       }                                                                       \
 156     else if (ch == SO)                                                        \
 157       {                                                                       \
 158         /* Switch to use ASCII.  */                                           \
 159         ++inptr;                                                              \
 160         set = ASCII_set;                                                      \
 161         continue;                                                             \
 162       }                                                                       \
 163                                                                               \
 164     if (set == ASCII_set || ch < 0x21 || ch == 0x7f)                          \
 165       /* Almost done, just advance the input pointer.  */                     \
 166       ++inptr;                                                                \
 167     else                                                                      \
 168       {                                                                       \
 169         assert (set == KSC5601_set);                                          \
 170                                                                               \
 171         /* Use the KSC 5601 table.  */                                        \
 172         ch = ksc5601_to_ucs4 (&inptr,                                         \
 173                               NEED_LENGTH_TEST ? inend - inptr : 2, 0);       \
 174                                                                               \
 175         if (NEED_LENGTH_TEST && ch == 0)                                      \
 176           {                                                                   \
 177             result = GCONV_EMPTY_INPUT;                                       \
 178             break;                                                            \
 179           }                                                                   \
 180         else if (ch == UNKNOWN_10646_CHAR)                                    \
 181           {                                                                   \
 182             result = GCONV_ILLEGAL_INPUT;                                     \
 183             break;                                                            \
 184           }                                                                   \
 185       }                                                                       \
 186                                                                               \
 187     *((uint32_t *) outptr)++ = ch;                                            \
 188   }
 189 #define EXTRA_LOOP_DECLS        , int set
 190 #include <iconv/loop.c>
 191
 192
 193 /* Next, define the other direction.  */
 194 #define MIN_NEEDED_INPUT        MIN_NEEDED_TO
 195 #define MIN_NEEDED_OUTPUT       MIN_NEEDED_FROM
 196 #define MAX_NEEDED_OUTPUT       MAX_NEEDED_FROM
 197 #define LOOPFCT                 TO_LOOP
 198 #define BODY \
 199   {                                                                           \
 200     unsigned char ch;                                                         \
 201     size_t written = 0;                                                       \
 202                                                                               \
 203     ch = *((uint32_t *) inptr);                                               \
 204                                                                               \
 205     /* First see whether we can write the character using the currently       \
 206        selected character set.  */                                            \
 207     if (set == ASCII_set || (ch >= 0x01 && (ch < 0x21 || ch == 0x7f)))        \
 208       {                                                                       \
 209         /* Please note that the NUL byte is *not* matched if we are not       \
 210            currently using the ASCII charset.  This is because we must        \
 211            switch to the initial state whenever a NUL byte is written.  */    \
 212         if (ch <= 0x7f)                                                       \
 213           {                                                                   \
 214             *outptr++ = ch;                                                   \
 215             written = 1;                                                      \
 216           }                                                                   \
 217       }                                                                       \
 218     else                                                                      \
 219       {                                                                       \
 220         assert (set == KSC5601_set);                                          \
 221                                                                               \
 222         written = ucs4_to_ksc5601 (ch, outptr,                                \
 223                                    (NEED_LENGTH_TEST ? outend - outptr : 2)); \
 224                                                                               \
 225         if (NEED_LENGTH_TEST && written == 0)                                 \
 226           {                                                                   \
 227             result = GCONV_FULL_OUTPUT;                                       \
 228             break;                                                            \
 229           }                                                                   \
 230         if (written == UNKNOWN_10646_CHAR)                                    \
 231           {                                                                   \
 232             /* Either this is an unknown character or we have to switch       \
 233                the currently selected character set.  The character sets      \
 234                do not code entirely separate parts of ISO 10646 and           \
 235                therefore there is no single correct result.  If we choose     \
 236                the character set to use wrong we might be end up with         \
 237                using yet another character set for the next character         \
 238                though the current and the next could be encoded with one      \
 239                character set.  We leave this kind of optimization for         \
 240                later and now simply use a fixed order in which we test for    \
 241                availability  */                                               \
 242                                                                               \
 243             if (ch <= 0x7f)                                                   \
 244               {                                                               \
 245                 /* We must encode using ASCII.  First write out the           \
 246                    escape sequence.  */                                       \
 247                 *outptr++ = SO;                                               \
 248                 set = ASCII_set;                                              \
 249                                                                               \
 250                 if (NEED_LENGTH_TEST && outptr == outend)                     \
 251                   {                                                           \
 252                     result = GCONV_FULL_OUTPUT;                               \
 253                     break;                                                    \
 254                   }                                                           \
 255                                                                               \
 256                 *outptr++ = ch;                                               \
 257               }                                                               \
 258             else                                                              \
 259               {                                                               \
 260                 char buf[2];                                                  \
 261                                                                               \
 262                 written = ucs4_to_ksc5601 (ch, buf, 2);                       \
 263                 if (written != UNKNOWN_10646_CHAR)                            \
 264                   {                                                           \
 265                     /* We use KSC 5601.  */                                   \
 266                     *outptr++ = SI;                                           \
 267                     set = KSC5601_set;                                        \
 268                                                                               \
 269                     if (NEED_LENGTH_TEST && outptr + 2 > outend)              \
 270                       {                                                       \
 271                         result = GCONV_FULL_OUTPUT;                           \
 272                         break;                                                \
 273                       }                                                       \
 274                                                                               \
 275                     *outptr++ = buf[0];                                       \
 276                     *outptr++ = buf[1];                                       \
 277                   }                                                           \
 278                 else                                                          \
 279                   {                                                           \
 280                     result = GCONV_ILLEGAL_INPUT;                             \
 281                     break;                                                    \
 282                   }                                                           \
 283               }                                                               \
 284           }                                                                   \
 285       }                                                                       \
 286                                                                               \
 287     /* Now that we wrote the output increment the input pointer.  */          \
 288     inptr += 4;                                                               \
 289   }
 290 #define EXTRA_LOOP_DECLS        , int set
 291 #include <iconv/loop.c>
 292
 293
 294 /* Now define the toplevel functions.  */
 295 #include <iconv/skeleton.c>