iconvdata/shift_jisx0213.c

   1 /* Conversion from and to Shift_JISX0213.
   2    Copyright (C) 2002-2014 Free Software Foundation, Inc.
   3    This file is part of the GNU C Library.
   4    Contributed by Bruno Haible <bruno@clisp.org>, 2002.
   5
   6    The GNU C Library is free software; you can redistribute it and/or
   7    modify it under the terms of the GNU Lesser General Public
   8    License as published by the Free Software Foundation; either
   9    version 2.1 of the 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    Lesser General Public License for more details.
  15
  16    You should have received a copy of the GNU Lesser General Public
  17    License along with the GNU C Library; if not, see
  18    <http://www.gnu.org/licenses/>.  */
  19
  20 #include <dlfcn.h>
  21 #include <stdint.h>
  22 #include <gconv.h>
  23
  24 /* The structure of Shift_JISX0213 is as follows:
  25
  26    0x00..0x7F: ISO646-JP, an ASCII variant
  27
  28    0x{A1..DF}: JISX0201 Katakana.
  29
  30    0x{81..9F,E0..EF}{40..7E,80..FC}: JISX0213 plane 1.
  31
  32    0x{F0..FC}{40..7E,80..FC}: JISX0213 plane 2, with irregular row mapping.
  33
  34    Note that some JISX0213 characters are not contained in Unicode 3.2
  35    and are therefore best represented as sequences of Unicode characters.
  36 */
  37
  38 #include "jisx0213.h"
  39
  40 /* Definitions used in the body of the `gconv' function.  */
  41 #define CHARSET_NAME            "SHIFT_JISX0213//"
  42 #define FROM_LOOP               from_shift_jisx0213
  43 #define TO_LOOP                 to_shift_jisx0213
  44 #define DEFINE_INIT             1
  45 #define DEFINE_FINI             1
  46 #define FROM_LOOP_MIN_NEEDED_FROM       1
  47 #define FROM_LOOP_MAX_NEEDED_FROM       2
  48 #define FROM_LOOP_MIN_NEEDED_TO         4
  49 #define FROM_LOOP_MAX_NEEDED_TO         8
  50 #define TO_LOOP_MIN_NEEDED_FROM         4
  51 #define TO_LOOP_MAX_NEEDED_FROM         4
  52 #define TO_LOOP_MIN_NEEDED_TO           1
  53 #define TO_LOOP_MAX_NEEDED_TO           2
  54 #define PREPARE_LOOP \
  55   int saved_state;                                                            \
  56   int *statep = &data->__statep->__count;
  57 #define EXTRA_LOOP_ARGS         , statep
  58
  59
  60 /* Since we might have to reset input pointer we must be able to save
  61    and restore the state.  */
  62 #define SAVE_RESET_STATE(Save) \
  63   if (Save)                                                                   \
  64     saved_state = *statep;                                                    \
  65   else                                                                        \
  66     *statep = saved_state
  67
  68
  69 /* During Shift_JISX0213 to UCS-4 conversion, the COUNT element of the state
  70    contains the last UCS-4 character, shifted by 3 bits.
  71    During UCS-4 to Shift_JISX0213 conversion, the COUNT element of the state
  72    contains the last two bytes to be output, shifted by 3 bits.  */
  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 != 0)                                           \
  79     {                                                                         \
  80       if (FROM_DIRECTION)                                                     \
  81         {                                                                     \
  82           if (__builtin_expect (outbuf + 4 <= outend, 1))                     \
  83             {                                                                 \
  84               /* Write out the last character.  */                            \
  85               *((uint32_t *) outbuf) = data->__statep->__count >> 3;          \
  86               outbuf += sizeof (uint32_t);                                    \
  87               data->__statep->__count = 0;                                    \
  88             }                                                                 \
  89           else                                                                \
  90             /* We don't have enough room in the output buffer.  */            \
  91             status = __GCONV_FULL_OUTPUT;                                     \
  92         }                                                                     \
  93       else                                                                    \
  94         {                                                                     \
  95           if (__builtin_expect (outbuf + 2 <= outend, 1))                     \
  96             {                                                                 \
  97               /* Write out the last character.  */                            \
  98               uint32_t lasttwo = data->__statep->__count >> 3;                \
  99               *outbuf++ = (lasttwo >> 8) & 0xff;                              \
 100               *outbuf++ = lasttwo & 0xff;                                     \
 101               data->__statep->__count = 0;                                    \
 102             }                                                                 \
 103           else                                                                \
 104             /* We don't have enough room in the output buffer.  */            \
 105             status = __GCONV_FULL_OUTPUT;                                     \
 106         }                                                                     \
 107     }
 108
 109
 110 /* First define the conversion function from Shift_JISX0213 to UCS-4.  */
 111 #define MIN_NEEDED_INPUT        FROM_LOOP_MIN_NEEDED_FROM
 112 #define MAX_NEEDED_INPUT        FROM_LOOP_MAX_NEEDED_FROM
 113 #define MIN_NEEDED_OUTPUT       FROM_LOOP_MIN_NEEDED_TO
 114 #define MAX_NEEDED_OUTPUT       FROM_LOOP_MAX_NEEDED_TO
 115 #define LOOPFCT                 FROM_LOOP
 116 #define BODY \
 117   {                                                                           \
 118     uint32_t ch;                                                              \
 119                                                                               \
 120     /* Determine whether there is a buffered character pending.  */           \
 121     ch = *statep >> 3;                                                        \
 122     if (__builtin_expect (ch == 0, 1))                                        \
 123       {                                                                       \
 124         /* No - so look at the next input byte.  */                           \
 125         ch = *inptr;                                                          \
 126                                                                               \
 127         if (ch < 0x80)                                                        \
 128           {                                                                   \
 129             /* Plain ISO646-JP character.  */                                 \
 130             if (__builtin_expect (ch == 0x5c, 0))                             \
 131               ch = 0xa5;                                                      \
 132             else if (__builtin_expect (ch == 0x7e, 0))                        \
 133               ch = 0x203e;                                                    \
 134             ++inptr;                                                          \
 135           }                                                                   \
 136         else if (ch >= 0xa1 && ch <= 0xdf)                                    \
 137           {                                                                   \
 138             /* Half-width katakana.  */                                       \
 139             ch += 0xfec0;                                                     \
 140             ++inptr;                                                          \
 141           }                                                                   \
 142         else if ((ch >= 0x81 && ch <= 0x9f) || (ch >= 0xe0 && ch <= 0xfc))    \
 143           {                                                                   \
 144             /* Two byte character.  */                                        \
 145             uint32_t ch2;                                                     \
 146                                                                               \
 147             if (__builtin_expect (inptr + 1 >= inend, 0))                     \
 148               {                                                               \
 149                 /* The second byte is not available.  */                      \
 150                 result = __GCONV_INCOMPLETE_INPUT;                            \
 151                 break;                                                        \
 152               }                                                               \
 153                                                                               \
 154             ch2 = inptr[1];                                                   \
 155                                                                               \
 156             /* The second byte must be in the range 0x{40..7E,80..FC}.  */    \
 157             if (__builtin_expect (ch2 < 0x40 || ch2 == 0x7f || ch2 > 0xfc, 0))\
 158               {                                                               \
 159                 /* This is an illegal character.  */                          \
 160                 STANDARD_FROM_LOOP_ERR_HANDLER (1);                           \
 161               }                                                               \
 162                                                                               \
 163             /* Convert to row and column.  */                                 \
 164             if (ch < 0xe0)                                                    \
 165               ch -= 0x81;                                                     \
 166             else                                                              \
 167               ch -= 0xc1;                                                     \
 168             if (ch2 < 0x80)                                                   \
 169               ch2 -= 0x40;                                                    \
 170             else                                                              \
 171               ch2 -= 0x41;                                                    \
 172             /* Now 0 <= ch <= 0x3b, 0 <= ch2 <= 0xbb.  */                     \
 173             ch = 2 * ch;                                                      \
 174             if (ch2 >= 0x5e)                                                  \
 175               ch2 -= 0x5e, ch++;                                              \
 176             ch2 += 0x21;                                                      \
 177             if (ch >= 0x5e)                                                   \
 178               {                                                               \
 179                 /* Handling of JISX 0213 plane 2 rows.  */                    \
 180                 if (ch >= 0x67)                                               \
 181                   ch += 230;                                                  \
 182                 else if (ch >= 0x63 || ch == 0x5f)                            \
 183                   ch += 168;                                                  \
 184                 else                                                          \
 185                   ch += 162;                                                  \
 186               }                                                               \
 187                                                                               \
 188             ch = jisx0213_to_ucs4 (0x121 + ch, ch2);                          \
 189                                                                               \
 190             if (ch == 0)                                                      \
 191               {                                                               \
 192                 /* This is an illegal character.  */                          \
 193                 STANDARD_FROM_LOOP_ERR_HANDLER (1);                           \
 194               }                                                               \
 195                                                                               \
 196             inptr += 2;                                                       \
 197                                                                               \
 198             if (ch < 0x80)                                                    \
 199               {                                                               \
 200                 /* It's a combining character.  */                            \
 201                 uint32_t u1 = __jisx0213_to_ucs_combining[ch - 1][0];         \
 202                 uint32_t u2 = __jisx0213_to_ucs_combining[ch - 1][1];         \
 203                                                                               \
 204                 put32 (outptr, u1);                                           \
 205                 outptr += 4;                                                  \
 206                                                                               \
 207                 /* See whether we have room for two characters.  */           \
 208                 if (outptr + 4 <= outend)                                     \
 209                   {                                                           \
 210                     put32 (outptr, u2);                                       \
 211                     outptr += 4;                                              \
 212                     continue;                                                 \
 213                   }                                                           \
 214                                                                               \
 215                 /* Otherwise store only the first character now, and          \
 216                    put the second one into the queue.  */                     \
 217                 *statep = u2 << 3;                                            \
 218                 /* Tell the caller why we terminate the loop.  */             \
 219                 result = __GCONV_FULL_OUTPUT;                                 \
 220                 break;                                                        \
 221               }                                                               \
 222           }                                                                   \
 223         else                                                                  \
 224           {                                                                   \
 225             /* This is illegal.  */                                           \
 226             STANDARD_FROM_LOOP_ERR_HANDLER (1);                               \
 227           }                                                                   \
 228       }                                                                       \
 229                                                                               \
 230     put32 (outptr, ch);                                                       \
 231     outptr += 4;                                                              \
 232   }
 233 #define LOOP_NEED_FLAGS
 234 #define EXTRA_LOOP_DECLS        , int *statep
 235 #define ONEBYTE_BODY \
 236   {                                                                           \
 237     if (c < 0x80)                                                             \
 238       {                                                                       \
 239         if (c == 0x5c)                                                        \
 240           return 0xa5;                                                        \
 241         if (c == 0x7e)                                                        \
 242           return 0x203e;                                                      \
 243         return c;                                                             \
 244       }                                                                       \
 245     if (c >= 0xa1 && c <= 0xdf)                                               \
 246       return 0xfec0 + c;                                                      \
 247     return WEOF;                                                              \
 248   }
 249 #include <iconv/loop.c>
 250
 251
 252 /* Next, define the other direction, from UCS-4 to Shift_JISX0213.  */
 253
 254 /* Composition tables for each of the relevant combining characters.  */
 255 static const struct
 256 {
 257   uint16_t base;
 258   uint16_t composed;
 259 } comp_table_data[] =
 260 {
 261 #define COMP_TABLE_IDX_02E5 0
 262 #define COMP_TABLE_LEN_02E5 1
 263   { 0x8684, 0x8685 }, /* 0x12B65 = 0x12B64 U+02E5 */
 264 #define COMP_TABLE_IDX_02E9 (COMP_TABLE_IDX_02E5 + COMP_TABLE_LEN_02E5)
 265 #define COMP_TABLE_LEN_02E9 1
 266   { 0x8680, 0x8686 }, /* 0x12B66 = 0x12B60 U+02E9 */
 267 #define COMP_TABLE_IDX_0300 (COMP_TABLE_IDX_02E9 + COMP_TABLE_LEN_02E9)
 268 #define COMP_TABLE_LEN_0300 5
 269   { 0x857b, 0x8663 }, /* 0x12B44 = 0x1295C U+0300 */
 270   { 0x8657, 0x8667 }, /* 0x12B48 = 0x12B38 U+0300 */
 271   { 0x8656, 0x8669 }, /* 0x12B4A = 0x12B37 U+0300 */
 272   { 0x864f, 0x866b }, /* 0x12B4C = 0x12B30 U+0300 */
 273   { 0x8662, 0x866d }, /* 0x12B4E = 0x12B43 U+0300 */
 274 #define COMP_TABLE_IDX_0301 (COMP_TABLE_IDX_0300 + COMP_TABLE_LEN_0300)
 275 #define COMP_TABLE_LEN_0301 4
 276   { 0x8657, 0x8668 }, /* 0x12B49 = 0x12B38 U+0301 */
 277   { 0x8656, 0x866a }, /* 0x12B4B = 0x12B37 U+0301 */
 278   { 0x864f, 0x866c }, /* 0x12B4D = 0x12B30 U+0301 */
 279   { 0x8662, 0x866e }, /* 0x12B4F = 0x12B43 U+0301 */
 280 #define COMP_TABLE_IDX_309A (COMP_TABLE_IDX_0301 + COMP_TABLE_LEN_0301)
 281 #define COMP_TABLE_LEN_309A 14
 282   { 0x82a9, 0x82f5 }, /* 0x12477 = 0x1242B U+309A */
 283   { 0x82ab, 0x82f6 }, /* 0x12478 = 0x1242D U+309A */
 284   { 0x82ad, 0x82f7 }, /* 0x12479 = 0x1242F U+309A */
 285   { 0x82af, 0x82f8 }, /* 0x1247A = 0x12431 U+309A */
 286   { 0x82b1, 0x82f9 }, /* 0x1247B = 0x12433 U+309A */
 287   { 0x834a, 0x8397 }, /* 0x12577 = 0x1252B U+309A */
 288   { 0x834c, 0x8398 }, /* 0x12578 = 0x1252D U+309A */
 289   { 0x834e, 0x8399 }, /* 0x12579 = 0x1252F U+309A */
 290   { 0x8350, 0x839a }, /* 0x1257A = 0x12531 U+309A */
 291   { 0x8352, 0x839b }, /* 0x1257B = 0x12533 U+309A */
 292   { 0x835a, 0x839c }, /* 0x1257C = 0x1253B U+309A */
 293   { 0x8363, 0x839d }, /* 0x1257D = 0x12544 U+309A */
 294   { 0x8367, 0x839e }, /* 0x1257E = 0x12548 U+309A */
 295   { 0x83f3, 0x83f6 }, /* 0x12678 = 0x12675 U+309A */
 296 };
 297
 298 #define MIN_NEEDED_INPUT        TO_LOOP_MIN_NEEDED_FROM
 299 #define MAX_NEEDED_INPUT        TO_LOOP_MAX_NEEDED_FROM
 300 #define MIN_NEEDED_OUTPUT       TO_LOOP_MIN_NEEDED_TO
 301 #define MAX_NEEDED_OUTPUT       TO_LOOP_MAX_NEEDED_TO
 302 #define LOOPFCT                 TO_LOOP
 303 #define BODY \
 304   {                                                                           \
 305     uint32_t ch = get32 (inptr);                                              \
 306                                                                               \
 307     if ((*statep >> 3) != 0)                                                  \
 308       {                                                                       \
 309         /* Attempt to combine the last character with this one.  */           \
 310         uint16_t lasttwo = *statep >> 3;                                      \
 311         unsigned int idx;                                                     \
 312         unsigned int len;                                                     \
 313                                                                               \
 314         if (ch == 0x02e5)                                                     \
 315           idx = COMP_TABLE_IDX_02E5, len = COMP_TABLE_LEN_02E5;               \
 316         else if (ch == 0x02e9)                                                \
 317           idx = COMP_TABLE_IDX_02E9, len = COMP_TABLE_LEN_02E9;               \
 318         else if (ch == 0x0300)                                                \
 319           idx = COMP_TABLE_IDX_0300, len = COMP_TABLE_LEN_0300;               \
 320         else if (ch == 0x0301)                                                \
 321           idx = COMP_TABLE_IDX_0301, len = COMP_TABLE_LEN_0301;               \
 322         else if (ch == 0x309a)                                                \
 323           idx = COMP_TABLE_IDX_309A, len = COMP_TABLE_LEN_309A;               \
 324         else                                                                  \
 325           goto not_combining;                                                 \
 326                                                                               \
 327         do                                                                    \
 328           if (comp_table_data[idx].base == lasttwo)                           \
 329             break;                                                            \
 330         while (++idx, --len > 0);                                             \
 331                                                                               \
 332         if (len > 0)                                                          \
 333           {                                                                   \
 334             /* Output the combined character.  */                             \
 335             if (__builtin_expect (outptr + 1 >= outend, 0))                   \
 336               {                                                               \
 337                 result = __GCONV_FULL_OUTPUT;                                 \
 338                 break;                                                        \
 339               }                                                               \
 340             lasttwo = comp_table_data[idx].composed;                          \
 341             *outptr++ = (lasttwo >> 8) & 0xff;                                \
 342             *outptr++ = lasttwo & 0xff;                                       \
 343             *statep = 0;                                                      \
 344             inptr += 4;                                                       \
 345             continue;                                                         \
 346           }                                                                   \
 347                                                                               \
 348       not_combining:                                                          \
 349         /* Output the buffered character.  */                                 \
 350         if (__builtin_expect (outptr + 1 >= outend, 0))                       \
 351           {                                                                   \
 352             result = __GCONV_FULL_OUTPUT;                                     \
 353             break;                                                            \
 354           }                                                                   \
 355         *outptr++ = (lasttwo >> 8) & 0xff;                                    \
 356         *outptr++ = lasttwo & 0xff;                                           \
 357         *statep = 0;                                                          \
 358         continue;                                                             \
 359       }                                                                       \
 360                                                                               \
 361     if (ch < 0x80)                                                            \
 362       /* Plain ISO646-JP character.  */                                       \
 363       *outptr++ = ch;                                                         \
 364     else if (ch == 0xa5)                                                      \
 365       *outptr++ = 0x5c;                                                       \
 366     else if (ch == 0x203e)                                                    \
 367       *outptr++ = 0x7e;                                                       \
 368     else if (ch >= 0xff61 && ch <= 0xff9f)                                    \
 369       /* Half-width katakana.  */                                             \
 370       *outptr++ = ch - 0xfec0;                                                \
 371     else                                                                      \
 372       {                                                                       \
 373         unsigned int s1, s2;                                                  \
 374         uint32_t jch = ucs4_to_jisx0213 (ch);                                 \
 375         if (jch == 0)                                                         \
 376           {                                                                   \
 377             UNICODE_TAG_HANDLER (ch, 4);                                      \
 378                                                                               \
 379             /* Illegal character.  */                                         \
 380             STANDARD_TO_LOOP_ERR_HANDLER (4);                                 \
 381           }                                                                   \
 382                                                                               \
 383         /* Convert it to shifted representation.  */                          \
 384         s1 = jch >> 8;                                                        \
 385         s2 = jch & 0x7f;                                                              \
 386         s1 -= 0x21;                                                           \
 387         s2 -= 0x21;                                                           \
 388         if (s1 >= 0x5e)                                                       \
 389           {                                                                   \
 390             /* Handling of JISX 0213 plane 2 rows.  */                        \
 391             if (s1 >= 0xcd) /* rows 0x26E..0x27E */                           \
 392               s1 -= 102;                                                      \
 393             else if (s1 >= 0x8b || s1 == 0x87) /* rows 0x228, 0x22C..0x22F */ \
 394               s1 -= 40;                                                       \
 395             else /* rows 0x221, 0x223..0x225 */                               \
 396               s1 -= 34;                                                       \
 397             /* Now 0x5e <= s1 <= 0x77.  */                                    \
 398           }                                                                   \
 399         if (s1 & 1)                                                           \
 400           s2 += 0x5e;                                                         \
 401         s1 = s1 >> 1;                                                         \
 402         if (s1 < 0x1f)                                                        \
 403           s1 += 0x81;                                                         \
 404         else                                                                  \
 405           s1 += 0xc1;                                                         \
 406         if (s2 < 0x3f)                                                        \
 407           s2 += 0x40;                                                         \
 408         else                                                                  \
 409           s2 += 0x41;                                                         \
 410                                                                               \
 411         if (jch & 0x0080)                                                     \
 412           {                                                                   \
 413             /* A possible match in comp_table_data.  We have to buffer it.  */\
 414                                                                               \
 415             /* We know it's a JISX 0213 plane 1 character.  */                \
 416             assert ((jch & 0x8000) == 0);                                     \
 417                                                                               \
 418             *statep = ((s1 << 8) | s2) << 3;                                  \
 419             inptr += 4;                                                       \
 420             continue;                                                         \
 421           }                                                                   \
 422                                                                               \
 423         /* Output the shifted representation.  */                             \
 424         if (__builtin_expect (outptr + 1 >= outend, 0))                       \
 425           {                                                                   \
 426             result = __GCONV_FULL_OUTPUT;                                     \
 427             break;                                                            \
 428           }                                                                   \
 429         *outptr++ = s1;                                                       \
 430         *outptr++ = s2;                                                       \
 431       }                                                                       \
 432                                                                               \
 433     inptr += 4;                                                               \
 434   }
 435 #define LOOP_NEED_FLAGS
 436 #define EXTRA_LOOP_DECLS        , int *statep
 437 #include <iconv/loop.c>
 438
 439
 440 /* Now define the toplevel functions.  */
 441 #include <iconv/skeleton.c>