iconvdata/cp1258.c

   1 /* Conversion from and to CP1258.
   2    Copyright (C) 1998, 2001 Free Software Foundation, Inc.
   3    This file is part of the GNU C Library.
   4    Contributed by Ulrich Drepper <drepper@cygnus.com>, 1998,
   5    and Bruno Haible <haible@clisp.cons.org>, 2001.
   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 <stdint.h>
  24 #include <assert.h>
  25 #include <stdlib.h>
  26
  27 #define NELEMS(arr) (sizeof (arr) / sizeof (arr[0]))
  28
  29 /* Definitions used in the body of the `gconv' function.  */
  30 #define CHARSET_NAME            "CP1258//"
  31 #define FROM_LOOP               from_cp1258
  32 #define TO_LOOP                 to_cp1258
  33 #define DEFINE_INIT             1
  34 #define DEFINE_FINI             1
  35 #define MIN_NEEDED_FROM         1
  36 #define MAX_NEEDED_FROM         1
  37 #define MIN_NEEDED_TO           4
  38 #define MAX_NEEDED_TO           4
  39 #define PREPARE_LOOP \
  40   int saved_state;                                                            \
  41   int *statep = &data->__statep->__count;
  42 #define EXTRA_LOOP_ARGS         , statep
  43
  44
  45 /* Since we might have to reset input pointer we must be able to save
  46    and restore the state.  */
  47 #define SAVE_RESET_STATE(Save) \
  48   if (Save)                                                                   \
  49     saved_state = *statep;                                                    \
  50   else                                                                        \
  51     *statep = saved_state
  52
  53
  54 /* During CP1258 to UCS4 conversion, the COUNT element of the state
  55    contains the last UCS4 character, shifted by 3 bits.  */
  56
  57
  58 /* Since this is a stateful encoding we have to provide code which resets
  59    the output state to the initial state.  This has to be done during the
  60    flushing.  */
  61 #define EMIT_SHIFT_TO_INIT \
  62   if (data->__statep->__count != 0)                                           \
  63     {                                                                         \
  64       if (FROM_DIRECTION)                                                     \
  65         {                                                                     \
  66           if (__builtin_expect (outbuf + 4 <= outend, 1))                     \
  67             {                                                                 \
  68               /* Write out the last character.  */                            \
  69               *((uint32_t *) outbuf)++ = data->__statep->__count >> 3;        \
  70               data->__statep->__count = 0;                                    \
  71             }                                                                 \
  72           else                                                                \
  73             /* We don't have enough room in the output buffer.  */            \
  74             status = __GCONV_FULL_OUTPUT;                                     \
  75         }                                                                     \
  76       else                                                                    \
  77         /* We don't use shift states in the TO_DIRECTION.  */                 \
  78         data->__statep->__count = 0;                                          \
  79     }
  80
  81
  82 /* First define the conversion function from CP1258 to UCS4.  */
  83
  84 static const uint16_t to_ucs4[128] = {
  85   /* 0x80 */
  86   0x20AC,      0, 0x201A, 0x0192, 0x201E, 0x2026, 0x2020, 0x2021,
  87   0x02C6, 0x2030,      0, 0x2039, 0x0152,      0,      0,      0,
  88   /* 0x90 */
  89        0, 0x2018, 0x2019, 0x201C, 0x201D, 0x2022, 0x2013, 0x2014,
  90   0x02DC, 0x2122,      0, 0x203A, 0x0153,      0,      0, 0x0178,
  91   /* 0xA0 */
  92   0x00A0, 0x00A1, 0x00A2, 0x00A3, 0x00A4, 0x00A5, 0x00A6, 0x00A7,
  93   0x00A8, 0x00A9, 0x00AA, 0x00AB, 0x00AC, 0x00AD, 0x00AE, 0x00AF,
  94   /* 0xB0 */
  95   0x00B0, 0x00B1, 0x00B2, 0x00B3, 0x00B4, 0x00B5, 0x00B6, 0x00B7,
  96   0x00B8, 0x00B9, 0x00BA, 0x00BB, 0x00BC, 0x00BD, 0x00BE, 0x00BF,
  97   /* 0xC0 */
  98   0x00C0, 0x00C1, 0x00C2, 0x0102, 0x00C4, 0x00C5, 0x00C6, 0x00C7,
  99   0x00C8, 0x00C9, 0x00CA, 0x00CB, 0x0300, 0x00CD, 0x00CE, 0x00CF,
 100   /* 0xD0 */
 101   0x0110, 0x00D1, 0x0309, 0x00D3, 0x00D4, 0x01A0, 0x00D6, 0x00D7,
 102   0x00D8, 0x00D9, 0x00DA, 0x00DB, 0x00DC, 0x01AF, 0x0303, 0x00DF,
 103   /* 0xE0 */
 104   0x00E0, 0x00E1, 0x00E2, 0x0103, 0x00E4, 0x00E5, 0x00E6, 0x00E7,
 105   0x00E8, 0x00E9, 0x00EA, 0x00EB, 0x0301, 0x00ED, 0x00EE, 0x00EF,
 106   /* 0xF0 */
 107   0x0111, 0x00F1, 0x0323, 0x00F3, 0x00F4, 0x01A1, 0x00F6, 0x00F7,
 108   0x00F8, 0x00F9, 0x00FA, 0x00FB, 0x00FC, 0x01B0, 0x20AB, 0x00FF,
 109 };
 110
 111 /* CP1258 contains five combining characters:
 112    0x0300, 0x0301, 0x0303, 0x0309, 0x0323.  */
 113
 114 /* Composition tables for each of the relevant combining characters.  */
 115 static const struct {
 116   uint16_t base;
 117   uint16_t composed;
 118 } comp_table_data[] = {
 119 #define COMP_TABLE_IDX_0300 0
 120 #define COMP_TABLE_LEN_0300 31
 121   { 0x0041, 0x00C0 },
 122   { 0x0045, 0x00C8 },
 123   { 0x0049, 0x00CC },
 124   { 0x004E, 0x01F8 },
 125   { 0x004F, 0x00D2 },
 126   { 0x0055, 0x00D9 },
 127   { 0x0057, 0x1E80 },
 128   { 0x0059, 0x1EF2 },
 129   { 0x0061, 0x00E0 },
 130   { 0x0065, 0x00E8 },
 131   { 0x0069, 0x00EC },
 132   { 0x006E, 0x01F9 },
 133   { 0x006F, 0x00F2 },
 134   { 0x0075, 0x00F9 },
 135   { 0x0077, 0x1E81 },
 136   { 0x0079, 0x1EF3 },
 137   { 0x00A8, 0x1FED },
 138   { 0x00C2, 0x1EA6 },
 139   { 0x00CA, 0x1EC0 },
 140   { 0x00D4, 0x1ED2 },
 141   { 0x00DC, 0x01DB },
 142   { 0x00E2, 0x1EA7 },
 143   { 0x00EA, 0x1EC1 },
 144   { 0x00F4, 0x1ED3 },
 145   { 0x00FC, 0x01DC },
 146   { 0x0102, 0x1EB0 },
 147   { 0x0103, 0x1EB1 },
 148 /*{ 0x0112, 0x1E14 },*/
 149 /*{ 0x0113, 0x1E15 },*/
 150 /*{ 0x014C, 0x1E50 },*/
 151 /*{ 0x014D, 0x1E51 },*/
 152   { 0x01A0, 0x1EDC },
 153   { 0x01A1, 0x1EDD },
 154   { 0x01AF, 0x1EEA },
 155   { 0x01B0, 0x1EEB },
 156 #define COMP_TABLE_IDX_0301 (COMP_TABLE_IDX_0300 + COMP_TABLE_LEN_0300)
 157 #define COMP_TABLE_LEN_0301 60
 158   { 0x0041, 0x00C1 },
 159   { 0x0043, 0x0106 },
 160   { 0x0045, 0x00C9 },
 161   { 0x0047, 0x01F4 },
 162   { 0x0049, 0x00CD },
 163   { 0x004B, 0x1E30 },
 164   { 0x004C, 0x0139 },
 165   { 0x004D, 0x1E3E },
 166   { 0x004E, 0x0143 },
 167   { 0x004F, 0x00D3 },
 168   { 0x0050, 0x1E54 },
 169   { 0x0052, 0x0154 },
 170   { 0x0053, 0x015A },
 171   { 0x0055, 0x00DA },
 172   { 0x0057, 0x1E82 },
 173   { 0x0059, 0x00DD },
 174   { 0x005A, 0x0179 },
 175   { 0x0061, 0x00E1 },
 176   { 0x0063, 0x0107 },
 177   { 0x0065, 0x00E9 },
 178   { 0x0067, 0x01F5 },
 179   { 0x0069, 0x00ED },
 180   { 0x006B, 0x1E31 },
 181   { 0x006C, 0x013A },
 182   { 0x006D, 0x1E3F },
 183   { 0x006E, 0x0144 },
 184   { 0x006F, 0x00F3 },
 185   { 0x0070, 0x1E55 },
 186   { 0x0072, 0x0155 },
 187   { 0x0073, 0x015B },
 188   { 0x0075, 0x00FA },
 189   { 0x0077, 0x1E83 },
 190   { 0x0079, 0x00FD },
 191   { 0x007A, 0x017A },
 192   { 0x00A5, 0x0385 },
 193   { 0x00A8, 0x1FEE },
 194   { 0x00C2, 0x1EA4 },
 195   { 0x00C5, 0x01FA },
 196   { 0x00C6, 0x01FC },
 197   { 0x00C7, 0x1E08 },
 198   { 0x00CA, 0x1EBE },
 199   { 0x00CF, 0x1E2E },
 200   { 0x00D4, 0x1ED0 },
 201 /*{ 0x00D5, 0x1E4C },*/
 202   { 0x00D8, 0x01FE },
 203   { 0x00DC, 0x01D7 },
 204   { 0x00E2, 0x1EA5 },
 205   { 0x00E5, 0x01FB },
 206   { 0x00E6, 0x01FD },
 207   { 0x00E7, 0x1E09 },
 208   { 0x00EA, 0x1EBF },
 209   { 0x00EF, 0x1E2F },
 210   { 0x00F4, 0x1ED1 },
 211 /*{ 0x00F5, 0x1E4D },*/
 212   { 0x00F8, 0x01FF },
 213   { 0x00FC, 0x01D8 },
 214   { 0x0102, 0x1EAE },
 215   { 0x0103, 0x1EAF },
 216 /*{ 0x0112, 0x1E16 },*/
 217 /*{ 0x0113, 0x1E17 },*/
 218 /*{ 0x014C, 0x1E52 },*/
 219 /*{ 0x014D, 0x1E53 },*/
 220 /*{ 0x0168, 0x1E78 },*/
 221 /*{ 0x0169, 0x1E79 },*/
 222   { 0x01A0, 0x1EDA },
 223   { 0x01A1, 0x1EDB },
 224   { 0x01AF, 0x1EE8 },
 225   { 0x01B0, 0x1EE9 },
 226 #define COMP_TABLE_IDX_0303 (COMP_TABLE_IDX_0301 + COMP_TABLE_LEN_0301)
 227 #define COMP_TABLE_LEN_0303 34
 228   { 0x0041, 0x00C3 },
 229   { 0x0045, 0x1EBC },
 230   { 0x0049, 0x0128 },
 231   { 0x004E, 0x00D1 },
 232   { 0x004F, 0x00D5 },
 233   { 0x0055, 0x0168 },
 234   { 0x0056, 0x1E7C },
 235   { 0x0059, 0x1EF8 },
 236   { 0x0061, 0x00E3 },
 237   { 0x0065, 0x1EBD },
 238   { 0x0069, 0x0129 },
 239   { 0x006E, 0x00F1 },
 240   { 0x006F, 0x00F5 },
 241   { 0x0075, 0x0169 },
 242   { 0x0076, 0x1E7D },
 243   { 0x0079, 0x1EF9 },
 244   { 0x00C2, 0x1EAA },
 245   { 0x00CA, 0x1EC4 },
 246   { 0x00D3, 0x1E4C },
 247   { 0x00D4, 0x1ED6 },
 248   { 0x00D6, 0x1E4E },
 249   { 0x00DA, 0x1E78 },
 250   { 0x00E2, 0x1EAB },
 251   { 0x00EA, 0x1EC5 },
 252   { 0x00F3, 0x1E4D },
 253   { 0x00F4, 0x1ED7 },
 254   { 0x00F6, 0x1E4F },
 255   { 0x00FA, 0x1E79 },
 256   { 0x0102, 0x1EB4 },
 257   { 0x0103, 0x1EB5 },
 258   { 0x01A0, 0x1EE0 },
 259   { 0x01A1, 0x1EE1 },
 260   { 0x01AF, 0x1EEE },
 261   { 0x01B0, 0x1EEF },
 262 #define COMP_TABLE_IDX_0309 (COMP_TABLE_IDX_0303 + COMP_TABLE_LEN_0303)
 263 #define COMP_TABLE_LEN_0309 24
 264   { 0x0041, 0x1EA2 },
 265   { 0x0045, 0x1EBA },
 266   { 0x0049, 0x1EC8 },
 267   { 0x004F, 0x1ECE },
 268   { 0x0055, 0x1EE6 },
 269   { 0x0059, 0x1EF6 },
 270   { 0x0061, 0x1EA3 },
 271   { 0x0065, 0x1EBB },
 272   { 0x0069, 0x1EC9 },
 273   { 0x006F, 0x1ECF },
 274   { 0x0075, 0x1EE7 },
 275   { 0x0079, 0x1EF7 },
 276   { 0x00C2, 0x1EA8 },
 277   { 0x00CA, 0x1EC2 },
 278   { 0x00D4, 0x1ED4 },
 279   { 0x00E2, 0x1EA9 },
 280   { 0x00EA, 0x1EC3 },
 281   { 0x00F4, 0x1ED5 },
 282   { 0x0102, 0x1EB2 },
 283   { 0x0103, 0x1EB3 },
 284   { 0x01A0, 0x1EDE },
 285   { 0x01A1, 0x1EDF },
 286   { 0x01AF, 0x1EEC },
 287   { 0x01B0, 0x1EED },
 288 #define COMP_TABLE_IDX_0323 (COMP_TABLE_IDX_0309 + COMP_TABLE_LEN_0309)
 289 #define COMP_TABLE_LEN_0323 50
 290   { 0x0041, 0x1EA0 },
 291   { 0x0042, 0x1E04 },
 292   { 0x0044, 0x1E0C },
 293   { 0x0045, 0x1EB8 },
 294   { 0x0048, 0x1E24 },
 295   { 0x0049, 0x1ECA },
 296   { 0x004B, 0x1E32 },
 297   { 0x004C, 0x1E36 },
 298   { 0x004D, 0x1E42 },
 299   { 0x004E, 0x1E46 },
 300   { 0x004F, 0x1ECC },
 301   { 0x0052, 0x1E5A },
 302   { 0x0053, 0x1E62 },
 303   { 0x0054, 0x1E6C },
 304   { 0x0055, 0x1EE4 },
 305   { 0x0056, 0x1E7E },
 306   { 0x0057, 0x1E88 },
 307   { 0x0059, 0x1EF4 },
 308   { 0x005A, 0x1E92 },
 309   { 0x0061, 0x1EA1 },
 310   { 0x0062, 0x1E05 },
 311   { 0x0064, 0x1E0D },
 312   { 0x0065, 0x1EB9 },
 313   { 0x0068, 0x1E25 },
 314   { 0x0069, 0x1ECB },
 315   { 0x006B, 0x1E33 },
 316   { 0x006C, 0x1E37 },
 317   { 0x006D, 0x1E43 },
 318   { 0x006E, 0x1E47 },
 319   { 0x006F, 0x1ECD },
 320   { 0x0072, 0x1E5B },
 321   { 0x0073, 0x1E63 },
 322   { 0x0074, 0x1E6D },
 323   { 0x0075, 0x1EE5 },
 324   { 0x0076, 0x1E7F },
 325   { 0x0077, 0x1E89 },
 326   { 0x0079, 0x1EF5 },
 327   { 0x007A, 0x1E93 },
 328   { 0x00C2, 0x1EAC },
 329   { 0x00CA, 0x1EC6 },
 330   { 0x00D4, 0x1ED8 },
 331   { 0x00E2, 0x1EAD },
 332   { 0x00EA, 0x1EC7 },
 333   { 0x00F4, 0x1ED9 },
 334   { 0x0102, 0x1EB6 },
 335   { 0x0103, 0x1EB7 },
 336   { 0x01A0, 0x1EE2 },
 337   { 0x01A1, 0x1EE3 },
 338   { 0x01AF, 0x1EF0 },
 339   { 0x01B0, 0x1EF1 },
 340 #define COMP_TABLE_IDX_END (COMP_TABLE_IDX_0323 + COMP_TABLE_LEN_0323)
 341 };
 342 /* Compile-time verification of table size.  */
 343 typedef int verify1[(NELEMS (comp_table_data) == COMP_TABLE_IDX_END) - 1];
 344
 345 static const struct { unsigned int idx; unsigned int len; } comp_table[5] = {
 346   { COMP_TABLE_IDX_0300, COMP_TABLE_LEN_0300 },
 347   { COMP_TABLE_IDX_0301, COMP_TABLE_LEN_0301 },
 348   { COMP_TABLE_IDX_0303, COMP_TABLE_LEN_0303 },
 349   { COMP_TABLE_IDX_0309, COMP_TABLE_LEN_0309 },
 350   { COMP_TABLE_IDX_0323, COMP_TABLE_LEN_0323 },
 351 };
 352
 353 #define MIN_NEEDED_INPUT        MIN_NEEDED_FROM
 354 #define MIN_NEEDED_OUTPUT       MIN_NEEDED_TO
 355 #define MAX_NEEDED_OUTPUT       8
 356 #define LOOPFCT                 FROM_LOOP
 357 #define BODY \
 358   {                                                                           \
 359     uint32_t ch = *inptr;                                                     \
 360     uint32_t last_ch;                                                         \
 361     int must_buffer_ch;                                                       \
 362                                                                               \
 363     if (ch >= 0x80)                                                           \
 364       {                                                                       \
 365         ch = to_ucs4[ch - 0x80];                                              \
 366         if (__builtin_expect (ch == L'\0', 0))                                \
 367           {                                                                   \
 368             /* This is an illegal character.  */                              \
 369             if (! ignore_errors_p ())                                         \
 370               {                                                               \
 371                 result = __GCONV_ILLEGAL_INPUT;                               \
 372                 break;                                                        \
 373               }                                                               \
 374                                                                               \
 375             ++inptr;                                                          \
 376             ++*irreversible;                                                  \
 377             continue;                                                         \
 378           }                                                                   \
 379       }                                                                       \
 380                                                                               \
 381     /* Determine whether there is a buffered character pending.  */           \
 382     last_ch = *statep >> 3;                                                   \
 383                                                                               \
 384     /* We have to buffer ch if it is a possible match in comp_table_data.  */ \
 385     must_buffer_ch = (ch >= 0x0041 && ch <= 0x01b0);                          \
 386                                                                               \
 387     if (last_ch)                                                              \
 388       {                                                                       \
 389         if (ch >= 0x0300 && ch < 0x0340)                                      \
 390           {                                                                   \
 391             /* See whether last_ch and ch can be combined.  */                \
 392             unsigned int i, i1, i2;                                           \
 393                                                                               \
 394             switch (ch)                                                       \
 395               {                                                               \
 396               case 0x0300:                                                    \
 397                 i = 0;                                                        \
 398                 break;                                                        \
 399               case 0x0301:                                                    \
 400                 i = 1;                                                        \
 401                 break;                                                        \
 402               case 0x0303:                                                    \
 403                 i = 2;                                                        \
 404                 break;                                                        \
 405               case 0x0309:                                                    \
 406                 i = 3;                                                        \
 407                 break;                                                        \
 408               case 0x0323:                                                    \
 409                 i = 4;                                                        \
 410                 break;                                                        \
 411               default:                                                        \
 412                 abort ();                                                     \
 413               }                                                               \
 414                                                                               \
 415             i1 = comp_table[i].idx;                                           \
 416             i2 = i1 + comp_table[i].len - 1;                                  \
 417                                                                               \
 418             if (last_ch >= comp_table_data[i1].base                           \
 419                 && last_ch <= comp_table_data[i2].base)                       \
 420               {                                                               \
 421                 for (;;)                                                      \
 422                   {                                                           \
 423                     i = (i1 + i2) >> 1;                                       \
 424                     if (last_ch == comp_table_data[i].base)                   \
 425                       break;                                                  \
 426                     if (last_ch < comp_table_data[i].base)                    \
 427                       {                                                       \
 428                         if (i1 == i)                                          \
 429                           goto not_combining;                                 \
 430                         i2 = i;                                               \
 431                       }                                                       \
 432                     else                                                      \
 433                       {                                                       \
 434                         if (i1 != i)                                          \
 435                           i1 = i;                                             \
 436                         else                                                  \
 437                           {                                                   \
 438                             i = i2;                                           \
 439                             if (last_ch == comp_table_data[i].base)           \
 440                               break;                                          \
 441                             goto not_combining;                               \
 442                           }                                                   \
 443                       }                                                       \
 444                   }                                                           \
 445                 last_ch = comp_table_data[i].composed;                        \
 446                 /* Output the combined character.  */                         \
 447                 put32 (outptr, last_ch);                                      \
 448                 outptr += 4;                                                  \
 449                 *statep = 0;                                                  \
 450                 ++inptr;                                                      \
 451                 continue;                                                     \
 452               }                                                               \
 453           }                                                                   \
 454                                                                               \
 455       not_combining:                                                          \
 456         /* Output the buffered character.  */                                 \
 457         put32 (outptr, last_ch);                                              \
 458         outptr += 4;                                                          \
 459         *statep = 0;                                                          \
 460                                                                               \
 461         /* If we don't have enough room to output ch as well, then deal       \
 462            with it in another round.  */                                      \
 463         if (!must_buffer_ch && __builtin_expect (outptr + 4 > outend, 0))     \
 464           continue;                                                           \
 465       }                                                                       \
 466                                                                               \
 467     if (must_buffer_ch)                                                       \
 468       *statep = ch << 3;                                                      \
 469     else                                                                      \
 470       {                                                                       \
 471         put32 (outptr, ch);                                                   \
 472         outptr += 4;                                                          \
 473       }                                                                       \
 474     ++inptr;                                                                  \
 475   }
 476 #define LOOP_NEED_FLAGS
 477 #define EXTRA_LOOP_DECLS        , int *statep
 478 #include <iconv/loop.c>
 479
 480
 481 /* Next, define the conversion function from UCS4 to CP1258.  */
 482
 483 static const unsigned char from_ucs4[] = {
 484 #define FROM_IDX_00 0
 485                           0xc4, 0xc5, 0xc6, 0xc7, /* 0x00c4-0x00c7 */
 486   0xc8, 0xc9, 0xca, 0xcb, 0x00, 0xcd, 0xce, 0xcf, /* 0x00c8-0x00cf */
 487   0x00, 0xd1, 0x00, 0xd3, 0xd4, 0x00, 0xd6, 0xd7, /* 0x00d0-0x00d7 */
 488   0xd8, 0xd9, 0xda, 0xdb, 0xdc, 0x00, 0x00, 0xdf, /* 0x00d8-0x00df */
 489   0xe0, 0xe1, 0xe2, 0x00, 0xe4, 0xe5, 0xe6, 0xe7, /* 0x00e0-0x00e7 */
 490   0xe8, 0xe9, 0xea, 0xeb, 0x00, 0xed, 0xee, 0xef, /* 0x00e8-0x00ef */
 491   0x00, 0xf1, 0x00, 0xf3, 0xf4, 0x00, 0xf6, 0xf7, /* 0x00f0-0x00f7 */
 492   0xf8, 0xf9, 0xfa, 0xfb, 0xfc, 0x00, 0x00, 0xff, /* 0x00f8-0x00ff */
 493   0x00, 0x00, 0xc3, 0xe3, 0x00, 0x00, 0x00, 0x00, /* 0x0100-0x0107 */
 494   0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x0108-0x010f */
 495   0xd0, 0xf0,                                     /* 0x0110-0x0111 */
 496 #define FROM_IDX_01 (FROM_IDX_00 + 78)
 497               0x8c, 0x9c, 0x00, 0x00, 0x00, 0x00, /* 0x0152-0x0157 */
 498   0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x0158-0x015f */
 499   0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x0160-0x0167 */
 500   0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x0168-0x016f */
 501   0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x0170-0x0177 */
 502   0x9f, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x0178-0x017f */
 503   0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x0180-0x0187 */
 504   0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x0188-0x018f */
 505   0x00, 0x00, 0x83, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x0190-0x0197 */
 506   0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x0198-0x019f */
 507   0xd5, 0xf5, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x01a0-0x01a7 */
 508   0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xdd, /* 0x01a8-0x01af */
 509   0xfd,                                           /* 0x01b0-0x01b0 */
 510 #define FROM_IDX_02 (FROM_IDX_01 + 95)
 511                                       0x88, 0x00, /* 0x02c6-0x02c7 */
 512   0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x02c8-0x02cf */
 513   0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x02d0-0x02d7 */
 514   0x00, 0x00, 0x00, 0x00, 0x98,                   /* 0x02d8-0x02dc */
 515 #define FROM_IDX_03 (FROM_IDX_02 + 23)
 516   0xcc, 0xec, 0x00, 0xde, 0x00, 0x00, 0x00, 0x00, /* 0x0300-0x0307 */
 517   0x00, 0xd2, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x0308-0x030f */
 518   0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x0310-0x0317 */
 519   0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x0318-0x031f */
 520   0x00, 0x00, 0x00, 0xf2,                         /* 0x0320-0x0323 */
 521 #define FROM_IDX_20 (FROM_IDX_03 + 36)
 522                     0x96, 0x97, 0x00, 0x00, 0x00, /* 0x2013-0x2017 */
 523   0x91, 0x92, 0x82, 0x00, 0x93, 0x94, 0x84, 0x00, /* 0x2018-0x201f */
 524   0x86, 0x87, 0x95, 0x00, 0x00, 0x00, 0x85, 0x00, /* 0x2020-0x2027 */
 525   0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x2028-0x202f */
 526   0x89, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x2030-0x2037 */
 527   0x00, 0x8b, 0x9b,                               /* 0x2038-0x203a */
 528 #define FROM_IDX_FF (FROM_IDX_20 + 40)
 529 };
 530 /* Compile-time verification of table size.  */
 531 typedef int verify2[(NELEMS (from_ucs4) == FROM_IDX_FF) - 1];
 532
 533 static const unsigned char comb_table[5] = {
 534   0xcc, 0xec, 0xde, 0xd2, 0xf2,
 535 };
 536
 537 /* Decomposition table for the relevant Unicode characters. */
 538 static const struct {
 539   uint16_t composed;
 540   uint32_t base : 12;
 541   uint32_t comb1 : 4;
 542 } decomp_table[] = {
 543   { 0x00c0, 0x0041, 0 },
 544   { 0x00c1, 0x0041, 1 },
 545   { 0x00c3, 0x0041, 2 },
 546   { 0x00c8, 0x0045, 0 },
 547   { 0x00c9, 0x0045, 1 },
 548   { 0x00cc, 0x0049, 0 },
 549   { 0x00cd, 0x0049, 1 },
 550   { 0x00d1, 0x004e, 2 },
 551   { 0x00d2, 0x004f, 0 },
 552   { 0x00d3, 0x004f, 1 },
 553   { 0x00d5, 0x004f, 2 },
 554   { 0x00d9, 0x0055, 0 },
 555   { 0x00da, 0x0055, 1 },
 556   { 0x00dd, 0x0059, 1 },
 557   { 0x00e0, 0x0061, 0 },
 558   { 0x00e1, 0x0061, 1 },
 559   { 0x00e3, 0x0061, 2 },
 560   { 0x00e8, 0x0065, 0 },
 561   { 0x00e9, 0x0065, 1 },
 562   { 0x00ec, 0x0069, 0 },
 563   { 0x00ed, 0x0069, 1 },
 564   { 0x00f1, 0x006e, 2 },
 565   { 0x00f2, 0x006f, 0 },
 566   { 0x00f3, 0x006f, 1 },
 567   { 0x00f5, 0x006f, 2 },
 568   { 0x00f9, 0x0075, 0 },
 569   { 0x00fa, 0x0075, 1 },
 570   { 0x00fd, 0x0079, 1 },
 571   { 0x0106, 0x0043, 1 },
 572   { 0x0107, 0x0063, 1 },
 573   { 0x0128, 0x0049, 2 },
 574   { 0x0129, 0x0069, 2 },
 575   { 0x0139, 0x004c, 1 },
 576   { 0x013a, 0x006c, 1 },
 577   { 0x0143, 0x004e, 1 },
 578   { 0x0144, 0x006e, 1 },
 579   { 0x0154, 0x0052, 1 },
 580   { 0x0155, 0x0072, 1 },
 581   { 0x015a, 0x0053, 1 },
 582   { 0x015b, 0x0073, 1 },
 583   { 0x0168, 0x0055, 2 },
 584   { 0x0169, 0x0075, 2 },
 585   { 0x0179, 0x005a, 1 },
 586   { 0x017a, 0x007a, 1 },
 587   { 0x01d7, 0x00dc, 1 },
 588   { 0x01d8, 0x00fc, 1 },
 589   { 0x01db, 0x00dc, 0 },
 590   { 0x01dc, 0x00fc, 0 },
 591   { 0x01f4, 0x0047, 1 },
 592   { 0x01f5, 0x0067, 1 },
 593   { 0x01f8, 0x004e, 0 },
 594   { 0x01f9, 0x006e, 0 },
 595   { 0x01fa, 0x00c5, 1 },
 596   { 0x01fb, 0x00e5, 1 },
 597   { 0x01fc, 0x00c6, 1 },
 598   { 0x01fd, 0x00e6, 1 },
 599   { 0x01fe, 0x00d8, 1 },
 600   { 0x01ff, 0x00f8, 1 },
 601   { 0x0385, 0x00a5, 1 },
 602   { 0x1e04, 0x0042, 4 },
 603   { 0x1e05, 0x0062, 4 },
 604   { 0x1e08, 0x00c7, 1 },
 605   { 0x1e09, 0x00e7, 1 },
 606   { 0x1e0c, 0x0044, 4 },
 607   { 0x1e0d, 0x0064, 4 },
 608   { 0x1e24, 0x0048, 4 },
 609   { 0x1e25, 0x0068, 4 },
 610   { 0x1e2e, 0x00cf, 1 },
 611   { 0x1e2f, 0x00ef, 1 },
 612   { 0x1e30, 0x004b, 1 },
 613   { 0x1e31, 0x006b, 1 },
 614   { 0x1e32, 0x004b, 4 },
 615   { 0x1e33, 0x006b, 4 },
 616   { 0x1e36, 0x004c, 4 },
 617   { 0x1e37, 0x006c, 4 },
 618   { 0x1e3e, 0x004d, 1 },
 619   { 0x1e3f, 0x006d, 1 },
 620   { 0x1e42, 0x004d, 4 },
 621   { 0x1e43, 0x006d, 4 },
 622   { 0x1e46, 0x004e, 4 },
 623   { 0x1e47, 0x006e, 4 },
 624   { 0x1e4c, 0x00d3, 2 }, /*{ 0x1e4c, 0x00d5, 1 }, { 0x1e4c, 0x004f, 1, 2 },*/
 625   { 0x1e4d, 0x00f3, 2 }, /*{ 0x1e4d, 0x00f5, 1 }, { 0x1e4d, 0x006f, 1, 2 },*/
 626   { 0x1e4e, 0x00d6, 2 },
 627   { 0x1e4f, 0x00f6, 2 },
 628   { 0x1e54, 0x0050, 1 },
 629   { 0x1e55, 0x0070, 1 },
 630   { 0x1e5a, 0x0052, 4 },
 631   { 0x1e5b, 0x0072, 4 },
 632   { 0x1e62, 0x0053, 4 },
 633   { 0x1e63, 0x0073, 4 },
 634   { 0x1e6c, 0x0054, 4 },
 635   { 0x1e6d, 0x0074, 4 },
 636   { 0x1e78, 0x00da, 2 }, /*{ 0x1e78, 0x0168, 1 }, { 0x1e78, 0x0055, 1, 2 },*/
 637   { 0x1e79, 0x00fa, 2 }, /*{ 0x1e79, 0x0169, 1 }, { 0x1e79, 0x0075, 1, 2 },*/
 638   { 0x1e7c, 0x0056, 2 },
 639   { 0x1e7d, 0x0076, 2 },
 640   { 0x1e7e, 0x0056, 4 },
 641   { 0x1e7f, 0x0076, 4 },
 642   { 0x1e80, 0x0057, 0 },
 643   { 0x1e81, 0x0077, 0 },
 644   { 0x1e82, 0x0057, 1 },
 645   { 0x1e83, 0x0077, 1 },
 646   { 0x1e88, 0x0057, 4 },
 647   { 0x1e89, 0x0077, 4 },
 648   { 0x1e92, 0x005a, 4 },
 649   { 0x1e93, 0x007a, 4 },
 650   { 0x1ea0, 0x0041, 4 },
 651   { 0x1ea1, 0x0061, 4 },
 652   { 0x1ea2, 0x0041, 3 },
 653   { 0x1ea3, 0x0061, 3 },
 654   { 0x1ea4, 0x00c2, 1 },
 655   { 0x1ea5, 0x00e2, 1 },
 656   { 0x1ea6, 0x00c2, 0 },
 657   { 0x1ea7, 0x00e2, 0 },
 658   { 0x1ea8, 0x00c2, 3 },
 659   { 0x1ea9, 0x00e2, 3 },
 660   { 0x1eaa, 0x00c2, 2 },
 661   { 0x1eab, 0x00e2, 2 },
 662   { 0x1eac, 0x00c2, 4 },
 663   { 0x1ead, 0x00e2, 4 },
 664   { 0x1eae, 0x0102, 1 },
 665   { 0x1eaf, 0x0103, 1 },
 666   { 0x1eb0, 0x0102, 0 },
 667   { 0x1eb1, 0x0103, 0 },
 668   { 0x1eb2, 0x0102, 3 },
 669   { 0x1eb3, 0x0103, 3 },
 670   { 0x1eb4, 0x0102, 2 },
 671   { 0x1eb5, 0x0103, 2 },
 672   { 0x1eb6, 0x0102, 4 },
 673   { 0x1eb7, 0x0103, 4 },
 674   { 0x1eb8, 0x0045, 4 },
 675   { 0x1eb9, 0x0065, 4 },
 676   { 0x1eba, 0x0045, 3 },
 677   { 0x1ebb, 0x0065, 3 },
 678   { 0x1ebc, 0x0045, 2 },
 679   { 0x1ebd, 0x0065, 2 },
 680   { 0x1ebe, 0x00ca, 1 },
 681   { 0x1ebf, 0x00ea, 1 },
 682   { 0x1ec0, 0x00ca, 0 },
 683   { 0x1ec1, 0x00ea, 0 },
 684   { 0x1ec2, 0x00ca, 3 },
 685   { 0x1ec3, 0x00ea, 3 },
 686   { 0x1ec4, 0x00ca, 2 },
 687   { 0x1ec5, 0x00ea, 2 },
 688   { 0x1ec6, 0x00ca, 4 },
 689   { 0x1ec7, 0x00ea, 4 },
 690   { 0x1ec8, 0x0049, 3 },
 691   { 0x1ec9, 0x0069, 3 },
 692   { 0x1eca, 0x0049, 4 },
 693   { 0x1ecb, 0x0069, 4 },
 694   { 0x1ecc, 0x004f, 4 },
 695   { 0x1ecd, 0x006f, 4 },
 696   { 0x1ece, 0x004f, 3 },
 697   { 0x1ecf, 0x006f, 3 },
 698   { 0x1ed0, 0x00d4, 1 },
 699   { 0x1ed1, 0x00f4, 1 },
 700   { 0x1ed2, 0x00d4, 0 },
 701   { 0x1ed3, 0x00f4, 0 },
 702   { 0x1ed4, 0x00d4, 3 },
 703   { 0x1ed5, 0x00f4, 3 },
 704   { 0x1ed6, 0x00d4, 2 },
 705   { 0x1ed7, 0x00f4, 2 },
 706   { 0x1ed8, 0x00d4, 4 },
 707   { 0x1ed9, 0x00f4, 4 },
 708   { 0x1eda, 0x01a0, 1 },
 709   { 0x1edb, 0x01a1, 1 },
 710   { 0x1edc, 0x01a0, 0 },
 711   { 0x1edd, 0x01a1, 0 },
 712   { 0x1ede, 0x01a0, 3 },
 713   { 0x1edf, 0x01a1, 3 },
 714   { 0x1ee0, 0x01a0, 2 },
 715   { 0x1ee1, 0x01a1, 2 },
 716   { 0x1ee2, 0x01a0, 4 },
 717   { 0x1ee3, 0x01a1, 4 },
 718   { 0x1ee4, 0x0055, 4 },
 719   { 0x1ee5, 0x0075, 4 },
 720   { 0x1ee6, 0x0055, 3 },
 721   { 0x1ee7, 0x0075, 3 },
 722   { 0x1ee8, 0x01af, 1 },
 723   { 0x1ee9, 0x01b0, 1 },
 724   { 0x1eea, 0x01af, 0 },
 725   { 0x1eeb, 0x01b0, 0 },
 726   { 0x1eec, 0x01af, 3 },
 727   { 0x1eed, 0x01b0, 3 },
 728   { 0x1eee, 0x01af, 2 },
 729   { 0x1eef, 0x01b0, 2 },
 730   { 0x1ef0, 0x01af, 4 },
 731   { 0x1ef1, 0x01b0, 4 },
 732   { 0x1ef2, 0x0059, 0 },
 733   { 0x1ef3, 0x0079, 0 },
 734   { 0x1ef4, 0x0059, 4 },
 735   { 0x1ef5, 0x0079, 4 },
 736   { 0x1ef6, 0x0059, 3 },
 737   { 0x1ef7, 0x0079, 3 },
 738   { 0x1ef8, 0x0059, 2 },
 739   { 0x1ef9, 0x0079, 2 },
 740   { 0x1fed, 0x00a8, 0 },
 741   { 0x1fee, 0x00a8, 1 },
 742 };
 743
 744 #define MIN_NEEDED_INPUT        MIN_NEEDED_TO
 745 #define MIN_NEEDED_OUTPUT       MIN_NEEDED_FROM
 746 #define MAX_NEEDED_OUTPUT       2
 747 #define LOOPFCT                 TO_LOOP
 748 #define BODY \
 749   {                                                                           \
 750     uint32_t ch = get32 (inptr);                                              \
 751                                                                               \
 752     if (ch < 0x0080 || (ch >= 0x00a0 && ch < 0x00c3))                         \
 753       {                                                                       \
 754         *outptr++ = ch;                                                       \
 755         inptr += 4;                                                           \
 756       }                                                                       \
 757     else                                                                      \
 758       {                                                                       \
 759         unsigned char res;                                                    \
 760                                                                               \
 761         if (ch >= 0x00c4 && ch < 0x0112)                                      \
 762           res = from_ucs4[ch - 0x00c4 + FROM_IDX_00];                         \
 763         else if (ch >= 0x0152 && ch < 0x01b1)                                 \
 764           res = from_ucs4[ch - 0x0152 + FROM_IDX_01];                         \
 765         else if (ch >= 0x02c6 && ch < 0x02dd)                                 \
 766           res = from_ucs4[ch - 0x02c6 + FROM_IDX_02];                         \
 767         else if (ch >= 0x0300 && ch < 0x0324)                                 \
 768           res = from_ucs4[ch - 0x0300 + FROM_IDX_03];                         \
 769         else if (ch >= 0x0340 && ch < 0x0342) /* Vietnamese tone marks */     \
 770           res = from_ucs4[ch - 0x0340 + FROM_IDX_03];                         \
 771         else if (ch >= 0x2013 && ch < 0x203b)                                 \
 772           res = from_ucs4[ch - 0x2013 + FROM_IDX_20];                         \
 773         else if (ch == 0x20ab)                                                \
 774           res = 0xfe;                                                         \
 775         else if (ch == 0x20ac)                                                \
 776           res = 0x80;                                                         \
 777         else if (ch == 0x2122)                                                \
 778           res = 0x99;                                                         \
 779         else                                                                  \
 780           {                                                                   \
 781             UNICODE_TAG_HANDLER (ch, 4);                                      \
 782             res = 0;                                                          \
 783           }                                                                   \
 784                                                                               \
 785         if (__builtin_expect (res != 0, 1))                                   \
 786           {                                                                   \
 787             *outptr++ = res;                                                  \
 788             inptr += 4;                                                       \
 789           }                                                                   \
 790         else                                                                  \
 791           {                                                                   \
 792             /* Try canonical decomposition.  */                               \
 793             unsigned int i1, i2;                                              \
 794                                                                               \
 795             i1 = 0;                                                           \
 796             i2 = sizeof (decomp_table) / sizeof (decomp_table[0]) - 1;        \
 797             if (ch >= decomp_table[i1].composed                               \
 798                 && ch <= decomp_table[i2].composed)                           \
 799               {                                                               \
 800                 unsigned int i;                                               \
 801                                                                               \
 802                 for (;;)                                                      \
 803                   {                                                           \
 804                     i = (i1 + i2) >> 1;                                       \
 805                     if (ch == decomp_table[i].composed)                       \
 806                       break;                                                  \
 807                     if (ch < decomp_table[i].composed)                        \
 808                       {                                                       \
 809                         if (i1 == i)                                          \
 810                           goto failed;                                        \
 811                         i2 = i;                                               \
 812                       }                                                       \
 813                     else                                                      \
 814                       {                                                       \
 815                         if (i1 != i)                                          \
 816                           i1 = i;                                             \
 817                         else                                                  \
 818                           {                                                   \
 819                             i = i2;                                           \
 820                             if (ch == decomp_table[i].composed)               \
 821                               break;                                          \
 822                             goto failed;                                      \
 823                           }                                                   \
 824                       }                                                       \
 825                   }                                                           \
 826                                                                               \
 827                 /* Found a canonical decomposition.  */                       \
 828                 ch = decomp_table[i].base;                                    \
 829                 /* ch is one of 0x0020, 0x0041..0x005a, 0x0061..0x007a,       \
 830                    0x00a5, 0x00a8, 0x00c2, 0x00c5..0x00c7, 0x00ca, 0x00cf,    \
 831                    0x00d3, 0x00d4, 0x00d6, 0x00d8, 0x00da, 0x00dc, 0x00e2,    \
 832                    0x00e5..0x00e7, 0x00ea, 0x00ef, 0x00f3, 0x00f4, 0x00f6,    \
 833                    0x00f8, 0x00fc, 0x0102, 0x0103, 0x01a0, 0x01a1, 0x01af,    \
 834                    0x01b0.  */                                                \
 835                 if (ch >= 0x0100)                                             \
 836                   {                                                           \
 837                     if (ch < 0x0112)                                          \
 838                       ch = from_ucs4[ch - 0x00c4 + FROM_IDX_00];              \
 839                     else                                                      \
 840                       ch = from_ucs4[ch - 0x0152 + FROM_IDX_01];              \
 841                   }                                                           \
 842                 assert (ch != 0);                                             \
 843                                                                               \
 844                 /* See whether we have room for two bytes.  */                \
 845                 if (__builtin_expect (outptr + 1 >= outend, 0))               \
 846                   {                                                           \
 847                     result = __GCONV_FULL_OUTPUT;                             \
 848                     break;                                                    \
 849                   }                                                           \
 850                                                                               \
 851                 *outptr++ = (unsigned char) ch;                               \
 852                 *outptr++ = comb_table[decomp_table[i].comb1];                \
 853                 inptr += 4;                                                   \
 854                 continue;                                                     \
 855               }                                                               \
 856                                                                               \
 857           failed:                                                             \
 858             /* This is an illegal character.  */                              \
 859             STANDARD_ERR_HANDLER (4);                                         \
 860           }                                                                   \
 861       }                                                                       \
 862   }
 863 #define LOOP_NEED_FLAGS
 864 #define EXTRA_LOOP_DECLS        , int *statep
 865 #include <iconv/loop.c>
 866
 867
 868 /* Now define the toplevel functions.  */
 869 #include <iconv/skeleton.c>