lib/vietcomb.h

   1 /*
   2  * Copyright (C) 2001, 2004, 2011 Free Software Foundation, Inc.
   3  * This file is part of the GNU LIBICONV Library.
   4  *
   5  * The GNU LIBICONV Library is free software; you can redistribute it
   6  * and/or modify it under the terms of the GNU Lesser General Public
   7  * License as published by the Free Software Foundation; either version 2.1
   8  * of the License, or (at your option) any later version.
   9  *
  10  * The GNU LIBICONV Library is distributed in the hope that it will be
  11  * useful, but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  13  * Lesser General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU Lesser General Public
  16  * License along with the GNU LIBICONV Library; see the file COPYING.LIB.
  17  * If not, see <https://www.gnu.org/licenses/>.
  18  */
  19
  20 /* Combining characters used in Vietnamese encodings CP1258, TCVN. */
  21
  22 #ifndef _VIETCOMB_H
  23 #define _VIETCOMB_H
  24
  25 /* Relevant combining characters:
  26    0x0300, 0x0301, 0x0303, 0x0309, 0x0323. */
  27
  28 /* Composition tables for each of the relevant combining characters. */
  29 static const struct { unsigned short base; unsigned short composed; } viet_comp_table_data[] = {
  30 #define viet_comp_table0300_idx 0
  31 #define viet_comp_table0300_len 31
  32   { 0x0041, 0x00C0 },
  33   { 0x0045, 0x00C8 },
  34   { 0x0049, 0x00CC },
  35   { 0x004E, 0x01F8 },
  36   { 0x004F, 0x00D2 },
  37   { 0x0055, 0x00D9 },
  38   { 0x0057, 0x1E80 },
  39   { 0x0059, 0x1EF2 },
  40   { 0x0061, 0x00E0 },
  41   { 0x0065, 0x00E8 },
  42   { 0x0069, 0x00EC },
  43   { 0x006E, 0x01F9 },
  44   { 0x006F, 0x00F2 },
  45   { 0x0075, 0x00F9 },
  46   { 0x0077, 0x1E81 },
  47   { 0x0079, 0x1EF3 },
  48   { 0x00A8, 0x1FED },
  49   { 0x00C2, 0x1EA6 },
  50   { 0x00CA, 0x1EC0 },
  51   { 0x00D4, 0x1ED2 },
  52   { 0x00DC, 0x01DB },
  53   { 0x00E2, 0x1EA7 },
  54   { 0x00EA, 0x1EC1 },
  55   { 0x00F4, 0x1ED3 },
  56   { 0x00FC, 0x01DC },
  57   { 0x0102, 0x1EB0 },
  58   { 0x0103, 0x1EB1 },
  59 /*{ 0x0112, 0x1E14 },*/
  60 /*{ 0x0113, 0x1E15 },*/
  61 /*{ 0x014C, 0x1E50 },*/
  62 /*{ 0x014D, 0x1E51 },*/
  63   { 0x01A0, 0x1EDC },
  64   { 0x01A1, 0x1EDD },
  65   { 0x01AF, 0x1EEA },
  66   { 0x01B0, 0x1EEB },
  67 #define viet_comp_table0301_idx (viet_comp_table0300_idx+viet_comp_table0300_len)
  68 #define viet_comp_table0301_len 63
  69   { 0x0041, 0x00C1 },
  70   { 0x0043, 0x0106 },
  71   { 0x0045, 0x00C9 },
  72   { 0x0047, 0x01F4 },
  73   { 0x0049, 0x00CD },
  74   { 0x004B, 0x1E30 },
  75   { 0x004C, 0x0139 },
  76   { 0x004D, 0x1E3E },
  77   { 0x004E, 0x0143 },
  78   { 0x004F, 0x00D3 },
  79   { 0x0050, 0x1E54 },
  80   { 0x0052, 0x0154 },
  81   { 0x0053, 0x015A },
  82   { 0x0055, 0x00DA },
  83   { 0x0057, 0x1E82 },
  84   { 0x0059, 0x00DD },
  85   { 0x005A, 0x0179 },
  86   { 0x0061, 0x00E1 },
  87   { 0x0063, 0x0107 },
  88   { 0x0065, 0x00E9 },
  89   { 0x0067, 0x01F5 },
  90   { 0x0069, 0x00ED },
  91   { 0x006B, 0x1E31 },
  92   { 0x006C, 0x013A },
  93   { 0x006D, 0x1E3F },
  94   { 0x006E, 0x0144 },
  95   { 0x006F, 0x00F3 },
  96   { 0x0070, 0x1E55 },
  97   { 0x0072, 0x0155 },
  98   { 0x0073, 0x015B },
  99   { 0x0075, 0x00FA },
 100   { 0x0077, 0x1E83 },
 101   { 0x0079, 0x00FD },
 102   { 0x007A, 0x017A },
 103   { 0x00A8, 0x0385 }, /* prefer U+0385 over U+1FEE */
 104   { 0x00C2, 0x1EA4 },
 105   { 0x00C5, 0x01FA },
 106   { 0x00C6, 0x01FC },
 107   { 0x00C7, 0x1E08 },
 108   { 0x00CA, 0x1EBE },
 109   { 0x00CF, 0x1E2E },
 110   { 0x00D4, 0x1ED0 },
 111   { 0x00D5, 0x1E4C },
 112   { 0x00D8, 0x01FE },
 113   { 0x00DC, 0x01D7 },
 114   { 0x00E2, 0x1EA5 },
 115   { 0x00E5, 0x01FB },
 116   { 0x00E6, 0x01FD },
 117   { 0x00E7, 0x1E09 },
 118   { 0x00EA, 0x1EBF },
 119   { 0x00EF, 0x1E2F },
 120   { 0x00F4, 0x1ED1 },
 121   { 0x00F5, 0x1E4D },
 122   { 0x00F8, 0x01FF },
 123   { 0x00FC, 0x01D8 },
 124   { 0x0102, 0x1EAE },
 125   { 0x0103, 0x1EAF },
 126 /*{ 0x0112, 0x1E16 },*/
 127 /*{ 0x0113, 0x1E17 },*/
 128 /*{ 0x014C, 0x1E52 },*/
 129 /*{ 0x014D, 0x1E53 },*/
 130   { 0x0168, 0x1E78 },
 131   { 0x0169, 0x1E79 },
 132   { 0x01A0, 0x1EDA },
 133   { 0x01A1, 0x1EDB },
 134   { 0x01AF, 0x1EE8 },
 135   { 0x01B0, 0x1EE9 },
 136 #define viet_comp_table0303_idx (viet_comp_table0301_idx+viet_comp_table0301_len)
 137 #define viet_comp_table0303_len 34
 138   { 0x0041, 0x00C3 },
 139   { 0x0045, 0x1EBC },
 140   { 0x0049, 0x0128 },
 141   { 0x004E, 0x00D1 },
 142   { 0x004F, 0x00D5 },
 143   { 0x0055, 0x0168 },
 144   { 0x0056, 0x1E7C },
 145   { 0x0059, 0x1EF8 },
 146   { 0x0061, 0x00E3 },
 147   { 0x0065, 0x1EBD },
 148   { 0x0069, 0x0129 },
 149   { 0x006E, 0x00F1 },
 150   { 0x006F, 0x00F5 },
 151   { 0x0075, 0x0169 },
 152   { 0x0076, 0x1E7D },
 153   { 0x0079, 0x1EF9 },
 154   { 0x00C2, 0x1EAA },
 155   { 0x00CA, 0x1EC4 },
 156   { 0x00D3, 0x1E4C },
 157   { 0x00D4, 0x1ED6 },
 158   { 0x00D6, 0x1E4E },
 159   { 0x00DA, 0x1E78 },
 160   { 0x00E2, 0x1EAB },
 161   { 0x00EA, 0x1EC5 },
 162   { 0x00F3, 0x1E4D },
 163   { 0x00F4, 0x1ED7 },
 164   { 0x00F6, 0x1E4F },
 165   { 0x00FA, 0x1E79 },
 166   { 0x0102, 0x1EB4 },
 167   { 0x0103, 0x1EB5 },
 168   { 0x01A0, 0x1EE0 },
 169   { 0x01A1, 0x1EE1 },
 170   { 0x01AF, 0x1EEE },
 171   { 0x01B0, 0x1EEF },
 172 #define viet_comp_table0309_idx (viet_comp_table0303_idx+viet_comp_table0303_len)
 173 #define viet_comp_table0309_len 24
 174   { 0x0041, 0x1EA2 },
 175   { 0x0045, 0x1EBA },
 176   { 0x0049, 0x1EC8 },
 177   { 0x004F, 0x1ECE },
 178   { 0x0055, 0x1EE6 },
 179   { 0x0059, 0x1EF6 },
 180   { 0x0061, 0x1EA3 },
 181   { 0x0065, 0x1EBB },
 182   { 0x0069, 0x1EC9 },
 183   { 0x006F, 0x1ECF },
 184   { 0x0075, 0x1EE7 },
 185   { 0x0079, 0x1EF7 },
 186   { 0x00C2, 0x1EA8 },
 187   { 0x00CA, 0x1EC2 },
 188   { 0x00D4, 0x1ED4 },
 189   { 0x00E2, 0x1EA9 },
 190   { 0x00EA, 0x1EC3 },
 191   { 0x00F4, 0x1ED5 },
 192   { 0x0102, 0x1EB2 },
 193   { 0x0103, 0x1EB3 },
 194   { 0x01A0, 0x1EDE },
 195   { 0x01A1, 0x1EDF },
 196   { 0x01AF, 0x1EEC },
 197   { 0x01B0, 0x1EED },
 198 #define viet_comp_table0323_idx (viet_comp_table0309_idx+viet_comp_table0309_len)
 199 #define viet_comp_table0323_len 50
 200   { 0x0041, 0x1EA0 },
 201   { 0x0042, 0x1E04 },
 202   { 0x0044, 0x1E0C },
 203   { 0x0045, 0x1EB8 },
 204   { 0x0048, 0x1E24 },
 205   { 0x0049, 0x1ECA },
 206   { 0x004B, 0x1E32 },
 207   { 0x004C, 0x1E36 },
 208   { 0x004D, 0x1E42 },
 209   { 0x004E, 0x1E46 },
 210   { 0x004F, 0x1ECC },
 211   { 0x0052, 0x1E5A },
 212   { 0x0053, 0x1E62 },
 213   { 0x0054, 0x1E6C },
 214   { 0x0055, 0x1EE4 },
 215   { 0x0056, 0x1E7E },
 216   { 0x0057, 0x1E88 },
 217   { 0x0059, 0x1EF4 },
 218   { 0x005A, 0x1E92 },
 219   { 0x0061, 0x1EA1 },
 220   { 0x0062, 0x1E05 },
 221   { 0x0064, 0x1E0D },
 222   { 0x0065, 0x1EB9 },
 223   { 0x0068, 0x1E25 },
 224   { 0x0069, 0x1ECB },
 225   { 0x006B, 0x1E33 },
 226   { 0x006C, 0x1E37 },
 227   { 0x006D, 0x1E43 },
 228   { 0x006E, 0x1E47 },
 229   { 0x006F, 0x1ECD },
 230   { 0x0072, 0x1E5B },
 231   { 0x0073, 0x1E63 },
 232   { 0x0074, 0x1E6D },
 233   { 0x0075, 0x1EE5 },
 234   { 0x0076, 0x1E7F },
 235   { 0x0077, 0x1E89 },
 236   { 0x0079, 0x1EF5 },
 237   { 0x007A, 0x1E93 },
 238   { 0x00C2, 0x1EAC },
 239   { 0x00CA, 0x1EC6 },
 240   { 0x00D4, 0x1ED8 },
 241   { 0x00E2, 0x1EAD },
 242   { 0x00EA, 0x1EC7 },
 243   { 0x00F4, 0x1ED9 },
 244   { 0x0102, 0x1EB6 },
 245   { 0x0103, 0x1EB7 },
 246   { 0x01A0, 0x1EE2 },
 247   { 0x01A1, 0x1EE3 },
 248   { 0x01AF, 0x1EF0 },
 249   { 0x01B0, 0x1EF1 },
 250 };
 251 static const struct { unsigned int len; unsigned int idx; } viet_comp_table[] = {
 252   { viet_comp_table0300_len, viet_comp_table0300_idx },
 253   { viet_comp_table0301_len, viet_comp_table0301_idx },
 254   { viet_comp_table0303_len, viet_comp_table0303_idx },
 255   { viet_comp_table0309_len, viet_comp_table0309_idx },
 256   { viet_comp_table0323_len, viet_comp_table0323_idx },
 257 };
 258
 259 /* Decomposition table for the relevant Unicode characters. */
 260 struct viet_decomp { unsigned short composed; unsigned int base : 12; int comb1 : 4; };
 261 static const struct viet_decomp viet_decomp_table[] = {
 262   { 0x00B4, 0x0020, 1 }, /* compatibility decomposition - for TCVN only */
 263   { 0x00C0, 0x0041, 0 },
 264   { 0x00C1, 0x0041, 1 },
 265   { 0x00C3, 0x0041, 2 },
 266   { 0x00C8, 0x0045, 0 },
 267   { 0x00C9, 0x0045, 1 },
 268   { 0x00CC, 0x0049, 0 },
 269   { 0x00CD, 0x0049, 1 },
 270   { 0x00D1, 0x004E, 2 },
 271   { 0x00D2, 0x004F, 0 },
 272   { 0x00D3, 0x004F, 1 },
 273   { 0x00D5, 0x004F, 2 },
 274   { 0x00D9, 0x0055, 0 },
 275   { 0x00DA, 0x0055, 1 },
 276   { 0x00DD, 0x0059, 1 },
 277   { 0x00E0, 0x0061, 0 },
 278   { 0x00E1, 0x0061, 1 },
 279   { 0x00E3, 0x0061, 2 },
 280   { 0x00E8, 0x0065, 0 },
 281   { 0x00E9, 0x0065, 1 },
 282   { 0x00EC, 0x0069, 0 },
 283   { 0x00ED, 0x0069, 1 },
 284   { 0x00F1, 0x006E, 2 },
 285   { 0x00F2, 0x006F, 0 },
 286   { 0x00F3, 0x006F, 1 },
 287   { 0x00F5, 0x006F, 2 },
 288   { 0x00F9, 0x0075, 0 },
 289   { 0x00FA, 0x0075, 1 },
 290   { 0x00FD, 0x0079, 1 },
 291   { 0x0106, 0x0043, 1 },
 292   { 0x0107, 0x0063, 1 },
 293   { 0x0128, 0x0049, 2 },
 294   { 0x0129, 0x0069, 2 },
 295   { 0x0139, 0x004C, 1 },
 296   { 0x013A, 0x006C, 1 },
 297   { 0x0143, 0x004E, 1 },
 298   { 0x0144, 0x006E, 1 },
 299   { 0x0154, 0x0052, 1 },
 300   { 0x0155, 0x0072, 1 },
 301   { 0x015A, 0x0053, 1 },
 302   { 0x015B, 0x0073, 1 },
 303   { 0x0168, 0x0055, 2 },
 304   { 0x0169, 0x0075, 2 },
 305   { 0x0179, 0x005A, 1 },
 306   { 0x017A, 0x007A, 1 },
 307   { 0x01D7, 0x00DC, 1 },
 308   { 0x01D8, 0x00FC, 1 },
 309   { 0x01DB, 0x00DC, 0 },
 310   { 0x01DC, 0x00FC, 0 },
 311   { 0x01F4, 0x0047, 1 },
 312   { 0x01F5, 0x0067, 1 },
 313   { 0x01F8, 0x004E, 0 },
 314   { 0x01F9, 0x006E, 0 },
 315   { 0x01FA, 0x00C5, 1 },
 316   { 0x01FB, 0x00E5, 1 },
 317   { 0x01FC, 0x00C6, 1 },
 318   { 0x01FD, 0x00E6, 1 },
 319   { 0x01FE, 0x00D8, 1 },
 320   { 0x01FF, 0x00F8, 1 },
 321   { 0x02DC, 0x0020, 2 }, /* compatibility decomposition - for TCVN only */
 322   { 0x0385, 0x00A8, 1 },
 323   { 0x1E04, 0x0042, 4 },
 324   { 0x1E05, 0x0062, 4 },
 325   { 0x1E08, 0x00C7, 1 },
 326   { 0x1E09, 0x00E7, 1 },
 327   { 0x1E0C, 0x0044, 4 },
 328   { 0x1E0D, 0x0064, 4 },
 329   { 0x1E24, 0x0048, 4 },
 330   { 0x1E25, 0x0068, 4 },
 331   { 0x1E2E, 0x00CF, 1 },
 332   { 0x1E2F, 0x00EF, 1 },
 333   { 0x1E30, 0x004B, 1 },
 334   { 0x1E31, 0x006B, 1 },
 335   { 0x1E32, 0x004B, 4 },
 336   { 0x1E33, 0x006B, 4 },
 337   { 0x1E36, 0x004C, 4 },
 338   { 0x1E37, 0x006C, 4 },
 339   { 0x1E3E, 0x004D, 1 },
 340   { 0x1E3F, 0x006D, 1 },
 341   { 0x1E42, 0x004D, 4 },
 342   { 0x1E43, 0x006D, 4 },
 343   { 0x1E46, 0x004E, 4 },
 344   { 0x1E47, 0x006E, 4 },
 345   { 0x1E4C, 0x00D3, 2 }, /*{ 0x1E4C, 0x00D5, 1 },*/ /*{ 0x1E4C, 0x004F, 1, 2 },*/
 346   { 0x1E4D, 0x00F3, 2 }, /*{ 0x1E4D, 0x00F5, 1 },*/ /*{ 0x1E4D, 0x006F, 1, 2 },*/
 347   { 0x1E4E, 0x00D6, 2 },
 348   { 0x1E4F, 0x00F6, 2 },
 349   { 0x1E54, 0x0050, 1 },
 350   { 0x1E55, 0x0070, 1 },
 351   { 0x1E5A, 0x0052, 4 },
 352   { 0x1E5B, 0x0072, 4 },
 353   { 0x1E62, 0x0053, 4 },
 354   { 0x1E63, 0x0073, 4 },
 355   { 0x1E6C, 0x0054, 4 },
 356   { 0x1E6D, 0x0074, 4 },
 357   { 0x1E78, 0x00DA, 2 }, /*{ 0x1E78, 0x0168, 1 },*/ /*{ 0x1E78, 0x0055, 1, 2 },*/
 358   { 0x1E79, 0x00FA, 2 }, /*{ 0x1E79, 0x0169, 1 },*/ /*{ 0x1E79, 0x0075, 1, 2 },*/
 359   { 0x1E7C, 0x0056, 2 },
 360   { 0x1E7D, 0x0076, 2 },
 361   { 0x1E7E, 0x0056, 4 },
 362   { 0x1E7F, 0x0076, 4 },
 363   { 0x1E80, 0x0057, 0 },
 364   { 0x1E81, 0x0077, 0 },
 365   { 0x1E82, 0x0057, 1 },
 366   { 0x1E83, 0x0077, 1 },
 367   { 0x1E88, 0x0057, 4 },
 368   { 0x1E89, 0x0077, 4 },
 369   { 0x1E92, 0x005A, 4 },
 370   { 0x1E93, 0x007A, 4 },
 371   { 0x1EA0, 0x0041, 4 },
 372   { 0x1EA1, 0x0061, 4 },
 373   { 0x1EA2, 0x0041, 3 },
 374   { 0x1EA3, 0x0061, 3 },
 375   { 0x1EA4, 0x00C2, 1 },
 376   { 0x1EA5, 0x00E2, 1 },
 377   { 0x1EA6, 0x00C2, 0 },
 378   { 0x1EA7, 0x00E2, 0 },
 379   { 0x1EA8, 0x00C2, 3 },
 380   { 0x1EA9, 0x00E2, 3 },
 381   { 0x1EAA, 0x00C2, 2 },
 382   { 0x1EAB, 0x00E2, 2 },
 383   { 0x1EAC, 0x00C2, 4 },
 384   { 0x1EAD, 0x00E2, 4 },
 385   { 0x1EAE, 0x0102, 1 },
 386   { 0x1EAF, 0x0103, 1 },
 387   { 0x1EB0, 0x0102, 0 },
 388   { 0x1EB1, 0x0103, 0 },
 389   { 0x1EB2, 0x0102, 3 },
 390   { 0x1EB3, 0x0103, 3 },
 391   { 0x1EB4, 0x0102, 2 },
 392   { 0x1EB5, 0x0103, 2 },
 393   { 0x1EB6, 0x0102, 4 },
 394   { 0x1EB7, 0x0103, 4 },
 395   { 0x1EB8, 0x0045, 4 },
 396   { 0x1EB9, 0x0065, 4 },
 397   { 0x1EBA, 0x0045, 3 },
 398   { 0x1EBB, 0x0065, 3 },
 399   { 0x1EBC, 0x0045, 2 },
 400   { 0x1EBD, 0x0065, 2 },
 401   { 0x1EBE, 0x00CA, 1 },
 402   { 0x1EBF, 0x00EA, 1 },
 403   { 0x1EC0, 0x00CA, 0 },
 404   { 0x1EC1, 0x00EA, 0 },
 405   { 0x1EC2, 0x00CA, 3 },
 406   { 0x1EC3, 0x00EA, 3 },
 407   { 0x1EC4, 0x00CA, 2 },
 408   { 0x1EC5, 0x00EA, 2 },
 409   { 0x1EC6, 0x00CA, 4 },
 410   { 0x1EC7, 0x00EA, 4 },
 411   { 0x1EC8, 0x0049, 3 },
 412   { 0x1EC9, 0x0069, 3 },
 413   { 0x1ECA, 0x0049, 4 },
 414   { 0x1ECB, 0x0069, 4 },
 415   { 0x1ECC, 0x004F, 4 },
 416   { 0x1ECD, 0x006F, 4 },
 417   { 0x1ECE, 0x004F, 3 },
 418   { 0x1ECF, 0x006F, 3 },
 419   { 0x1ED0, 0x00D4, 1 },
 420   { 0x1ED1, 0x00F4, 1 },
 421   { 0x1ED2, 0x00D4, 0 },
 422   { 0x1ED3, 0x00F4, 0 },
 423   { 0x1ED4, 0x00D4, 3 },
 424   { 0x1ED5, 0x00F4, 3 },
 425   { 0x1ED6, 0x00D4, 2 },
 426   { 0x1ED7, 0x00F4, 2 },
 427   { 0x1ED8, 0x00D4, 4 },
 428   { 0x1ED9, 0x00F4, 4 },
 429   { 0x1EDA, 0x01A0, 1 },
 430   { 0x1EDB, 0x01A1, 1 },
 431   { 0x1EDC, 0x01A0, 0 },
 432   { 0x1EDD, 0x01A1, 0 },
 433   { 0x1EDE, 0x01A0, 3 },
 434   { 0x1EDF, 0x01A1, 3 },
 435   { 0x1EE0, 0x01A0, 2 },
 436   { 0x1EE1, 0x01A1, 2 },
 437   { 0x1EE2, 0x01A0, 4 },
 438   { 0x1EE3, 0x01A1, 4 },
 439   { 0x1EE4, 0x0055, 4 },
 440   { 0x1EE5, 0x0075, 4 },
 441   { 0x1EE6, 0x0055, 3 },
 442   { 0x1EE7, 0x0075, 3 },
 443   { 0x1EE8, 0x01AF, 1 },
 444   { 0x1EE9, 0x01B0, 1 },
 445   { 0x1EEA, 0x01AF, 0 },
 446   { 0x1EEB, 0x01B0, 0 },
 447   { 0x1EEC, 0x01AF, 3 },
 448   { 0x1EED, 0x01B0, 3 },
 449   { 0x1EEE, 0x01AF, 2 },
 450   { 0x1EEF, 0x01B0, 2 },
 451   { 0x1EF0, 0x01AF, 4 },
 452   { 0x1EF1, 0x01B0, 4 },
 453   { 0x1EF2, 0x0059, 0 },
 454   { 0x1EF3, 0x0079, 0 },
 455   { 0x1EF4, 0x0059, 4 },
 456   { 0x1EF5, 0x0079, 4 },
 457   { 0x1EF6, 0x0059, 3 },
 458   { 0x1EF7, 0x0079, 3 },
 459   { 0x1EF8, 0x0059, 2 },
 460   { 0x1EF9, 0x0079, 2 },
 461   { 0x1FED, 0x00A8, 0 },
 462   { 0x1FEE, 0x00A8, 1 }, /* U+1FEE => U+0385 => U+00A8 U+0301 */
 463 };
 464
 465 #endif /* _VIETCOMB_H */