i18npool/source/nativenumber/data/numberchar.h

   1 /*************************************************************************
   2  *
   3  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
   4  *
   5  * Copyright 2008 by Sun Microsystems, Inc.
   6  *
   7  * OpenOffice.org - a multi-platform office productivity suite
   8  *
   9  * $RCSfile: numberchar.h,v $
  10  * $Revision: 1.14 $
  11  *
  12  * This file is part of OpenOffice.org.
  13  *
  14  * OpenOffice.org is free software: you can redistribute it and/or modify
  15  * it under the terms of the GNU Lesser General Public License version 3
  16  * only, as published by the Free Software Foundation.
  17  *
  18  * OpenOffice.org is distributed in the hope that it will be useful,
  19  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  20  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  21  * GNU Lesser General Public License version 3 for more details
  22  * (a copy is included in the LICENSE file that accompanied this code).
  23  *
  24  * You should have received a copy of the GNU Lesser General Public License
  25  * version 3 along with OpenOffice.org.  If not, see
  26  * <http://www.openoffice.org/license.html>
  27  * for a copy of the LGPLv3 License.
  28  *
  29  ************************************************************************/
  30 #ifndef _L10N_TRANSLITERATION_NUMTOCHAR_H_
  31 #define _L10N_TRANSLITERATION_NUMTOCHAR_H_
  32
  33 namespace com { namespace sun { namespace star { namespace i18n {
  34
  35 static const sal_Int16 NumberChar_HalfWidth     = 0;
  36 static const sal_Int16 NumberChar_FullWidth     = 1;
  37 static const sal_Int16 NumberChar_Lower_zh      = 2;
  38 static const sal_Int16 NumberChar_Upper_zh      = 3;
  39 static const sal_Int16 NumberChar_Upper_zh_TW   = 4;
  40 static const sal_Int16 NumberChar_Modern_ja     = 5;
  41 static const sal_Int16 NumberChar_Traditional_ja= 6;
  42 static const sal_Int16 NumberChar_Lower_ko      = 7;
  43 static const sal_Int16 NumberChar_Upper_ko      = 8;
  44 static const sal_Int16 NumberChar_Hangul_ko     = 9;
  45 static const sal_Int16 NumberChar_Indic_ar      = 10;
  46 static const sal_Int16 NumberChar_EastIndic_ar  = 11;
  47 static const sal_Int16 NumberChar_hi            = 12;
  48 static const sal_Int16 NumberChar_th            = 13;
  49 static const sal_Int16 NumberChar_or            = 14;
  50 static const sal_Int16 NumberChar_mr            = 15;
  51 static const sal_Int16 NumberChar_bn            = 16;
  52 static const sal_Int16 NumberChar_pa            = 17;
  53 static const sal_Int16 NumberChar_gu            = 18;
  54 static const sal_Int16 NumberChar_ta            = 19;
  55 static const sal_Int16 NumberChar_te            = 20;
  56 static const sal_Int16 NumberChar_kn            = 21;
  57 static const sal_Int16 NumberChar_ml            = 22;
  58 static const sal_Int16 NumberChar_lo            = 23;
  59 static const sal_Int16 NumberChar_bo            = 24;
  60 static const sal_Int16 NumberChar_my            = 25;
  61 static const sal_Int16 NumberChar_km            = 26;
  62 static const sal_Int16 NumberChar_mn            = 27;
  63 static const sal_Int16 NumberChar_he            = 28;
  64 static const sal_Int16 NumberChar_ne            = 29;
  65 static const sal_Int16 NumberChar_dz            = 30;
  66 static const sal_Int16 NumberChar_Count         = 31;
  67
  68 static sal_Unicode NumberChar[][10] = {
  69 //      0       1       2       3       4       5       6       7       8       9
  70     { 0x0030, 0x0031, 0x0032, 0x0033, 0x0034, 0x0035, 0x0036, 0x0037, 0x0038, 0x0039 }, // Half Width (Ascii)
  71     { 0xFF10, 0xFF11, 0xFF12, 0xFF13, 0xFF14, 0xFF15, 0xFF16, 0xFF17, 0xFF18, 0xFF19 }, // Full Width
  72     { 0x3007, 0x4E00, 0x4E8c, 0x4E09, 0x56DB, 0x4E94, 0x516D, 0x4E03, 0x516B, 0x4E5D }, // Chinese Lower
  73     { 0x96F6, 0x58F9, 0x8D30, 0x53C1, 0x8086, 0x4F0D, 0x9646, 0x67D2, 0x634C, 0x7396 }, // S. Chinese Upper
  74     { 0x96F6, 0x58F9, 0x8CB3, 0x53C3, 0x8086, 0x4F0D, 0x9678, 0x67D2, 0x634C, 0x7396 }, // T. Chinese Upper
  75     { 0x3007, 0x4E00, 0x4E8C, 0x4E09, 0x56DB, 0x4E94, 0x516D, 0x4E03, 0x516B, 0x4E5D }, // Japanese Modern
  76     { 0x96F6, 0x58F1, 0x5F10, 0x53C2, 0x56DB, 0x4F0D, 0x516D, 0x4E03, 0x516B, 0x4E5D }, // Japanese Trad.
  77     { 0x3007, 0x4E00, 0x4E8C, 0x4E09, 0x56DB, 0x4E94, 0x516D, 0x4E03, 0x516B, 0x4E5D }, // Korean Lower
  78     { 0xF9B2, 0x58F9, 0x8CB3, 0x53C3, 0x56DB, 0x4E94, 0x516D, 0x4E03, 0x516B, 0x4E5D }, // Korean Upper
  79     { 0xC601, 0xC77C, 0xC774, 0xC0BC, 0xC0AC, 0xC624, 0xC721, 0xCE60, 0xD314, 0xAD6C }, // Korean Hangul
  80     { 0x0660, 0x0661, 0x0662, 0x0663, 0x0664, 0x0665, 0x0666, 0x0667, 0x0668, 0x0669 }, // Arabic Indic
  81     { 0x06F0, 0x06F1, 0x06F2, 0x06F3, 0x06F4, 0x06F5, 0x06F6, 0x06F7, 0x06F8, 0x06F9 }, // Est. Arabic Indic
  82     { 0x0966, 0x0967, 0x0968, 0x0969, 0x096A, 0x096B, 0x096C, 0x096D, 0x096E, 0x096F }, // Indic (Devanagari)
  83     { 0x0E50, 0x0E51, 0x0E52, 0x0E53, 0x0E54, 0x0E55, 0x0E56, 0x0E57, 0x0E58, 0x0E59 }, // Thai
  84     { 0x0866, 0x0867, 0x0868, 0x0869, 0x086A, 0x086B, 0x086C, 0x086D, 0x086E, 0x086F }, // Oriya
  85     { 0x0966, 0x0967, 0x0968, 0x0969, 0x096A, 0x096B, 0x096C, 0x096D, 0x096E, 0x096F }, // Marathi
  86     { 0x09E6, 0x09E7, 0x09E8, 0x09E9, 0x09EA, 0x09EB, 0x09EC, 0x09ED, 0x09EE, 0x09EF }, // Bengali
  87     { 0x0A66, 0x0A67, 0x0A68, 0x0A69, 0x0A6A, 0x0A6B, 0x0A6C, 0x0A6D, 0x0A6E, 0x0A6F }, // Punjabi (Gurmukhi)
  88     { 0x0AE6, 0x0AE7, 0x0AE8, 0x0AE9, 0x0AEA, 0x0AEB, 0x0AEC, 0x0AED, 0x0AEE, 0x0AEF }, // Gujarali
  89     { 0x0030, 0x0BE7, 0x0BE8, 0x0BE9, 0x0BEA, 0x0BEB, 0x0BEC, 0x0BED, 0x0BEE, 0x0BEF }, // Tamil
  90     { 0x0C66, 0x0C67, 0x0C68, 0x0C69, 0x0C6A, 0x0C6B, 0x0C6C, 0x0C6D, 0x0C6E, 0x0C6F }, // Telugu
  91     { 0x0CE6, 0x0CE7, 0x0CE8, 0x0CE9, 0x0CEA, 0x0CEB, 0x0CEC, 0x0CED, 0x0CEE, 0x0CEF }, // Kannada
  92     { 0x0DE6, 0x0DE7, 0x0DE8, 0x0DE9, 0x0DEA, 0x0DEB, 0x0DEC, 0x0DED, 0x0DEE, 0x0DEF }, // Malayalam
  93     { 0x0ED0, 0x0ED1, 0x0ED2, 0x0ED3, 0x0ED4, 0x0ED5, 0x0ED6, 0x0ED7, 0x0ED8, 0x0ED9 }, // Lao
  94     { 0x0F20, 0x0F21, 0x0F22, 0x0F23, 0x0F24, 0x0F25, 0x0F26, 0x0F27, 0x0F28, 0x0F29 }, // Tibetan
  95     { 0x1040, 0x1041, 0x1042, 0x1043, 0x1044, 0x1045, 0x1046, 0x1047, 0x1048, 0x1049 }, // Myammar
  96     { 0x17E0, 0x17E1, 0x17E2, 0x17E3, 0x17E4, 0x17E5, 0x17E6, 0x17E7, 0x17E8, 0x17E9 }, // Cambodian (Khmer)
  97     { 0x1810, 0x1811, 0x1812, 0x1813, 0x1814, 0x1815, 0x1816, 0x1817, 0x1818, 0x1819 }, // Mongolian
  98     { 0x0020, 0x05D0, 0x05D1, 0x05D2, 0x05D3, 0x05D4, 0x05D5, 0x05D6, 0x05D7, 0x05D8 }, // Hebrew
  99     { 0x0966, 0x0967, 0x0968, 0x0969, 0x096A, 0x096B, 0x096C, 0x096D, 0x096E, 0x096F }, // Nepali
 100     { 0x0F20, 0x0F21, 0x0F22, 0x0F23, 0x0F24, 0x0F25, 0x0F26, 0x0F27, 0x0F28, 0x0F29 }, // Dzongkha
 101 };
 102
 103 static sal_Unicode DecimalChar[] = {
 104     0x002E, // Half Width (Ascii)
 105     0xFF0E, // Full Width
 106     0xFF0E, // Chinese Lower
 107     0x70B9, // S. Chinese Upper
 108     0x9EDE, // T. Chinese Upper
 109     0x30FB, // Japanese Modern
 110     0x30FB, // Japanese Trad.
 111     0xFF0E, // Korean Lower
 112     0x9EDE, // Korean Upper
 113     0xC810, // Korean Hangul
 114     0x066B, // Arabic Indic
 115     0x066B, // Est. Arabic Indic
 116     0x0000, // Indic (Devanagari)
 117     0x0000, // Thai
 118     0x0000, // Oriya
 119     0x0000, // Marathi
 120     0x0000, // Bengali
 121     0x0000, // Punjabi (Gurmukhi)
 122     0x0000, // Gujarali
 123     0x0000, // Tamil
 124     0x0000, // Telugu
 125     0x0000, // Kannada
 126     0x0000, // Malayalam
 127     0x0000, // Lao
 128     0x0000, // Tibetan
 129     0x0000, // Myammar
 130     0x0000, // Cambodian (Khmer)
 131     0x0000, // Mongolian
 132     0x0000, // Hebrew
 133     0x0000, // Nepali
 134     0x0000, // Dzongkha
 135 };
 136
 137 static sal_Unicode MinusChar[] = {
 138     0x002D, // Half Width (Ascii)
 139     0xFF0D, // Full Width
 140     0xFF0D, // Chinese Lower
 141     0x8D1F, // S. Chinese Upper
 142     0x5069, // T. Chinese Upper
 143     0x2212, // Japanese Modern
 144     0x2212, // Japanese Trad.
 145     0xFF0D, // Korean Lower
 146     0x5069, // Korean Upper
 147     0xFF0D, // Korean Hangul ???
 148     0x0000, // Arabic Indic
 149     0x2212, // Est. Arabic Indic
 150     0x0000, // Indic
 151     0x0000, // Thai
 152     0x0000, // Oriya
 153     0x0000, // Marathi
 154     0x0000, // Bengali
 155     0x0000, // Punjabi
 156     0x0000, // Gujarali
 157     0x0000, // Tamil
 158     0x0000, // Telugu
 159     0x0000, // Kannada
 160     0x0000, // Malayalam
 161     0x0000, // Lao
 162     0x0000, // Tibetan
 163     0x0000, // Myammar
 164     0x0000, // Cambodian (Khmer)
 165     0x0000, // Mongolian
 166     0x0000, // Hebrew
 167     0x0000, // Nepali
 168     0x0000, // Dzongkha
 169 };
 170
 171 static sal_uInt16 SeparatorChar[] = {
 172     0x002C, // Half Width (Ascii)
 173     0xFF0C, // Full Width
 174     0x3001, // Chinese Lower
 175     0x3001, // S. Chinese Upper
 176     0x3001, // T. Chinese Upper
 177     0x3001, // Japanese Modern
 178     0x3001, // Japanese Trad.
 179     0x3001, // Korean Lower
 180     0x3001, // Korean Upper
 181     0x3001, // Korean Hangul
 182     0x0000, // Arabic Indic
 183     0x066C, // Est. Arabic Indic
 184     0x0000, // Indic
 185     0x0000, // Thai
 186     0x0000, // Oriya
 187     0x0000, // Marathi
 188     0x0000, // Bengali
 189     0x0000, // Punjabi
 190     0x0000, // Gujarali
 191     0x0000, // Tamil
 192     0x0000, // Telugu
 193     0x0000, // Kannada
 194     0x0000, // Malayalam
 195     0x0000, // Lao
 196     0x0000, // Tibetan
 197     0x0000, // Myammar
 198     0x0000, // Cambodian (Khmer)
 199     0x0000, // Mongolian
 200     0x0000, // Hebrew
 201     0x0000, // Nepali
 202     0x0000, // Dzongkha
 203 };
 204
 205 #define NUMBER_ZERO     NumberChar[NumberChar_HalfWidth][0]     // 0x0030
 206 #define NUMBER_ONE      NumberChar[NumberChar_HalfWidth][1]     // 0x0031
 207 #define NUMBER_NINE     NumberChar[NumberChar_HalfWidth][9]     // 0x0039
 208 #define isNumber(n)     ( NUMBER_ZERO <= n && n <= NUMBER_NINE )
 209 #define isDecimal(n)    ( n == DecimalChar[NumberChar_HalfWidth] )
 210 #define isMinus(n)          ( n == MinusChar[NumberChar_HalfWidth] )
 211 #define isSeparator(n)  ( n == SeparatorChar[NumberChar_HalfWidth] )
 212
 213 const sal_Int16 Multiplier_Lower_zh     = 0;
 214 const sal_Int16 Multiplier_Upper_zh     = 1;
 215 const sal_Int16 Multiplier_Lower_zh_TW  = 2;
 216 const sal_Int16 Multiplier_Upper_zh_TW  = 3;
 217 const sal_Int16 Multiplier_Lower_ko     = 4;
 218 const sal_Int16 Multiplier_Hangul_ko    = 5;
 219 const sal_Int16 Multiplier_Modern_ja    = 6;
 220 const sal_Int16 Multiplier_Traditional_ja = 7;
 221 const sal_Int16 Multiplier_Count        = 8;
 222
 223 const sal_Int16 ExponentCount_6_CJK     = 6;
 224
 225 static sal_Int16 MultiplierExponent_6_CJK[ExponentCount_6_CJK] = {
 226         12,     8,      4,      3,      2,      1
 227 };
 228 static sal_Unicode MultiplierChar_6_CJK[][ExponentCount_6_CJK] = {
 229     {0x5146, 0x4EBF, 0x4E07, 0x5343, 0x767E, 0x5341},   // S. Chinese Lower
 230     {0x5146, 0x4EBF, 0x4E07, 0x4EDF, 0x4F70, 0x62FE},   // S. Chinese Upper
 231     {0x5146, 0x5104, 0x842C, 0x5343, 0x767E, 0x5341},   // T. Chinese Lower
 232     {0x5146, 0x5104, 0x842C, 0x4EDF, 0x4F70, 0x62FE},   // T. Chinese & Korean Upper
 233     {0x5146, 0x5104, 0x4E07, 0x5343, 0x767E, 0x5341},   // Korean Lower
 234     {0xC870, 0xC5B5, 0xB9CC, 0xCC9C, 0xBC31, 0xC2ED},   // Korean Hangul
 235     {0x5146, 0x5104, 0x4E07, 0x5343, 0x767E, 0x5341},   // Japanese Modern
 236     {0x5146, 0x5104, 0x842C, 0x9621, 0x767E, 0x62FE}    // Japanese Traditional
 237 };
 238
 239 const sal_Int16 ExponentCount_2_CJK     = 2;
 240
 241 static sal_Int16 MultiplierExponent_2_CJK[ExponentCount_2_CJK] = {
 242         8,      4,
 243 };
 244
 245 static sal_Unicode MultiplierChar_2_CJK[][ExponentCount_2_CJK] = {
 246     {0x4EBF, 0x4E07}, // S. Chinese Lower
 247     {0x4EBF, 0x4E07}, // S. Chinese Upper
 248     {0x5104, 0x842C}, // T. Chinese Lower
 249     {0x5104, 0x842C}, // T. Chinese & Korean Upper
 250     {0x5104, 0x4E07}, // Korean Lower
 251     {0xC5B5, 0xB9CC}, // Korean Hangul
 252     {0x5104, 0x4E07}, // Japanese Modern
 253     {0x5104, 0x842C}  // Japanese Traditional
 254 };
 255
 256 const sal_Int16 ExponentCount_7_CJK     = 7;
 257
 258 static sal_Int16 MultiplierExponent_7_CJK[ExponentCount_7_CJK] = {
 259         16,     12,     8,      4,      3,      2,      1
 260 };
 261 static sal_Unicode MultiplierChar_7_CJK[][ExponentCount_7_CJK] = {
 262     {0x4EAC, 0x5146, 0x4EBF, 0x4E07, 0x5343, 0x767E, 0x5341},   // S. Chinese Lower
 263     {0x4EAC, 0x5146, 0x4EBF, 0x4E07, 0x4EDF, 0x4F70, 0x62FE},   // S. Chinese Upper
 264     {0x4EAC, 0x5146, 0x5104, 0x842C, 0x5343, 0x767E, 0x5341},   // T. Chinese Lower
 265     {0x4EAC, 0x5146, 0x5104, 0x842C, 0x4EDF, 0x4F70, 0x62FE},   // T. Chinese & Korean Upper
 266     {0x4EAC, 0x5146, 0x5104, 0x4E07, 0x5343, 0x767E, 0x5341},   // Korean Lower
 267     {0x4EAC, 0xC870, 0xC5B5, 0xB9CC, 0xCC9C, 0xBC31, 0xC2ED},   // Korean Hangul
 268     {0x4EAC, 0x5146, 0x5104, 0x4E07, 0x5343, 0x767E, 0x5341},   // Japanese Modern
 269     {0x4EAC, 0x5146, 0x5104, 0x842C, 0x9621, 0x767E, 0x62FE}    // Japanese Traditional
 270 };
 271
 272 typedef struct {
 273     sal_Int16 ExponentCount;
 274     sal_Int16* MultiplierExponent;
 275     sal_Unicode *MultiplierChar;
 276 } Multiplier;
 277
 278 } } } }
 279
 280 #endif // _L10N_TRANSLITERATION_NUMTOCHAR_H_