i18nutil/source/utility/widthfolding.cxx

   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: widthfolding.cxx,v $
  10  * $Revision: 1.7 $
  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
  31 // prevent internal compiler error with MSVC6SP3
  32 #include <utility>
  33 #include <i18nutil/widthfolding.hxx>
  34 #include <i18nutil/x_rtl_ustring.h>
  35 #include "widthfolding_data.h"
  36
  37 using namespace com::sun::star::uno;
  38 using namespace rtl;
  39
  40 namespace com { namespace sun { namespace star { namespace i18n {
  41
  42 sal_Unicode widthfolding::decompose_ja_voiced_sound_marksChar2Char (sal_Unicode inChar)
  43 {
  44     if (0x30a0 <= inChar && inChar <= 0x30ff) {
  45       sal_Int16 i = inChar - 0x3040;
  46       if (decomposition_table[i].decomposited_character_1)
  47           return 0xFFFF;
  48     }
  49     return inChar;
  50 }
  51
  52 /**
  53  * Decompose Japanese specific voiced and semi-voiced sound marks.
  54  */
  55 OUString widthfolding::decompose_ja_voiced_sound_marks (const OUString& inStr, sal_Int32 startPos, sal_Int32 nCount, Sequence< sal_Int32 >& offset, sal_Bool useOffset )
  56 {
  57   // Create a string buffer which can hold nCount * 2 + 1 characters.
  58   // Its size may become double of nCount.
  59   rtl_uString * newStr;
  60   x_rtl_uString_new_WithLength( &newStr, nCount * 2 ); // defined in x_rtl_ustring.h  The reference count is 0 now.
  61
  62   sal_Int32 *p = NULL;
  63   sal_Int32 position = 0;
  64   if (useOffset) {
  65       // Allocate double of nCount length to offset argument.
  66       offset.realloc( nCount * 2 );
  67       p = offset.getArray();
  68       position = startPos;
  69   }
  70
  71   // Prepare pointers of unicode character arrays.
  72   const sal_Unicode* src = inStr.getStr() + startPos;
  73   sal_Unicode* dst = newStr->buffer;
  74
  75   // Decomposition: GA --> KA + voice-mark
  76   while (nCount -- > 0) {
  77     sal_Unicode c = *src++;
  78     // see http://charts.unicode.org/Web/U3040.html Hiragana (U+3040..U+309F)
  79     // see http://charts.unicode.org/Web/U30A0.html Katakana (U+30A0..U+30FF)
  80     // Hiragana is not applied to decomposition.
  81     // Only Katakana is applied to decomposition
  82     if (0x30a0 <= c && c <= 0x30ff) {
  83       int i = int(c - 0x3040);
  84       sal_Unicode first = decomposition_table[i].decomposited_character_1;
  85       if (first != 0x0000) {
  86     *dst ++ = first;
  87     *dst ++ = decomposition_table[i].decomposited_character_2; // second
  88         if (useOffset) {
  89             *p ++ = position;
  90             *p ++ = position ++;
  91         }
  92     continue;
  93       }
  94     }
  95     *dst ++ = c;
  96     if (useOffset)
  97         *p ++ = position ++;
  98   }
  99   *dst = (sal_Unicode) 0;
 100
 101   newStr->length = sal_Int32(dst - newStr->buffer);
 102   if (useOffset)
 103       offset.realloc(newStr->length);
 104   return OUString( newStr ); // defined in rtl/usrting. The reference count is increased from 0 to 1.
 105 }
 106
 107 oneToOneMapping& widthfolding::getfull2halfTable(void)
 108 {
 109     static oneToOneMappingWithFlag table(full2half, sizeof(full2half), FULL2HALF_NORMAL);
 110     table.makeIndex();
 111     return table;
 112 }
 113
 114 /**
 115  * Compose Japanese specific voiced and semi-voiced sound marks.
 116  */
 117 OUString widthfolding::compose_ja_voiced_sound_marks (const OUString& inStr, sal_Int32 startPos, sal_Int32 nCount, Sequence< sal_Int32 >& offset, sal_Bool useOffset, sal_Int32 nFlags )
 118 {
 119   // Create a string buffer which can hold nCount + 1 characters.
 120   // Its size may become equal to nCount or smaller.
 121   // The reference count is 0 now.
 122   rtl_uString * newStr = x_rtl_uString_new_WithLength( nCount ); // defined in x_rtl_ustring.h
 123
 124   // Prepare pointers of unicode character arrays.
 125   const sal_Unicode* src = inStr.getStr() + startPos;
 126   sal_Unicode* dst = newStr->buffer;
 127
 128   // This conversion algorithm requires at least one character.
 129  if (nCount > 0) {
 130
 131   // .. .. KA         VOICE .. ..
 132   //       ^          ^
 133   //       previousChar   currentChar
 134   //       ^
 135   //       position
 136   //
 137   // will be converted to
 138   // .. .. GA       .. ..
 139
 140   sal_Int32 *p = NULL;
 141   sal_Int32 position = 0;
 142   if (useOffset) {
 143       // Allocate nCount length to offset argument.
 144       offset.realloc( nCount );
 145       p = offset.getArray();
 146       position = startPos;
 147   }
 148
 149   //
 150   sal_Unicode previousChar = *src ++;
 151   sal_Unicode currentChar;
 152
 153   // Composition: KA + voice-mark --> GA
 154   while (-- nCount > 0) {
 155     currentChar = *src ++;
 156     // see http://charts.unicode.org/Web/U3040.html Hiragana (U+3040..U+309F)
 157     // see http://charts.unicode.org/Web/U30A0.html Katakana (U+30A0..U+30FF)
 158     // 0x3099 COMBINING KATAKANA-HIRAGANA VOICED SOUND MARK
 159     // 0x309a COMBINING KATAKANA-HIRAGANA SEMI-VOICED SOUND MARK
 160     // 0x309b KATAKANA-HIRAGANA VOICED SOUND MARK
 161     // 0x309c KATAKANA-HIRAGANA SEMI-VOICED SOUND MARK
 162     int j = currentChar - 0x3099; // 0x3099, 0x309a, 0x309b, 0x309c ?
 163
 164     if (2 <= j && j <= 3) // 0x309b or 0x309c
 165         j -= 2;
 166
 167     if (0 <= j && j <= 1) {
 168       // 0 addresses a code point regarding 0x3099 or 0x309b (voiced sound mark),
 169       // 1 is 0x309a or 0x309c (semi-voiced sound mark)
 170       int i = int(previousChar - 0x3040); // i acts as an index of array
 171       sal_Bool bCompose = sal_False;
 172
 173       if (0 <= i && i <= (0x30ff - 0x3040) && composition_table[i][j])
 174         bCompose = sal_True;
 175
 176       // not to use combined KATAKANA LETTER VU
 177       if ( previousChar == 0x30a6 && (nFlags & WIDTHFOLDNIG_DONT_USE_COMBINED_VU) )
 178         bCompose = sal_False;
 179
 180       if( bCompose ){
 181         if (useOffset) {
 182             position ++;
 183             *p ++ = position ++;
 184         }
 185     *dst ++ =  composition_table[i][j];
 186     previousChar = *src ++;
 187     nCount --;
 188     continue;
 189       }
 190     }
 191     if (useOffset)
 192         *p ++ = position ++;
 193     *dst ++ = previousChar;
 194     previousChar = currentChar;
 195   }
 196
 197   if (nCount == 0) {
 198     if (useOffset)
 199         *p = position;
 200     *dst ++ = previousChar;
 201   }
 202
 203   *dst = (sal_Unicode) 0;
 204
 205   newStr->length = sal_Int32(dst - newStr->buffer);
 206  }
 207   if (useOffset)
 208       offset.realloc(newStr->length);
 209   return OUString( newStr ); // defined in rtl/usrting. The reference count is increased from 0 to 1.
 210 }
 211
 212 oneToOneMapping& widthfolding::gethalf2fullTable(void)
 213 {
 214     static oneToOneMappingWithFlag table(half2full, sizeof(half2full), HALF2FULL_NORMAL);
 215     table.makeIndex();
 216     return table;
 217 }
 218
 219 sal_Unicode widthfolding::getCompositionChar(sal_Unicode c1, sal_Unicode c2)
 220 {
 221     return composition_table[c1 - 0x3040][c2 - 0x3099];
 222 }
 223
 224
 225 oneToOneMapping& widthfolding::getfull2halfTableForASC()
 226 {
 227     static oneToOneMappingWithFlag table(full2half, sizeof(full2half), FULL2HALF_ASC_FUNCTION);
 228     table.makeIndex();
 229
 230     // bluedwarf: dirty hack!
 231     // There is an exception. Additional conversion is required following:
 232     //  0xFFE5 (FULLWIDTH YEN SIGN)  --> 0x005C (REVERSE SOLIDUS)
 233     //
 234     //  See the following page for detail:
 235     // http://wiki.services.openoffice.org/wiki/Calc/Features/JIS_and_ASC_functions
 236     int i, j, high, low;
 237     int n = sizeof(full2halfASCException) / sizeof(UnicodePairWithFlag);
 238     for( i = 0; i < n; i++ )
 239     {
 240         high = (full2halfASCException[i].first >> 8) & 0xFF;
 241         low  = (full2halfASCException[i].first)      & 0xFF;
 242
 243         if( !table.mpIndex[high] )
 244         {
 245             table.mpIndex[high] = new UnicodePairWithFlag*[256];
 246
 247             for( j = 0; j < 256; j++ )
 248                 table.mpIndex[high][j] = NULL;
 249         }
 250         table.mpIndex[high][low] = &full2halfASCException[i];
 251     }
 252
 253     return table;
 254 }
 255
 256 oneToOneMapping& widthfolding::gethalf2fullTableForJIS()
 257 {
 258     static oneToOneMappingWithFlag table(half2full, sizeof(half2full), HALF2FULL_JIS_FUNCTION);
 259     table.makeIndex();
 260
 261     // bluedwarf: dirty hack!
 262     //  There are some exceptions. Additional conversion are required following:
 263     //  0x0022 (QUOTATION MARK)  --> 0x201D (RIGHT DOUBLE QUOTATION MARK)
 264     //  0x0027 (APOSTROPHE)      --> 0x2019 (RIGHT SINGLE QUOTATION MARK)
 265     //  0x005C (REVERSE SOLIDUS) --> 0xFFE5 (FULLWIDTH YEN SIGN)
 266     //  0x0060 (GRAVE ACCENT)    --> 0x2018 (LEFT SINGLE QUOTATION MARK)
 267     //
 268     //  See the following page for detail:
 269     // http://wiki.services.openoffice.org/wiki/Calc/Features/JIS_and_ASC_functions
 270     int i, j, high, low;
 271     int n = sizeof(half2fullJISException) / sizeof(UnicodePairWithFlag);
 272     for( i = 0; i < n; i++ )
 273     {
 274         high = (half2fullJISException[i].first >> 8) & 0xFF;
 275         low  = (half2fullJISException[i].first)      & 0xFF;
 276
 277         if( !table.mpIndex[high] )
 278         {
 279             table.mpIndex[high] = new UnicodePairWithFlag*[256];
 280
 281             for( j = 0; j < 256; j++ )
 282                 table.mpIndex[high][j] = NULL;
 283         }
 284         table.mpIndex[high][low] = &half2fullJISException[i];
 285     }
 286
 287     return table;
 288 }
 289
 290 oneToOneMapping& widthfolding::getfullKana2halfKanaTable()
 291 {
 292     static oneToOneMappingWithFlag table(full2half, sizeof(full2half), FULL2HALF_KATAKANA_ONLY);
 293     table.makeIndex();
 294     return table;
 295 }
 296
 297 oneToOneMapping& widthfolding::gethalfKana2fullKanaTable()
 298 {
 299     static oneToOneMappingWithFlag table(half2full, sizeof(half2full), HALF2FULL_KATAKANA_ONLY);
 300     table.makeIndex();
 301     return table;
 302 }
 303
 304 } } } }