i18npool/source/transliteration/ignoreProlongedSoundMark_ja_JP.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: ignoreProlongedSoundMark_ja_JP.cxx,v $
  10  * $Revision: 1.10 $
  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 // MARKER(update_precomp.py): autogen include statement, do not remove
  32 #include "precompiled_i18npool.hxx"
  33
  34 // prevent internal compiler error with MSVC6SP3
  35 #include <utility>
  36
  37 #define TRANSLITERATION_ProlongedSoundMark_ja_JP
  38 #include <transliteration_Ignore.hxx>
  39
  40 using namespace com::sun::star::uno;
  41 using namespace com::sun::star::lang;
  42 using namespace rtl;
  43
  44 namespace com { namespace sun { namespace star { namespace i18n {
  45
  46 static sal_Unicode table_normalwidth[] = {
  47     //  0x0000,   // 0x3040
  48     0x3041,       // 0x3041 HIRAGANA LETTER SMALL A
  49     0x3042,       // 0x3042 HIRAGANA LETTER A
  50     0x3043,       // 0x3043 HIRAGANA LETTER SMALL I
  51     0x3044,       // 0x3044 HIRAGANA LETTER I
  52     0x3045,       // 0x3045 HIRAGANA LETTER SMALL U
  53     0x3046,       // 0x3046 HIRAGANA LETTER U
  54     0x3047,       // 0x3047 HIRAGANA LETTER SMALL E
  55     0x3048,       // 0x3048 HIRAGANA LETTER E
  56     0x3049,       // 0x3049 HIRAGANA LETTER SMALL O
  57     0x304a,       // 0x304a HIRAGANA LETTER O
  58     0x3042,       // 0x304b HIRAGANA LETTER KA
  59     0x3042,       // 0x304c HIRAGANA LETTER GA
  60     0x3044,       // 0x304d HIRAGANA LETTER KI
  61     0x3044,       // 0x304e HIRAGANA LETTER GI
  62     0x3046,       // 0x304f HIRAGANA LETTER KU
  63     0x3046,       // 0x3050 HIRAGANA LETTER GU
  64     0x3048,       // 0x3051 HIRAGANA LETTER KE
  65     0x3048,       // 0x3052 HIRAGANA LETTER GE
  66     0x304a,       // 0x3053 HIRAGANA LETTER KO
  67     0x304a,       // 0x3054 HIRAGANA LETTER GO
  68     0x3042,       // 0x3055 HIRAGANA LETTER SA
  69     0x3042,       // 0x3056 HIRAGANA LETTER ZA
  70     0x3044,       // 0x3057 HIRAGANA LETTER SI
  71     0x3044,       // 0x3058 HIRAGANA LETTER ZI
  72     0x3046,       // 0x3059 HIRAGANA LETTER SU
  73     0x3046,       // 0x305a HIRAGANA LETTER ZU
  74     0x3048,       // 0x305b HIRAGANA LETTER SE
  75     0x3048,       // 0x305c HIRAGANA LETTER ZE
  76     0x304a,       // 0x305d HIRAGANA LETTER SO
  77     0x304a,       // 0x305e HIRAGANA LETTER ZO
  78     0x3042,       // 0x305f HIRAGANA LETTER TA
  79     0x3042,       // 0x3060 HIRAGANA LETTER DA
  80     0x3044,       // 0x3061 HIRAGANA LETTER TI
  81     0x3044,       // 0x3062 HIRAGANA LETTER DI
  82     0x3045,       // 0x3063 HIRAGANA LETTER SMALL TU
  83     0x3046,       // 0x3064 HIRAGANA LETTER TU
  84     0x3046,       // 0x3065 HIRAGANA LETTER DU
  85     0x3048,       // 0x3066 HIRAGANA LETTER TE
  86     0x3048,       // 0x3067 HIRAGANA LETTER DE
  87     0x304a,       // 0x3068 HIRAGANA LETTER TO
  88     0x304a,       // 0x3069 HIRAGANA LETTER DO
  89     0x3042,       // 0x306a HIRAGANA LETTER NA
  90     0x3044,       // 0x306b HIRAGANA LETTER NI
  91     0x3046,       // 0x306c HIRAGANA LETTER NU
  92     0x3048,       // 0x306d HIRAGANA LETTER NE
  93     0x304a,       // 0x306e HIRAGANA LETTER NO
  94     0x3042,       // 0x306f HIRAGANA LETTER HA
  95     0x3042,       // 0x3070 HIRAGANA LETTER BA
  96     0x3042,       // 0x3071 HIRAGANA LETTER PA
  97     0x3044,       // 0x3072 HIRAGANA LETTER HI
  98     0x3044,       // 0x3073 HIRAGANA LETTER BI
  99     0x3044,       // 0x3074 HIRAGANA LETTER PI
 100     0x3046,       // 0x3075 HIRAGANA LETTER HU
 101     0x3046,       // 0x3076 HIRAGANA LETTER BU
 102     0x3046,       // 0x3077 HIRAGANA LETTER PU
 103     0x3048,       // 0x3078 HIRAGANA LETTER HE
 104     0x3048,       // 0x3079 HIRAGANA LETTER BE
 105     0x3048,       // 0x307a HIRAGANA LETTER PE
 106     0x304a,       // 0x307b HIRAGANA LETTER HO
 107     0x304a,       // 0x307c HIRAGANA LETTER BO
 108     0x304a,       // 0x307d HIRAGANA LETTER PO
 109     0x3042,       // 0x307e HIRAGANA LETTER MA
 110     0x3044,       // 0x307f HIRAGANA LETTER MI
 111     0x3046,       // 0x3080 HIRAGANA LETTER MU
 112     0x3048,       // 0x3081 HIRAGANA LETTER ME
 113     0x304a,       // 0x3082 HIRAGANA LETTER MO
 114     0x3041,       // 0x3083 HIRAGANA LETTER SMALL YA
 115     0x3042,       // 0x3084 HIRAGANA LETTER YA
 116     0x3045,       // 0x3085 HIRAGANA LETTER SMALL YU
 117     0x3046,       // 0x3086 HIRAGANA LETTER YU
 118     0x3049,       // 0x3087 HIRAGANA LETTER SMALL YO
 119     0x304a,       // 0x3088 HIRAGANA LETTER YO
 120     0x3042,       // 0x3089 HIRAGANA LETTER RA
 121     0x3044,       // 0x308a HIRAGANA LETTER RI
 122     0x3046,       // 0x308b HIRAGANA LETTER RU
 123     0x3048,       // 0x308c HIRAGANA LETTER RE
 124     0x304a,       // 0x308d HIRAGANA LETTER RO
 125     0x3041,       // 0x308e HIRAGANA LETTER SMALL WA
 126     0x3042,       // 0x308f HIRAGANA LETTER WA
 127     0x3044,       // 0x3090 HIRAGANA LETTER WI
 128     0x3048,       // 0x3091 HIRAGANA LETTER WE
 129     0x304a,       // 0x3092 HIRAGANA LETTER WO
 130     0x0000,       // 0x3093 HIRAGANA LETTER N
 131     0x3046,       // 0x3094 HIRAGANA LETTER VU
 132     0x0000,       // 0x3095
 133     0x0000,       // 0x3096
 134     0x0000,       // 0x3097
 135     0x0000,       // 0x3098
 136     0x0000,       // 0x3099 COMBINING KATAKANA-HIRAGANA VOICED SOUND MARK
 137     0x0000,       // 0x309a COMBINING KATAKANA-HIRAGANA SEMI-VOICED SOUND MARK
 138     0x0000,       // 0x309b KATAKANA-HIRAGANA VOICED SOUND MARK
 139     0x0000,       // 0x309c KATAKANA-HIRAGANA SEMI-VOICED SOUND MARK
 140     0x0000,       // 0x309d HIRAGANA ITERATION MARK
 141     0x0000,       // 0x309e HIRAGANA VOICED ITERATION MARK
 142     0x0000,       // 0x309f
 143     0x0000,       // 0x30a0
 144     0x30a1,       // 0x30a1 KATAKANA LETTER SMALL A
 145     0x30a2,       // 0x30a2 KATAKANA LETTER A
 146     0x30a3,       // 0x30a3 KATAKANA LETTER SMALL I
 147     0x30a4,       // 0x30a4 KATAKANA LETTER I
 148     0x30a5,       // 0x30a5 KATAKANA LETTER SMALL U
 149     0x30a6,       // 0x30a6 KATAKANA LETTER U
 150     0x30a7,       // 0x30a7 KATAKANA LETTER SMALL E
 151     0x30a8,       // 0x30a8 KATAKANA LETTER E
 152     0x30a9,       // 0x30a9 KATAKANA LETTER SMALL O
 153     0x30aa,       // 0x30aa KATAKANA LETTER O
 154     0x30a2,       // 0x30ab KATAKANA LETTER KA
 155     0x30a2,       // 0x30ac KATAKANA LETTER GA
 156     0x30a4,       // 0x30ad KATAKANA LETTER KI
 157     0x30a4,       // 0x30ae KATAKANA LETTER GI
 158     0x30a6,       // 0x30af KATAKANA LETTER KU
 159     0x30a6,       // 0x30b0 KATAKANA LETTER GU
 160     0x30a8,       // 0x30b1 KATAKANA LETTER KE
 161     0x30a8,       // 0x30b2 KATAKANA LETTER GE
 162     0x30aa,       // 0x30b3 KATAKANA LETTER KO
 163     0x30aa,       // 0x30b4 KATAKANA LETTER GO
 164     0x30a2,       // 0x30b5 KATAKANA LETTER SA
 165     0x30a2,       // 0x30b6 KATAKANA LETTER ZA
 166     0x30a4,       // 0x30b7 KATAKANA LETTER SI
 167     0x30a4,       // 0x30b8 KATAKANA LETTER ZI
 168     0x30a6,       // 0x30b9 KATAKANA LETTER SU
 169     0x30a6,       // 0x30ba KATAKANA LETTER ZU
 170     0x30a8,       // 0x30bb KATAKANA LETTER SE
 171     0x30a8,       // 0x30bc KATAKANA LETTER ZE
 172     0x30aa,       // 0x30bd KATAKANA LETTER SO
 173     0x30aa,       // 0x30be KATAKANA LETTER ZO
 174     0x30a2,       // 0x30bf KATAKANA LETTER TA
 175     0x30a2,       // 0x30c0 KATAKANA LETTER DA
 176     0x30a4,       // 0x30c1 KATAKANA LETTER TI
 177     0x30a4,       // 0x30c2 KATAKANA LETTER DI
 178     0x30a5,       // 0x30c3 KATAKANA LETTER SMALL TU
 179     0x30a6,       // 0x30c4 KATAKANA LETTER TU
 180     0x30a6,       // 0x30c5 KATAKANA LETTER DU
 181     0x30a8,       // 0x30c6 KATAKANA LETTER TE
 182     0x30a8,       // 0x30c7 KATAKANA LETTER DE
 183     0x30aa,       // 0x30c8 KATAKANA LETTER TO
 184     0x30aa,       // 0x30c9 KATAKANA LETTER DO
 185     0x30a2,       // 0x30ca KATAKANA LETTER NA
 186     0x30a4,       // 0x30cb KATAKANA LETTER NI
 187     0x30a6,       // 0x30cc KATAKANA LETTER NU
 188     0x30a8,       // 0x30cd KATAKANA LETTER NE
 189     0x30aa,       // 0x30ce KATAKANA LETTER NO
 190     0x30a2,       // 0x30cf KATAKANA LETTER HA
 191     0x30a2,       // 0x30d0 KATAKANA LETTER BA
 192     0x30a2,       // 0x30d1 KATAKANA LETTER PA
 193     0x30a4,       // 0x30d2 KATAKANA LETTER HI
 194     0x30a4,       // 0x30d3 KATAKANA LETTER BI
 195     0x30a4,       // 0x30d4 KATAKANA LETTER PI
 196     0x30a6,       // 0x30d5 KATAKANA LETTER HU
 197     0x30a6,       // 0x30d6 KATAKANA LETTER BU
 198     0x30a6,       // 0x30d7 KATAKANA LETTER PU
 199     0x30a8,       // 0x30d8 KATAKANA LETTER HE
 200     0x30a8,       // 0x30d9 KATAKANA LETTER BE
 201     0x30a8,       // 0x30da KATAKANA LETTER PE
 202     0x30aa,       // 0x30db KATAKANA LETTER HO
 203     0x30aa,       // 0x30dc KATAKANA LETTER BO
 204     0x30aa,       // 0x30dd KATAKANA LETTER PO
 205     0x30a2,       // 0x30de KATAKANA LETTER MA
 206     0x30a4,       // 0x30df KATAKANA LETTER MI
 207     0x30a6,       // 0x30e0 KATAKANA LETTER MU
 208     0x30a8,       // 0x30e1 KATAKANA LETTER ME
 209     0x30aa,       // 0x30e2 KATAKANA LETTER MO
 210     0x30a1,       // 0x30e3 KATAKANA LETTER SMALL YA
 211     0x30a2,       // 0x30e4 KATAKANA LETTER YA
 212     0x30a5,       // 0x30e5 KATAKANA LETTER SMALL YU
 213     0x30a6,       // 0x30e6 KATAKANA LETTER YU
 214     0x30a9,       // 0x30e7 KATAKANA LETTER SMALL YO
 215     0x30aa,       // 0x30e8 KATAKANA LETTER YO
 216     0x30a2,       // 0x30e9 KATAKANA LETTER RA
 217     0x30a4,       // 0x30ea KATAKANA LETTER RI
 218     0x30a6,       // 0x30eb KATAKANA LETTER RU
 219     0x30a8,       // 0x30ec KATAKANA LETTER RE
 220     0x30aa,       // 0x30ed KATAKANA LETTER RO
 221     0x30a1,       // 0x30ee KATAKANA LETTER SMALL WA
 222     0x30a2,       // 0x30ef KATAKANA LETTER WA
 223     0x30a4,       // 0x30f0 KATAKANA LETTER WI
 224     0x30a8,       // 0x30f1 KATAKANA LETTER WE
 225     0x30aa,       // 0x30f2 KATAKANA LETTER WO
 226     0x0000,       // 0x30f3 KATAKANA LETTER N
 227     0x30a6,       // 0x30f4 KATAKANA LETTER VU
 228     0x30a1,       // 0x30f5 KATAKANA LETTER SMALL KA
 229     0x30a7,       // 0x30f6 KATAKANA LETTER SMALL KE
 230     0x30a2,       // 0x30f7 KATAKANA LETTER VA
 231     0x30a4,       // 0x30f8 KATAKANA LETTER VI
 232     0x30a8,       // 0x30f9 KATAKANA LETTER VE
 233     0x30aa        // 0x30fa KATAKANA LETTER VO
 234     //  0x0000,   // 0x30fb KATAKANA MIDDLE DOT
 235     //  0x0000,   // 0x30fc KATAKANA-HIRAGANA PROLONGED SOUND MARK
 236     //  0x0000,   // 0x30fd KATAKANA ITERATION MARK
 237     //  0x0000,   // 0x30fe KATAKANA VOICED ITERATION MARK
 238     //  0x0000    // 0x30ff
 239 };
 240
 241 static sal_Unicode table_halfwidth[] = {
 242     //  0x0000,   // 0xff61 HALFWIDTH IDEOGRAPHIC FULL STOP
 243     //  0x0000,   // 0xff62 HALFWIDTH LEFT CORNER BRACKET
 244     //  0x0000,   // 0xff63 HALFWIDTH RIGHT CORNER BRACKET
 245     //  0x0000,   // 0xff64 HALFWIDTH IDEOGRAPHIC COMMA
 246     //  0x0000,   // 0xff65 HALFWIDTH KATAKANA MIDDLE DOT
 247     0xff75,       // 0xff66 HALFWIDTH KATAKANA LETTER WO
 248     0xff67,       // 0xff67 HALFWIDTH KATAKANA LETTER SMALL A
 249     0xff68,       // 0xff68 HALFWIDTH KATAKANA LETTER SMALL I
 250     0xff69,       // 0xff69 HALFWIDTH KATAKANA LETTER SMALL U
 251     0xff6a,       // 0xff6a HALFWIDTH KATAKANA LETTER SMALL E
 252     0xff6b,       // 0xff6b HALFWIDTH KATAKANA LETTER SMALL O
 253     0xff67,       // 0xff6c HALFWIDTH KATAKANA LETTER SMALL YA
 254     0xff69,       // 0xff6d HALFWIDTH KATAKANA LETTER SMALL YU
 255     0xff6b,       // 0xff6e HALFWIDTH KATAKANA LETTER SMALL YO
 256     0xff69,       // 0xff6f HALFWIDTH KATAKANA LETTER SMALL TU
 257     0x0000,       // 0xff70 HALFWIDTH KATAKANA-HIRAGANA PROLONGED SOUND MARK
 258     0xff71,       // 0xff71 HALFWIDTH KATAKANA LETTER A
 259     0xff72,       // 0xff72 HALFWIDTH KATAKANA LETTER I
 260     0xff73,       // 0xff73 HALFWIDTH KATAKANA LETTER U
 261     0xff74,       // 0xff74 HALFWIDTH KATAKANA LETTER E
 262     0xff75,       // 0xff75 HALFWIDTH KATAKANA LETTER O
 263     0xff71,       // 0xff76 HALFWIDTH KATAKANA LETTER KA
 264     0xff72,       // 0xff77 HALFWIDTH KATAKANA LETTER KI
 265     0xff73,       // 0xff78 HALFWIDTH KATAKANA LETTER KU
 266     0xff74,       // 0xff79 HALFWIDTH KATAKANA LETTER KE
 267     0xff75,       // 0xff7a HALFWIDTH KATAKANA LETTER KO
 268     0xff71,       // 0xff7b HALFWIDTH KATAKANA LETTER SA
 269     0xff72,       // 0xff7c HALFWIDTH KATAKANA LETTER SI
 270     0xff73,       // 0xff7d HALFWIDTH KATAKANA LETTER SU
 271     0xff74,       // 0xff7e HALFWIDTH KATAKANA LETTER SE
 272     0xff75,       // 0xff7f HALFWIDTH KATAKANA LETTER SO
 273     0xff71,       // 0xff80 HALFWIDTH KATAKANA LETTER TA
 274     0xff72,       // 0xff81 HALFWIDTH KATAKANA LETTER TI
 275     0xff73,       // 0xff82 HALFWIDTH KATAKANA LETTER TU
 276     0xff74,       // 0xff83 HALFWIDTH KATAKANA LETTER TE
 277     0xff75,       // 0xff84 HALFWIDTH KATAKANA LETTER TO
 278     0xff71,       // 0xff85 HALFWIDTH KATAKANA LETTER NA
 279     0xff72,       // 0xff86 HALFWIDTH KATAKANA LETTER NI
 280     0xff73,       // 0xff87 HALFWIDTH KATAKANA LETTER NU
 281     0xff74,       // 0xff88 HALFWIDTH KATAKANA LETTER NE
 282     0xff75,       // 0xff89 HALFWIDTH KATAKANA LETTER NO
 283     0xff71,       // 0xff8a HALFWIDTH KATAKANA LETTER HA
 284     0xff72,       // 0xff8b HALFWIDTH KATAKANA LETTER HI
 285     0xff73,       // 0xff8c HALFWIDTH KATAKANA LETTER HU
 286     0xff74,       // 0xff8d HALFWIDTH KATAKANA LETTER HE
 287     0xff75,       // 0xff8e HALFWIDTH KATAKANA LETTER HO
 288     0xff71,       // 0xff8f HALFWIDTH KATAKANA LETTER MA
 289     0xff72,       // 0xff90 HALFWIDTH KATAKANA LETTER MI
 290     0xff73,       // 0xff91 HALFWIDTH KATAKANA LETTER MU
 291     0xff74,       // 0xff92 HALFWIDTH KATAKANA LETTER ME
 292     0xff75,       // 0xff93 HALFWIDTH KATAKANA LETTER MO
 293     0xff71,       // 0xff94 HALFWIDTH KATAKANA LETTER YA
 294     0xff73,       // 0xff95 HALFWIDTH KATAKANA LETTER YU
 295     0xff75,       // 0xff96 HALFWIDTH KATAKANA LETTER YO
 296     0xff71,       // 0xff97 HALFWIDTH KATAKANA LETTER RA
 297     0xff72,       // 0xff98 HALFWIDTH KATAKANA LETTER RI
 298     0xff73,       // 0xff99 HALFWIDTH KATAKANA LETTER RU
 299     0xff74,       // 0xff9a HALFWIDTH KATAKANA LETTER RE
 300     0xff75,       // 0xff9b HALFWIDTH KATAKANA LETTER RO
 301     0xff71        // 0xff9c HALFWIDTH KATAKANA LETTER WA
 302     //  0x0000,   // 0xff9d HALFWIDTH KATAKANA LETTER N
 303     //  0x0000,   // 0xff9e HALFWIDTH KATAKANA VOICED SOUND MARK
 304     //  0x0000    // 0xff9f HALFWIDTH KATAKANA SEMI-VOICED SOUND MARK
 305 };
 306
 307
 308 OUString SAL_CALL
 309 ignoreProlongedSoundMark_ja_JP::folding( const OUString& inStr, sal_Int32 startPos, sal_Int32 nCount, Sequence< sal_Int32 >& offset )
 310   throw(RuntimeException)
 311 {
 312     // Create a string buffer which can hold nCount + 1 characters.
 313     // The reference count is 0 now.
 314     rtl_uString * newStr = x_rtl_uString_new_WithLength( nCount ); // defined in x_rtl_ustring.h
 315     sal_Unicode * dst = newStr->buffer;
 316     const sal_Unicode * src = inStr.getStr() + startPos;
 317
 318     sal_Int32 *p = 0;
 319     sal_Int32 position = 0;
 320
 321     if (useOffset) {
 322         // Allocate nCount length to offset argument.
 323         offset.realloc( nCount );
 324         p = offset.getArray();
 325         position = startPos;
 326     }
 327
 328     //
 329     sal_Unicode previousChar = *src ++;
 330     sal_Unicode currentChar;
 331
 332     // Conversion
 333     while (-- nCount > 0) {
 334         currentChar = *src ++;
 335
 336         if (currentChar == 0x30fc || // KATAKANA-HIRAGANA PROLONGED SOUND MARK
 337                 currentChar == 0xff70) { // HALFWIDTH KATAKANA-HIRAGANA PROLONGED SOUND MARK
 338
 339             if (0x3041 <= previousChar && previousChar <= 0x30fa) {
 340                 currentChar = table_normalwidth[ previousChar - 0x3041 ];
 341             }
 342             else if (0xff66 <= previousChar && previousChar <= 0xff9c) {
 343                 currentChar = table_halfwidth[ previousChar - 0xff66 ];
 344             }
 345         }
 346
 347         if (useOffset)
 348             *p ++ = position ++;
 349         *dst ++ = previousChar;
 350         previousChar = currentChar;
 351     }
 352
 353     if (nCount == 0) {
 354         if (useOffset)
 355             *p = position;
 356         *dst ++ = previousChar;
 357     }
 358
 359     *dst = (sal_Unicode) 0;
 360
 361     newStr->length = sal_Int32(dst - newStr->buffer);
 362     if (useOffset)
 363         offset.realloc(newStr->length);
 364     return OUString( newStr ); // defined in rtl/usrting. The reference count is increased from 0 to 1.
 365
 366 }
 367
 368 } } } }