i18npool/source/textconversion/textconversion_ko.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: textconversion_ko.cxx,v $
  10  * $Revision: 1.15.22.1 $
  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 #include <assert.h>
  34 #include <textconversion.hxx>
  35 #include <com/sun/star/i18n/TextConversionType.hpp>
  36 #include <com/sun/star/i18n/TextConversionOption.hpp>
  37 #include <com/sun/star/linguistic2/ConversionDirection.hpp>
  38 #include <com/sun/star/linguistic2/ConversionDictionaryType.hpp>
  39 #include <rtl/ustrbuf.hxx>
  40 #include <i18nutil/x_rtl_ustring.h>
  41 #include <unicode/uchar.h>
  42
  43 using namespace com::sun::star::lang;
  44 using namespace com::sun::star::i18n;
  45 using namespace com::sun::star::linguistic2;
  46 using namespace com::sun::star::uno;
  47 using namespace rtl;
  48
  49 namespace com { namespace sun { namespace star { namespace i18n {
  50
  51 #define SCRIPT_OTHERS   0
  52 #define SCRIPT_HANJA    1
  53 #define SCRIPT_HANGUL   2
  54
  55 TextConversion_ko::TextConversion_ko( const Reference < XMultiServiceFactory >& xMSF )
  56 {
  57     Reference < XInterface > xI;
  58
  59     xI = xMSF->createInstance(
  60         OUString::createFromAscii("com.sun.star.i18n.ConversionDictionary_ko"));
  61
  62     if ( xI.is() )
  63         xI->queryInterface( getCppuType((const Reference< XConversionDictionary>*)0) ) >>= xCD;
  64
  65     xI = xMSF->createInstance(
  66         OUString::createFromAscii( "com.sun.star.linguistic2.ConversionDictionaryList" ));
  67
  68     if ( xI.is() )
  69         xI->queryInterface( getCppuType((const Reference< XConversionDictionaryList>*)0) ) >>= xCDL;
  70
  71     maxLeftLength = maxRightLength = 1;
  72
  73     // get maximum length of word in dictionary
  74     if (xCDL.is()) {
  75         Locale loc(OUString::createFromAscii("ko"),
  76                     OUString::createFromAscii("KR"),
  77                     OUString());
  78         maxLeftLength = xCDL->queryMaxCharCount(loc,
  79                         ConversionDictionaryType::HANGUL_HANJA,
  80                         ConversionDirection_FROM_LEFT);
  81         maxRightLength = xCDL->queryMaxCharCount(loc,
  82                         ConversionDictionaryType::HANGUL_HANJA,
  83                         ConversionDirection_FROM_RIGHT);
  84         if (xCD.is()) {
  85             sal_Int32 tmp = xCD->getMaxCharCount(ConversionDirection_FROM_LEFT);
  86             if (tmp > maxLeftLength)
  87                 maxLeftLength = tmp;
  88             tmp = xCD->getMaxCharCount(ConversionDirection_FROM_RIGHT);
  89             if (tmp > maxRightLength)
  90                 maxRightLength = tmp;
  91         }
  92     } else if (xCD.is()) {
  93         maxLeftLength = xCD->getMaxCharCount(ConversionDirection_FROM_LEFT);
  94         maxRightLength = xCD->getMaxCharCount(ConversionDirection_FROM_RIGHT);
  95     }
  96
  97     implementationName = "com.sun.star.i18n.TextConversion_ko";
  98 }
  99
 100 sal_Int16 SAL_CALL checkScriptType(sal_Unicode c)
 101 {
 102     typedef struct {
 103         UBlockCode from;
 104         UBlockCode to;
 105         sal_Int16 script;
 106     } UBlock2Script;
 107
 108     static UBlock2Script scriptList[] = {
 109         {UBLOCK_HANGUL_JAMO, UBLOCK_HANGUL_JAMO, SCRIPT_HANGUL},
 110         {UBLOCK_CJK_RADICALS_SUPPLEMENT, UBLOCK_BOPOMOFO, SCRIPT_HANJA},
 111         {UBLOCK_HANGUL_COMPATIBILITY_JAMO, UBLOCK_HANGUL_COMPATIBILITY_JAMO, SCRIPT_HANGUL},
 112         {UBLOCK_KANBUN, UBLOCK_YI_RADICALS, SCRIPT_HANJA},
 113         {UBLOCK_HANGUL_SYLLABLES, UBLOCK_HANGUL_SYLLABLES, SCRIPT_HANGUL},
 114         {UBLOCK_CJK_COMPATIBILITY_IDEOGRAPHS, UBLOCK_CJK_COMPATIBILITY_IDEOGRAPHS, SCRIPT_HANJA},
 115         {UBLOCK_COMBINING_HALF_MARKS, UBLOCK_SMALL_FORM_VARIANTS, SCRIPT_HANJA},
 116         {UBLOCK_HALFWIDTH_AND_FULLWIDTH_FORMS, UBLOCK_HALFWIDTH_AND_FULLWIDTH_FORMS, SCRIPT_HANJA},
 117     };
 118
 119 #define scriptListCount sizeof (scriptList) / sizeof (UBlock2Script)
 120
 121     UBlockCode block=ublock_getCode((sal_uInt32) c);
 122     sal_uInt16 i;
 123     for ( i = 0; i < scriptListCount; i++) {
 124         if (block <= scriptList[i].to) break;
 125     }
 126     return (i < scriptListCount && block >= scriptList[i].from) ? scriptList[i].script : SCRIPT_OTHERS;
 127 }
 128
 129 Sequence< OUString > SAL_CALL
 130 TextConversion_ko::getCharConversions(const OUString& aText, sal_Int32 nStartPos, sal_Int32 nLength, sal_Bool toHanja)
 131 {
 132     sal_Unicode ch;
 133     Sequence< OUString > output;
 134     const sal_Unicode* (*getHangul2HanjaData)() = (const sal_Unicode* (*)())getFunctionBySymbol("getHangul2HanjaData");
 135     const Hangul_Index* (*getHangul2HanjaIndex)() = (const Hangul_Index* (*)()) getFunctionBySymbol("getHangul2HanjaIndex");
 136     sal_Int16 (*getHangul2HanjaIndexCount)() = (sal_Int16 (*)()) getFunctionBySymbol("getHangul2HanjaIndexCount");
 137     const sal_uInt16* (*getHanja2HangulIndex)() = (const sal_uInt16* (*)()) getFunctionBySymbol("getHanja2HangulIndex");
 138     const sal_Unicode* (*getHanja2HangulData)() = (const sal_Unicode* (*)()) getFunctionBySymbol("getHanja2HangulData");
 139     if (toHanja && getHangul2HanjaIndex && getHangul2HanjaIndexCount && getHangul2HanjaData) {
 140         ch = aText[nStartPos];
 141         const Hangul_Index *Hangul_ko = getHangul2HanjaIndex();
 142         sal_Int16 top =  getHangul2HanjaIndexCount();
 143         --top;
 144         sal_Int16 bottom = 0;
 145
 146         while (bottom <= top) {
 147             sal_Int16 current = (top + bottom) / 2;
 148             sal_Unicode current_ch = Hangul_ko[current].code;
 149             if (ch < current_ch)
 150                 top = current - 1;
 151             else if (ch > current_ch)
 152                 bottom = current + 1;
 153             else {
 154                 const sal_Unicode *ptr = getHangul2HanjaData() + Hangul_ko[current].address;
 155                 sal_Int16 count = Hangul_ko[current].count;
 156                 output.realloc(count);
 157                 for (sal_Int16 i = 0; i < count; i++)
 158                     output[i] = OUString(ptr + i, 1);
 159                 break;
 160             }
 161         }
 162     } else if (! toHanja && getHanja2HangulIndex && getHanja2HangulData) {
 163         rtl_uString * newStr = x_rtl_uString_new_WithLength( nLength ); // defined in x_rtl_ustring.h
 164         sal_Int32 count = 0;
 165         while (count < nLength) {
 166             ch = aText[nStartPos + count];
 167             sal_Unicode address = getHanja2HangulIndex()[ch>>8];
 168             if (address != 0xFFFF)
 169                 address = getHanja2HangulData()[address + (ch & 0xFF)];
 170
 171             if (address != 0xFFFF)
 172                 newStr->buffer[count++] = address;
 173             else
 174                 break;
 175         }
 176         if (count > 0) {
 177             output.realloc(1);
 178             output[0] = OUString( newStr->buffer, count);
 179         }
 180     }
 181     return output;
 182 }
 183
 184 static Sequence< OUString >& operator += (Sequence< OUString > &rSeq1, Sequence< OUString > &rSeq2 )
 185 {
 186     if (! rSeq1.hasElements() && rSeq2.hasElements())
 187         rSeq1 = rSeq2;
 188     else if (rSeq2.hasElements()) {
 189         sal_Int32 i, j, k, l;
 190         k = l = rSeq1.getLength();
 191         rSeq1.realloc(l + rSeq2.getLength());
 192
 193         for (i = 0; i < rSeq2.getLength(); i++) {
 194             for (j = 0; j < l; j++)
 195                 if (rSeq1[j] == rSeq2[i])
 196                     break;
 197             if (j == l)
 198                 rSeq1[k++] = rSeq2[i];
 199         }
 200         if (rSeq1.getLength() > k)
 201             rSeq1.realloc(k);
 202     }
 203     return rSeq1;
 204 }
 205
 206 TextConversionResult SAL_CALL
 207 TextConversion_ko::getConversions( const OUString& aText, sal_Int32 nStartPos, sal_Int32 nLength,
 208     const Locale& aLocale, sal_Int16 nConversionType, sal_Int32 nConversionOptions)
 209     throw(  RuntimeException, IllegalArgumentException, NoSupportException )
 210 {
 211     TextConversionResult result;
 212     Sequence <OUString> candidates;
 213     result.Boundary.startPos = result.Boundary.endPos = 0;
 214
 215     // do conversion only when there are right conversion type and dictionary services.
 216     if (nConversionType == TextConversionType::TO_HANGUL ||
 217             nConversionType == TextConversionType::TO_HANJA) {
 218         sal_Int32 start, end, length = aText.getLength() - nStartPos;
 219
 220         if (length < 0 || nStartPos < 0)
 221             length = 0;
 222         else if (length > nLength)
 223             length = nLength;
 224
 225         sal_Int16 scriptType = SCRIPT_OTHERS;
 226         sal_Int32 len = 1;
 227         sal_Bool toHanja = (nConversionType == TextConversionType::TO_HANJA);
 228         // FROM_LEFT:  Hangul -> Hanja
 229         // FROM_RIGHT: Hanja  -> Hangul
 230         ConversionDirection eDirection = toHanja ? ConversionDirection_FROM_LEFT : ConversionDirection_FROM_RIGHT;
 231         sal_Int32 maxLength = toHanja ? maxLeftLength : maxRightLength;
 232         if (maxLength == 0) maxLength = 1;
 233
 234         // search for a max length of convertible text
 235         for (start = 0, end = 0; start < length; start++) {
 236             if (end <= start) {
 237                 scriptType = checkScriptType(aText[nStartPos + start]);
 238                 if (nConversionType == TextConversionType::TO_HANJA) {
 239                     if (scriptType != SCRIPT_HANGUL) // skip non-Hangul characters
 240                         continue;
 241                 } else {
 242                     if (scriptType != SCRIPT_HANJA) // skip non-Hanja characters
 243                         continue;
 244                 }
 245                 end = start + 1;
 246             }
 247             if (nConversionOptions & TextConversionOption::CHARACTER_BY_CHARACTER) {
 248                 result.Candidates = getCharConversions(aText, nStartPos + start, len, toHanja); // char2char conversion
 249             } else {
 250                 for (; end < length && end - start < maxLength; end++)
 251                     if (checkScriptType(aText[nStartPos + end]) != scriptType)
 252                         break;
 253
 254                 for (len = end - start; len > 0; len--) {
 255                     if (len > 1) {
 256                         try {
 257                             if (xCDL.is())
 258                                 result.Candidates = xCDL->queryConversions(aText, start + nStartPos, len,
 259                                     aLocale, ConversionDictionaryType::HANGUL_HANJA, eDirection, nConversionOptions); // user dictionary
 260                         }
 261                         catch ( NoSupportException & ) {
 262                             // clear reference (when there is no user dictionary) in order
 263                             // to not always have to catch this exception again
 264                             // in further calls. (save time)
 265                             xCDL = 0;
 266                         }
 267                         catch (...) {
 268                             // catch all other exceptions to allow
 269                             // querying the system dictionary in the next line
 270                         }
 271                         if (xCD.is() && toHanja) { // System dictionary would not do Hanja_to_Hangul conversion.
 272                             candidates = xCD->getConversions(aText, start + nStartPos, len, eDirection, nConversionOptions);
 273                             result.Candidates += candidates;
 274                         }
 275                     } else if (! toHanja) { // do whole word character 2 character conversion for Hanja to Hangul conversion
 276                         result.Candidates = getCharConversions(aText, nStartPos + start, length - start, toHanja);
 277                         if (result.Candidates.hasElements())
 278                             len = result.Candidates[0].getLength();
 279                     }
 280                     if (result.Candidates.hasElements())
 281                         break;
 282                 }
 283             }
 284             // found match
 285             if (result.Candidates.hasElements()) {
 286                 result.Boundary.startPos = start + nStartPos;;
 287                 result.Boundary.endPos = start + len + nStartPos;
 288                 return result;
 289             }
 290         }
 291     } else
 292         throw NoSupportException(); // Conversion type is not supported in this service.
 293     return result;
 294 }
 295
 296 OUString SAL_CALL
 297 TextConversion_ko::getConversion( const OUString& aText, sal_Int32 nStartPos, sal_Int32 nLength,
 298     const Locale& aLocale, sal_Int16 nConversionType, sal_Int32 nConversionOptions)
 299     throw(  RuntimeException, IllegalArgumentException, NoSupportException )
 300 {
 301     sal_Int32 length = aText.getLength() - nStartPos;
 302
 303     if (length <= 0 || nStartPos < 0)
 304         return OUString();
 305     else if (length > nLength)
 306         length = nLength;
 307
 308     OUStringBuffer aBuf(length + 1);
 309     TextConversionResult result;
 310     const sal_Unicode *str = aText.getStr();
 311
 312     for (sal_Int32 start = nStartPos; length + nStartPos > start; start = result.Boundary.endPos) {
 313
 314         result = getConversions(aText, start, length + nStartPos - start, aLocale, nConversionType, nConversionOptions);
 315
 316         if (result.Boundary.endPos > 0) {
 317             if (result.Boundary.startPos > start)
 318                 aBuf.append(str + start, result.Boundary.startPos - start); // append skip portion
 319             aBuf.append(result.Candidates[0]); // append converted portion
 320         } else {
 321             if (length + nStartPos > start)
 322                 aBuf.append(str + start, length + nStartPos - start); // append last portion
 323             break;
 324         }
 325     }
 326
 327     return aBuf.makeStringAndClear();
 328 }
 329
 330 OUString SAL_CALL
 331 TextConversion_ko::getConversionWithOffset( const OUString& aText, sal_Int32 nStartPos, sal_Int32 nLength,
 332     const Locale& rLocale, sal_Int16 nConversionType, sal_Int32 nConversionOptions, Sequence<sal_Int32>& offset)
 333     throw(  RuntimeException, IllegalArgumentException, NoSupportException )
 334 {
 335     offset.realloc(0);
 336     return getConversion(aText, nStartPos, nLength, rLocale, nConversionType, nConversionOptions);
 337 }
 338
 339 sal_Bool SAL_CALL
 340 TextConversion_ko::interactiveConversion( const Locale& /*rLocale*/, sal_Int16 /*nTextConversionType*/, sal_Int32 /*nTextConversionOptions*/ )
 341     throw(  RuntimeException, IllegalArgumentException, NoSupportException )
 342 {
 343     return sal_True;
 344 }
 345
 346 } } } }