i18npool/source/characterclassification/cclass_unicode.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: cclass_unicode.cxx,v $
  10  * $Revision: 1.13 $
  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 #include <cclass_unicode.hxx>
  35 #include <com/sun/star/i18n/UnicodeScript.hpp>
  36 #include <com/sun/star/i18n/UnicodeType.hpp>
  37 #include <com/sun/star/i18n/KCharacterType.hpp>
  38 #include <unicode/uchar.h>
  39 #include <i18nutil/x_rtl_ustring.h>
  40 #include <breakiteratorImpl.hxx>
  41
  42 using namespace ::com::sun::star::uno;
  43 using namespace ::com::sun::star::lang;
  44 using namespace ::rtl;
  45
  46 namespace com { namespace sun { namespace star { namespace i18n {
  47 //      ----------------------------------------------------
  48 //      class cclass_Unicode
  49 //      ----------------------------------------------------;
  50
  51 cclass_Unicode::cclass_Unicode( uno::Reference < XMultiServiceFactory > xSMgr ) : xMSF( xSMgr ),
  52         pTable( NULL ),
  53         pStart( NULL ),
  54         pCont( NULL ),
  55         nStartTypes( 0 ),
  56         nContTypes( 0 ),
  57         eState( ssGetChar ),
  58         cGroupSep( ',' ),
  59         cDecimalSep( '.' )
  60 {
  61     trans = new Transliteration_casemapping();
  62     cClass = "com.sun.star.i18n.CharacterClassification_Unicode";
  63 }
  64
  65 cclass_Unicode::~cclass_Unicode() {
  66     destroyParserTable();
  67     delete trans;
  68 }
  69
  70
  71 OUString SAL_CALL
  72 cclass_Unicode::toUpper( const OUString& Text, sal_Int32 nPos, sal_Int32 nCount, const Locale& rLocale ) throw(RuntimeException) {
  73     sal_Int32 len = Text.getLength();
  74     if (nPos >= len)
  75         return OUString();
  76     if (nCount + nPos > len)
  77         nCount = len - nPos;
  78
  79     trans->setMappingType(MappingTypeToUpper, rLocale);
  80     return trans->transliterateString2String(Text, nPos, nCount);
  81 }
  82
  83 OUString SAL_CALL
  84 cclass_Unicode::toLower( const OUString& Text, sal_Int32 nPos, sal_Int32 nCount, const Locale& rLocale ) throw(RuntimeException) {
  85     sal_Int32 len = Text.getLength();
  86     if (nPos >= len)
  87         return OUString();
  88     if (nCount + nPos > len)
  89         nCount = len - nPos;
  90
  91     trans->setMappingType(MappingTypeToLower, rLocale);
  92     return trans->transliterateString2String(Text, nPos, nCount);
  93 }
  94
  95 OUString SAL_CALL
  96 cclass_Unicode::toTitle( const OUString& Text, sal_Int32 nPos, sal_Int32 nCount, const Locale& rLocale ) throw(RuntimeException) {
  97     sal_Int32 len = Text.getLength();
  98     if (nPos >= len)
  99         return OUString();
 100     if (nCount + nPos > len)
 101         nCount = len - nPos;
 102
 103     trans->setMappingType(MappingTypeToTitle, rLocale);
 104     rtl_uString* pStr = x_rtl_uString_new_WithLength( nCount, 1 );
 105     sal_Unicode* out = pStr->buffer;
 106     BreakIteratorImpl brk(xMSF);
 107     Boundary bdy = brk.getWordBoundary(Text, nPos, rLocale,
 108                 WordType::ANYWORD_IGNOREWHITESPACES, sal_True);
 109     for (sal_Int32 i = nPos; i < nCount + nPos; i++, out++) {
 110         if (i >= bdy.endPos)
 111             bdy = brk.nextWord(Text, bdy.endPos, rLocale,
 112                         WordType::ANYWORD_IGNOREWHITESPACES);
 113         *out = (i == bdy.startPos) ?
 114             trans->transliterateChar2Char(Text[i]) : Text[i];
 115     }
 116     *out = 0;
 117     return OUString( pStr, SAL_NO_ACQUIRE );
 118 }
 119
 120 sal_Int16 SAL_CALL
 121 cclass_Unicode::getType( const OUString& Text, sal_Int32 nPos ) throw(RuntimeException) {
 122     if ( nPos < 0 || Text.getLength() <= nPos ) return 0;
 123     return (sal_Int16) u_charType(Text.iterateCodePoints(&nPos, 0));
 124 }
 125
 126 sal_Int16 SAL_CALL
 127 cclass_Unicode::getCharacterDirection( const OUString& Text, sal_Int32 nPos ) throw(RuntimeException) {
 128     if ( nPos < 0 || Text.getLength() <= nPos ) return 0;
 129     return (sal_Int16) u_charDirection(Text.iterateCodePoints(&nPos, 0));
 130 }
 131
 132
 133 sal_Int16 SAL_CALL
 134 cclass_Unicode::getScript( const OUString& Text, sal_Int32 nPos ) throw(RuntimeException) {
 135     if ( nPos < 0 || Text.getLength() <= nPos ) return 0;
 136     // ICU Unicode script type UBlockCode starts from 1 for Basci Latin,
 137     // while OO.o enum UnicideScript starts from 0.
 138     // To map ICU UBlockCode to OO.o UnicodeScript, it needs to shift 1.
 139     return (sal_Int16) ublock_getCode(Text.iterateCodePoints(&nPos, 0))-1;
 140 }
 141
 142
 143 sal_Int32 SAL_CALL
 144 cclass_Unicode::getCharType( const OUString& Text, sal_Int32* nPos, sal_Int32 increment) {
 145     using namespace ::com::sun::star::i18n::KCharacterType;
 146
 147     sal_uInt32 ch = Text.iterateCodePoints(nPos, increment);
 148     if (increment > 0) ch = Text.iterateCodePoints(nPos, 0);
 149     switch ( u_charType(ch) ) {
 150     // Upper
 151     case U_UPPERCASE_LETTER :
 152         return UPPER|LETTER|PRINTABLE|BASE_FORM;
 153
 154     // Lower
 155     case U_LOWERCASE_LETTER :
 156         return LOWER|LETTER|PRINTABLE|BASE_FORM;
 157
 158     // Title
 159     case U_TITLECASE_LETTER :
 160         return TITLE_CASE|LETTER|PRINTABLE|BASE_FORM;
 161
 162     // Letter
 163     case U_MODIFIER_LETTER :
 164     case U_OTHER_LETTER :
 165         return LETTER|PRINTABLE|BASE_FORM;
 166
 167     // Digit
 168     case U_DECIMAL_DIGIT_NUMBER:
 169     case U_LETTER_NUMBER:
 170     case U_OTHER_NUMBER:
 171         return DIGIT|PRINTABLE|BASE_FORM;
 172
 173     // Base
 174     case U_NON_SPACING_MARK:
 175     case U_ENCLOSING_MARK:
 176     case U_COMBINING_SPACING_MARK:
 177         return BASE_FORM|PRINTABLE;
 178
 179     // Print
 180     case U_SPACE_SEPARATOR:
 181
 182     case U_DASH_PUNCTUATION:
 183     case U_INITIAL_PUNCTUATION:
 184     case U_FINAL_PUNCTUATION:
 185     case U_CONNECTOR_PUNCTUATION:
 186     case U_OTHER_PUNCTUATION:
 187
 188     case U_MATH_SYMBOL:
 189     case U_CURRENCY_SYMBOL:
 190     case U_MODIFIER_SYMBOL:
 191     case U_OTHER_SYMBOL:
 192         return PRINTABLE;
 193
 194     // Control
 195     case U_CONTROL_CHAR:
 196     case U_FORMAT_CHAR:
 197         return CONTROL;
 198
 199     case U_LINE_SEPARATOR:
 200     case U_PARAGRAPH_SEPARATOR:
 201         return CONTROL|PRINTABLE;
 202
 203     // for all others
 204     default:
 205         return U_GENERAL_OTHER_TYPES;
 206     }
 207 }
 208
 209 sal_Int32 SAL_CALL
 210 cclass_Unicode::getCharacterType( const OUString& Text, sal_Int32 nPos, const Locale& /*rLocale*/ ) throw(RuntimeException) {
 211     if ( nPos < 0 || Text.getLength() <= nPos ) return 0;
 212     return getCharType(Text, &nPos, 0);
 213
 214 }
 215
 216 sal_Int32 SAL_CALL
 217 cclass_Unicode::getStringType( const OUString& Text, sal_Int32 nPos, sal_Int32 nCount, const Locale& /*rLocale*/ ) throw(RuntimeException) {
 218     if ( nPos < 0 || Text.getLength() <= nPos ) return 0;
 219
 220     sal_Int32 result = getCharType(Text, &nPos, 0);
 221     for (sal_Int32 i = 1; i < nCount && nPos < Text.getLength(); i++)
 222         result |= getCharType(Text, &nPos, 1);
 223     return result;
 224 }
 225
 226 ParseResult SAL_CALL cclass_Unicode::parseAnyToken(
 227             const OUString& Text,
 228             sal_Int32 nPos,
 229             const Locale& rLocale,
 230             sal_Int32 startCharTokenType,
 231             const OUString& userDefinedCharactersStart,
 232             sal_Int32 contCharTokenType,
 233             const OUString& userDefinedCharactersCont )
 234                 throw(RuntimeException)
 235 {
 236     ParseResult r;
 237     if ( Text.getLength() <= nPos )
 238         return r;
 239
 240     setupParserTable( rLocale,
 241         startCharTokenType, userDefinedCharactersStart,
 242         contCharTokenType, userDefinedCharactersCont );
 243     parseText( r, Text, nPos );
 244
 245     return r;
 246 }
 247
 248
 249 ParseResult SAL_CALL cclass_Unicode::parsePredefinedToken(
 250             sal_Int32 nTokenType,
 251             const OUString& Text,
 252             sal_Int32 nPos,
 253             const Locale& rLocale,
 254             sal_Int32 startCharTokenType,
 255             const OUString& userDefinedCharactersStart,
 256             sal_Int32 contCharTokenType,
 257             const OUString& userDefinedCharactersCont )
 258                 throw(RuntimeException)
 259 {
 260     ParseResult r;
 261     if ( Text.getLength() <= nPos )
 262         return r;
 263
 264     setupParserTable( rLocale,
 265         startCharTokenType, userDefinedCharactersStart,
 266         contCharTokenType, userDefinedCharactersCont );
 267     parseText( r, Text, nPos, nTokenType );
 268
 269     return r;
 270 }
 271
 272 OUString SAL_CALL cclass_Unicode::getImplementationName() throw( RuntimeException )
 273 {
 274     return OUString::createFromAscii(cClass);
 275 }
 276
 277
 278 sal_Bool SAL_CALL cclass_Unicode::supportsService(const OUString& rServiceName) throw( RuntimeException )
 279 {
 280     return !rServiceName.compareToAscii(cClass);
 281 }
 282
 283 Sequence< OUString > SAL_CALL cclass_Unicode::getSupportedServiceNames() throw( RuntimeException )
 284 {
 285     Sequence< OUString > aRet(1);
 286     aRet[0] = OUString::createFromAscii(cClass);
 287     return aRet;
 288 }
 289
 290 } } } }
 291