1*b1cdbd2cSJim Jagielski /************************************************************** 2*b1cdbd2cSJim Jagielski * 3*b1cdbd2cSJim Jagielski * Licensed to the Apache Software Foundation (ASF) under one 4*b1cdbd2cSJim Jagielski * or more contributor license agreements. See the NOTICE file 5*b1cdbd2cSJim Jagielski * distributed with this work for additional information 6*b1cdbd2cSJim Jagielski * regarding copyright ownership. The ASF licenses this file 7*b1cdbd2cSJim Jagielski * to you under the Apache License, Version 2.0 (the 8*b1cdbd2cSJim Jagielski * "License"); you may not use this file except in compliance 9*b1cdbd2cSJim Jagielski * with the License. You may obtain a copy of the License at 10*b1cdbd2cSJim Jagielski * 11*b1cdbd2cSJim Jagielski * http://www.apache.org/licenses/LICENSE-2.0 12*b1cdbd2cSJim Jagielski * 13*b1cdbd2cSJim Jagielski * Unless required by applicable law or agreed to in writing, 14*b1cdbd2cSJim Jagielski * software distributed under the License is distributed on an 15*b1cdbd2cSJim Jagielski * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY 16*b1cdbd2cSJim Jagielski * KIND, either express or implied. See the License for the 17*b1cdbd2cSJim Jagielski * specific language governing permissions and limitations 18*b1cdbd2cSJim Jagielski * under the License. 19*b1cdbd2cSJim Jagielski * 20*b1cdbd2cSJim Jagielski *************************************************************/ 21*b1cdbd2cSJim Jagielski 22*b1cdbd2cSJim Jagielski 23*b1cdbd2cSJim Jagielski #ifndef _I18N_CCLASS_UNICODE_HXX_ 24*b1cdbd2cSJim Jagielski #define _I18N_CCLASS_UNICODE_HXX_ 25*b1cdbd2cSJim Jagielski 26*b1cdbd2cSJim Jagielski #include <com/sun/star/i18n/XNativeNumberSupplier.hpp> 27*b1cdbd2cSJim Jagielski #include <com/sun/star/i18n/XCharacterClassification.hpp> 28*b1cdbd2cSJim Jagielski #include <com/sun/star/i18n/XLocaleData.hpp> 29*b1cdbd2cSJim Jagielski #include <com/sun/star/lang/XMultiServiceFactory.hpp> 30*b1cdbd2cSJim Jagielski #include <cppuhelper/implbase1.hxx> // helper for implementations 31*b1cdbd2cSJim Jagielski #include <com/sun/star/lang/XServiceInfo.hpp> 32*b1cdbd2cSJim Jagielski 33*b1cdbd2cSJim Jagielski #define TRANSLITERATION_casemapping 34*b1cdbd2cSJim Jagielski #include <transliteration_body.hxx> 35*b1cdbd2cSJim Jagielski 36*b1cdbd2cSJim Jagielski namespace com { namespace sun { namespace star { namespace i18n { 37*b1cdbd2cSJim Jagielski 38*b1cdbd2cSJim Jagielski typedef sal_uInt32 UPT_FLAG_TYPE; 39*b1cdbd2cSJim Jagielski 40*b1cdbd2cSJim Jagielski class cclass_Unicode : public cppu::WeakImplHelper1 < XCharacterClassification > 41*b1cdbd2cSJim Jagielski { 42*b1cdbd2cSJim Jagielski public: 43*b1cdbd2cSJim Jagielski cclass_Unicode(com::sun::star::uno::Reference < com::sun::star::lang::XMultiServiceFactory > xSMgr ); 44*b1cdbd2cSJim Jagielski ~cclass_Unicode(); 45*b1cdbd2cSJim Jagielski 46*b1cdbd2cSJim Jagielski virtual rtl::OUString SAL_CALL toUpper( const rtl::OUString& Text, sal_Int32 nPos, sal_Int32 nCount, 47*b1cdbd2cSJim Jagielski const com::sun::star::lang::Locale& rLocale ) throw(com::sun::star::uno::RuntimeException); 48*b1cdbd2cSJim Jagielski virtual rtl::OUString SAL_CALL toLower( const rtl::OUString& Text, sal_Int32 nPos, sal_Int32 nCount, 49*b1cdbd2cSJim Jagielski const com::sun::star::lang::Locale& rLocale ) throw(com::sun::star::uno::RuntimeException); 50*b1cdbd2cSJim Jagielski virtual rtl::OUString SAL_CALL toTitle( const rtl::OUString& Text, sal_Int32 nPos, sal_Int32 nCount, 51*b1cdbd2cSJim Jagielski const com::sun::star::lang::Locale& rLocale ) throw(com::sun::star::uno::RuntimeException); 52*b1cdbd2cSJim Jagielski virtual sal_Int16 SAL_CALL getType( const rtl::OUString& Text, sal_Int32 nPos ) throw(com::sun::star::uno::RuntimeException); 53*b1cdbd2cSJim Jagielski virtual sal_Int16 SAL_CALL getCharacterDirection( const rtl::OUString& Text, sal_Int32 nPos ) 54*b1cdbd2cSJim Jagielski throw(com::sun::star::uno::RuntimeException); 55*b1cdbd2cSJim Jagielski virtual sal_Int16 SAL_CALL getScript( const rtl::OUString& Text, sal_Int32 nPos ) throw(com::sun::star::uno::RuntimeException); 56*b1cdbd2cSJim Jagielski virtual sal_Int32 SAL_CALL getCharacterType( const rtl::OUString& text, sal_Int32 nPos, 57*b1cdbd2cSJim Jagielski const com::sun::star::lang::Locale& rLocale ) throw(com::sun::star::uno::RuntimeException); 58*b1cdbd2cSJim Jagielski virtual sal_Int32 SAL_CALL getStringType( const rtl::OUString& text, sal_Int32 nPos, sal_Int32 nCount, 59*b1cdbd2cSJim Jagielski const com::sun::star::lang::Locale& rLocale ) throw(com::sun::star::uno::RuntimeException); 60*b1cdbd2cSJim Jagielski virtual ParseResult SAL_CALL parseAnyToken( const rtl::OUString& Text, sal_Int32 nPos, 61*b1cdbd2cSJim Jagielski const com::sun::star::lang::Locale& rLocale, sal_Int32 nStartCharFlags, const rtl::OUString& userDefinedCharactersStart, 62*b1cdbd2cSJim Jagielski sal_Int32 nContCharFlags, const rtl::OUString& userDefinedCharactersCont ) throw(com::sun::star::uno::RuntimeException); 63*b1cdbd2cSJim Jagielski virtual ParseResult SAL_CALL parsePredefinedToken( sal_Int32 nTokenType, const rtl::OUString& Text, 64*b1cdbd2cSJim Jagielski sal_Int32 nPos, const com::sun::star::lang::Locale& rLocale, sal_Int32 nStartCharFlags, 65*b1cdbd2cSJim Jagielski const rtl::OUString& userDefinedCharactersStart, sal_Int32 nContCharFlags, 66*b1cdbd2cSJim Jagielski const rtl::OUString& userDefinedCharactersCont ) throw(com::sun::star::uno::RuntimeException); 67*b1cdbd2cSJim Jagielski 68*b1cdbd2cSJim Jagielski //XServiceInfo 69*b1cdbd2cSJim Jagielski virtual rtl::OUString SAL_CALL getImplementationName() throw( com::sun::star::uno::RuntimeException ); 70*b1cdbd2cSJim Jagielski virtual sal_Bool SAL_CALL supportsService(const rtl::OUString& ServiceName) throw( com::sun::star::uno::RuntimeException ); 71*b1cdbd2cSJim Jagielski virtual com::sun::star::uno::Sequence< rtl::OUString > SAL_CALL getSupportedServiceNames() throw( com::sun::star::uno::RuntimeException ); 72*b1cdbd2cSJim Jagielski 73*b1cdbd2cSJim Jagielski protected: 74*b1cdbd2cSJim Jagielski const sal_Char *cClass; 75*b1cdbd2cSJim Jagielski 76*b1cdbd2cSJim Jagielski private: 77*b1cdbd2cSJim Jagielski Transliteration_casemapping *trans; 78*b1cdbd2cSJim Jagielski 79*b1cdbd2cSJim Jagielski // --- parser specific (implemented in cclass_unicode_parser.cxx) --- 80*b1cdbd2cSJim Jagielski 81*b1cdbd2cSJim Jagielski enum ScanState 82*b1cdbd2cSJim Jagielski { 83*b1cdbd2cSJim Jagielski ssGetChar, 84*b1cdbd2cSJim Jagielski ssGetValue, 85*b1cdbd2cSJim Jagielski ssGetWord, 86*b1cdbd2cSJim Jagielski ssGetWordFirstChar, 87*b1cdbd2cSJim Jagielski ssGetString, 88*b1cdbd2cSJim Jagielski ssGetBool, 89*b1cdbd2cSJim Jagielski ssRewindFromValue, 90*b1cdbd2cSJim Jagielski ssIgnoreLeadingInRewind, 91*b1cdbd2cSJim Jagielski ssStopBack, 92*b1cdbd2cSJim Jagielski ssBounce, 93*b1cdbd2cSJim Jagielski ssStop 94*b1cdbd2cSJim Jagielski }; 95*b1cdbd2cSJim Jagielski 96*b1cdbd2cSJim Jagielski static const sal_uInt8 nDefCnt; 97*b1cdbd2cSJim Jagielski static const UPT_FLAG_TYPE pDefaultParserTable[]; 98*b1cdbd2cSJim Jagielski static const sal_Int32 pParseTokensType[]; 99*b1cdbd2cSJim Jagielski 100*b1cdbd2cSJim Jagielski /// Flag values of table. 101*b1cdbd2cSJim Jagielski static const UPT_FLAG_TYPE TOKEN_ILLEGAL; 102*b1cdbd2cSJim Jagielski static const UPT_FLAG_TYPE TOKEN_CHAR; 103*b1cdbd2cSJim Jagielski static const UPT_FLAG_TYPE TOKEN_CHAR_BOOL; 104*b1cdbd2cSJim Jagielski static const UPT_FLAG_TYPE TOKEN_CHAR_WORD; 105*b1cdbd2cSJim Jagielski static const UPT_FLAG_TYPE TOKEN_CHAR_VALUE; 106*b1cdbd2cSJim Jagielski static const UPT_FLAG_TYPE TOKEN_CHAR_STRING; 107*b1cdbd2cSJim Jagielski static const UPT_FLAG_TYPE TOKEN_CHAR_DONTCARE; 108*b1cdbd2cSJim Jagielski static const UPT_FLAG_TYPE TOKEN_BOOL; 109*b1cdbd2cSJim Jagielski static const UPT_FLAG_TYPE TOKEN_WORD; 110*b1cdbd2cSJim Jagielski static const UPT_FLAG_TYPE TOKEN_WORD_SEP; 111*b1cdbd2cSJim Jagielski static const UPT_FLAG_TYPE TOKEN_VALUE; 112*b1cdbd2cSJim Jagielski static const UPT_FLAG_TYPE TOKEN_VALUE_SEP; 113*b1cdbd2cSJim Jagielski static const UPT_FLAG_TYPE TOKEN_VALUE_EXP; 114*b1cdbd2cSJim Jagielski static const UPT_FLAG_TYPE TOKEN_VALUE_SIGN; 115*b1cdbd2cSJim Jagielski static const UPT_FLAG_TYPE TOKEN_VALUE_EXP_VALUE; 116*b1cdbd2cSJim Jagielski static const UPT_FLAG_TYPE TOKEN_VALUE_DIGIT; 117*b1cdbd2cSJim Jagielski static const UPT_FLAG_TYPE TOKEN_NAME_SEP; 118*b1cdbd2cSJim Jagielski static const UPT_FLAG_TYPE TOKEN_STRING_SEP; 119*b1cdbd2cSJim Jagielski static const UPT_FLAG_TYPE TOKEN_EXCLUDED; 120*b1cdbd2cSJim Jagielski 121*b1cdbd2cSJim Jagielski /// If and where c occurs in pStr 122*b1cdbd2cSJim Jagielski static const sal_Unicode* StrChr( const sal_Unicode* pStr, sal_Unicode c ); 123*b1cdbd2cSJim Jagielski 124*b1cdbd2cSJim Jagielski 125*b1cdbd2cSJim Jagielski com::sun::star::uno::Reference < com::sun::star::lang::XMultiServiceFactory > xMSF; 126*b1cdbd2cSJim Jagielski 127*b1cdbd2cSJim Jagielski /// used for parser only 128*b1cdbd2cSJim Jagielski com::sun::star::lang::Locale aParserLocale; 129*b1cdbd2cSJim Jagielski com::sun::star::uno::Reference < XLocaleData > xLocaleData; 130*b1cdbd2cSJim Jagielski com::sun::star::uno::Reference < com::sun::star::i18n::XNativeNumberSupplier > xNatNumSup; 131*b1cdbd2cSJim Jagielski rtl::OUString aStartChars; 132*b1cdbd2cSJim Jagielski rtl::OUString aContChars; 133*b1cdbd2cSJim Jagielski UPT_FLAG_TYPE* pTable; 134*b1cdbd2cSJim Jagielski UPT_FLAG_TYPE* pStart; 135*b1cdbd2cSJim Jagielski UPT_FLAG_TYPE* pCont; 136*b1cdbd2cSJim Jagielski sal_Int32 nStartTypes; 137*b1cdbd2cSJim Jagielski sal_Int32 nContTypes; 138*b1cdbd2cSJim Jagielski ScanState eState; 139*b1cdbd2cSJim Jagielski sal_Unicode cGroupSep; 140*b1cdbd2cSJim Jagielski sal_Unicode cDecimalSep; 141*b1cdbd2cSJim Jagielski 142*b1cdbd2cSJim Jagielski /// Get corresponding KParseTokens flag for a character 143*b1cdbd2cSJim Jagielski sal_Int32 getParseTokensType( const sal_Unicode* aStr, sal_Int32 nPos ); 144*b1cdbd2cSJim Jagielski 145*b1cdbd2cSJim Jagielski /// Access parser table flags. 146*b1cdbd2cSJim Jagielski UPT_FLAG_TYPE getFlags( const sal_Unicode* aStr, sal_Int32 nPos ); 147*b1cdbd2cSJim Jagielski 148*b1cdbd2cSJim Jagielski /// Access parser flags via International and special definitions. 149*b1cdbd2cSJim Jagielski UPT_FLAG_TYPE getFlagsExtended( const sal_Unicode* aStr, sal_Int32 nPos ); 150*b1cdbd2cSJim Jagielski 151*b1cdbd2cSJim Jagielski /// Access parser table flags for user defined start characters. 152*b1cdbd2cSJim Jagielski UPT_FLAG_TYPE getStartCharsFlags( sal_Unicode c ); 153*b1cdbd2cSJim Jagielski 154*b1cdbd2cSJim Jagielski /// Access parser table flags for user defined continuation characters. 155*b1cdbd2cSJim Jagielski UPT_FLAG_TYPE getContCharsFlags( sal_Unicode c ); 156*b1cdbd2cSJim Jagielski 157*b1cdbd2cSJim Jagielski /// Setup parser table. Calls initParserTable() only if needed. 158*b1cdbd2cSJim Jagielski void setupParserTable( const com::sun::star::lang::Locale& rLocale, sal_Int32 startCharTokenType, 159*b1cdbd2cSJim Jagielski const rtl::OUString& userDefinedCharactersStart, sal_Int32 contCharTokenType, 160*b1cdbd2cSJim Jagielski const rtl::OUString& userDefinedCharactersCont ); 161*b1cdbd2cSJim Jagielski 162*b1cdbd2cSJim Jagielski /// Init parser table. 163*b1cdbd2cSJim Jagielski void initParserTable( const com::sun::star::lang::Locale& rLocale, sal_Int32 startCharTokenType, 164*b1cdbd2cSJim Jagielski const rtl::OUString& userDefinedCharactersStart, sal_Int32 contCharTokenType, 165*b1cdbd2cSJim Jagielski const rtl::OUString& userDefinedCharactersCont ); 166*b1cdbd2cSJim Jagielski 167*b1cdbd2cSJim Jagielski /// Destroy parser table. 168*b1cdbd2cSJim Jagielski void destroyParserTable(); 169*b1cdbd2cSJim Jagielski 170*b1cdbd2cSJim Jagielski /// Parse a text. 171*b1cdbd2cSJim Jagielski void parseText( ParseResult& r, const rtl::OUString& rText, sal_Int32 nPos, 172*b1cdbd2cSJim Jagielski sal_Int32 nTokenType = 0xffffffff ); 173*b1cdbd2cSJim Jagielski 174*b1cdbd2cSJim Jagielski /// Setup International class, new'ed only if different from existing. 175*b1cdbd2cSJim Jagielski sal_Bool setupInternational( const com::sun::star::lang::Locale& rLocale ); 176*b1cdbd2cSJim Jagielski 177*b1cdbd2cSJim Jagielski /// Implementation of getCharacterType() for one single character 178*b1cdbd2cSJim Jagielski sal_Int32 SAL_CALL getCharType( const rtl::OUString& Text, sal_Int32 *nPos, sal_Int32 increment); 179*b1cdbd2cSJim Jagielski 180*b1cdbd2cSJim Jagielski }; 181*b1cdbd2cSJim Jagielski 182*b1cdbd2cSJim Jagielski } } } } 183*b1cdbd2cSJim Jagielski 184*b1cdbd2cSJim Jagielski #endif 185