i18nutil/source/utility/casefolding.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: casefolding.cxx,v $
  10  * $Revision: 1.8 $
  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 #include "i18nutil/casefolding.hxx"
  32 #include "casefolding_data.h"
  33 #include "i18nutil/widthfolding.hxx"
  34
  35 using namespace com::sun::star::lang;
  36 using namespace com::sun::star::uno;
  37
  38 namespace com { namespace sun { namespace star { namespace i18n {
  39
  40 static Mapping mapping_03a3[] = {{0, 1, {0x03c2, 0, 0}},{0, 1, {0x03c3, 0, 0}}};
  41 static Mapping mapping_0307[] = {{0, 0, {0, 0, 0}},{0, 1, {0x0307, 0, 0}}};
  42 static Mapping mapping_004a[] = {{0, 2, {0x006a, 0x0307, 0}},{0, 1, {0x006a, 0, 0}}};
  43 static Mapping mapping_012e[] = {{0, 2, {0x012f, 0x0307, 0}},{0, 1, {0x012f, 0, 0}}};
  44 static Mapping mapping_00cc[] = {{0, 3, {0x0069, 0x0307, 0x0300}},{0, 1, {0x00ec, 0, 0}}};
  45 static Mapping mapping_00cd[] = {{0, 3, {0x0069, 0x0307, 0x0301}},{0, 1, {0x00ed, 0, 0}}};
  46 static Mapping mapping_0128[] = {{0, 3, {0x0069, 0x0307, 0x0303}},{0, 1, {0x0129, 0, 0}}};
  47 static Mapping mapping_0049[] = {{0, 2, {0x0069, 0x0307, 0}},{0, 1, {0x0131, 0, 0}},{0, 1, {0x0069, 0, 0}}};
  48 static Mapping mapping_0069[] = {{0, 1, {0x0130, 0, 0}},{0, 1, {0x0049, 0, 0}}};
  49 static Mapping mapping_0130[] = {{0, 1, {0x0069, 0, 0}},{0, 1, {0x0130, 0, 0}}};
  50
  51 #define langIs(lang) (aLocale.Language.compareToAscii(lang) == 0)
  52
  53 // only check simple case, there is more complicated case need to be checked.
  54 #define type_i(ch) ((ch) == 0x0069 || (ch) == 0x006a)
  55
  56 #define cased_letter(ch) (CaseMappingIndex[(ch)>>8] >= 0 && (CaseMappingValue[(CaseMappingIndex[(ch)>>8] << 8) + ((ch)&0xff)].type & CasedLetter))
  57
  58 // for Lithuanian, condition to make explicit dot above when lowercasing capital I's and J's
  59 // whenever there are more accents above.
  60 #define accent_above(ch) (((ch) >= 0x0300 && (ch) <= 0x0314) || ((ch) >= 0x033D && (ch) <= 0x0344) || (ch) == 0x0346 || ((ch) >= 0x034A && (ch) <= 0x034C))
  61
  62 Mapping& casefolding::getConditionalValue(const sal_Unicode* str, sal_Int32 pos, sal_Int32 len, Locale& aLocale, sal_uInt8 nMappingType) throw (RuntimeException)
  63 {
  64         switch(str[pos]) {
  65         case 0x03a3:
  66             // final_sigma (not followed by cased and preceded by cased character)
  67             // DOES NOT check ignorable sequence yet (more complicated implementation).
  68             return !(pos < len && cased_letter(str[pos+1])) && (pos > 0 && cased_letter(str[pos-1])) ?
  69                 mapping_03a3[0] : mapping_03a3[1];
  70         case 0x0307:
  71             return (((nMappingType == MappingTypeLowerToUpper && langIs("lt")) ||
  72                 (nMappingType == MappingTypeUpperToLower && (langIs("tr") || langIs("az")))) &&
  73                 (pos > 0 && type_i(str[pos-1]))) ?      // after_i
  74                     mapping_0307[0] : mapping_0307[1];
  75         case 0x0130:
  76             return (langIs("tr") || langIs("az")) ? mapping_0130[0] : mapping_0130[1];
  77         case 0x0069:
  78             return (langIs("tr") || langIs("az")) ? mapping_0069[0] : mapping_0069[1];
  79         case 0x0049: return langIs("lt") && pos > len && accent_above(str[pos+1]) ? mapping_0049[0] :
  80                     (langIs("tr") || langIs("az")) ? mapping_0049[1] : mapping_0049[2];
  81         case 0x004a: return langIs("lt") && pos > len && accent_above(str[pos+1]) ? mapping_004a[0] : mapping_004a[1];
  82         case 0x012e: return langIs("lt") && pos > len && accent_above(str[pos+1]) ? mapping_012e[0] : mapping_012e[1];
  83         case 0x00cc: return langIs("lt") ? mapping_00cc[0] : mapping_00cc[1];
  84         case 0x00cd: return langIs("lt") ? mapping_00cd[0] : mapping_00cd[1];
  85         case 0x0128: return langIs("lt") ? mapping_0128[0] : mapping_0128[1];
  86         }
  87         // Should not come here
  88         throw RuntimeException();
  89 }
  90
  91 Mapping& casefolding::getValue(const sal_Unicode* str, sal_Int32 pos, sal_Int32 len, Locale& aLocale, sal_uInt8 nMappingType) throw (RuntimeException)
  92 {
  93         static Mapping dummy = { 0, 1, { 0, 0, 0 } };
  94         sal_Int16 address = CaseMappingIndex[str[pos] >> 8] << 8;
  95
  96         dummy.map[0] = str[pos];
  97
  98         if (address >= 0 && (CaseMappingValue[address += (str[pos] & 0xFF)].type & nMappingType)) {
  99             sal_uInt8 type = CaseMappingValue[address].type;
 100             if (type & ValueTypeNotValue) {
 101                 if (CaseMappingValue[address].value == 0)
 102                     return getConditionalValue(str, pos, len, aLocale, nMappingType);
 103                 else {
 104                     for (int map = CaseMappingValue[address].value;
 105                             map < CaseMappingValue[address].value + MaxCaseMappingExtras; map++) {
 106                         if (CaseMappingExtra[map].type & nMappingType) {
 107                             if (CaseMappingExtra[map].type & ValueTypeNotValue)
 108                                 return getConditionalValue(str, pos, len, aLocale, nMappingType);
 109                             else
 110                                 return CaseMappingExtra[map];
 111                         }
 112                     }
 113                     // Should not come here
 114                     throw RuntimeException();
 115                 }
 116             } else
 117                 dummy.map[0] = CaseMappingValue[address].value;
 118         }
 119         return dummy;
 120 }
 121
 122 inline sal_Bool SAL_CALL
 123 is_ja_voice_sound_mark(sal_Unicode& current, sal_Unicode next)
 124 {
 125         sal_Unicode c = 0;
 126
 127         if ((next == 0x3099 || next == 0x309a) && ( (c = widthfolding::getCompositionChar(current, next)) != 0 ))
 128             current = c;
 129         return c != 0;
 130 }
 131
 132 sal_Unicode casefolding::getNextChar(const sal_Unicode *str, sal_Int32& idx, sal_Int32 len, MappingElement& e, Locale& aLocale, sal_uInt8 nMappingType, TransliterationModules moduleLoaded) throw (RuntimeException)
 133 {
 134         if( idx >= len )
 135         {
 136             e = MappingElement();
 137             return 0;
 138         }
 139
 140         sal_Unicode c;
 141
 142         if (moduleLoaded & TransliterationModules_IGNORE_CASE) {
 143             if( e.current >= e.element.nmap ) {
 144                 e.element = getValue(str, idx++, len, aLocale, nMappingType);
 145                 e.current = 0;
 146             }
 147             c = e.element.map[e.current++];
 148         } else {
 149             c = *(str + idx++);
 150         }
 151
 152         if (moduleLoaded & TransliterationModules_IGNORE_KANA) {
 153             if ((0x3040 <= c && c <= 0x3094) || (0x309d <= c && c <= 0x309f))
 154                 c += 0x60;
 155         }
 156
 157         // composition: KA + voice-mark --> GA. see halfwidthToFullwidth.cxx for detail
 158         if (moduleLoaded & TransliterationModules_IGNORE_WIDTH) {
 159             static oneToOneMapping& half2fullTable = widthfolding::gethalf2fullTable();
 160             c = half2fullTable[c];
 161             if (0x3040 <= c && c <= 0x30ff && idx < len &&
 162                     is_ja_voice_sound_mark(c, half2fullTable[*(str + idx)]))
 163                 idx++;
 164         }
 165
 166         return c;
 167 }
 168
 169 } } } }
 170