diff options
author | Kurt Zenker <kz@openoffice.org> | 2006-01-31 17:32:54 +0000 |
---|---|---|
committer | Kurt Zenker <kz@openoffice.org> | 2006-01-31 17:32:54 +0000 |
commit | 33ced467ac52c02b28a003873aba57b6a743df4c (patch) | |
tree | ce0047f98fabc30fe6f7fb2076a700c37441a955 /i18nutil | |
parent | 89b81d503dce96b24292513b51fdac839b0098bd (diff) |
INTEGRATION: CWS i18n24 (1.4.6); FILE MERGED
2006/01/13 01:05:01 khong 1.4.6.1: #i43395 stop i without dot fold to i with dot in case ignore match
Diffstat (limited to 'i18nutil')
-rw-r--r-- | i18nutil/source/utility/casefolding.cxx | 161 |
1 files changed, 82 insertions, 79 deletions
diff --git a/i18nutil/source/utility/casefolding.cxx b/i18nutil/source/utility/casefolding.cxx index 1bb3665ff85f..4d79a40389b4 100644 --- a/i18nutil/source/utility/casefolding.cxx +++ b/i18nutil/source/utility/casefolding.cxx @@ -4,9 +4,9 @@ * * $RCSfile: casefolding.cxx,v $ * - * $Revision: 1.4 $ + * $Revision: 1.5 $ * - * last change: $Author: rt $ $Date: 2005-09-07 17:40:21 $ + * last change: $Author: kz $ $Date: 2006-01-31 18:32:54 $ * * The Contents of this file are made available subject to * the terms of GNU Lesser General Public License Version 2.1. @@ -44,13 +44,14 @@ namespace com { namespace sun { namespace star { namespace i18n { static Mapping mapping_03a3[] = {{0, 1, 0x03c2, 0, 0 },{0, 1, 0x03c3, 0, 0}}; static Mapping mapping_0307[] = {{0, 0, 0, 0, 0 },{0, 1, 0x0307, 0, 0}}; -static Mapping mapping_0049[] = {{0, 2, 0x0069, 0x0307, 0},{0, 1, 0x0131, 0, 0},{0, 1, 0x0069, 0, 0}}; static Mapping mapping_004a[] = {{0, 2, 0x006a, 0x0307, 0},{0, 1, 0x006a, 0, 0}}; static Mapping mapping_012e[] = {{0, 2, 0x012f, 0x0307, 0},{0, 1, 0x012f, 0, 0}}; static Mapping mapping_00cc[] = {{0, 3, 0x0069, 0x0307, 0x0300},{0, 1, 0x00ec, 0, 0}}; static Mapping mapping_00cd[] = {{0, 3, 0x0069, 0x0307, 0x0301},{0, 1, 0x00ed, 0, 0}}; static Mapping mapping_0128[] = {{0, 3, 0x0069, 0x0307, 0x0303},{0, 1, 0x0129, 0, 0}}; +static Mapping mapping_0049[] = {{0, 2, 0x0069, 0x0307, 0},{0, 1, 0x0131, 0, 0},{0, 1, 0x0069, 0, 0}}; static Mapping mapping_0069[] = {{0, 1, 0x0130, 0, 0},{0, 1, 0x0049, 0, 0}}; +static Mapping mapping_0130[] = {{0, 1, 0x0069, 0, 0},{0, 1, 0x0130, 0, 0}}; #define langIs(lang) (aLocale.Language.compareToAscii(lang) == 0) @@ -61,98 +62,100 @@ static Mapping mapping_0069[] = {{0, 1, 0x0130, 0, 0},{0, 1, 0x0049, 0, 0}}; Mapping& casefolding::getConditionalValue(const sal_Unicode* str, sal_Int32 pos, sal_Int32 len, Locale& aLocale, sal_uInt8 nMappingType) throw (RuntimeException) { - switch(str[pos]) { - case 0x03a3: - // final_sigma (not followed by cased and preceded by cased character) - // DOES NOT check ignorable sequence yet (more complicated implementation). - return !(pos < len && cased_letter(str[pos+1])) && (pos > 0 && cased_letter(str[pos-1])) ? - mapping_03a3[0] : mapping_03a3[1]; - case 0x0307: - return ((nMappingType == MappingTypeLowerToUpper && langIs("lt") || - nMappingType == MappingTypeUpperToLower && (langIs("tr") || langIs("az"))) && - (pos > 0 && type_i(str[pos-1]))) ? // after_i - mapping_0307[0] : mapping_0307[1]; - case 0x0069: - return (langIs("tr") || langIs("az")) ? mapping_0069[0] : mapping_0069[1]; - case 0x0049: return langIs("lt") ? mapping_0049[0] : - (langIs("tr") || langIs("az")) ? mapping_0049[1] : mapping_0049[2]; - case 0x004a: return langIs("lt") ? mapping_004a[0] : mapping_004a[1]; - case 0x012e: return langIs("lt") ? mapping_012e[0] : mapping_012e[1]; - case 0x00cc: return langIs("lt") ? mapping_00cc[0] : mapping_00cc[1]; - case 0x00cd: return langIs("lt") ? mapping_00cd[0] : mapping_00cd[1]; - case 0x0128: return langIs("lt") ? mapping_0128[0] : mapping_0128[1]; - } - // Should not come here - throw RuntimeException(); + switch(str[pos]) { + case 0x03a3: + // final_sigma (not followed by cased and preceded by cased character) + // DOES NOT check ignorable sequence yet (more complicated implementation). + return !(pos < len && cased_letter(str[pos+1])) && (pos > 0 && cased_letter(str[pos-1])) ? + mapping_03a3[0] : mapping_03a3[1]; + case 0x0307: + return ((nMappingType == MappingTypeLowerToUpper && langIs("lt") || + nMappingType == MappingTypeUpperToLower && (langIs("tr") || langIs("az"))) && + (pos > 0 && type_i(str[pos-1]))) ? // after_i + mapping_0307[0] : mapping_0307[1]; + case 0x0130: + return (langIs("tr") || langIs("az")) ? mapping_0130[0] : mapping_0130[1]; + case 0x0069: + return (langIs("tr") || langIs("az")) ? mapping_0069[0] : mapping_0069[1]; + case 0x0049: return langIs("lt") ? mapping_0049[0] : + (langIs("tr") || langIs("az")) ? mapping_0049[1] : mapping_0049[2]; + case 0x004a: return langIs("lt") ? mapping_004a[0] : mapping_004a[1]; + case 0x012e: return langIs("lt") ? mapping_012e[0] : mapping_012e[1]; + case 0x00cc: return langIs("lt") ? mapping_00cc[0] : mapping_00cc[1]; + case 0x00cd: return langIs("lt") ? mapping_00cd[0] : mapping_00cd[1]; + case 0x0128: return langIs("lt") ? mapping_0128[0] : mapping_0128[1]; + } + // Should not come here + throw RuntimeException(); } Mapping& casefolding::getValue(const sal_Unicode* str, sal_Int32 pos, sal_Int32 len, Locale& aLocale, sal_uInt8 nMappingType) throw (RuntimeException) { - static Mapping dummy = { 0, 1, 0, 0, 0 }; - sal_Int16 address = CaseMappingIndex[str[pos] >> 8] << 8; - - dummy.map[0] = str[pos]; - - if (address >= 0 && (CaseMappingValue[address += (str[pos] & 0xFF)].type & nMappingType)) { - sal_uInt8 type = CaseMappingValue[address].type; - if (type & ValueTypeNotValue) { - if (CaseMappingValue[address].value == 0) - return getConditionalValue(str, pos, len, aLocale, nMappingType); - else { - for (int map = CaseMappingValue[address].value; - map < CaseMappingValue[address].value + MaxCaseMappingExtras; map++) { - if (CaseMappingExtra[map].type & nMappingType) { - if (CaseMappingExtra[map].type & ValueTypeNotValue) - return getConditionalValue(str, pos, len, aLocale, nMappingType); - else - return CaseMappingExtra[map]; - } - } - // Should not come here - throw RuntimeException(); + static Mapping dummy = { 0, 1, 0, 0, 0 }; + sal_Int16 address = CaseMappingIndex[str[pos] >> 8] << 8; + + dummy.map[0] = str[pos]; + + if (address >= 0 && (CaseMappingValue[address += (str[pos] & 0xFF)].type & nMappingType)) { + sal_uInt8 type = CaseMappingValue[address].type; + if (type & ValueTypeNotValue) { + if (CaseMappingValue[address].value == 0) + return getConditionalValue(str, pos, len, aLocale, nMappingType); + else { + for (int map = CaseMappingValue[address].value; + map < CaseMappingValue[address].value + MaxCaseMappingExtras; map++) { + if (CaseMappingExtra[map].type & nMappingType) { + if (CaseMappingExtra[map].type & ValueTypeNotValue) + return getConditionalValue(str, pos, len, aLocale, nMappingType); + else + return CaseMappingExtra[map]; + } + } + // Should not come here + throw RuntimeException(); + } + } else + dummy.map[0] = CaseMappingValue[address].value; } - } else - dummy.map[0] = CaseMappingValue[address].value; - } - return dummy; + return dummy; } inline sal_Bool SAL_CALL is_ja_voice_sound_mark(sal_Unicode& current, sal_Unicode next) { - sal_Unicode c = 0; + sal_Unicode c = 0; - if ((next == 0x3099 || next == 0x309a) && (c = widthfolding::getCompositionChar(current, next))) - current = c; - return c != 0; + if ((next == 0x3099 || next == 0x309a) && (c = widthfolding::getCompositionChar(current, next))) + current = c; + return c != 0; } 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) { - sal_Unicode c; - if (moduleLoaded & TransliterationModules_IGNORE_CASE) { - if (e.current >= e.element.nmap && idx < len ) { - e.element = getValue(str, idx++, len, aLocale, nMappingType); - e.current = 0; + sal_Unicode c; + if (moduleLoaded & TransliterationModules_IGNORE_CASE) { + if (e.current >= e.element.nmap && idx < len ) { + e.element = getValue(str, idx++, len, aLocale, nMappingType); + e.current = 0; + } + c = e.element.map[e.current++]; + } else { + c = *(str + idx++); + } + if (moduleLoaded & TransliterationModules_IGNORE_KANA) { + if (0x3040 <= c && c <= 0x3094 || 0x309d <= c && c <= 0x309f) + c += 0x60; + } + + // composition: KA + voice-mark --> GA. see halfwidthToFullwidth.cxx for detail + if (moduleLoaded & TransliterationModules_IGNORE_WIDTH) { + static oneToOneMapping& half2fullTable = widthfolding::gethalf2fullTable(); + c = half2fullTable[c]; + if (0x3040 <= c && c <= 0x30ff && idx < len && + is_ja_voice_sound_mark(c, half2fullTable[*(str + idx)])) + idx++; } - c = e.element.map[e.current++]; - } else { - c = *(str + idx++); - } - if (moduleLoaded & TransliterationModules_IGNORE_KANA) { - if (0x3040 <= c && c <= 0x3094 || 0x309d <= c && c <= 0x309f) - c += 0x60; - } - - // composition: KA + voice-mark --> GA. see halfwidthToFullwidth.cxx for detail - if (moduleLoaded & TransliterationModules_IGNORE_WIDTH) { - static oneToOneMapping& half2fullTable = widthfolding::gethalf2fullTable(); - c = half2fullTable[c]; - if (0x3040 <= c && c <= 0x30ff && idx < len && - is_ja_voice_sound_mark(c, half2fullTable[*(str + idx)])) - idx++; - } - return c; + return c; } } } } } |