// © 2016 and later: Unicode, Inc. and others. // License & terms of use: http://www.unicode.org/copyright.html /* ******************************************************************************* * * Copyright (C) 2005-2016, International Business Machines * Corporation and others. All Rights Reserved. * ******************************************************************************* * file name: ucasemap.cpp * encoding: UTF-8 * tab size: 8 (not used) * indentation:4 * * created on: 2005may06 * created by: Markus W. Scherer * * Case mapping service object and functions using it. */ #include "unicode/utypes.h" #include "unicode/brkiter.h" #include "unicode/bytestream.h" #include "unicode/casemap.h" #include "unicode/edits.h" #include "unicode/stringoptions.h" #include "unicode/stringpiece.h" #include "unicode/ubrk.h" #include "unicode/uloc.h" #include "unicode/ustring.h" #include "unicode/ucasemap.h" #if !UCONFIG_NO_BREAK_ITERATION #include "unicode/utext.h" #endif #include "unicode/utf.h" #include "unicode/utf8.h" #include "unicode/utf16.h" #include "bytesinkutil.h" #include "cmemory.h" #include "cstring.h" #include "uassert.h" #include "ucase.h" #include "ucasemap_imp.h" #include "ustr_imp.h" U_NAMESPACE_USE /* UCaseMap service object -------------------------------------------------- */ UCaseMap::UCaseMap(const char *localeID, uint32_t opts, UErrorCode *pErrorCode) : … { … } UCaseMap::~UCaseMap() { … } U_CAPI UCaseMap * U_EXPORT2 ucasemap_open(const char *locale, uint32_t options, UErrorCode *pErrorCode) { … } U_CAPI void U_EXPORT2 ucasemap_close(UCaseMap *csm) { … } U_CAPI const char * U_EXPORT2 ucasemap_getLocale(const UCaseMap *csm) { … } U_CAPI uint32_t U_EXPORT2 ucasemap_getOptions(const UCaseMap *csm) { … } U_CAPI void U_EXPORT2 ucasemap_setLocale(UCaseMap *csm, const char *locale, UErrorCode *pErrorCode) { … } U_CAPI void U_EXPORT2 ucasemap_setOptions(UCaseMap *csm, uint32_t options, UErrorCode *pErrorCode) { … } /* UTF-8 string case mappings ----------------------------------------------- */ /* TODO(markus): Move to a new, separate utf8case.cpp file. */ namespace { /* append a full case mapping result, see UCASE_MAX_STRING_LENGTH */ inline UBool appendResult(int32_t cpLength, int32_t result, const char16_t *s, ByteSink &sink, uint32_t options, icu::Edits *edits, UErrorCode &errorCode) { … } // See unicode/utf8.h U8_APPEND_UNSAFE(). inline uint8_t getTwoByteLead(UChar32 c) { … } inline uint8_t getTwoByteTrail(UChar32 c) { … } UChar32 U_CALLCONV utf8_caseContextIterator(void *context, int8_t dir) { … } /** * caseLocale >= 0: Lowercases [srcStart..srcLimit[ but takes context [0..srcLength[ into account. * caseLocale < 0: Case-folds [srcStart..srcLimit[. */ void toLower(int32_t caseLocale, uint32_t options, const uint8_t *src, UCaseContext *csc, int32_t srcStart, int32_t srcLimit, icu::ByteSink &sink, icu::Edits *edits, UErrorCode &errorCode) { … } void toUpper(int32_t caseLocale, uint32_t options, const uint8_t *src, UCaseContext *csc, int32_t srcLength, icu::ByteSink &sink, icu::Edits *edits, UErrorCode &errorCode) { … } } // namespace #if !UCONFIG_NO_BREAK_ITERATION namespace { constexpr uint8_t ACUTE_BYTE0 = …; constexpr uint8_t ACUTE_BYTE1 = …; /** * Input: c is a letter I with or without acute accent. * start is the index in src after c, and is less than segmentLimit. * If a plain i/I is followed by a plain j/J, * or an i/I with acute (precomposed or decomposed) is followed by a j/J with acute, * then we output accordingly. * * @return the src index after the titlecased sequence, or the start index if no Dutch IJ */ int32_t maybeTitleDutchIJ(const uint8_t *src, UChar32 c, int32_t start, int32_t segmentLimit, ByteSink &sink, uint32_t options, icu::Edits *edits, UErrorCode &errorCode) { … } } // namespace U_CFUNC void U_CALLCONV ucasemap_internalUTF8ToTitle( int32_t caseLocale, uint32_t options, BreakIterator *iter, const uint8_t *src, int32_t srcLength, ByteSink &sink, icu::Edits *edits, UErrorCode &errorCode) { … } #endif U_NAMESPACE_BEGIN namespace GreekUpper { UBool isFollowedByCasedLetter(const uint8_t *s, int32_t i, int32_t length) { … } // Keep this consistent with the UTF-16 version in ustrcase.cpp and the Java version in CaseMap.java. void toUpper(uint32_t options, const uint8_t *src, int32_t srcLength, ByteSink &sink, Edits *edits, UErrorCode &errorCode) { … } } // namespace GreekUpper U_NAMESPACE_END static void U_CALLCONV ucasemap_internalUTF8ToLower(int32_t caseLocale, uint32_t options, UCASEMAP_BREAK_ITERATOR_UNUSED const uint8_t *src, int32_t srcLength, icu::ByteSink &sink, icu::Edits *edits, UErrorCode &errorCode) { … } static void U_CALLCONV ucasemap_internalUTF8ToUpper(int32_t caseLocale, uint32_t options, UCASEMAP_BREAK_ITERATOR_UNUSED const uint8_t *src, int32_t srcLength, icu::ByteSink &sink, icu::Edits *edits, UErrorCode &errorCode) { … } static void U_CALLCONV ucasemap_internalUTF8Fold(int32_t /* caseLocale */, uint32_t options, UCASEMAP_BREAK_ITERATOR_UNUSED const uint8_t *src, int32_t srcLength, icu::ByteSink &sink, icu::Edits *edits, UErrorCode &errorCode) { … } void ucasemap_mapUTF8(int32_t caseLocale, uint32_t options, UCASEMAP_BREAK_ITERATOR_PARAM const char *src, int32_t srcLength, UTF8CaseMapper *stringCaseMapper, icu::ByteSink &sink, icu::Edits *edits, UErrorCode &errorCode) { … } int32_t ucasemap_mapUTF8(int32_t caseLocale, uint32_t options, UCASEMAP_BREAK_ITERATOR_PARAM char *dest, int32_t destCapacity, const char *src, int32_t srcLength, UTF8CaseMapper *stringCaseMapper, icu::Edits *edits, UErrorCode &errorCode) { … } /* public API functions */ U_CAPI int32_t U_EXPORT2 ucasemap_utf8ToLower(const UCaseMap *csm, char *dest, int32_t destCapacity, const char *src, int32_t srcLength, UErrorCode *pErrorCode) { … } U_CAPI int32_t U_EXPORT2 ucasemap_utf8ToUpper(const UCaseMap *csm, char *dest, int32_t destCapacity, const char *src, int32_t srcLength, UErrorCode *pErrorCode) { … } U_CAPI int32_t U_EXPORT2 ucasemap_utf8FoldCase(const UCaseMap *csm, char *dest, int32_t destCapacity, const char *src, int32_t srcLength, UErrorCode *pErrorCode) { … } U_NAMESPACE_BEGIN void CaseMap::utf8ToLower( const char *locale, uint32_t options, StringPiece src, ByteSink &sink, Edits *edits, UErrorCode &errorCode) { … } void CaseMap::utf8ToUpper( const char *locale, uint32_t options, StringPiece src, ByteSink &sink, Edits *edits, UErrorCode &errorCode) { … } void CaseMap::utf8Fold( uint32_t options, StringPiece src, ByteSink &sink, Edits *edits, UErrorCode &errorCode) { … } int32_t CaseMap::utf8ToLower( const char *locale, uint32_t options, const char *src, int32_t srcLength, char *dest, int32_t destCapacity, Edits *edits, UErrorCode &errorCode) { … } int32_t CaseMap::utf8ToUpper( const char *locale, uint32_t options, const char *src, int32_t srcLength, char *dest, int32_t destCapacity, Edits *edits, UErrorCode &errorCode) { … } int32_t CaseMap::utf8Fold( uint32_t options, const char *src, int32_t srcLength, char *dest, int32_t destCapacity, Edits *edits, UErrorCode &errorCode) { … } U_NAMESPACE_END
Codebase Browser
chromium