ICU 71.1: normalizer2.h Source File

00001 // © 2016 and later: Unicode, Inc. and others.
00002 // License & terms of use: http://www.unicode.org/copyright.html
00003 /*
00004 *******************************************************************************
00005 *
00006 *   Copyright (C) 2009-2013, International Business Machines
00007 *   Corporation and others.  All Rights Reserved.
00008 *
00009 *******************************************************************************
00010 *   file name:  normalizer2.h
00011 *   encoding:   UTF-8
00012 *   tab size:   8 (not used)
00013 *   indentation:4
00014 *
00015 *   created on: 2009nov22
00016 *   created by: Markus W. Scherer
00017 */
00018 
00019 #ifndef __NORMALIZER2_H__
00020 #define __NORMALIZER2_H__
00021 
00027 #include "unicode/utypes.h"
00028 
00029 #if U_SHOW_CPLUSPLUS_API
00030 
00031 #if !UCONFIG_NO_NORMALIZATION
00032 
00033 #include "unicode/stringpiece.h"
00034 #include "unicode/uniset.h"
00035 #include "unicode/unistr.h"
00036 #include "unicode/unorm2.h"
00037 
00038 U_NAMESPACE_BEGIN
00039 
00040 class ByteSink;
00041 
00085 class U_COMMON_API Normalizer2 : public UObject {
00086 public:
00091     ~Normalizer2();
00092 
00104     static const Normalizer2 *
00105     getNFCInstance(UErrorCode &errorCode);
00106 
00118     static const Normalizer2 *
00119     getNFDInstance(UErrorCode &errorCode);
00120 
00132     static const Normalizer2 *
00133     getNFKCInstance(UErrorCode &errorCode);
00134 
00146     static const Normalizer2 *
00147     getNFKDInstance(UErrorCode &errorCode);
00148 
00160     static const Normalizer2 *
00161     getNFKCCasefoldInstance(UErrorCode &errorCode);
00162 
00184     static const Normalizer2 *
00185     getInstance(const char *packageName,
00186                 const char *name,
00187                 UNormalization2Mode mode,
00188                 UErrorCode &errorCode);
00189 
00200     UnicodeString
00201     normalize(const UnicodeString &src, UErrorCode &errorCode) const {
00202         UnicodeString result;
00203         normalize(src, result, errorCode);
00204         return result;
00205     }
00219     virtual UnicodeString &
00220     normalize(const UnicodeString &src,
00221               UnicodeString &dest,
00222               UErrorCode &errorCode) const = 0;
00223 
00246     virtual void
00247     normalizeUTF8(uint32_t options, StringPiece src, ByteSink &sink,
00248                   Edits *edits, UErrorCode &errorCode) const;
00249 
00264     virtual UnicodeString &
00265     normalizeSecondAndAppend(UnicodeString &first,
00266                              const UnicodeString &second,
00267                              UErrorCode &errorCode) const = 0;
00282     virtual UnicodeString &
00283     append(UnicodeString &first,
00284            const UnicodeString &second,
00285            UErrorCode &errorCode) const = 0;
00286 
00300     virtual UBool
00301     getDecomposition(UChar32 c, UnicodeString &decomposition) const = 0;
00302 
00327     virtual UBool
00328     getRawDecomposition(UChar32 c, UnicodeString &decomposition) const;
00329 
00345     virtual UChar32
00346     composePair(UChar32 a, UChar32 b) const;
00347 
00356     virtual uint8_t
00357     getCombiningClass(UChar32 c) const;
00358 
00373     virtual UBool
00374     isNormalized(const UnicodeString &s, UErrorCode &errorCode) const = 0;
00394     virtual UBool
00395     isNormalizedUTF8(StringPiece s, UErrorCode &errorCode) const;
00396 
00397 
00413     virtual UNormalizationCheckResult
00414     quickCheck(const UnicodeString &s, UErrorCode &errorCode) const = 0;
00415 
00438     virtual int32_t
00439     spanQuickCheckYes(const UnicodeString &s, UErrorCode &errorCode) const = 0;
00440 
00454     virtual UBool hasBoundaryBefore(UChar32 c) const = 0;
00455 
00470     virtual UBool hasBoundaryAfter(UChar32 c) const = 0;
00471 
00485     virtual UBool isInert(UChar32 c) const = 0;
00486 };
00487 
00499 class U_COMMON_API FilteredNormalizer2 : public Normalizer2 {
00500 public:
00511     FilteredNormalizer2(const Normalizer2 &n2, const UnicodeSet &filterSet) :
00512             norm2(n2), set(filterSet) {}
00513 
00518     ~FilteredNormalizer2();
00519 
00533     virtual UnicodeString &
00534     normalize(const UnicodeString &src,
00535               UnicodeString &dest,
00536               UErrorCode &errorCode) const U_OVERRIDE;
00537 
00560     virtual void
00561     normalizeUTF8(uint32_t options, StringPiece src, ByteSink &sink,
00562                   Edits *edits, UErrorCode &errorCode) const U_OVERRIDE;
00563 
00578     virtual UnicodeString &
00579     normalizeSecondAndAppend(UnicodeString &first,
00580                              const UnicodeString &second,
00581                              UErrorCode &errorCode) const U_OVERRIDE;
00596     virtual UnicodeString &
00597     append(UnicodeString &first,
00598            const UnicodeString &second,
00599            UErrorCode &errorCode) const U_OVERRIDE;
00600 
00612     virtual UBool
00613     getDecomposition(UChar32 c, UnicodeString &decomposition) const U_OVERRIDE;
00614 
00626     virtual UBool
00627     getRawDecomposition(UChar32 c, UnicodeString &decomposition) const U_OVERRIDE;
00628 
00639     virtual UChar32
00640     composePair(UChar32 a, UChar32 b) const U_OVERRIDE;
00641 
00650     virtual uint8_t
00651     getCombiningClass(UChar32 c) const U_OVERRIDE;
00652 
00664     virtual UBool
00665     isNormalized(const UnicodeString &s, UErrorCode &errorCode) const U_OVERRIDE;
00685     virtual UBool
00686     isNormalizedUTF8(StringPiece s, UErrorCode &errorCode) const U_OVERRIDE;
00698     virtual UNormalizationCheckResult
00699     quickCheck(const UnicodeString &s, UErrorCode &errorCode) const U_OVERRIDE;
00711     virtual int32_t
00712     spanQuickCheckYes(const UnicodeString &s, UErrorCode &errorCode) const U_OVERRIDE;
00713 
00722     virtual UBool hasBoundaryBefore(UChar32 c) const U_OVERRIDE;
00723 
00732     virtual UBool hasBoundaryAfter(UChar32 c) const U_OVERRIDE;
00733 
00741     virtual UBool isInert(UChar32 c) const U_OVERRIDE;
00742 private:
00743     UnicodeString &
00744     normalize(const UnicodeString &src,
00745               UnicodeString &dest,
00746               USetSpanCondition spanCondition,
00747               UErrorCode &errorCode) const;
00748 
00749     void
00750     normalizeUTF8(uint32_t options, const char *src, int32_t length,
00751                   ByteSink &sink, Edits *edits,
00752                   USetSpanCondition spanCondition,
00753                   UErrorCode &errorCode) const;
00754 
00755     UnicodeString &
00756     normalizeSecondAndAppend(UnicodeString &first,
00757                              const UnicodeString &second,
00758                              UBool doNormalize,
00759                              UErrorCode &errorCode) const;
00760 
00761     const Normalizer2 &norm2;
00762     const UnicodeSet &set;
00763 };
00764 
00765 U_NAMESPACE_END
00766 
00767 #endif  // !UCONFIG_NO_NORMALIZATION
00768 
00769 #endif /* U_SHOW_CPLUSPLUS_API */
00770 
00771 #endif  // __NORMALIZER2_H__