ICU 67.1: 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 
00248     virtual void
00249     normalizeUTF8(uint32_t options, StringPiece src, ByteSink &sink,
00250                   Edits *edits, UErrorCode &errorCode) const;
00251 
00266     virtual UnicodeString &
00267     normalizeSecondAndAppend(UnicodeString &first,
00268                              const UnicodeString &second,
00269                              UErrorCode &errorCode) const = 0;
00284     virtual UnicodeString &
00285     append(UnicodeString &first,
00286            const UnicodeString &second,
00287            UErrorCode &errorCode) const = 0;
00288 
00302     virtual UBool
00303     getDecomposition(UChar32 c, UnicodeString &decomposition) const = 0;
00304 
00329     virtual UBool
00330     getRawDecomposition(UChar32 c, UnicodeString &decomposition) const;
00331 
00347     virtual UChar32
00348     composePair(UChar32 a, UChar32 b) const;
00349 
00358     virtual uint8_t
00359     getCombiningClass(UChar32 c) const;
00360 
00375     virtual UBool
00376     isNormalized(const UnicodeString &s, UErrorCode &errorCode) const = 0;
00398     virtual UBool
00399     isNormalizedUTF8(StringPiece s, UErrorCode &errorCode) const;
00400 
00401 
00417     virtual UNormalizationCheckResult
00418     quickCheck(const UnicodeString &s, UErrorCode &errorCode) const = 0;
00419 
00442     virtual int32_t
00443     spanQuickCheckYes(const UnicodeString &s, UErrorCode &errorCode) const = 0;
00444 
00458     virtual UBool hasBoundaryBefore(UChar32 c) const = 0;
00459 
00474     virtual UBool hasBoundaryAfter(UChar32 c) const = 0;
00475 
00489     virtual UBool isInert(UChar32 c) const = 0;
00490 };
00491 
00503 class U_COMMON_API FilteredNormalizer2 : public Normalizer2 {
00504 public:
00515     FilteredNormalizer2(const Normalizer2 &n2, const UnicodeSet &filterSet) :
00516             norm2(n2), set(filterSet) {}
00517 
00522     ~FilteredNormalizer2();
00523 
00537     virtual UnicodeString &
00538     normalize(const UnicodeString &src,
00539               UnicodeString &dest,
00540               UErrorCode &errorCode) const U_OVERRIDE;
00541 
00566     virtual void
00567     normalizeUTF8(uint32_t options, StringPiece src, ByteSink &sink,
00568                   Edits *edits, UErrorCode &errorCode) const U_OVERRIDE;
00569 
00584     virtual UnicodeString &
00585     normalizeSecondAndAppend(UnicodeString &first,
00586                              const UnicodeString &second,
00587                              UErrorCode &errorCode) const U_OVERRIDE;
00602     virtual UnicodeString &
00603     append(UnicodeString &first,
00604            const UnicodeString &second,
00605            UErrorCode &errorCode) const U_OVERRIDE;
00606 
00618     virtual UBool
00619     getDecomposition(UChar32 c, UnicodeString &decomposition) const U_OVERRIDE;
00620 
00632     virtual UBool
00633     getRawDecomposition(UChar32 c, UnicodeString &decomposition) const U_OVERRIDE;
00634 
00645     virtual UChar32
00646     composePair(UChar32 a, UChar32 b) const U_OVERRIDE;
00647 
00656     virtual uint8_t
00657     getCombiningClass(UChar32 c) const U_OVERRIDE;
00658 
00670     virtual UBool
00671     isNormalized(const UnicodeString &s, UErrorCode &errorCode) const U_OVERRIDE;
00693     virtual UBool
00694     isNormalizedUTF8(StringPiece s, UErrorCode &errorCode) const U_OVERRIDE;
00706     virtual UNormalizationCheckResult
00707     quickCheck(const UnicodeString &s, UErrorCode &errorCode) const U_OVERRIDE;
00719     virtual int32_t
00720     spanQuickCheckYes(const UnicodeString &s, UErrorCode &errorCode) const U_OVERRIDE;
00721 
00730     virtual UBool hasBoundaryBefore(UChar32 c) const U_OVERRIDE;
00731 
00740     virtual UBool hasBoundaryAfter(UChar32 c) const U_OVERRIDE;
00741 
00749     virtual UBool isInert(UChar32 c) const U_OVERRIDE;
00750 private:
00751     UnicodeString &
00752     normalize(const UnicodeString &src,
00753               UnicodeString &dest,
00754               USetSpanCondition spanCondition,
00755               UErrorCode &errorCode) const;
00756 
00757     void
00758     normalizeUTF8(uint32_t options, const char *src, int32_t length,
00759                   ByteSink &sink, Edits *edits,
00760                   USetSpanCondition spanCondition,
00761                   UErrorCode &errorCode) const;
00762 
00763     UnicodeString &
00764     normalizeSecondAndAppend(UnicodeString &first,
00765                              const UnicodeString &second,
00766                              UBool doNormalize,
00767                              UErrorCode &errorCode) const;
00768 
00769     const Normalizer2 &norm2;
00770     const UnicodeSet &set;
00771 };
00772 
00773 U_NAMESPACE_END
00774 
00775 #endif  // !UCONFIG_NO_NORMALIZATION
00776 
00777 #endif /* U_SHOW_CPLUSPLUS_API */
00778 
00779 #endif  // __NORMALIZER2_H__