ICU 63.1: 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 !UCONFIG_NO_NORMALIZATION
00030 
00031 #include "unicode/stringpiece.h"
00032 #include "unicode/uniset.h"
00033 #include "unicode/unistr.h"
00034 #include "unicode/unorm2.h"
00035 
00036 U_NAMESPACE_BEGIN
00037 
00038 class ByteSink;
00039 
00083 class U_COMMON_API Normalizer2 : public UObject {
00084 public:
00089     ~Normalizer2();
00090 
00102     static const Normalizer2 *
00103     getNFCInstance(UErrorCode &errorCode);
00104 
00116     static const Normalizer2 *
00117     getNFDInstance(UErrorCode &errorCode);
00118 
00130     static const Normalizer2 *
00131     getNFKCInstance(UErrorCode &errorCode);
00132 
00144     static const Normalizer2 *
00145     getNFKDInstance(UErrorCode &errorCode);
00146 
00158     static const Normalizer2 *
00159     getNFKCCasefoldInstance(UErrorCode &errorCode);
00160 
00182     static const Normalizer2 *
00183     getInstance(const char *packageName,
00184                 const char *name,
00185                 UNormalization2Mode mode,
00186                 UErrorCode &errorCode);
00187 
00198     UnicodeString
00199     normalize(const UnicodeString &src, UErrorCode &errorCode) const {
00200         UnicodeString result;
00201         normalize(src, result, errorCode);
00202         return result;
00203     }
00217     virtual UnicodeString &
00218     normalize(const UnicodeString &src,
00219               UnicodeString &dest,
00220               UErrorCode &errorCode) const = 0;
00221 
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;
00396     virtual UBool
00397     isNormalizedUTF8(StringPiece s, UErrorCode &errorCode) const;
00398 
00399 
00415     virtual UNormalizationCheckResult
00416     quickCheck(const UnicodeString &s, UErrorCode &errorCode) const = 0;
00417 
00440     virtual int32_t
00441     spanQuickCheckYes(const UnicodeString &s, UErrorCode &errorCode) const = 0;
00442 
00456     virtual UBool hasBoundaryBefore(UChar32 c) const = 0;
00457 
00472     virtual UBool hasBoundaryAfter(UChar32 c) const = 0;
00473 
00487     virtual UBool isInert(UChar32 c) const = 0;
00488 };
00489 
00501 class U_COMMON_API FilteredNormalizer2 : public Normalizer2 {
00502 public:
00513     FilteredNormalizer2(const Normalizer2 &n2, const UnicodeSet &filterSet) :
00514             norm2(n2), set(filterSet) {}
00515 
00520     ~FilteredNormalizer2();
00521 
00535     virtual UnicodeString &
00536     normalize(const UnicodeString &src,
00537               UnicodeString &dest,
00538               UErrorCode &errorCode) const U_OVERRIDE;
00539 
00564     virtual void
00565     normalizeUTF8(uint32_t options, StringPiece src, ByteSink &sink,
00566                   Edits *edits, UErrorCode &errorCode) const U_OVERRIDE;
00567 
00582     virtual UnicodeString &
00583     normalizeSecondAndAppend(UnicodeString &first,
00584                              const UnicodeString &second,
00585                              UErrorCode &errorCode) const U_OVERRIDE;
00600     virtual UnicodeString &
00601     append(UnicodeString &first,
00602            const UnicodeString &second,
00603            UErrorCode &errorCode) const U_OVERRIDE;
00604 
00616     virtual UBool
00617     getDecomposition(UChar32 c, UnicodeString &decomposition) const U_OVERRIDE;
00618 
00630     virtual UBool
00631     getRawDecomposition(UChar32 c, UnicodeString &decomposition) const U_OVERRIDE;
00632 
00643     virtual UChar32
00644     composePair(UChar32 a, UChar32 b) const U_OVERRIDE;
00645 
00654     virtual uint8_t
00655     getCombiningClass(UChar32 c) const U_OVERRIDE;
00656 
00668     virtual UBool
00669     isNormalized(const UnicodeString &s, UErrorCode &errorCode) const U_OVERRIDE;
00691     virtual UBool
00692     isNormalizedUTF8(StringPiece s, UErrorCode &errorCode) const U_OVERRIDE;
00704     virtual UNormalizationCheckResult
00705     quickCheck(const UnicodeString &s, UErrorCode &errorCode) const U_OVERRIDE;
00717     virtual int32_t
00718     spanQuickCheckYes(const UnicodeString &s, UErrorCode &errorCode) const U_OVERRIDE;
00719 
00728     virtual UBool hasBoundaryBefore(UChar32 c) const U_OVERRIDE;
00729 
00738     virtual UBool hasBoundaryAfter(UChar32 c) const U_OVERRIDE;
00739 
00747     virtual UBool isInert(UChar32 c) const U_OVERRIDE;
00748 private:
00749     UnicodeString &
00750     normalize(const UnicodeString &src,
00751               UnicodeString &dest,
00752               USetSpanCondition spanCondition,
00753               UErrorCode &errorCode) const;
00754 
00755     void
00756     normalizeUTF8(uint32_t options, const char *src, int32_t length,
00757                   ByteSink &sink, Edits *edits,
00758                   USetSpanCondition spanCondition,
00759                   UErrorCode &errorCode) const;
00760 
00761     UnicodeString &
00762     normalizeSecondAndAppend(UnicodeString &first,
00763                              const UnicodeString &second,
00764                              UBool doNormalize,
00765                              UErrorCode &errorCode) const;
00766 
00767     const Normalizer2 &norm2;
00768     const UnicodeSet &set;
00769 };
00770 
00771 U_NAMESPACE_END
00772 
00773 #endif  // !UCONFIG_NO_NORMALIZATION
00774 #endif  // __NORMALIZER2_H__