ICU 4.6 4.6
uniset.h
Go to the documentation of this file.
00001 /*
00002 ***************************************************************************
00003 * Copyright (C) 1999-2010, International Business Machines Corporation
00004 * and others. All Rights Reserved.
00005 ***************************************************************************
00006 *   Date        Name        Description
00007 *   10/20/99    alan        Creation.
00008 ***************************************************************************
00009 */
00010 
00011 #ifndef UNICODESET_H
00012 #define UNICODESET_H
00013 
00014 #include "unicode/unifilt.h"
00015 #include "unicode/unistr.h"
00016 #include "unicode/uset.h"
00017 
00023 U_NAMESPACE_BEGIN
00024 
00025 class BMPSet;
00026 class ParsePosition;
00027 class SymbolTable;
00028 class UnicodeSetStringSpan;
00029 class UVector;
00030 class RuleCharacterIterator;
00031 
00272 class U_COMMON_API UnicodeSet : public UnicodeFilter {
00273 
00274     int32_t len; // length of list used; 0 <= len <= capacity
00275     int32_t capacity; // capacity of list
00276     UChar32* list; // MUST be terminated with HIGH
00277     BMPSet *bmpSet; // The set is frozen iff either bmpSet or stringSpan is not NULL.
00278     UChar32* buffer; // internal buffer, may be NULL
00279     int32_t bufferCapacity; // capacity of buffer
00280     int32_t patLen;
00281 
00291     UChar *pat;
00292     UVector* strings; // maintained in sorted order
00293     UnicodeSetStringSpan *stringSpan;
00294 
00295 private:
00296     enum { // constants
00297         kIsBogus = 1       // This set is bogus (i.e. not valid)
00298     };
00299     uint8_t fFlags;         // Bit flag (see constants above)
00300 public:
00310     inline UBool isBogus(void) const;
00311     
00328     void setToBogus();
00329 
00330 public:
00331 
00332     enum {
00337         MIN_VALUE = 0,
00338 
00343         MAX_VALUE = 0x10ffff
00344     };
00345 
00346     //----------------------------------------------------------------
00347     // Constructors &c
00348     //----------------------------------------------------------------
00349 
00350 public:
00351 
00356     UnicodeSet();
00357 
00366     UnicodeSet(UChar32 start, UChar32 end);
00367 
00376     UnicodeSet(const UnicodeString& pattern,
00377                UErrorCode& status);
00378 
00391     UnicodeSet(const UnicodeString& pattern,
00392                uint32_t options,
00393                const SymbolTable* symbols,
00394                UErrorCode& status);
00395 
00409     UnicodeSet(const UnicodeString& pattern, ParsePosition& pos,
00410                uint32_t options,
00411                const SymbolTable* symbols,
00412                UErrorCode& status);
00413 
00418     UnicodeSet(const UnicodeSet& o);
00419 
00424     virtual ~UnicodeSet();
00425 
00431     UnicodeSet& operator=(const UnicodeSet& o);
00432 
00444     virtual UBool operator==(const UnicodeSet& o) const;
00445 
00451     UBool operator!=(const UnicodeSet& o) const;
00452 
00462     virtual UnicodeFunctor* clone() const;
00463 
00471     virtual int32_t hashCode(void) const;
00472 
00481     inline static UnicodeSet *fromUSet(USet *uset);
00482 
00491     inline static const UnicodeSet *fromUSet(const USet *uset);
00492     
00500     inline USet *toUSet();
00501 
00502 
00510     inline const USet * toUSet() const;
00511 
00512 
00513     //----------------------------------------------------------------
00514     // Freezable API
00515     //----------------------------------------------------------------
00516 
00525     inline UBool isFrozen() const;
00526 
00540     UnicodeFunctor *freeze();
00541 
00550     UnicodeFunctor *cloneAsThawed() const;
00551 
00552     //----------------------------------------------------------------
00553     // Public API
00554     //----------------------------------------------------------------
00555 
00566     UnicodeSet& set(UChar32 start, UChar32 end);
00567 
00573     static UBool resemblesPattern(const UnicodeString& pattern,
00574                                   int32_t pos);
00575 
00588     UnicodeSet& applyPattern(const UnicodeString& pattern,
00589                              UErrorCode& status);
00590 
00607     UnicodeSet& applyPattern(const UnicodeString& pattern,
00608                              uint32_t options,
00609                              const SymbolTable* symbols,
00610                              UErrorCode& status);
00611 
00643     UnicodeSet& applyPattern(const UnicodeString& pattern,
00644                              ParsePosition& pos,
00645                              uint32_t options,
00646                              const SymbolTable* symbols,
00647                              UErrorCode& status);
00648 
00662     virtual UnicodeString& toPattern(UnicodeString& result,
00663                              UBool escapeUnprintable = FALSE) const;
00664 
00687     UnicodeSet& applyIntPropertyValue(UProperty prop,
00688                                       int32_t value,
00689                                       UErrorCode& ec);
00690 
00720     UnicodeSet& applyPropertyAlias(const UnicodeString& prop,
00721                                    const UnicodeString& value,
00722                                    UErrorCode& ec);
00723 
00732     virtual int32_t size(void) const;
00733 
00740     virtual UBool isEmpty(void) const;
00741 
00749     virtual UBool contains(UChar32 c) const;
00750 
00759     virtual UBool contains(UChar32 start, UChar32 end) const;
00760 
00768     UBool contains(const UnicodeString& s) const;
00769 
00777     virtual UBool containsAll(const UnicodeSet& c) const;
00778 
00786     UBool containsAll(const UnicodeString& s) const;
00787 
00796     UBool containsNone(UChar32 start, UChar32 end) const;
00797 
00805     UBool containsNone(const UnicodeSet& c) const;
00806 
00814     UBool containsNone(const UnicodeString& s) const;
00815 
00824     inline UBool containsSome(UChar32 start, UChar32 end) const;
00825 
00833     inline UBool containsSome(const UnicodeSet& s) const;
00834 
00842     inline UBool containsSome(const UnicodeString& s) const;
00843 
00862     int32_t span(const UChar *s, int32_t length, USetSpanCondition spanCondition) const;
00863 
00876     inline int32_t span(const UnicodeString &s, int32_t start, USetSpanCondition spanCondition) const;
00877 
00895     int32_t spanBack(const UChar *s, int32_t length, USetSpanCondition spanCondition) const;
00896 
00910     inline int32_t spanBack(const UnicodeString &s, int32_t limit, USetSpanCondition spanCondition) const;
00911 
00930     int32_t spanUTF8(const char *s, int32_t length, USetSpanCondition spanCondition) const;
00931 
00949     int32_t spanBackUTF8(const char *s, int32_t length, USetSpanCondition spanCondition) const;
00950 
00955     virtual UMatchDegree matches(const Replaceable& text,
00956                          int32_t& offset,
00957                          int32_t limit,
00958                          UBool incremental);
00959 
00960 private:
00982     static int32_t matchRest(const Replaceable& text,
00983                              int32_t start, int32_t limit,
00984                              const UnicodeString& s);
00985 
00995     int32_t findCodePoint(UChar32 c) const;
00996 
00997 public:
00998 
01006     virtual void addMatchSetTo(UnicodeSet& toUnionTo) const;
01007 
01016     int32_t indexOf(UChar32 c) const;
01017 
01027     UChar32 charAt(int32_t index) const;
01028 
01043     virtual UnicodeSet& add(UChar32 start, UChar32 end);
01044 
01052     UnicodeSet& add(UChar32 c);
01053 
01065     UnicodeSet& add(const UnicodeString& s);
01066 
01067  private:
01073     static int32_t getSingleCP(const UnicodeString& s);
01074 
01075     void _add(const UnicodeString& s);
01076 
01077  public:
01086     UnicodeSet& addAll(const UnicodeString& s);
01087 
01096     UnicodeSet& retainAll(const UnicodeString& s);
01097 
01106     UnicodeSet& complementAll(const UnicodeString& s);
01107 
01116     UnicodeSet& removeAll(const UnicodeString& s);
01117 
01126     static UnicodeSet* U_EXPORT2 createFrom(const UnicodeString& s);
01127 
01128 
01136     static UnicodeSet* U_EXPORT2 createFromAll(const UnicodeString& s);
01137 
01151     virtual UnicodeSet& retain(UChar32 start, UChar32 end);
01152 
01153 
01159     UnicodeSet& retain(UChar32 c);
01160 
01174     virtual UnicodeSet& remove(UChar32 start, UChar32 end);
01175 
01183     UnicodeSet& remove(UChar32 c);
01184 
01194     UnicodeSet& remove(const UnicodeString& s);
01195 
01203     virtual UnicodeSet& complement(void);
01204 
01219     virtual UnicodeSet& complement(UChar32 start, UChar32 end);
01220 
01228     UnicodeSet& complement(UChar32 c);
01229 
01240     UnicodeSet& complement(const UnicodeString& s);
01241 
01254     virtual UnicodeSet& addAll(const UnicodeSet& c);
01255 
01267     virtual UnicodeSet& retainAll(const UnicodeSet& c);
01268 
01280     virtual UnicodeSet& removeAll(const UnicodeSet& c);
01281 
01292     virtual UnicodeSet& complementAll(const UnicodeSet& c);
01293 
01300     virtual UnicodeSet& clear(void);
01301 
01327     UnicodeSet& closeOver(int32_t attribute);
01328 
01335     virtual UnicodeSet &removeAllStrings();
01336 
01344     virtual int32_t getRangeCount(void) const;
01345 
01353     virtual UChar32 getRangeStart(int32_t index) const;
01354 
01362     virtual UChar32 getRangeEnd(int32_t index) const;
01363 
01412     int32_t serialize(uint16_t *dest, int32_t destCapacity, UErrorCode& ec) const;
01413 
01420     virtual UnicodeSet& compact();
01421 
01433     static UClassID U_EXPORT2 getStaticClassID(void);
01434 
01443     virtual UClassID getDynamicClassID(void) const;
01444 
01445 private:
01446 
01447     // Private API for the USet API
01448 
01449     friend class USetAccess;
01450 
01451     int32_t getStringCount() const;
01452 
01453     const UnicodeString* getString(int32_t index) const;
01454 
01455     //----------------------------------------------------------------
01456     // RuleBasedTransliterator support
01457     //----------------------------------------------------------------
01458 
01459 private:
01460 
01466     virtual UBool matchesIndexValue(uint8_t v) const;
01467 
01468 private:
01469 
01470     //----------------------------------------------------------------
01471     // Implementation: Clone as thawed (see ICU4J Freezable)
01472     //----------------------------------------------------------------
01473 
01474     UnicodeSet(const UnicodeSet& o, UBool /* asThawed */);
01475 
01476     //----------------------------------------------------------------
01477     // Implementation: Pattern parsing
01478     //----------------------------------------------------------------
01479 
01480     void applyPattern(RuleCharacterIterator& chars,
01481                       const SymbolTable* symbols,
01482                       UnicodeString& rebuiltPat,
01483                       uint32_t options,
01484                       UErrorCode& ec);
01485 
01486     //----------------------------------------------------------------
01487     // Implementation: Utility methods
01488     //----------------------------------------------------------------
01489 
01490     void ensureCapacity(int32_t newLen, UErrorCode& ec);
01491 
01492     void ensureBufferCapacity(int32_t newLen, UErrorCode& ec);
01493 
01494     void swapBuffers(void);
01495 
01496     UBool allocateStrings(UErrorCode &status);
01497 
01498     UnicodeString& _toPattern(UnicodeString& result,
01499                               UBool escapeUnprintable) const;
01500 
01501     UnicodeString& _generatePattern(UnicodeString& result,
01502                                     UBool escapeUnprintable) const;
01503 
01504     static void _appendToPat(UnicodeString& buf, const UnicodeString& s, UBool escapeUnprintable);
01505 
01506     static void _appendToPat(UnicodeString& buf, UChar32 c, UBool escapeUnprintable);
01507 
01508     //----------------------------------------------------------------
01509     // Implementation: Fundamental operators
01510     //----------------------------------------------------------------
01511 
01512     void exclusiveOr(const UChar32* other, int32_t otherLen, int8_t polarity);
01513 
01514     void add(const UChar32* other, int32_t otherLen, int8_t polarity);
01515 
01516     void retain(const UChar32* other, int32_t otherLen, int8_t polarity);
01517 
01523     static UBool resemblesPropertyPattern(const UnicodeString& pattern,
01524                                           int32_t pos);
01525 
01526     static UBool resemblesPropertyPattern(RuleCharacterIterator& chars,
01527                                           int32_t iterOpts);
01528 
01567     UnicodeSet& applyPropertyPattern(const UnicodeString& pattern,
01568                                      ParsePosition& ppos,
01569                                      UErrorCode &ec);
01570 
01571     void applyPropertyPattern(RuleCharacterIterator& chars,
01572                               UnicodeString& rebuiltPat,
01573                               UErrorCode& ec);
01574 
01575     static const UnicodeSet* getInclusions(int32_t src, UErrorCode &status);
01576 
01581     typedef UBool (*Filter)(UChar32 codePoint, void* context);
01582 
01592     void applyFilter(Filter filter,
01593                      void* context,
01594                      int32_t src,
01595                      UErrorCode &status);
01596 
01600     void setPattern(const UnicodeString& newPat);
01604     void releasePattern();
01605 
01606     friend class UnicodeSetIterator;
01607 };
01608 
01609 
01610 
01611 inline UBool UnicodeSet::operator!=(const UnicodeSet& o) const {
01612     return !operator==(o);
01613 }
01614 
01615 inline UBool UnicodeSet::isFrozen() const {
01616     return (UBool)(bmpSet!=NULL || stringSpan!=NULL);
01617 }
01618 
01619 inline UBool UnicodeSet::containsSome(UChar32 start, UChar32 end) const {
01620     return !containsNone(start, end);
01621 }
01622 
01623 inline UBool UnicodeSet::containsSome(const UnicodeSet& s) const {
01624     return !containsNone(s);
01625 }
01626 
01627 inline UBool UnicodeSet::containsSome(const UnicodeString& s) const {
01628     return !containsNone(s);
01629 }
01630 
01631 inline UBool UnicodeSet::isBogus() const {
01632     return (UBool)(fFlags & kIsBogus);
01633 }
01634 
01635 inline UnicodeSet *UnicodeSet::fromUSet(USet *uset) {
01636     return reinterpret_cast<UnicodeSet *>(uset);
01637 }
01638 
01639 inline const UnicodeSet *UnicodeSet::fromUSet(const USet *uset) {
01640     return reinterpret_cast<const UnicodeSet *>(uset);
01641 }
01642 
01643 inline USet *UnicodeSet::toUSet() {
01644     return reinterpret_cast<USet *>(this);
01645 }
01646 
01647 inline const USet *UnicodeSet::toUSet() const {
01648     return reinterpret_cast<const USet *>(this);
01649 }
01650 
01651 inline int32_t UnicodeSet::span(const UnicodeString &s, int32_t start, USetSpanCondition spanCondition) const {
01652     int32_t sLength=s.length();
01653     if(start<0) {
01654         start=0;
01655     } else if(start>sLength) {
01656         start=sLength;
01657     }
01658     return start+span(s.getBuffer()+start, sLength-start, spanCondition);
01659 }
01660 
01661 inline int32_t UnicodeSet::spanBack(const UnicodeString &s, int32_t limit, USetSpanCondition spanCondition) const {
01662     int32_t sLength=s.length();
01663     if(limit<0) {
01664         limit=0;
01665     } else if(limit>sLength) {
01666         limit=sLength;
01667     }
01668     return spanBack(s.getBuffer(), limit, spanCondition);
01669 }
01670 
01671 U_NAMESPACE_END
01672 
01673 #endif
 All Data Structures Files Functions Variables Typedefs Enumerations Enumerator Friends Defines