00001
00002
00003
00004
00005
00006
00007
00008
00009
00010
00011
00012
00013
00014
00015
00016
00017
00018 #ifndef REGEX_H
00019 #define REGEX_H
00020
00021
00022
00047 #include "unicode/utypes.h"
00048
00049 #if !UCONFIG_NO_REGULAR_EXPRESSIONS
00050
00051 #include "unicode/uobject.h"
00052 #include "unicode/unistr.h"
00053 #include "unicode/utext.h"
00054 #include "unicode/parseerr.h"
00055
00056 #include "unicode/uregex.h"
00057
00058
00059
00060 struct UHashtable;
00061
00062 U_NAMESPACE_BEGIN
00063
00064 struct Regex8BitSet;
00065 class RegexCImpl;
00066 class RegexMatcher;
00067 class RegexPattern;
00068 struct REStackFrame;
00069 class RuleBasedBreakIterator;
00070 class UnicodeSet;
00071 class UVector;
00072 class UVector32;
00073 class UVector64;
00074
00075
00087 class U_I18N_API RegexPattern U_FINAL : public UObject {
00088 public:
00089
00097 RegexPattern();
00098
00105 RegexPattern(const RegexPattern &source);
00106
00112 virtual ~RegexPattern();
00113
00122 UBool operator==(const RegexPattern& that) const;
00123
00132 inline UBool operator!=(const RegexPattern& that) const {return ! operator ==(that);}
00133
00139 RegexPattern &operator =(const RegexPattern &source);
00140
00148 virtual RegexPattern *clone() const;
00149
00150
00175 static RegexPattern * U_EXPORT2 compile( const UnicodeString ®ex,
00176 UParseError &pe,
00177 UErrorCode &status);
00178
00205 static RegexPattern * U_EXPORT2 compile( UText *regex,
00206 UParseError &pe,
00207 UErrorCode &status);
00208
00233 static RegexPattern * U_EXPORT2 compile( const UnicodeString ®ex,
00234 uint32_t flags,
00235 UParseError &pe,
00236 UErrorCode &status);
00237
00264 static RegexPattern * U_EXPORT2 compile( UText *regex,
00265 uint32_t flags,
00266 UParseError &pe,
00267 UErrorCode &status);
00268
00291 static RegexPattern * U_EXPORT2 compile( const UnicodeString ®ex,
00292 uint32_t flags,
00293 UErrorCode &status);
00294
00319 static RegexPattern * U_EXPORT2 compile( UText *regex,
00320 uint32_t flags,
00321 UErrorCode &status);
00322
00328 virtual uint32_t flags() const;
00329
00347 virtual RegexMatcher *matcher(const UnicodeString &input,
00348 UErrorCode &status) const;
00349
00350 private:
00363 RegexMatcher *matcher(const char16_t *input,
00364 UErrorCode &status) const;
00365 public:
00366
00367
00379 virtual RegexMatcher *matcher(UErrorCode &status) const;
00380
00381
00396 static UBool U_EXPORT2 matches(const UnicodeString ®ex,
00397 const UnicodeString &input,
00398 UParseError &pe,
00399 UErrorCode &status);
00400
00415 static UBool U_EXPORT2 matches(UText *regex,
00416 UText *input,
00417 UParseError &pe,
00418 UErrorCode &status);
00419
00428 virtual UnicodeString pattern() const;
00429
00430
00441 virtual UText *patternText(UErrorCode &status) const;
00442
00443
00457 virtual int32_t groupNumberFromName(const UnicodeString &groupName, UErrorCode &status) const;
00458
00459
00476 virtual int32_t groupNumberFromName(const char *groupName, int32_t nameLength, UErrorCode &status) const;
00477
00478
00517 virtual int32_t split(const UnicodeString &input,
00518 UnicodeString dest[],
00519 int32_t destCapacity,
00520 UErrorCode &status) const;
00521
00522
00561 virtual int32_t split(UText *input,
00562 UText *dest[],
00563 int32_t destCapacity,
00564 UErrorCode &status) const;
00565
00566
00572 virtual UClassID getDynamicClassID() const;
00573
00579 static UClassID U_EXPORT2 getStaticClassID();
00580
00581 private:
00582
00583
00584
00585 UText *fPattern;
00586 UnicodeString *fPatternString;
00587 uint32_t fFlags;
00588
00589 UVector64 *fCompiledPat;
00590 UnicodeString fLiteralText;
00591
00592
00593 UVector *fSets;
00594 Regex8BitSet *fSets8;
00595
00596
00597 UErrorCode fDeferredStatus;
00598
00599
00600 int32_t fMinMatchLen;
00601
00602
00603
00604
00605 int32_t fFrameSize;
00606
00607
00608 int32_t fDataSize;
00609
00610
00611
00612 UVector32 *fGroupMap;
00613
00614
00615 UnicodeSet **fStaticSets;
00616
00617
00618 Regex8BitSet *fStaticSets8;
00619
00620
00621 int32_t fStartType;
00622 int32_t fInitialStringIdx;
00623 int32_t fInitialStringLen;
00624 UnicodeSet *fInitialChars;
00625 UChar32 fInitialChar;
00626 Regex8BitSet *fInitialChars8;
00627 UBool fNeedsAltInput;
00628
00629 UHashtable *fNamedCaptureMap;
00630
00631 friend class RegexCompile;
00632 friend class RegexMatcher;
00633 friend class RegexCImpl;
00634
00635
00636
00637
00638 void init();
00639 void zap();
00640
00641 void dumpOp(int32_t index) const;
00642
00643 public:
00644 #ifndef U_HIDE_INTERNAL_API
00645
00649 void dumpPattern() const;
00650 #endif
00651 };
00652
00653
00654
00664 class U_I18N_API RegexMatcher U_FINAL : public UObject {
00665 public:
00666
00681 RegexMatcher(const UnicodeString ®exp, uint32_t flags, UErrorCode &status);
00682
00698 RegexMatcher(UText *regexp, uint32_t flags, UErrorCode &status);
00699
00721 RegexMatcher(const UnicodeString ®exp, const UnicodeString &input,
00722 uint32_t flags, UErrorCode &status);
00723
00745 RegexMatcher(UText *regexp, UText *input,
00746 uint32_t flags, UErrorCode &status);
00747
00748 private:
00761 RegexMatcher(const UnicodeString ®exp, const char16_t *input,
00762 uint32_t flags, UErrorCode &status);
00763 public:
00764
00765
00771 virtual ~RegexMatcher();
00772
00773
00780 virtual UBool matches(UErrorCode &status);
00781
00782
00793 virtual UBool matches(int64_t startIndex, UErrorCode &status);
00794
00795
00809 virtual UBool lookingAt(UErrorCode &status);
00810
00811
00825 virtual UBool lookingAt(int64_t startIndex, UErrorCode &status);
00826
00827
00840 virtual UBool find();
00841
00842
00856 virtual UBool find(UErrorCode &status);
00857
00867 virtual UBool find(int64_t start, UErrorCode &status);
00868
00869
00879 virtual UnicodeString group(UErrorCode &status) const;
00880
00881
00899 virtual UnicodeString group(int32_t groupNum, UErrorCode &status) const;
00900
00906 virtual int32_t groupCount() const;
00907
00908
00923 virtual UText *group(UText *dest, int64_t &group_len, UErrorCode &status) const;
00924
00945 virtual UText *group(int32_t groupNum, UText *dest, int64_t &group_len, UErrorCode &status) const;
00946
00954 virtual int32_t start(UErrorCode &status) const;
00955
00963 virtual int64_t start64(UErrorCode &status) const;
00964
00965
00979 virtual int32_t start(int32_t group, UErrorCode &status) const;
00980
00994 virtual int64_t start64(int32_t group, UErrorCode &status) const;
00995
01009 virtual int32_t end(UErrorCode &status) const;
01010
01024 virtual int64_t end64(UErrorCode &status) const;
01025
01026
01044 virtual int32_t end(int32_t group, UErrorCode &status) const;
01045
01063 virtual int64_t end64(int32_t group, UErrorCode &status) const;
01064
01073 virtual RegexMatcher &reset();
01074
01075
01091 virtual RegexMatcher &reset(int64_t index, UErrorCode &status);
01092
01093
01111 virtual RegexMatcher &reset(const UnicodeString &input);
01112
01113
01127 virtual RegexMatcher &reset(UText *input);
01128
01129
01154 virtual RegexMatcher &refreshInputText(UText *input, UErrorCode &status);
01155
01156 private:
01169 RegexMatcher &reset(const char16_t *input);
01170 public:
01171
01179 virtual const UnicodeString &input() const;
01180
01189 virtual UText *inputText() const;
01190
01201 virtual UText *getInput(UText *dest, UErrorCode &status) const;
01202
01203
01222 virtual RegexMatcher ®ion(int64_t start, int64_t limit, UErrorCode &status);
01223
01235 virtual RegexMatcher ®ion(int64_t regionStart, int64_t regionLimit, int64_t startIndex, UErrorCode &status);
01236
01245 virtual int32_t regionStart() const;
01246
01255 virtual int64_t regionStart64() const;
01256
01257
01266 virtual int32_t regionEnd() const;
01267
01276 virtual int64_t regionEnd64() const;
01277
01286 virtual UBool hasTransparentBounds() const;
01287
01306 virtual RegexMatcher &useTransparentBounds(UBool b);
01307
01308
01316 virtual UBool hasAnchoringBounds() const;
01317
01318
01331 virtual RegexMatcher &useAnchoringBounds(UBool b);
01332
01333
01346 virtual UBool hitEnd() const;
01347
01357 virtual UBool requireEnd() const;
01358
01359
01365 virtual const RegexPattern &pattern() const;
01366
01367
01384 virtual UnicodeString replaceAll(const UnicodeString &replacement, UErrorCode &status);
01385
01386
01407 virtual UText *replaceAll(UText *replacement, UText *dest, UErrorCode &status);
01408
01409
01430 virtual UnicodeString replaceFirst(const UnicodeString &replacement, UErrorCode &status);
01431
01432
01457 virtual UText *replaceFirst(UText *replacement, UText *dest, UErrorCode &status);
01458
01459
01487 virtual RegexMatcher &appendReplacement(UnicodeString &dest,
01488 const UnicodeString &replacement, UErrorCode &status);
01489
01490
01518 virtual RegexMatcher &appendReplacement(UText *dest,
01519 UText *replacement, UErrorCode &status);
01520
01521
01532 virtual UnicodeString &appendTail(UnicodeString &dest);
01533
01534
01548 virtual UText *appendTail(UText *dest, UErrorCode &status);
01549
01550
01574 virtual int32_t split(const UnicodeString &input,
01575 UnicodeString dest[],
01576 int32_t destCapacity,
01577 UErrorCode &status);
01578
01579
01603 virtual int32_t split(UText *input,
01604 UText *dest[],
01605 int32_t destCapacity,
01606 UErrorCode &status);
01607
01629 virtual void setTimeLimit(int32_t limit, UErrorCode &status);
01630
01637 virtual int32_t getTimeLimit() const;
01638
01660 virtual void setStackLimit(int32_t limit, UErrorCode &status);
01661
01669 virtual int32_t getStackLimit() const;
01670
01671
01685 virtual void setMatchCallback(URegexMatchCallback *callback,
01686 const void *context,
01687 UErrorCode &status);
01688
01689
01700 virtual void getMatchCallback(URegexMatchCallback *&callback,
01701 const void *&context,
01702 UErrorCode &status);
01703
01704
01718 virtual void setFindProgressCallback(URegexFindProgressCallback *callback,
01719 const void *context,
01720 UErrorCode &status);
01721
01722
01733 virtual void getFindProgressCallback(URegexFindProgressCallback *&callback,
01734 const void *&context,
01735 UErrorCode &status);
01736
01737 #ifndef U_HIDE_INTERNAL_API
01738
01743 void setTrace(UBool state);
01744 #endif
01745
01751 static UClassID U_EXPORT2 getStaticClassID();
01752
01758 virtual UClassID getDynamicClassID() const;
01759
01760 private:
01761
01762
01763 RegexMatcher();
01764 RegexMatcher(const RegexPattern *pat);
01765 RegexMatcher(const RegexMatcher &other);
01766 RegexMatcher &operator =(const RegexMatcher &rhs);
01767 void init(UErrorCode &status);
01768 void init2(UText *t, UErrorCode &e);
01769
01770 friend class RegexPattern;
01771 friend class RegexCImpl;
01772 public:
01773 #ifndef U_HIDE_INTERNAL_API
01774
01775 void resetPreserveRegion();
01776 #endif
01777 private:
01778
01779
01780
01781
01782
01783 void MatchAt(int64_t startIdx, UBool toEnd, UErrorCode &status);
01784 inline void backTrack(int64_t &inputIdx, int32_t &patIdx);
01785 UBool isWordBoundary(int64_t pos);
01786 UBool isUWordBoundary(int64_t pos);
01787 REStackFrame *resetStack();
01788 inline REStackFrame *StateSave(REStackFrame *fp, int64_t savePatIdx, UErrorCode &status);
01789 void IncrementTime(UErrorCode &status);
01790
01791
01792 inline UBool findProgressInterrupt(int64_t matchIndex, UErrorCode &status);
01793
01794 int64_t appendGroup(int32_t groupNum, UText *dest, UErrorCode &status) const;
01795
01796 UBool findUsingChunk(UErrorCode &status);
01797 void MatchChunkAt(int32_t startIdx, UBool toEnd, UErrorCode &status);
01798 UBool isChunkWordBoundary(int32_t pos);
01799
01800 const RegexPattern *fPattern;
01801 RegexPattern *fPatternOwned;
01802
01803
01804 const UnicodeString *fInput;
01805 UText *fInputText;
01806 UText *fAltInputText;
01807
01808 int64_t fInputLength;
01809 int32_t fFrameSize;
01810
01811 int64_t fRegionStart;
01812 int64_t fRegionLimit;
01813
01814 int64_t fAnchorStart;
01815 int64_t fAnchorLimit;
01816
01817 int64_t fLookStart;
01818 int64_t fLookLimit;
01819
01820
01821 int64_t fActiveStart;
01822 int64_t fActiveLimit;
01823
01824
01825
01826 UBool fTransparentBounds;
01827 UBool fAnchoringBounds;
01828
01829 UBool fMatch;
01830 int64_t fMatchStart;
01831 int64_t fMatchEnd;
01832
01833
01834 int64_t fLastMatchEnd;
01835
01836 int64_t fAppendPosition;
01837
01838
01839
01840 UBool fHitEnd;
01841 UBool fRequireEnd;
01842
01843
01844 UVector64 *fStack;
01845 REStackFrame *fFrame;
01846
01847
01848
01849 int64_t *fData;
01850 int64_t fSmallData[8];
01851
01852 int32_t fTimeLimit;
01853
01854
01855 int32_t fTime;
01856 int32_t fTickCounter;
01857
01858
01859
01860
01861 int32_t fStackLimit;
01862
01863
01864 URegexMatchCallback *fCallbackFn;
01865
01866 const void *fCallbackContext;
01867
01868 URegexFindProgressCallback *fFindProgressCallbackFn;
01869
01870 const void *fFindProgressCallbackContext;
01871
01872
01873 UBool fInputUniStrMaybeMutable;
01874
01875 UBool fTraceDebug;
01876
01877 UErrorCode fDeferredStatus;
01878
01879
01880 RuleBasedBreakIterator *fWordBreakItr;
01881 };
01882
01883 U_NAMESPACE_END
01884 #endif // UCONFIG_NO_REGULAR_EXPRESSIONS
01885 #endif