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
00045 #include "unicode/utypes.h"
00046
00047 #if U_SHOW_CPLUSPLUS_API
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
00680 RegexMatcher(const UnicodeString ®exp, uint32_t flags, UErrorCode &status);
00681
00696 RegexMatcher(UText *regexp, uint32_t flags, UErrorCode &status);
00697
00718 RegexMatcher(const UnicodeString ®exp, const UnicodeString &input,
00719 uint32_t flags, UErrorCode &status);
00720
00741 RegexMatcher(UText *regexp, UText *input,
00742 uint32_t flags, UErrorCode &status);
00743
00744 private:
00756 RegexMatcher(const UnicodeString ®exp, const char16_t *input,
00757 uint32_t flags, UErrorCode &status);
00758 public:
00759
00760
00766 virtual ~RegexMatcher();
00767
00768
00775 virtual UBool matches(UErrorCode &status);
00776
00777
00788 virtual UBool matches(int64_t startIndex, UErrorCode &status);
00789
00790
00804 virtual UBool lookingAt(UErrorCode &status);
00805
00806
00820 virtual UBool lookingAt(int64_t startIndex, UErrorCode &status);
00821
00822
00835 virtual UBool find();
00836
00837
00852 virtual UBool find(UErrorCode &status);
00853
00863 virtual UBool find(int64_t start, UErrorCode &status);
00864
00865
00875 virtual UnicodeString group(UErrorCode &status) const;
00876
00877
00895 virtual UnicodeString group(int32_t groupNum, UErrorCode &status) const;
00896
00902 virtual int32_t groupCount() const;
00903
00904
00919 virtual UText *group(UText *dest, int64_t &group_len, UErrorCode &status) const;
00920
00941 virtual UText *group(int32_t groupNum, UText *dest, int64_t &group_len, UErrorCode &status) const;
00942
00950 virtual int32_t start(UErrorCode &status) const;
00951
00959 virtual int64_t start64(UErrorCode &status) const;
00960
00961
00975 virtual int32_t start(int32_t group, UErrorCode &status) const;
00976
00990 virtual int64_t start64(int32_t group, UErrorCode &status) const;
00991
01005 virtual int32_t end(UErrorCode &status) const;
01006
01020 virtual int64_t end64(UErrorCode &status) const;
01021
01022
01040 virtual int32_t end(int32_t group, UErrorCode &status) const;
01041
01059 virtual int64_t end64(int32_t group, UErrorCode &status) const;
01060
01069 virtual RegexMatcher &reset();
01070
01071
01087 virtual RegexMatcher &reset(int64_t index, UErrorCode &status);
01088
01089
01107 virtual RegexMatcher &reset(const UnicodeString &input);
01108
01109
01123 virtual RegexMatcher &reset(UText *input);
01124
01125
01150 virtual RegexMatcher &refreshInputText(UText *input, UErrorCode &status);
01151
01152 private:
01165 RegexMatcher &reset(const char16_t *input);
01166 public:
01167
01175 virtual const UnicodeString &input() const;
01176
01185 virtual UText *inputText() const;
01186
01197 virtual UText *getInput(UText *dest, UErrorCode &status) const;
01198
01199
01218 virtual RegexMatcher ®ion(int64_t start, int64_t limit, UErrorCode &status);
01219
01231 virtual RegexMatcher ®ion(int64_t regionStart, int64_t regionLimit, int64_t startIndex, UErrorCode &status);
01232
01241 virtual int32_t regionStart() const;
01242
01251 virtual int64_t regionStart64() const;
01252
01253
01262 virtual int32_t regionEnd() const;
01263
01272 virtual int64_t regionEnd64() const;
01273
01282 virtual UBool hasTransparentBounds() const;
01283
01302 virtual RegexMatcher &useTransparentBounds(UBool b);
01303
01304
01312 virtual UBool hasAnchoringBounds() const;
01313
01314
01327 virtual RegexMatcher &useAnchoringBounds(UBool b);
01328
01329
01342 virtual UBool hitEnd() const;
01343
01353 virtual UBool requireEnd() const;
01354
01355
01361 virtual const RegexPattern &pattern() const;
01362
01363
01380 virtual UnicodeString replaceAll(const UnicodeString &replacement, UErrorCode &status);
01381
01382
01403 virtual UText *replaceAll(UText *replacement, UText *dest, UErrorCode &status);
01404
01405
01426 virtual UnicodeString replaceFirst(const UnicodeString &replacement, UErrorCode &status);
01427
01428
01453 virtual UText *replaceFirst(UText *replacement, UText *dest, UErrorCode &status);
01454
01455
01483 virtual RegexMatcher &appendReplacement(UnicodeString &dest,
01484 const UnicodeString &replacement, UErrorCode &status);
01485
01486
01514 virtual RegexMatcher &appendReplacement(UText *dest,
01515 UText *replacement, UErrorCode &status);
01516
01517
01528 virtual UnicodeString &appendTail(UnicodeString &dest);
01529
01530
01544 virtual UText *appendTail(UText *dest, UErrorCode &status);
01545
01546
01570 virtual int32_t split(const UnicodeString &input,
01571 UnicodeString dest[],
01572 int32_t destCapacity,
01573 UErrorCode &status);
01574
01575
01599 virtual int32_t split(UText *input,
01600 UText *dest[],
01601 int32_t destCapacity,
01602 UErrorCode &status);
01603
01625 virtual void setTimeLimit(int32_t limit, UErrorCode &status);
01626
01633 virtual int32_t getTimeLimit() const;
01634
01656 virtual void setStackLimit(int32_t limit, UErrorCode &status);
01657
01665 virtual int32_t getStackLimit() const;
01666
01667
01681 virtual void setMatchCallback(URegexMatchCallback *callback,
01682 const void *context,
01683 UErrorCode &status);
01684
01685
01696 virtual void getMatchCallback(URegexMatchCallback *&callback,
01697 const void *&context,
01698 UErrorCode &status);
01699
01700
01714 virtual void setFindProgressCallback(URegexFindProgressCallback *callback,
01715 const void *context,
01716 UErrorCode &status);
01717
01718
01729 virtual void getFindProgressCallback(URegexFindProgressCallback *&callback,
01730 const void *&context,
01731 UErrorCode &status);
01732
01733 #ifndef U_HIDE_INTERNAL_API
01734
01739 void setTrace(UBool state);
01740 #endif
01741
01747 static UClassID U_EXPORT2 getStaticClassID();
01748
01754 virtual UClassID getDynamicClassID() const;
01755
01756 private:
01757
01758
01759 RegexMatcher();
01760 RegexMatcher(const RegexPattern *pat);
01761 RegexMatcher(const RegexMatcher &other);
01762 RegexMatcher &operator =(const RegexMatcher &rhs);
01763 void init(UErrorCode &status);
01764 void init2(UText *t, UErrorCode &e);
01765
01766 friend class RegexPattern;
01767 friend class RegexCImpl;
01768 public:
01769 #ifndef U_HIDE_INTERNAL_API
01770
01771 void resetPreserveRegion();
01772 #endif
01773 private:
01774
01775
01776
01777
01778
01779 void MatchAt(int64_t startIdx, UBool toEnd, UErrorCode &status);
01780 inline void backTrack(int64_t &inputIdx, int32_t &patIdx);
01781 UBool isWordBoundary(int64_t pos);
01782 UBool isUWordBoundary(int64_t pos);
01783 REStackFrame *resetStack();
01784 inline REStackFrame *StateSave(REStackFrame *fp, int64_t savePatIdx, UErrorCode &status);
01785 void IncrementTime(UErrorCode &status);
01786
01787
01788 inline UBool findProgressInterrupt(int64_t matchIndex, UErrorCode &status);
01789
01790 int64_t appendGroup(int32_t groupNum, UText *dest, UErrorCode &status) const;
01791
01792 UBool findUsingChunk(UErrorCode &status);
01793 void MatchChunkAt(int32_t startIdx, UBool toEnd, UErrorCode &status);
01794 UBool isChunkWordBoundary(int32_t pos);
01795
01796 const RegexPattern *fPattern;
01797 RegexPattern *fPatternOwned;
01798
01799
01800 const UnicodeString *fInput;
01801 UText *fInputText;
01802 UText *fAltInputText;
01803
01804 int64_t fInputLength;
01805 int32_t fFrameSize;
01806
01807 int64_t fRegionStart;
01808 int64_t fRegionLimit;
01809
01810 int64_t fAnchorStart;
01811 int64_t fAnchorLimit;
01812
01813 int64_t fLookStart;
01814 int64_t fLookLimit;
01815
01816
01817 int64_t fActiveStart;
01818 int64_t fActiveLimit;
01819
01820
01821
01822 UBool fTransparentBounds;
01823 UBool fAnchoringBounds;
01824
01825 UBool fMatch;
01826 int64_t fMatchStart;
01827 int64_t fMatchEnd;
01828
01829
01830 int64_t fLastMatchEnd;
01831
01832 int64_t fAppendPosition;
01833
01834
01835
01836 UBool fHitEnd;
01837 UBool fRequireEnd;
01838
01839
01840 UVector64 *fStack;
01841 REStackFrame *fFrame;
01842
01843
01844
01845 int64_t *fData;
01846 int64_t fSmallData[8];
01847
01848 int32_t fTimeLimit;
01849
01850
01851 int32_t fTime;
01852 int32_t fTickCounter;
01853
01854
01855
01856
01857 int32_t fStackLimit;
01858
01859
01860 URegexMatchCallback *fCallbackFn;
01861
01862 const void *fCallbackContext;
01863
01864 URegexFindProgressCallback *fFindProgressCallbackFn;
01865
01866 const void *fFindProgressCallbackContext;
01867
01868
01869 UBool fInputUniStrMaybeMutable;
01870
01871 UBool fTraceDebug;
01872
01873 UErrorCode fDeferredStatus;
01874
01875
01876 RuleBasedBreakIterator *fWordBreakItr;
01877 };
01878
01879 U_NAMESPACE_END
01880 #endif // UCONFIG_NO_REGULAR_EXPRESSIONS
01881
01882 #endif
01883
01884 #endif