summaryrefslogtreecommitdiffabout
path: root/shared-code/RegEx.h
authorMichael Krelin <hacker@klever.net>2004-07-05 01:53:09 (UTC)
committer Michael Krelin <hacker@klever.net>2004-07-05 01:53:09 (UTC)
commit955d4e00adc9f39ab93bf21f07506eb75b013c70 (patch) (unidiff)
tree92493a2c9ac206b822e24a9e5a6f6b1589be6afb /shared-code/RegEx.h
downloadT42-955d4e00adc9f39ab93bf21f07506eb75b013c70.zip
T42-955d4e00adc9f39ab93bf21f07506eb75b013c70.tar.gz
T42-955d4e00adc9f39ab93bf21f07506eb75b013c70.tar.bz2
initial commit into svn repository
git-svn-id: http://svn.klever.net/kin/T42/trunk@1 fe716a7a-6dde-0310-88d9-d003556173a8
Diffstat (limited to 'shared-code/RegEx.h') (more/less context) (ignore whitespace changes)
-rw-r--r--shared-code/RegEx.h158
1 files changed, 158 insertions, 0 deletions
diff --git a/shared-code/RegEx.h b/shared-code/RegEx.h
new file mode 100644
index 0000000..2534768
--- a/dev/null
+++ b/shared-code/RegEx.h
@@ -0,0 +1,158 @@
1#ifndef __REGEX_H
2#define __REGEX_H
3
4 class CRegEx{
5public:
6 CString GetMatch(int match=0);
7 CString m_Input;
8 struct CMatch{
9 CMatch() : m_Begin(-1), m_End(-1) {}
10 int GetLength() { return m_End-m_Begin; }
11 int m_Begin;
12 int m_End;
13 };
14 typedefCArray<CMatch,CMatch&> CMatchBox;
15 enum{
16 matchMatch = 0,
17 matchPreMatch = -1,
18 matchPostMatch = -2
19 };
20 CMatchBox m_Matches;
21 enum{
22 charOut = 256,
23 charBOL, charEOL, charBOLEOL, charNothing, charBOW, charEOW,
24 charMaxCode = charEOW,
25 charNNChars = (charMaxCode-255)
26 };
27 int m_mFlags;
28 enum{
29 regeSuccess = 0,
30 regeNoMatch = 1, regeBadPattern, regeCollate, regeCType, regeEscape, regeSubReg, regeBracket,
31 regeParen, regeBrace, regeBadBrace, regeRange, regeSpace, regeBadRepeat, regeEmpty, regeAssert,
32 regeInvArg
33 };
34 int m_Error;
35 CRegEx();
36 BOOL m_bCompiled;
37 CString m_Pattern;
38 BOOL m_bBackRefs;
39 int m_Pluses;
40 CString m_Must;
41 BYTE m_Category[CHAR_MAX-CHAR_MIN+1];
42 int m_Categories;
43 int m_EOLs;
44 int m_BOLs;
45 int m_iFlags;
46 int m_Subexps;
47 struct CSop {
48 void Dump(CDumpContext& dc);
49 CSop(){}
50 CSop(BYTE op,DWORD opnd=0) : m_Operator(op), m_Operand(opnd) {}
51 BOOL operator==(CSop& other) {return m_Operator==other.m_Operator && m_Operand==other.m_Operand;}
52 BOOL operator!=(CSop& other) { return !((*this)==other);}
53 enum{
54 opEnd = 1, opChar, opBOL, opEOL, opAny, opAnyOf, opBackRef0, opBackRef1,
55 opPlus0, opPlus1, opQuest0, opQuest1, opLeftParen, opRightParen, opChoice0,
56 opOr0, opOr1, opChoice1, opBOW, opEOW
57 };
58 BYTE m_Operator;
59 DWORD m_Operand;
60 enum{
61 stCurrent = 1, stFresh = 2, stTemp = 4, stEmpty = 8
62 };
63 BYTE m_MatchData;
64 };
65 typedef CArray<CSop,CSop&> CStrip;
66 CStrip m_Strip;
67 int m_Flags;
68 struct CSet{
69 CSet() : m_Hash(0) { memset(m_Set,0,sizeof(m_Set)); }
70 CSet(CSet& src) { (*this)=src; }
71 CSet& operator=(CSet& src) { memmove(this,&src,sizeof(*this)); return *this; }
72 BOOL operator==(CSet& other) { if(m_Hash!=other.m_Hash)return FALSE;return !memcmp(m_Set,other.m_Set,sizeof(m_Set)); }
73 enum{
74 size = (CHAR_MAX-CHAR_MIN+1)
75 };
76 BOOL m_Set[size];
77 BYTE m_Hash;
78 public:
79 UCHAR GetOnly();
80 void Sub(UCHAR c);
81 BOOL IsIn(UCHAR c);
82 void Add(UCHAR c);
83 };
84 typedef CArray<CSet,CSet&> CSets;
85 CSets m_Sets;
86 enum{
87 // Compile flags
88 regBasic = 0, regExtended = 1,
89 regIgnoreCase = 2,
90 regNoSubExpressions = 4,// Also works for matching.
91 regNewLine = 16,
92 regLiteral = 32,
93 // Match Flags
94 regNotBOL = 1,
95 regNotEOL = 2,
96 regOneMatch=64,
97 regBackRefs=128,
98 // iFlags
99 iflagsUseBOL=1, iflagsUseEOL=2, iflagsBad=4
100 };
101 CString Replace(LPCTSTR src,LPCTSTR rep,int flags=0);
102 BOOL Match(LPCTSTR src,int flags=0);
103 BOOL Compile(LPCTSTR regex,int flags=0);
104private:
105 #ifdef_DEBUG
106 void DumpStrip(CDumpContext& dc);
107#endif
108 LPCTSTR MatchBackRef(LPCTSTR begin,LPCTSTR end,int from,int to,int level);
109 typedef CArray<LPCTSTR,LPCTSTR> CStrPtrArray;
110 CStrPtrArray m_mLastPos;
111 LPCTSTR MatchDissect(LPCTSTR begin,LPCTSTR end,int from,int to);
112 LPCTSTR MatchSlow(LPCTSTR begin,LPCTSTR end,int from,int to);
113 LPCTSTR m_cOldP;
114 BOOL MatchStatesEqual(BYTE m1,BYTE m2);
115 LPCTSTR m_mBegin;
116 void MatchStatesCopy(BYTE dst,BYTE src);
117 void MatchStep(int from,int to,BYTE maskBefore,int charCode,BYTE maskAfter);
118 void MatchStatesClear(BYTE mask);
119 LPCTSTR MatchFast(LPCTSTR begin);
120 LPCTSTR m_mEnd;
121 LPCTSTR m_mPointer;
122 BOOL ParseBREexp(BOOL ordinaryStar);
123 void ParseBRE(int stopa=0,int stopb=0);
124 void ParseLiteral();
125 int CountPluses();
126 void FigureMust();
127 BOOL IsInSameSets(UCHAR c1,UCHAR c2);
128 BOOL IsInSets(UCHAR c);
129 void Categorize();
130 int StripDuplicate(int from,int to);
131 void EmitRepeat(int pos,int from,int to);
132 UCHAR ParseBracketSymbol();
133 UCHAR ParseBracketCollatingElement(UCHAR term);
134 void ParseBracketEClass(CSet& cset);
135 void ParseBracketCClass(CSet& cset);
136 void ParseBracketTerm(CSet& cset);
137 int StoreSet(CSet& cset);
138 void ParseBracket();
139 int ParseCount();
140 void EmitNonNewLineAny();
141 void EmitOrdinary(UCHAR c);
142 void StripInsert(int pos,CSop& sop);
143 void ParseEREexp();
144 void ParseERE(int stop=0);
145 struct CParenthesis{
146 long m_Begin;
147 long m_End;
148 CParenthesis(long b=0,long e=0) : m_Begin(b), m_End(e) {}
149 };
150 typedef CArray<CParenthesis,CParenthesis&> CParens;
151 CParens m_ParseParens;
152 int m_ParsePointer;
153};
154 #ifdef_DEBUG
155inline CDumpContext& operator<<(CDumpContext& dc, CRegEx::CSop& sop) { sop.Dump(dc); return dc; }
156#endif
157
158#endif // __REGEX_H \ No newline at end of file