SF Area/Download Docs: User Internal/Devel		csregex.h Go to the documentation of this file. //--------------------------------------------------------------------------- // // Author: Cleo Saulnier // Date: July 14, 2005 // // Regular expression engine. // // See LICENSE file for legal information. // Basically public domain, but author keeps moral rights. // Original author's name above must remain in source. // Simply put, you can do what you want, but don't misrepresent // the origins, and contents, of this code. // // http://sourceforge.net/projects/csregex/ // //--------------------------------------------------------------------------- #ifndef csregexH #define csregexH #include <vector> #ifdef _UNICODE #include <tchar.h> #endif using namespace std; #ifndef _UNICODE #ifndef __TCHAR_DEFINED #define _TCHAR char #define _TUCHAR unsigned char #endif #endif // Internal compiled regular expression format. /* GRP chars ranges : Begin Group GRN chars ranges : Begin Group Negated DOT : match anything CHR char : match CHaR STR count char+ : match STRing BCK : match BaCKstring (0-9) OPA next : open parentheses (this OP has 10 values for each index). CPA : close parentheses OPN next : Open Parentheses, No backreference. CPN : Close Parentheses, No backreference. // Qualifiers RPT prev start end : Repeat matching. STA prev : * 0-inf PLS prev : + 1-inf QES prev : ? 0-1 RPT_LAZY prev start end : Repeat matching. STA_LAZY prev : * 0-inf PLS_LAZY prev : + 1-inf QES_LAZY prev : ? 0-1 ORM next : or matching HED : match beginning of string. TAL : match 0 TAiL END : end matching */ class CSMatch { public: unsigned int buf_idx; unsigned int start; unsigned int end; int rpt; }; class CSPar { public: unsigned int buf_idx; unsigned int match_idx; }; class CSRegEx { private: enum RGTok { rg_end = 0, rg_str = 1, rg_grp = 2, rg_grn = 3, rg_hed = 32, rg_tal = 33, rg_dot = 34, rg_cpn = 35, rg_cpa = 36, rg_cpa1 = 37, rg_cpa2 = 38, rg_cpa3 = 39, rg_cpa4 = 40, rg_cpa5 = 41, rg_cpa6 = 42, rg_cpa7 = 43, rg_cpa8 = 44, rg_cpa9 = 45, rg_bck = 46, rg_bck1 = 47, rg_bck2 = 48, rg_bck3 = 49, rg_bck4 = 50, rg_bck5 = 51, rg_bck6 = 52, rg_bck7 = 53, rg_bck8 = 54, rg_bck9 = 55, rg_chr = 64, rg_opa = 96, rg_opa1 = 97, rg_opa2 = 98, rg_opa3 = 99, rg_opa4 = 100, rg_opa5 = 101, rg_opa6 = 102, rg_opa7 = 103, rg_opa8 = 104, rg_opa9 = 105, rg_orm = 106, rg_opn = 108, rg_sta = 109, rg_pls = 110, rg_qes = 111, rg_sta_lazy = 112, rg_pls_lazy = 113, rg_qes_lazy = 114, rg_rpt = 224, rg_rpt_lazy = 225, }; const _TUCHAR *str; int str_len; _TUCHAR *buf; int buf_len; int buf_size; bool tal; int opa_cnt; vector<int> last_par; vector<CSMatch> matches; vector<CSPar> par; vector<int> match_start[10]; vector<int> match_end[10]; void Put(_TCHAR c); #ifndef _UNICODE void PutValue16(unsigned short c); #endif int Compile(int pos); int Escape(int pos, _TUCHAR &c); int CompileModifiers(int pos, int st_pos); bool RangeAddChar(vector<_TUCHAR> &chars, vector<_TUCHAR> &start, vector<_TUCHAR> &end, _TUCHAR c); bool RangeAddRange(vector<_TUCHAR> &chars, vector<_TUCHAR> &start, vector<_TUCHAR> &end, _TUCHAR cstart, _TUCHAR cend); void CompressRange(vector<_TUCHAR> &chars, vector<_TUCHAR> &start, vector<_TUCHAR> &end); int CompileRange(int pos); void CheckString(int str_idx); unsigned int Next(unsigned int pos) const; unsigned int GetRpt(unsigned int buf_idx, int &low, int &high, bool &lazy, unsigned int &prev) const; bool Match1(const _TUCHAR *str); public: enum RGError { rge_ok=0, rge_too_many_refs=1, rge_missing_round_bracket=2, rge_overlapping_chars=3, rge_esc_eof=4, rge_missing_square_bracket=5, rge_invalid_esc_hex=6, rge_invalid_repeat_format=7, rge_invalid_repeat_range=8, rge_unbalanced_round_bracket=9, rge_invalid_range=10, rge_invalid_backreference=11, rge_regex_too_long=12 }; int error; enum RGError error_code; _TCHAR *error_str; bool bMatchHead; int MatchStart; int MatchEnd; int BackStart[10]; int BackEnd[10]; // There's no copy constructor, so don't use it. CSRegEx::CSRegEx(); CSRegEx::~CSRegEx(); bool Compile(const _TUCHAR *str); #ifndef _UNICODE bool Compile(const char *str); #endif bool MatchRE(const _TUCHAR *str, const _TUCHAR *re); #ifndef _UNICODE bool MatchRE(const char *str, const char *re); bool MatchRE(const unsigned char *str, const char *re); bool MatchRE(const char *str, const unsigned char *re); #endif bool Match(const _TUCHAR *str); #ifndef _UNICODE bool Match(const char *str); #endif bool Match(const _TUCHAR *str, const _TUCHAR *cmp); #ifndef _UNICODE bool Match(const char *str, const char *cmp); bool Match(const unsigned char *str, const char*cmp); bool Match(const char *str, const unsigned char *cmp); #endif _TUCHAR* GetCompiledString() const; void SetCompiledString(_TUCHAR *str); }; #endif Docs for CSRegEx created on Tue Dec 11 14:36:53 2007 by Doxygen 1.4.3 Webmaster: Cléo Saulnier