Chris@16: /* Chris@16: * Chris@16: * Copyright (c) 1998-2002 Chris@16: * John Maddock Chris@16: * Chris@16: * Use, modification and distribution are subject to the Chris@16: * Boost Software License, Version 1.0. (See accompanying file Chris@16: * LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt) Chris@16: * Chris@16: */ Chris@16: Chris@16: /* Chris@16: * LOCATION: see http://www.boost.org for most recent version. Chris@16: * FILE cregex.cpp Chris@16: * VERSION see Chris@16: * DESCRIPTION: Declares POSIX API functions Chris@16: * + boost::RegEx high level wrapper. Chris@16: */ Chris@16: Chris@16: #ifndef BOOST_RE_CREGEX_HPP_INCLUDED Chris@16: #define BOOST_RE_CREGEX_HPP_INCLUDED Chris@16: Chris@16: #ifndef BOOST_REGEX_CONFIG_HPP Chris@16: #include Chris@16: #endif Chris@16: #include Chris@16: #include Chris@16: Chris@16: #ifdef __cplusplus Chris@16: #include Chris@16: #else Chris@16: #include Chris@16: #endif Chris@16: Chris@16: #ifdef BOOST_MSVC Chris@16: #pragma warning(push) Chris@16: #pragma warning(disable: 4103) Chris@16: #endif Chris@16: #ifdef BOOST_HAS_ABI_HEADERS Chris@16: # include BOOST_ABI_PREFIX Chris@16: #endif Chris@16: #ifdef BOOST_MSVC Chris@16: #pragma warning(pop) Chris@16: #endif Chris@16: Chris@16: /* include these defs only for POSIX compatablity */ Chris@16: #ifdef __cplusplus Chris@16: namespace boost{ Chris@16: extern "C" { Chris@16: #endif Chris@16: Chris@16: #if defined(__cplusplus) && !defined(BOOST_NO_STDC_NAMESPACE) Chris@16: typedef std::ptrdiff_t regoff_t; Chris@16: typedef std::size_t regsize_t; Chris@16: #else Chris@16: typedef ptrdiff_t regoff_t; Chris@16: typedef size_t regsize_t; Chris@16: #endif Chris@16: Chris@16: typedef struct Chris@16: { Chris@16: unsigned int re_magic; Chris@16: #ifdef __cplusplus Chris@16: std::size_t re_nsub; /* number of parenthesized subexpressions */ Chris@16: #else Chris@16: size_t re_nsub; Chris@16: #endif Chris@16: const char* re_endp; /* end pointer for REG_PEND */ Chris@16: void* guts; /* none of your business :-) */ Chris@16: match_flag_type eflags; /* none of your business :-) */ Chris@16: } regex_tA; Chris@16: Chris@16: #ifndef BOOST_NO_WREGEX Chris@16: typedef struct Chris@16: { Chris@16: unsigned int re_magic; Chris@16: #ifdef __cplusplus Chris@16: std::size_t re_nsub; /* number of parenthesized subexpressions */ Chris@16: #else Chris@16: size_t re_nsub; Chris@16: #endif Chris@16: const wchar_t* re_endp; /* end pointer for REG_PEND */ Chris@16: void* guts; /* none of your business :-) */ Chris@16: match_flag_type eflags; /* none of your business :-) */ Chris@16: } regex_tW; Chris@16: #endif Chris@16: Chris@16: typedef struct Chris@16: { Chris@16: regoff_t rm_so; /* start of match */ Chris@16: regoff_t rm_eo; /* end of match */ Chris@16: } regmatch_t; Chris@16: Chris@16: /* regcomp() flags */ Chris@16: typedef enum{ Chris@16: REG_BASIC = 0000, Chris@16: REG_EXTENDED = 0001, Chris@16: REG_ICASE = 0002, Chris@16: REG_NOSUB = 0004, Chris@16: REG_NEWLINE = 0010, Chris@16: REG_NOSPEC = 0020, Chris@16: REG_PEND = 0040, Chris@16: REG_DUMP = 0200, Chris@16: REG_NOCOLLATE = 0400, Chris@16: REG_ESCAPE_IN_LISTS = 01000, Chris@16: REG_NEWLINE_ALT = 02000, Chris@16: REG_PERLEX = 04000, Chris@16: Chris@16: REG_PERL = REG_EXTENDED | REG_NOCOLLATE | REG_ESCAPE_IN_LISTS | REG_PERLEX, Chris@16: REG_AWK = REG_EXTENDED | REG_ESCAPE_IN_LISTS, Chris@16: REG_GREP = REG_BASIC | REG_NEWLINE_ALT, Chris@16: REG_EGREP = REG_EXTENDED | REG_NEWLINE_ALT, Chris@16: Chris@16: REG_ASSERT = 15, Chris@16: REG_INVARG = 16, Chris@16: REG_ATOI = 255, /* convert name to number (!) */ Chris@16: REG_ITOA = 0400 /* convert number to name (!) */ Chris@16: } reg_comp_flags; Chris@16: Chris@16: /* regexec() flags */ Chris@16: typedef enum{ Chris@16: REG_NOTBOL = 00001, Chris@16: REG_NOTEOL = 00002, Chris@16: REG_STARTEND = 00004 Chris@16: } reg_exec_flags; Chris@16: Chris@16: /* Chris@16: * POSIX error codes: Chris@16: */ Chris@16: typedef unsigned reg_error_t; Chris@16: typedef reg_error_t reg_errcode_t; /* backwards compatibility */ Chris@16: Chris@16: static const reg_error_t REG_NOERROR = 0; /* Success. */ Chris@16: static const reg_error_t REG_NOMATCH = 1; /* Didn't find a match (for regexec). */ Chris@16: Chris@16: /* POSIX regcomp return error codes. (In the order listed in the Chris@16: standard.) */ Chris@16: static const reg_error_t REG_BADPAT = 2; /* Invalid pattern. */ Chris@16: static const reg_error_t REG_ECOLLATE = 3; /* Undefined collating element. */ Chris@16: static const reg_error_t REG_ECTYPE = 4; /* Invalid character class name. */ Chris@16: static const reg_error_t REG_EESCAPE = 5; /* Trailing backslash. */ Chris@16: static const reg_error_t REG_ESUBREG = 6; /* Invalid back reference. */ Chris@16: static const reg_error_t REG_EBRACK = 7; /* Unmatched left bracket. */ Chris@16: static const reg_error_t REG_EPAREN = 8; /* Parenthesis imbalance. */ Chris@16: static const reg_error_t REG_EBRACE = 9; /* Unmatched \{. */ Chris@16: static const reg_error_t REG_BADBR = 10; /* Invalid contents of \{\}. */ Chris@16: static const reg_error_t REG_ERANGE = 11; /* Invalid range end. */ Chris@16: static const reg_error_t REG_ESPACE = 12; /* Ran out of memory. */ Chris@16: static const reg_error_t REG_BADRPT = 13; /* No preceding re for repetition op. */ Chris@16: static const reg_error_t REG_EEND = 14; /* unexpected end of expression */ Chris@16: static const reg_error_t REG_ESIZE = 15; /* expression too big */ Chris@16: static const reg_error_t REG_ERPAREN = 8; /* = REG_EPAREN : unmatched right parenthesis */ Chris@16: static const reg_error_t REG_EMPTY = 17; /* empty expression */ Chris@16: static const reg_error_t REG_E_MEMORY = 15; /* = REG_ESIZE : out of memory */ Chris@16: static const reg_error_t REG_ECOMPLEXITY = 18; /* complexity too high */ Chris@16: static const reg_error_t REG_ESTACK = 19; /* out of stack space */ Chris@16: static const reg_error_t REG_E_PERL = 20; /* Perl (?...) error */ Chris@16: static const reg_error_t REG_E_UNKNOWN = 21; /* unknown error */ Chris@16: static const reg_error_t REG_ENOSYS = 21; /* = REG_E_UNKNOWN : Reserved. */ Chris@16: Chris@16: BOOST_REGEX_DECL int BOOST_REGEX_CCALL regcompA(regex_tA*, const char*, int); Chris@16: BOOST_REGEX_DECL regsize_t BOOST_REGEX_CCALL regerrorA(int, const regex_tA*, char*, regsize_t); Chris@16: BOOST_REGEX_DECL int BOOST_REGEX_CCALL regexecA(const regex_tA*, const char*, regsize_t, regmatch_t*, int); Chris@16: BOOST_REGEX_DECL void BOOST_REGEX_CCALL regfreeA(regex_tA*); Chris@16: Chris@16: #ifndef BOOST_NO_WREGEX Chris@16: BOOST_REGEX_DECL int BOOST_REGEX_CCALL regcompW(regex_tW*, const wchar_t*, int); Chris@16: BOOST_REGEX_DECL regsize_t BOOST_REGEX_CCALL regerrorW(int, const regex_tW*, wchar_t*, regsize_t); Chris@16: BOOST_REGEX_DECL int BOOST_REGEX_CCALL regexecW(const regex_tW*, const wchar_t*, regsize_t, regmatch_t*, int); Chris@16: BOOST_REGEX_DECL void BOOST_REGEX_CCALL regfreeW(regex_tW*); Chris@16: #endif Chris@16: Chris@16: #ifdef UNICODE Chris@16: #define regcomp regcompW Chris@16: #define regerror regerrorW Chris@16: #define regexec regexecW Chris@16: #define regfree regfreeW Chris@16: #define regex_t regex_tW Chris@16: #else Chris@16: #define regcomp regcompA Chris@16: #define regerror regerrorA Chris@16: #define regexec regexecA Chris@16: #define regfree regfreeA Chris@16: #define regex_t regex_tA Chris@16: #endif Chris@16: Chris@16: #ifdef BOOST_MSVC Chris@16: #pragma warning(push) Chris@16: #pragma warning(disable: 4103) Chris@16: #endif Chris@16: #ifdef BOOST_HAS_ABI_HEADERS Chris@16: # include BOOST_ABI_SUFFIX Chris@16: #endif Chris@16: #ifdef BOOST_MSVC Chris@16: #pragma warning(pop) Chris@16: #endif Chris@16: Chris@16: #ifdef __cplusplus Chris@16: } /* extern "C" */ Chris@16: } /* namespace */ Chris@16: #endif Chris@16: Chris@16: #if defined(__cplusplus) Chris@16: /* Chris@16: * C++ high level wrapper goes here: Chris@16: */ Chris@16: #include Chris@16: #include Chris@16: namespace boost{ Chris@16: Chris@16: #ifdef BOOST_MSVC Chris@16: #pragma warning(push) Chris@16: #pragma warning(disable: 4103) Chris@16: #endif Chris@16: #ifdef BOOST_HAS_ABI_HEADERS Chris@16: # include BOOST_ABI_PREFIX Chris@16: #endif Chris@16: #ifdef BOOST_MSVC Chris@16: #pragma warning(pop) Chris@16: #endif Chris@16: Chris@16: class RegEx; Chris@16: Chris@16: namespace re_detail{ Chris@16: Chris@16: class RegExData; Chris@16: struct pred1; Chris@16: struct pred2; Chris@16: struct pred3; Chris@16: struct pred4; Chris@16: Chris@16: } /* namespace re_detail */ Chris@16: Chris@16: #if (defined(BOOST_MSVC) || defined(__BORLANDC__)) && !defined(BOOST_DISABLE_WIN32) Chris@16: typedef bool (__cdecl *GrepCallback)(const RegEx& expression); Chris@16: typedef bool (__cdecl *GrepFileCallback)(const char* file, const RegEx& expression); Chris@16: typedef bool (__cdecl *FindFilesCallback)(const char* file); Chris@16: #else Chris@16: typedef bool (*GrepCallback)(const RegEx& expression); Chris@16: typedef bool (*GrepFileCallback)(const char* file, const RegEx& expression); Chris@16: typedef bool (*FindFilesCallback)(const char* file); Chris@16: #endif Chris@16: Chris@16: class BOOST_REGEX_DECL RegEx Chris@16: { Chris@16: private: Chris@16: re_detail::RegExData* pdata; Chris@16: public: Chris@16: RegEx(); Chris@16: RegEx(const RegEx& o); Chris@16: ~RegEx(); Chris@16: explicit RegEx(const char* c, bool icase = false); Chris@16: explicit RegEx(const std::string& s, bool icase = false); Chris@16: RegEx& operator=(const RegEx& o); Chris@16: RegEx& operator=(const char* p); Chris@16: RegEx& operator=(const std::string& s){ return this->operator=(s.c_str()); } Chris@16: unsigned int SetExpression(const char* p, bool icase = false); Chris@16: unsigned int SetExpression(const std::string& s, bool icase = false){ return SetExpression(s.c_str(), icase); } Chris@16: std::string Expression()const; Chris@16: unsigned int error_code()const; Chris@16: /* Chris@16: * now matching operators: Chris@16: */ Chris@16: bool Match(const char* p, match_flag_type flags = match_default); Chris@16: bool Match(const std::string& s, match_flag_type flags = match_default) { return Match(s.c_str(), flags); } Chris@16: bool Search(const char* p, match_flag_type flags = match_default); Chris@16: bool Search(const std::string& s, match_flag_type flags = match_default) { return Search(s.c_str(), flags); } Chris@16: unsigned int Grep(GrepCallback cb, const char* p, match_flag_type flags = match_default); Chris@16: unsigned int Grep(GrepCallback cb, const std::string& s, match_flag_type flags = match_default) { return Grep(cb, s.c_str(), flags); } Chris@16: unsigned int Grep(std::vector& v, const char* p, match_flag_type flags = match_default); Chris@16: unsigned int Grep(std::vector& v, const std::string& s, match_flag_type flags = match_default) { return Grep(v, s.c_str(), flags); } Chris@16: unsigned int Grep(std::vector& v, const char* p, match_flag_type flags = match_default); Chris@16: unsigned int Grep(std::vector& v, const std::string& s, match_flag_type flags = match_default) { return Grep(v, s.c_str(), flags); } Chris@16: #ifndef BOOST_REGEX_NO_FILEITER Chris@16: unsigned int GrepFiles(GrepFileCallback cb, const char* files, bool recurse = false, match_flag_type flags = match_default); Chris@16: unsigned int GrepFiles(GrepFileCallback cb, const std::string& files, bool recurse = false, match_flag_type flags = match_default) { return GrepFiles(cb, files.c_str(), recurse, flags); } Chris@16: unsigned int FindFiles(FindFilesCallback cb, const char* files, bool recurse = false, match_flag_type flags = match_default); Chris@16: unsigned int FindFiles(FindFilesCallback cb, const std::string& files, bool recurse = false, match_flag_type flags = match_default) { return FindFiles(cb, files.c_str(), recurse, flags); } Chris@16: #endif Chris@16: Chris@16: std::string Merge(const std::string& in, const std::string& fmt, Chris@16: bool copy = true, match_flag_type flags = match_default); Chris@16: std::string Merge(const char* in, const char* fmt, Chris@16: bool copy = true, match_flag_type flags = match_default); Chris@16: Chris@16: std::size_t Split(std::vector& v, std::string& s, match_flag_type flags = match_default, unsigned max_count = ~0); Chris@16: /* Chris@16: * now operators for returning what matched in more detail: Chris@16: */ Chris@16: std::size_t Position(int i = 0)const; Chris@16: std::size_t Length(int i = 0)const; Chris@16: bool Matched(int i = 0)const; Chris@16: std::size_t Marks()const; Chris@16: std::string What(int i = 0)const; Chris@16: std::string operator[](int i)const { return What(i); } Chris@16: Chris@16: static const std::size_t npos; Chris@16: Chris@16: friend struct re_detail::pred1; Chris@16: friend struct re_detail::pred2; Chris@16: friend struct re_detail::pred3; Chris@16: friend struct re_detail::pred4; Chris@16: }; Chris@16: Chris@16: #ifdef BOOST_MSVC Chris@16: #pragma warning(push) Chris@16: #pragma warning(disable: 4103) Chris@16: #endif Chris@16: #ifdef BOOST_HAS_ABI_HEADERS Chris@16: # include BOOST_ABI_SUFFIX Chris@16: #endif Chris@16: #ifdef BOOST_MSVC Chris@16: #pragma warning(pop) Chris@16: #endif Chris@16: Chris@16: } /* namespace boost */ Chris@16: Chris@16: #endif /* __cplusplus */ Chris@16: Chris@16: #endif /* include guard */ Chris@16: Chris@16: Chris@16: Chris@16: Chris@16: Chris@16: Chris@16: Chris@16: Chris@16: Chris@16: