X-Git-Url: https://git.saurik.com/wxWidgets.git/blobdiff_plain/ab0f0eddc4acd89223c79e38047447419ead583a..4b5d2be39657bb364c9111c4d13b1eecb8f3b69f:/src/common/regex.cpp?ds=sidebyside diff --git a/src/common/regex.cpp b/src/common/regex.cpp index 12c790b915..7a7029c8b5 100644 --- a/src/common/regex.cpp +++ b/src/common/regex.cpp @@ -1,11 +1,11 @@ /////////////////////////////////////////////////////////////////////////////// // Name: src/common/regex.cpp // Purpose: regular expression matching -// Author: Karsten Ballüder and Vadim Zeitlin +// Author: Karsten Ballueder and Vadim Zeitlin // Modified by: // Created: 13.07.01 // RCS-ID: $Id$ -// Copyright: (c) 2000 Karsten Ballüder +// Copyright: (c) 2000 Karsten Ballueder // 2001 Vadim Zeitlin // Licence: wxWindows licence /////////////////////////////////////////////////////////////////////////////// @@ -32,6 +32,7 @@ #include "wx/string.h" #include "wx/log.h" #include "wx/intl.h" + #include "wx/crt.h" #endif //WX_PRECOMP // FreeBSD, Watcom and DMars require this, CW doesn't have nor need it. @@ -45,34 +46,42 @@ #include #include "wx/regex.h" -// WXREGEX_USING_BUILTIN defined when using the built-in regex lib -// WXREGEX_IF_NEED_LEN() wrap the len parameter only used with the built-in -// or GNU regex -// WXREGEX_CONVERT_TO_MB defined when the regex lib is using chars and -// wxChar is wide, so conversion must be done -// WXREGEX_CHAR(x) Convert wxChar to wxRegChar +// WXREGEX_USING_BUILTIN defined when using the built-in regex lib +// WXREGEX_USING_RE_SEARCH defined when using re_search in the GNU regex lib +// WXREGEX_IF_NEED_LEN() wrap the len parameter only used with the built-in +// or GNU regex +// WXREGEX_CONVERT_TO_MB defined when the regex lib is using chars and +// wxChar is wide, so conversion must be done +// WXREGEX_CHAR(x) Convert wxChar to wxRegChar // #ifdef __REG_NOFRONT # define WXREGEX_USING_BUILTIN # define WXREGEX_IF_NEED_LEN(x) ,x -# define WXREGEX_CHAR(x) x +# if wxUSE_UNICODE +# define WXREGEX_CHAR(x) (x).wc_str() +# else +# define WXREGEX_CHAR(x) (x).mb_str() +# endif #else # ifdef HAVE_RE_SEARCH # define WXREGEX_IF_NEED_LEN(x) ,x -# else +# define WXREGEX_USING_RE_SEARCH +# else # define WXREGEX_IF_NEED_LEN(x) # endif # if wxUSE_UNICODE # define WXREGEX_CONVERT_TO_MB # endif -# define WXREGEX_CHAR(x) wxConvertWX2MB(x) +# define WXREGEX_CHAR(x) (x).mb_str() +# define wx_regfree regfree +# define wx_regerror regerror #endif // ---------------------------------------------------------------------------- // private classes // ---------------------------------------------------------------------------- -#ifndef HAVE_RE_SEARCH +#ifndef WXREGEX_USING_RE_SEARCH // the array of offsets for the matches, the usual POSIX regmatch_t array. class wxRegExMatches @@ -80,11 +89,22 @@ class wxRegExMatches public: typedef regmatch_t *match_type; - wxRegExMatches(size_t n) { m_matches = new regmatch_t[n]; } + wxRegExMatches(size_t n) { m_matches = new regmatch_t[n]; } ~wxRegExMatches() { delete [] m_matches; } - size_t Start(size_t n) const { return m_matches[n].rm_so; } - size_t End(size_t n) const { return m_matches[n].rm_eo; } + // we just use casts here because the fields of regmatch_t struct may be 64 + // bit but we're limited to size_t in our public API and are not going to + // change it because operating on strings longer than 4GB using it is + // absolutely impractical anyhow + size_t Start(size_t n) const + { + return wx_truncate_cast(size_t, m_matches[n].rm_so); + } + + size_t End(size_t n) const + { + return wx_truncate_cast(size_t, m_matches[n].rm_eo); + } regmatch_t *get() const { return m_matches; } @@ -92,7 +112,7 @@ private: regmatch_t *m_matches; }; -#else // HAVE_RE_SEARCH +#else // WXREGEX_USING_RE_SEARCH // the array of offsets for the matches, the struct used by the GNU lib class wxRegExMatches @@ -122,7 +142,7 @@ private: re_registers m_matches; }; -#endif // HAVE_RE_SEARCH +#endif // WXREGEX_USING_RE_SEARCH // the character type used by the regular expression engine #ifndef WXREGEX_CONVERT_TO_MB @@ -168,7 +188,7 @@ private: { if ( IsValid() ) { - regfree(&m_RegEx); + wx_regfree(&m_RegEx); } delete m_Matches; @@ -227,14 +247,14 @@ wxString wxRegExImpl::GetErrorMsg(int errorcode, bool badconv) const wxString szError; // first get the string length needed - int len = regerror(errorcode, &m_RegEx, NULL, 0); + int len = wx_regerror(errorcode, &m_RegEx, NULL, 0); if ( len > 0 ) { char* szcmbError = new char[++len]; - (void)regerror(errorcode, &m_RegEx, szcmbError, len); + (void)wx_regerror(errorcode, &m_RegEx, szcmbError, len); - szError = wxConvertMB2WX(szcmbError); + szError = wxConvLibc.cMB2WX(szcmbError); delete [] szcmbError; } else // regerror() returned 0 @@ -262,12 +282,14 @@ bool wxRegExImpl::Compile(const wxString& expr, int flags) // translate our flags to regcomp() ones int flagsRE = 0; if ( !(flags & wxRE_BASIC) ) + { #ifndef WX_NO_REGEX_ADVANCED if (flags & wxRE_ADVANCED) flagsRE |= REG_ADVANCED; else #endif flagsRE |= REG_EXTENDED; + } if ( flags & wxRE_ICASE ) flagsRE |= REG_ICASE; if ( flags & wxRE_NOSUB ) @@ -278,8 +300,11 @@ bool wxRegExImpl::Compile(const wxString& expr, int flags) // compile it #ifdef WXREGEX_USING_BUILTIN bool conv = true; - int errorcode = wx_re_comp(&m_RegEx, expr, expr.length(), flagsRE); + // FIXME-UTF8: use wc_str() after removing ANSI build + int errorcode = wx_re_comp(&m_RegEx, expr.c_str(), expr.length(), flagsRE); #else + // FIXME-UTF8: this is potentially broken, we shouldn't even try it + // and should always use builtin regex library (or PCRE?) const wxWX2MBbuf conv = expr.mbc_str(); int errorcode = conv ? regcomp(&m_RegEx, conv, flagsRE) : REG_BADPAT; #endif @@ -338,7 +363,7 @@ bool wxRegExImpl::Compile(const wxString& expr, int flags) return IsValid(); } -#ifdef HAVE_RE_SEARCH +#ifdef WXREGEX_USING_RE_SEARCH // On GNU, regexec is implemented as a wrapper around re_search. re_search // requires a length parameter which the POSIX regexec does not have, @@ -363,7 +388,7 @@ static int ReSearch(const regex_t *preg, return ret >= 0 ? 0 : REG_NOMATCH; } -#endif // HAVE_RE_SEARCH +#endif // WXREGEX_USING_RE_SEARCH bool wxRegExImpl::Matches(const wxRegChar *str, int flags @@ -393,7 +418,7 @@ bool wxRegExImpl::Matches(const wxRegChar *str, // do match it #if defined WXREGEX_USING_BUILTIN int rc = wx_re_exec(&self->m_RegEx, str, len, NULL, m_nMatches, matches, flagsRE); -#elif defined HAVE_RE_SEARCH +#elif defined WXREGEX_USING_RE_SEARCH int rc = str ? ReSearch(&self->m_RegEx, str, len, matches, flagsRE) : REG_BADPAT; #else int rc = str ? regexec(&self->m_RegEx, str, m_nMatches, matches, flagsRE) : REG_BADPAT; @@ -490,7 +515,12 @@ int wxRegExImpl::Replace(wxString *text, // note that "^" shouldn't match after the first call to Matches() so we // use wxRE_NOTBOL to prevent it from happening while ( (!maxMatches || countRepl < maxMatches) && - Matches(textstr + matchStart, + Matches( +#ifndef WXREGEX_CONVERT_TO_MB + textstr + matchStart, +#else + textstr.data() + matchStart, +#endif countRepl ? wxRE_NOTBOL : 0 WXREGEX_IF_NEED_LEN(textlen - matchStart)) ) { @@ -536,7 +566,12 @@ int wxRegExImpl::Replace(wxString *text, } else { +#ifndef WXREGEX_CONVERT_TO_MB textNew += wxString(textstr + matchStart + start, +#else + textNew += wxString(textstr.data() + matchStart + + start, +#endif *wxConvCurrent, len); mayHaveBackrefs = true; @@ -566,7 +601,8 @@ int wxRegExImpl::Replace(wxString *text, #ifndef WXREGEX_CONVERT_TO_MB result.append(*text, matchStart, start); #else - result.append(wxString(textstr + matchStart, *wxConvCurrent, start)); + result.append(wxString(textstr.data() + matchStart, *wxConvCurrent, + start)); #endif matchStart += start; result.append(textNew); @@ -579,7 +615,7 @@ int wxRegExImpl::Replace(wxString *text, #ifndef WXREGEX_CONVERT_TO_MB result.append(*text, matchStart, wxString::npos); #else - result.append(wxString(textstr + matchStart, *wxConvCurrent)); + result.append(wxString(textstr.data() + matchStart, *wxConvCurrent)); #endif *text = result; @@ -619,21 +655,12 @@ bool wxRegEx::Compile(const wxString& expr, int flags) return true; } -bool wxRegEx::Matches(const wxChar *str, int flags, size_t len) const -{ - wxCHECK_MSG( IsValid(), false, _T("must successfully Compile() first") ); - (void)len; - - return m_impl->Matches(WXREGEX_CHAR(str), flags WXREGEX_IF_NEED_LEN(len)); -} - -bool wxRegEx::Matches(const wxChar *str, int flags) const +bool wxRegEx::Matches(const wxString& str, int flags) const { wxCHECK_MSG( IsValid(), false, _T("must successfully Compile() first") ); - return m_impl->Matches(WXREGEX_CHAR(str), - flags - WXREGEX_IF_NEED_LEN(wxStrlen(str))); + return m_impl->Matches(WXREGEX_CHAR(str), flags + WXREGEX_IF_NEED_LEN(str.length())); } bool wxRegEx::GetMatch(size_t *start, size_t *len, size_t index) const