X-Git-Url: https://git.saurik.com/wxWidgets.git/blobdiff_plain/00e6c2bd41efbbdd158c9fb609450ebe001a27cc..75a2645e11fe191cc812cd768b61322456b5f8b8:/src/common/regex.cpp diff --git a/src/common/regex.cpp b/src/common/regex.cpp index ec57e48df3..896d2ad248 100644 --- a/src/common/regex.cpp +++ b/src/common/regex.cpp @@ -18,7 +18,7 @@ // headers // ---------------------------------------------------------------------------- -#ifdef __GNUG__ +#if defined(__GNUG__) && !defined(NO_GCC_PRAGMA) #pragma implementation "regex.h" #endif @@ -38,8 +38,15 @@ #include "wx/intl.h" #endif //WX_PRECOMP -#include +// FreeBSD, Watcom and DMars require this, CW doesn't have nor need it. +// Others also don't seem to need it. If you have an error related to +// (not) including please report details to +// wx-dev@lists.wxwindows.org +#if defined(__UNIX__) || defined(__WATCOMC__) || defined(__DIGITALMARS__) +# include +#endif +#include #include "wx/regex.h" // ---------------------------------------------------------------------------- @@ -54,18 +61,28 @@ public: wxRegExImpl(); ~wxRegExImpl(); - // return TRUE if Compile() had been called successfully + // return true if Compile() had been called successfully bool IsValid() const { return m_isCompiled; } // RE operations - bool Compile(const wxString& expr, int flags); - bool Matches(const wxString& str, int flags) const; - bool GetMatch(size_t *start, size_t *len, size_t index) const; - int Replace(wxString *pattern, const wxString& replacement) const; + bool Compile(const wxString& expr, int flags = 0); + bool Matches(const wxChar *str, int flags = 0) const; + bool GetMatch(size_t *start, size_t *len, size_t index = 0) const; + size_t GetMatchCount() const; + int Replace(wxString *pattern, const wxString& replacement, + size_t maxMatches = 0) const; private: // return the string containing the error message for the given err code - wxString GetErrorMsg(int errorcode) const; + wxString GetErrorMsg(int errorcode, bool badconv) const; + + // init the members + void Init() + { + m_isCompiled = false; + m_Matches = NULL; + m_nMatches = 0; + } // free the RE if compiled void Free() @@ -73,11 +90,19 @@ private: if ( IsValid() ) { regfree(&m_RegEx); - - m_isCompiled = FALSE; } + + delete [] m_Matches; + } + + // free the RE if any and reinit the members + void Reinit() + { + Free(); + Init(); } + // compiled RE regex_t m_RegEx; @@ -85,7 +110,7 @@ private: regmatch_t *m_Matches; size_t m_nMatches; - // TRUE if m_RegEx is valid + // true if m_RegEx is valid bool m_isCompiled; }; @@ -99,51 +124,71 @@ private: wxRegExImpl::wxRegExImpl() { - m_isCompiled = FALSE; - m_Matches = NULL; + Init(); } wxRegExImpl::~wxRegExImpl() { Free(); - - delete [] m_Matches; } -wxString wxRegExImpl::GetErrorMsg(int errorcode) const +wxString wxRegExImpl::GetErrorMsg(int errorcode, bool badconv) const { - wxString msg; +#if wxUSE_UNICODE && !defined(__REG_NOFRONT) + // currently only needed when using system library in Unicode mode + if ( badconv ) + { + return _("conversion to 8-bit encoding failed"); + } +#else + // 'use' badconv to avoid a compiler warning + (void)badconv; +#endif + + wxString szError; // first get the string length needed int len = regerror(errorcode, &m_RegEx, NULL, 0); if ( len > 0 ) { - len++; + char* szcmbError = new char[++len]; - (void)regerror(errorcode, &m_RegEx, msg.GetWriteBuf(len), len); + (void)regerror(errorcode, &m_RegEx, szcmbError, len); - msg.UngetWriteBuf(); + szError = wxConvertMB2WX(szcmbError); + delete [] szcmbError; } else // regerror() returned 0 { - msg = _("unknown error"); + szError = _("unknown error"); } - return msg; + return szError; } bool wxRegExImpl::Compile(const wxString& expr, int flags) { - Free(); - - // translate our flags to regcomp() ones - wxASSERT_MSG( !(flags & - ~(wxRE_BASIC | wxRE_ICASE | wxRE_NOSUB | wxRE_NEWLINE)), + Reinit(); + +#ifdef WX_NO_REGEX_ADVANCED +# define FLAVORS wxRE_BASIC +#else +# define FLAVORS (wxRE_ADVANCED | wxRE_BASIC) + wxASSERT_MSG( (flags & FLAVORS) != FLAVORS, + _T("incompatible flags in wxRegEx::Compile") ); +#endif + wxASSERT_MSG( !(flags & ~(FLAVORS | wxRE_ICASE | wxRE_NOSUB | wxRE_NEWLINE)), _T("unrecognized flags in wxRegEx::Compile") ); + // translate our flags to regcomp() ones int flagsRE = 0; if ( !(flags & wxRE_BASIC) ) - flagsRE |= REG_EXTENDED; +#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 ) @@ -152,13 +197,20 @@ bool wxRegExImpl::Compile(const wxString& expr, int flags) flagsRE |= REG_NEWLINE; // compile it - int errorcode = regcomp(&m_RegEx, expr, flagsRE); +#ifdef __REG_NOFRONT + bool conv = true; + int errorcode = wx_re_comp(&m_RegEx, expr, expr.length(), flagsRE); +#else + const wxWX2MBbuf conv = expr.mbc_str(); + int errorcode = conv ? regcomp(&m_RegEx, conv, flagsRE) : REG_BADPAT; +#endif + if ( errorcode ) { wxLogError(_("Invalid regular expression '%s': %s"), - expr.c_str(), GetErrorMsg(errorcode).c_str()); + expr.c_str(), GetErrorMsg(errorcode, !conv).c_str()); - m_isCompiled = FALSE; + m_isCompiled = false; } else // ok { @@ -170,19 +222,46 @@ bool wxRegExImpl::Compile(const wxString& expr, int flags) } else { - // will alloc later - m_nMatches = WX_REGEX_MAXMATCHES; + // we will alloc the array later (only if really needed) but count + // the number of sub-expressions in the regex right now + + // there is always one for the whole expression + m_nMatches = 1; + + // and some more for bracketed subexperessions + for ( const wxChar *cptr = expr.c_str(); *cptr; cptr++ ) + { + if ( *cptr == _T('\\') ) + { + // in basic RE syntax groups are inside \(...\) + if ( *++cptr == _T('(') && (flags & wxRE_BASIC) ) + { + m_nMatches++; + } + } + else if ( *cptr == _T('(') && !(flags & wxRE_BASIC) ) + { + // we know that the previous character is not an unquoted + // backslash because it would have been eaten above, so we + // have a bare '(' and this indicates a group start for the + // extended syntax. '(?' is used for extensions by perl- + // like REs (e.g. advanced), and is not valid for POSIX + // extended, so ignore them always. + if ( cptr[1] != _T('?') ) + m_nMatches++; + } + } } - m_isCompiled = TRUE; + m_isCompiled = true; } return IsValid(); } -bool wxRegExImpl::Matches(const wxString& str, int flags) const +bool wxRegExImpl::Matches(const wxChar *str, int flags) const { - wxCHECK_MSG( IsValid(), FALSE, _T("must successfully Compile() first") ); + wxCHECK_MSG( IsValid(), false, _T("must successfully Compile() first") ); // translate our flags to regexec() ones wxASSERT_MSG( !(flags & ~(wxRE_NOTBOL | wxRE_NOTEOL)), @@ -202,69 +281,166 @@ bool wxRegExImpl::Matches(const wxString& str, int flags) const } // do match it - int rc = regexec(&self->m_RegEx, str, m_nMatches, m_Matches, flagsRE); +#ifdef __REG_NOFRONT + bool conv = true; + int rc = wx_re_exec(&self->m_RegEx, str, wxStrlen(str), NULL, m_nMatches, m_Matches, flagsRE); +#else + const wxWX2MBbuf conv = wxConvertWX2MB(str); + int rc = conv ? regexec(&self->m_RegEx, conv, m_nMatches, m_Matches, flagsRE) : REG_BADPAT; +#endif switch ( rc ) { case 0: // matched successfully - return TRUE; + return true; default: // an error occured wxLogError(_("Failed to match '%s' in regular expression: %s"), - str.c_str(), GetErrorMsg(rc).c_str()); + str, GetErrorMsg(rc, !conv).c_str()); // fall through case REG_NOMATCH: // no match - return FALSE; + return false; } } bool wxRegExImpl::GetMatch(size_t *start, size_t *len, size_t index) const { - wxCHECK_MSG( IsValid(), FALSE, _T("must successfully Compile() first") ); - wxCHECK_MSG( m_Matches, FALSE, _T("can't use with wxRE_NOSUB") ); - wxCHECK_MSG( index < m_nMatches, FALSE, _T("invalid match index") ); + wxCHECK_MSG( IsValid(), false, _T("must successfully Compile() first") ); + wxCHECK_MSG( m_nMatches, false, _T("can't use with wxRE_NOSUB") ); + wxCHECK_MSG( m_Matches, false, _T("must call Matches() first") ); + wxCHECK_MSG( index < m_nMatches, false, _T("invalid match index") ); const regmatch_t& match = m_Matches[index]; - if ( match.rm_so == -1 ) - return FALSE; if ( start ) *start = match.rm_so; if ( len ) *len = match.rm_eo - match.rm_so; - return TRUE; + return true; +} + +size_t wxRegExImpl::GetMatchCount() const +{ + wxCHECK_MSG( IsValid(), 0, _T("must successfully Compile() first") ); + wxCHECK_MSG( m_nMatches, 0, _T("can't use with wxRE_NOSUB") ); + + return m_nMatches; } -int wxRegExImpl::Replace(wxString *pattern, const wxString& replacement) const +int wxRegExImpl::Replace(wxString *text, + const wxString& replacement, + size_t maxMatches) const { - wxCHECK_MSG( pattern, -1, _T("NULL pattern in wxRegEx::Replace") ); + wxCHECK_MSG( text, wxNOT_FOUND, _T("NULL text in wxRegEx::Replace") ); + wxCHECK_MSG( IsValid(), wxNOT_FOUND, _T("must successfully Compile() first") ); + + // the replacement text + wxString textNew; - wxCHECK_MSG( IsValid(), FALSE, _T("must successfully Compile() first") ); + // attempt at optimization: don't iterate over the string if it doesn't + // contain back references at all + bool mayHaveBackrefs = + replacement.find_first_of(_T("\\&")) != wxString::npos; - int replaced = 0; - size_t lastpos = 0; - wxString newstring; + if ( !mayHaveBackrefs ) + { + textNew = replacement; + } - for ( size_t idx = 0; - m_Matches[idx].rm_so != -1 && idx < m_nMatches; - idx++ ) + // the position where we start looking for the match + // + // NB: initial version had a nasty bug because it used a wxChar* instead of + // an index but the problem is that replace() in the loop invalidates + // all pointers into the string so we have to use indices instead + size_t matchStart = 0; + + // number of replacement made: we won't make more than maxMatches of them + // (unless maxMatches is 0 which doesn't limit the number of replacements) + size_t countRepl = 0; + + // 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(text->c_str() + matchStart, countRepl ? wxRE_NOTBOL : 0) ) { - // copy non-matching bits: - newstring << pattern->Mid(lastpos, m_Matches[idx].rm_so - lastpos); - // copy replacement: - newstring << replacement; - // remember how far we got: - lastpos = m_Matches[idx].rm_eo; - replaced ++; + // the string possibly contains back references: we need to calculate + // the replacement text anew after each match + if ( mayHaveBackrefs ) + { + mayHaveBackrefs = false; + textNew.clear(); + textNew.reserve(replacement.length()); + + for ( const wxChar *p = replacement.c_str(); *p; p++ ) + { + size_t index = (size_t)-1; + + if ( *p == _T('\\') ) + { + if ( wxIsdigit(*++p) ) + { + // back reference + wxChar *end; + index = (size_t)wxStrtoul(p, &end, 10); + p = end - 1; // -1 to compensate for p++ in the loop + } + //else: backslash used as escape character + } + else if ( *p == _T('&') ) + { + // treat this as "\0" for compatbility with ed and such + index = 0; + } + + // do we have a back reference? + if ( index != (size_t)-1 ) + { + // yes, get its text + size_t start, len; + if ( !GetMatch(&start, &len, index) ) + { + wxFAIL_MSG( _T("invalid back reference") ); + + // just eat it... + } + else + { + textNew += wxString(text->c_str() + matchStart + start, + len); + + mayHaveBackrefs = true; + } + } + else // ordinary character + { + textNew += *p; + } + } + } + + size_t start, len; + if ( !GetMatch(&start, &len) ) + { + // we did have match as Matches() returned true above! + wxFAIL_MSG( _T("internal logic error in wxRegEx::Replace") ); + + return wxNOT_FOUND; + } + + matchStart += start; + text->replace(matchStart, len, textNew); + + countRepl++; + + matchStart += textNew.length(); } - if(replaced > 0) - *pattern = newstring; - return replaced; + + return countRepl; } // ---------------------------------------------------------------------------- @@ -295,22 +471,22 @@ bool wxRegEx::Compile(const wxString& expr, int flags) delete m_impl; m_impl = NULL; - return FALSE; + return false; } - return TRUE; + return true; } -bool wxRegEx::Matches(const wxString& str, int flags) const +bool wxRegEx::Matches(const wxChar *str, int flags) const { - wxCHECK_MSG( IsValid(), FALSE, _T("must successfully Compile() first") ); + wxCHECK_MSG( IsValid(), false, _T("must successfully Compile() first") ); return m_impl->Matches(str, flags); } bool wxRegEx::GetMatch(size_t *start, size_t *len, size_t index) const { - wxCHECK_MSG( IsValid(), -1, _T("must successfully Compile() first") ); + wxCHECK_MSG( IsValid(), false, _T("must successfully Compile() first") ); return m_impl->GetMatch(start, len, index); } @@ -324,11 +500,20 @@ wxString wxRegEx::GetMatch(const wxString& text, size_t index) const return text.Mid(start, len); } -int wxRegEx::Replace(wxString *pattern, const wxString& replacement) const +size_t wxRegEx::GetMatchCount() const +{ + wxCHECK_MSG( IsValid(), 0, _T("must successfully Compile() first") ); + + return m_impl->GetMatchCount(); +} + +int wxRegEx::Replace(wxString *pattern, + const wxString& replacement, + size_t maxMatches) const { - wxCHECK_MSG( IsValid(), -1, _T("must successfully Compile() first") ); + wxCHECK_MSG( IsValid(), wxNOT_FOUND, _T("must successfully Compile() first") ); - return m_impl->Replace(pattern, replacement); + return m_impl->Replace(pattern, replacement, maxMatches); } #endif // wxUSE_REGEX