X-Git-Url: https://git.saurik.com/wxWidgets.git/blobdiff_plain/00e6c2bd41efbbdd158c9fb609450ebe001a27cc..bb80bb5b30d7b16f39e6c1c63637f98fdf4ff63a:/src/common/regex.cpp?ds=sidebyside diff --git a/src/common/regex.cpp b/src/common/regex.cpp index ec57e48df3..7b480f24c8 100644 --- a/src/common/regex.cpp +++ b/src/common/regex.cpp @@ -38,6 +38,11 @@ #include "wx/intl.h" #endif //WX_PRECOMP +// FreeBSD & Watcom require this, it probably doesn't hurt for others +#if defined(__UNIX__) || defined(__WATCOMC__) + #include +#endif + #include #include "wx/regex.h" @@ -58,10 +63,11 @@ public: 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; + 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 @@ -101,6 +107,7 @@ wxRegExImpl::wxRegExImpl() { m_isCompiled = FALSE; m_Matches = NULL; + m_nMatches = 0; } wxRegExImpl::~wxRegExImpl() @@ -120,9 +127,17 @@ wxString wxRegExImpl::GetErrorMsg(int errorcode) const { len++; +#if wxUSE_UNICODE + wxCharBuffer buf(len); + + (void)regerror(errorcode, &m_RegEx, (char *)buf.data(), len); + + msg = buf.data(); +#else // !Unicode (void)regerror(errorcode, &m_RegEx, msg.GetWriteBuf(len), len); msg.UngetWriteBuf(); +#endif // Unicode/!Unicode } else // regerror() returned 0 { @@ -152,7 +167,7 @@ bool wxRegExImpl::Compile(const wxString& expr, int flags) flagsRE |= REG_NEWLINE; // compile it - int errorcode = regcomp(&m_RegEx, expr, flagsRE); + int errorcode = regcomp(&m_RegEx, expr.mb_str(), flagsRE); if ( errorcode ) { wxLogError(_("Invalid regular expression '%s': %s"), @@ -170,8 +185,29 @@ 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 + const wxChar *cptr = expr.c_str(); + wxChar prev = _T('\0'); + while ( *cptr != _T('\0') ) + { + // is this a subexpr start, i.e. "(" for extended regex or + // "\(" for a basic one? + if ( *cptr == _T('(') && + (flags & wxRE_BASIC ? prev == _T('\\') + : prev != _T('\\')) ) + { + m_nMatches++; + } + + prev = *cptr; + cptr++; + } } m_isCompiled = TRUE; @@ -180,7 +216,7 @@ bool wxRegExImpl::Compile(const wxString& expr, int flags) 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") ); @@ -202,7 +238,7 @@ bool wxRegExImpl::Matches(const wxString& str, int flags) const } // do match it - int rc = regexec(&self->m_RegEx, str, m_nMatches, m_Matches, flagsRE); + int rc = regexec(&self->m_RegEx, wxConvertWX2MB(str), m_nMatches, m_Matches, flagsRE); switch ( rc ) { @@ -213,7 +249,7 @@ bool wxRegExImpl::Matches(const wxString& str, int flags) const default: // an error occured wxLogError(_("Failed to match '%s' in regular expression: %s"), - str.c_str(), GetErrorMsg(rc).c_str()); + str, GetErrorMsg(rc).c_str()); // fall through case REG_NOMATCH: @@ -229,8 +265,6 @@ bool wxRegExImpl::GetMatch(size_t *start, size_t *len, size_t index) const 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; @@ -240,31 +274,115 @@ bool wxRegExImpl::GetMatch(size_t *start, size_t *len, size_t index) const return TRUE; } -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, -1, _T("NULL text in wxRegEx::Replace") ); + wxCHECK_MSG( IsValid(), -1, _T("must successfully Compile() first") ); - wxCHECK_MSG( IsValid(), FALSE, _T("must successfully Compile() first") ); + // the replacement text + wxString textNew; - int replaced = 0; - size_t lastpos = 0; - wxString newstring; + // 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; - for ( size_t idx = 0; - m_Matches[idx].rm_so != -1 && idx < m_nMatches; - idx++ ) + if ( !mayHaveBackrefs ) { - // 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 ++; + textNew = replacement; + } + + // 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) ) + { + // 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 -1; + } + + matchStart += start; + text->replace(matchStart, len, textNew); + + countRepl++; + + matchStart += textNew.length(); } - if(replaced > 0) - *pattern = newstring; - return replaced; + + return countRepl; } // ---------------------------------------------------------------------------- @@ -301,7 +419,7 @@ bool wxRegEx::Compile(const wxString& expr, int flags) 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") ); @@ -310,7 +428,7 @@ bool wxRegEx::Matches(const wxString& str, int flags) const 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 +442,13 @@ wxString wxRegEx::GetMatch(const wxString& text, size_t index) const return text.Mid(start, len); } -int wxRegEx::Replace(wxString *pattern, const wxString& replacement) const +int wxRegEx::Replace(wxString *pattern, + const wxString& replacement, + size_t maxMatches) const { wxCHECK_MSG( IsValid(), -1, _T("must successfully Compile() first") ); - return m_impl->Replace(pattern, replacement); + return m_impl->Replace(pattern, replacement, maxMatches); } #endif // wxUSE_REGEX