From: Vadim Zeitlin Date: Sat, 14 Jul 2001 16:13:32 +0000 (+0000) Subject: rewrote wxRegEx::Replace() to do something useful X-Git-Url: https://git.saurik.com/wxWidgets.git/commitdiff_plain/765624f7350bf32b1c78a10ab1a1e10729c52cd1 rewrote wxRegEx::Replace() to do something useful git-svn-id: https://svn.wxwidgets.org/svn/wx/wxWidgets/trunk@11045 c3d73ce0-8a6f-49c7-b76d-6d57e0e08775 --- diff --git a/include/wx/regex.h b/include/wx/regex.h index 70443b2964..e024d0e52a 100644 --- a/include/wx/regex.h +++ b/include/wx/regex.h @@ -99,7 +99,7 @@ public: // flags may be combination of wxRE_NOTBOL and wxRE_NOTEOL // // may only be called after successful call to Compile() - bool Matches(const wxString& str, int flags = 0) const; + bool Matches(const wxChar *text, int flags = 0) const; // get the start index and the length of the match of the expression // (index 0) or a bracketed subexpression (index != 0) @@ -118,7 +118,24 @@ public: // replaces the current regular expression in the string pointed to by // pattern, with the text in replacement and return number of matches // replaced (maybe 0 if none found) or -1 on error - int Replace(wxString *str, const wxString& replacement) const; + // + // the replacement text may contain backreferences (\number) which will be + // replaced with the value of the corresponding subexpression in the + // pattern match + // + // maxMatches may be used to limit the number of replacements made, setting + // it to 1, for example, will only replace first occurence (if any) of the + // pattern in the text while default value of 0 means replace all + int Replace(wxString *text, const wxString& replacement, + size_t maxMatches = 0) const; + + // replace the first occurence + int ReplaceFirst(wxString *text, const wxString& replacement) const + { return Replace(text, replacement, 1); } + + // replace all occurences: this is actually a synonym for Replace() + int ReplaceAll(wxString *text, const wxString& replacement) const + { return Replace(text, replacement, 0); } // dtor not virtual, don't derive from this class ~wxRegEx(); diff --git a/samples/console/console.cpp b/samples/console/console.cpp index 8c89e05dbf..67a1b7f2b2 100644 --- a/samples/console/console.cpp +++ b/samples/console/console.cpp @@ -1827,6 +1827,55 @@ static void TestRegExSubmatch() } } +static void TestRegExReplacement() +{ + wxPuts(_T("*** Testing RE replacement ***")); + + static struct RegExReplTestData + { + const wxChar *text; + const wxChar *repl; + const wxChar *result; + size_t count; + } regExReplTestData[] = + { + { _T("foo123"), _T("bar"), _T("bar"), 1 }, + { _T("foo123"), _T("\\2\\1"), _T("123foo"), 1 }, + { _T("foo_123"), _T("\\2\\1"), _T("123foo"), 1 }, + { _T("123foo"), _T("bar"), _T("123foo"), 0 }, + { _T("123foo456foo"), _T("&&"), _T("123foo456foo456foo"), 1 }, + { _T("foo123foo123"), _T("bar"), _T("barbar"), 2 }, + { _T("foo123_foo456_foo789"), _T("bar"), _T("bar_bar_bar"), 3 }, + }; + + const wxChar *pattern = _T("([a-z]+)[^0-9]*([0-9]+)"); + wxRegEx re = pattern; + + wxPrintf(_T("Using pattern '%s' for replacement.\n"), pattern); + + for ( size_t n = 0; n < WXSIZEOF(regExReplTestData); n++ ) + { + const RegExReplTestData& data = regExReplTestData[n]; + + wxString text = data.text; + size_t nRepl = re.Replace(&text, data.repl); + + wxPrintf(_T("%s =~ s/RE/%s/g: %u match%s, result = '%s' ("), + data.text, data.repl, + nRepl, nRepl == 1 ? _T("") : _T("es"), + text.c_str()); + if ( text == data.result && nRepl == data.count ) + { + wxPuts(_T("ok)")); + } + else + { + wxPrintf(_T("ERROR: should be %u and '%s')\n"), + data.count, data.result); + } + } +} + static void TestRegExInteractive() { wxPuts(_T("*** Testing RE interactively ***")); @@ -5074,8 +5123,9 @@ int main(int argc, char **argv) TestRegExCompile(); TestRegExMatch(); TestRegExSubmatch(); + TestRegExInteractive(); } - TestRegExInteractive(); + TestRegExReplacement(); #endif // TEST_REGEX #ifdef TEST_REGISTRY diff --git a/src/common/regex.cpp b/src/common/regex.cpp index ec57e48df3..46eb372cfb 100644 --- a/src/common/regex.cpp +++ b/src/common/regex.cpp @@ -58,10 +58,11 @@ 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 @@ -180,7 +181,7 @@ 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") ); @@ -213,7 +214,7 @@ 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: @@ -240,31 +241,115 @@ 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; } - if(replaced > 0) - *pattern = newstring; - return replaced; + + // 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(); + } + + return countRepl; } // ---------------------------------------------------------------------------- @@ -301,7 +386,7 @@ 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 +395,7 @@ 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 +409,13 @@ 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