stricter reply code checking: verify that we get the expected reply, not just that...

[wxWidgets.git] / src / common / regex.cpp

diff --git a/src/common/regex.cpp b/src/common/regex.cpp
index d608a5773d45eec10457409dcc99433421e2026c..40dff4df7073d021a788530f5c15462fe759a2db 100644 (file)
--- a/src/common/regex.cpp
+++ b/src/common/regex.cpp
@@ -1,11 +1,11 @@
 ///////////////////////////////////////////////////////////////////////////////
 // Name:        src/common/regex.cpp
 // Purpose:     regular expression matching
 ///////////////////////////////////////////////////////////////////////////////
 // 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:      $Id$
 // Modified by:
 // Created:     13.07.01
 // RCS-ID:      $Id$

-// Copyright:   (c) 2000 Karsten Ballüder <ballueder@gmx.net>
+// Copyright:   (c) 2000 Karsten Ballueder <ballueder@gmx.net>

 //                  2001 Vadim Zeitlin <vadim@wxwindows.org>
 // Licence:     wxWindows licence
 ///////////////////////////////////////////////////////////////////////////////
 //                  2001 Vadim Zeitlin <vadim@wxwindows.org>
 // Licence:     wxWindows licence
 ///////////////////////////////////////////////////////////////////////////////


@@ -32,6 +32,7 @@
     #include "wx/string.h"
     #include "wx/log.h"
     #include "wx/intl.h"
     #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.
 #endif //WX_PRECOMP
 
 // FreeBSD, Watcom and DMars require this, CW doesn't have nor need it.


@@ -56,7 +57,11 @@
 #ifdef __REG_NOFRONT
 #   define WXREGEX_USING_BUILTIN
 #   define WXREGEX_IF_NEED_LEN(x) ,x
 #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
 #   ifdef HAVE_RE_SEARCH
 #       define WXREGEX_IF_NEED_LEN(x) ,x


@@ -67,7 +72,7 @@
 #   if wxUSE_UNICODE
 #       define WXREGEX_CONVERT_TO_MB
 #   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
 #   define wx_regfree regfree
 #   define wx_regerror regerror
 #endif


@@ -90,16 +95,14 @@ public:
     // 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
     // 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, but still check at least in debug
+    // absolutely impractical anyhow

     size_t Start(size_t n) const
     {
     size_t Start(size_t n) const
     {

-        wxASSERT_MSG( m_matches[n].rm_so < UINT_MAX, _T("regex offset overflow") );

         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_so);
     }
 
     size_t End(size_t n) const
     {

-        wxASSERT_MSG( m_matches[n].rm_eo < UINT_MAX, _T("regex offset overflow") );

         return wx_truncate_cast(size_t, m_matches[n].rm_eo);
     }
 
         return wx_truncate_cast(size_t, m_matches[n].rm_eo);
     }
 


@@ -251,7 +254,7 @@ wxString wxRegExImpl::GetErrorMsg(int errorcode, bool badconv) const
 
         (void)wx_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
         delete [] szcmbError;
     }
     else // regerror() returned 0


@@ -279,12 +282,14 @@ bool wxRegExImpl::Compile(const wxString& expr, int flags)
     // translate our flags to regcomp() ones
     int flagsRE = 0;
     if ( !(flags & wxRE_BASIC) )
     // 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;
 #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 )
     if ( flags & wxRE_ICASE )
         flagsRE |= REG_ICASE;
     if ( flags & wxRE_NOSUB )


@@ -295,8 +300,11 @@ bool wxRegExImpl::Compile(const wxString& expr, int flags)
     // compile it
 #ifdef WXREGEX_USING_BUILTIN
     bool conv = true;
     // 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
 #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
     const wxWX2MBbuf conv = expr.mbc_str();
     int errorcode = conv ? regcomp(&m_RegEx, conv, flagsRE) : REG_BADPAT;
 #endif


@@ -370,7 +378,7 @@ static int ReSearch(const regex_t *preg,
                     re_registers *matches,
                     int eflags)
 {
                     re_registers *matches,
                     int eflags)
 {

-    regex_t *pattern = wx_const_cast(regex_t*, preg);
+    regex_t *pattern = const_cast<regex_t*>(preg);

 
     pattern->not_bol = (eflags & REG_NOTBOL) != 0;
     pattern->not_eol = (eflags & REG_NOTEOL) != 0;
 
     pattern->not_bol = (eflags & REG_NOTBOL) != 0;
     pattern->not_eol = (eflags & REG_NOTEOL) != 0;


@@ -507,7 +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) &&
     // 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)) )
     {
                     countRepl ? wxRE_NOTBOL : 0
                     WXREGEX_IF_NEED_LEN(textlen - matchStart)) )
     {


@@ -553,7 +566,12 @@ int wxRegExImpl::Replace(wxString *text,
                     }
                     else
                     {
                     }
                     else
                     {

+#ifndef WXREGEX_CONVERT_TO_MB

                         textNew += wxString(textstr + matchStart + start,
                         textNew += wxString(textstr + matchStart + start,

+#else
+                        textNew += wxString(textstr.data() + matchStart +
+                                           start,
+#endif

                                             *wxConvCurrent, len);
 
                         mayHaveBackrefs = true;
                                             *wxConvCurrent, len);
 
                         mayHaveBackrefs = true;


@@ -583,7 +601,8 @@ int wxRegExImpl::Replace(wxString *text,
 #ifndef WXREGEX_CONVERT_TO_MB
         result.append(*text, matchStart, start);
 #else
 #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);
 #endif
         matchStart += start;
         result.append(textNew);


@@ -596,7 +615,7 @@ int wxRegExImpl::Replace(wxString *text,
 #ifndef WXREGEX_CONVERT_TO_MB
     result.append(*text, matchStart, wxString::npos);
 #else
 #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;
 
 #endif
     *text = result;
 


@@ -636,21 +655,12 @@ bool wxRegEx::Compile(const wxString& expr, int flags)
     return true;
 }
 
     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") );
 
 {
     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
 }
 
 bool wxRegEx::GetMatch(size_t *start, size_t *len, size_t index) const