wxMac uses wxStandardPathsCF to mean wxStandardPaths in its wxBase

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

diff --git a/src/common/strconv.cpp b/src/common/strconv.cpp
index 59fdd4af985bfbac0e21586dc875a6ec35378ea0..83c06de9c69e3d7cc7f1a5c05d1ecd1394a3234e 100644 (file)
--- a/src/common/strconv.cpp
+++ b/src/common/strconv.cpp
@@ -1,12 +1,15 @@
 /////////////////////////////////////////////////////////////////////////////
 // Name:        strconv.cpp
 // Purpose:     Unicode conversion classes
 /////////////////////////////////////////////////////////////////////////////
 // Name:        strconv.cpp
 // Purpose:     Unicode conversion classes

-// Author:      Ove Kaaven, Robert Roebling, Vadim Zeitlin, Vaclav Slavik
+// Author:      Ove Kaaven, Robert Roebling, Vadim Zeitlin, Vaclav Slavik,
+//              Ryan Norton, Fredrik Roubert (UTF7)

 // Modified by:
 // Created:     29/01/98
 // RCS-ID:      $Id$
 // Modified by:
 // Created:     29/01/98
 // RCS-ID:      $Id$

-// Copyright:   (c) 1999 Ove Kaaven, Robert Roebling, Vadim Zeitlin, Vaclav Slavik
-// Licence:     wxWindows license
+// Copyright:   (c) 1999 Ove Kaaven, Robert Roebling, Vaclav Slavik
+//              (c) 2000-2003 Vadim Zeitlin
+//              (c) 2004 Ryan Norton, Fredrik Roubert
+// Licence:     wxWindows licence

 /////////////////////////////////////////////////////////////////////////////
 
 // ============================================================================
 /////////////////////////////////////////////////////////////////////////////
 
 // ============================================================================


@@ -17,7 +20,7 @@
 // headers
 // ----------------------------------------------------------------------------
 
 // headers
 // ----------------------------------------------------------------------------
 

-#ifdef __GNUG__
+#if defined(__GNUG__) && !defined(NO_GCC_PRAGMA)

   #pragma implementation "strconv.h"
 #endif
 
   #pragma implementation "strconv.h"
 #endif
 


@@ -33,61 +36,34 @@
     #include "wx/log.h"
 #endif // WX_PRECOMP
 
     #include "wx/log.h"
 #endif // WX_PRECOMP
 

-#ifdef __WXMSW__
-    #include "wx/msw/private.h"
-#endif
-
-#include <errno.h>
-#include <ctype.h>
-#include <string.h>
-#include <stdlib.h>
-
-#include "wx/module.h"

 #include "wx/strconv.h"
 
 #include "wx/strconv.h"
 

-// ----------------------------------------------------------------------------
-// globals
-// ----------------------------------------------------------------------------
-

 #if wxUSE_WCHAR_T
 #if wxUSE_WCHAR_T

-    WXDLLEXPORT_DATA(wxMBConv) wxConvLibc;
-    WXDLLEXPORT_DATA(wxCSConv) wxConvLocal((const wxChar *)NULL);
-    WXDLLEXPORT_DATA(wxCSConv) wxConvISO8859_1(_T("iso-8859-1"));
-#else
-    // stand-ins in absence of wchar_t
-    WXDLLEXPORT_DATA(wxMBConv) wxConvLibc,
-                               wxConvFile,
-                               wxConvISO8859_1,
-                               wxConvLocal;
-#endif // wxUSE_WCHAR_T

 
 

-WXDLLEXPORT_DATA(wxMBConv *) wxConvCurrent = &wxConvLibc;
+#ifdef __WXMSW__
+    #include "wx/msw/private.h"
+#endif

 
 

-class wxStrConvModule: public wxModule
-{
-public:
-    wxStrConvModule() : wxModule() { }
-    virtual bool OnInit() { return TRUE; }
-    virtual void OnExit()
-    {
-#if wxUSE_WCHAR_T
-         wxConvLocal.Clear();
-         wxConvISO8859_1.Clear();
+#ifdef __WINDOWS__
+    #include "wx/msw/missing.h"

 #endif
 #endif

-    }

 
 

-    DECLARE_DYNAMIC_CLASS(wxStrConvModule)
-};
+#ifndef __WXWINCE__
+#include <errno.h>
+#endif

 
 

-IMPLEMENT_DYNAMIC_CLASS(wxStrConvModule, wxModule)
+#include <ctype.h>
+#include <string.h>
+#include <stdlib.h>

 
 

+#if defined(__WIN32__) && !defined(__WXMICROWIN__)
+    #define wxHAVE_WIN32_MB2WC
+#endif // __WIN32__ but !__WXMICROWIN__

 
 // ----------------------------------------------------------------------------
 // headers
 // ----------------------------------------------------------------------------
 
 
 // ----------------------------------------------------------------------------
 // headers
 // ----------------------------------------------------------------------------
 

-#if wxUSE_WCHAR_T
-

 #ifdef __SALFORDC__
     #include <clib.h>
 #endif
 #ifdef __SALFORDC__
     #include <clib.h>
 #endif


@@ -98,7 +74,15 @@ IMPLEMENT_DYNAMIC_CLASS(wxStrConvModule, wxModule)
 
 #include "wx/encconv.h"
 #include "wx/fontmap.h"
 
 #include "wx/encconv.h"
 #include "wx/fontmap.h"

+#include "wx/utils.h"
+
+#ifdef __WXMAC__
+#include <ATSUnicode.h>
+#include <TextCommon.h>
+#include <TextEncodingConverter.h>

 
 

+#include  "wx/mac/private.h"  // includes mac headers
+#endif

 // ----------------------------------------------------------------------------
 // macros
 // ----------------------------------------------------------------------------
 // ----------------------------------------------------------------------------
 // macros
 // ----------------------------------------------------------------------------


@@ -106,12 +90,6 @@ IMPLEMENT_DYNAMIC_CLASS(wxStrConvModule, wxModule)
 #define BSWAP_UCS4(str, len) { unsigned _c; for (_c=0; _c<len; _c++) str[_c]=wxUINT32_SWAP_ALWAYS(str[_c]); }
 #define BSWAP_UTF16(str, len) { unsigned _c; for (_c=0; _c<len; _c++) str[_c]=wxUINT16_SWAP_ALWAYS(str[_c]); }
 
 #define BSWAP_UCS4(str, len) { unsigned _c; for (_c=0; _c<len; _c++) str[_c]=wxUINT32_SWAP_ALWAYS(str[_c]); }
 #define BSWAP_UTF16(str, len) { unsigned _c; for (_c=0; _c<len; _c++) str[_c]=wxUINT16_SWAP_ALWAYS(str[_c]); }
 

-// under Unix SIZEOF_WCHAR_T is defined by configure, but under other platforms
-// it might be not defined - assume the most common value
-#ifndef SIZEOF_WCHAR_T
-    #define SIZEOF_WCHAR_T 2
-#endif // !defined(SIZEOF_WCHAR_T)
-

 #if SIZEOF_WCHAR_T == 4
     #define WC_NAME         "UCS4"
     #define WC_BSWAP         BSWAP_UCS4
 #if SIZEOF_WCHAR_T == 4
     #define WC_NAME         "UCS4"
     #define WC_BSWAP         BSWAP_UCS4


@@ -130,8 +108,8 @@ IMPLEMENT_DYNAMIC_CLASS(wxStrConvModule, wxModule)
       #define WC_NAME_BEST  "UTF-16LE"
     #endif
 #else // sizeof(wchar_t) != 2 nor 4
       #define WC_NAME_BEST  "UTF-16LE"
     #endif
 #else // sizeof(wchar_t) != 2 nor 4

-    // I don't know what to do about this
-    #error "Weird sizeof(wchar_t): please report your platform details to wx-users mailing list"
+    // does this ever happen?
+    #error "Unknown sizeof(wchar_t): please report this to wx-dev@lists.wxwindows.org"

 #endif
 
 // ============================================================================
 #endif
 
 // ============================================================================


@@ -139,16 +117,16 @@ IMPLEMENT_DYNAMIC_CLASS(wxStrConvModule, wxModule)
 // ============================================================================
 
 // ----------------------------------------------------------------------------
 // ============================================================================
 
 // ----------------------------------------------------------------------------

-// UTF-16 en/decoding
+// UTF-16 en/decoding to/from UCS-4

 // ----------------------------------------------------------------------------
 
 // ----------------------------------------------------------------------------
 

-#ifdef WC_UTF16

 
 

-static size_t encode_utf16(wxUint32 input, wchar_t *output)
+static size_t encode_utf16(wxUint32 input, wxUint16 *output)

 {
     if (input<=0xffff)
     {
 {
     if (input<=0xffff)
     {

-        if (output) *output++ = (wchar_t) input;
+        if (output)
+            *output = (wxUint16) input;

         return 1;
     }
     else if (input>=0x110000)
         return 1;
     }
     else if (input>=0x110000)


@@ -159,14 +137,14 @@ static size_t encode_utf16(wxUint32 input, wchar_t *output)
     {
         if (output)
         {
     {
         if (output)
         {

-            *output++ = (wchar_t) ((input >> 10)+0xd7c0);
-            *output++ = (wchar_t) ((input&0x3ff)+0xdc00);
+            *output++ = (wxUint16) ((input >> 10)+0xd7c0);
+            *output = (wxUint16) ((input&0x3ff)+0xdc00);

         }
         return 2;
     }
 }
 
         }
         return 2;
     }
 }
 

-static size_t decode_utf16(const wchar_t* input, wxUint32& output)
+static size_t decode_utf16(const wxUint16* input, wxUint32& output)

 {
     if ((*input<0xd800) || (*input>0xdfff))
     {
 {
     if ((*input<0xd800) || (*input>0xdfff))
     {


@@ -185,53 +163,14 @@ static size_t decode_utf16(const wchar_t* input, wxUint32& output)
     }
 }
 
     }
 }
 

-#endif // WC_UTF16

 
 // ----------------------------------------------------------------------------
 // wxMBConv
 // ----------------------------------------------------------------------------
 
 
 // ----------------------------------------------------------------------------
 // wxMBConv
 // ----------------------------------------------------------------------------
 

-#define IGNORE_LIBC 0
-

 wxMBConv::~wxMBConv()
 {
 wxMBConv::~wxMBConv()
 {

-    // nothing to do here
-}
-
-size_t wxMBConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
-{
-#if IGNORE_LIBC
-    if (buf)
-    {
-        for (size_t i = 0; i < strlen( psz )+1; i++)
-            buf[i] = (wchar_t) psz[i];
-        return strlen( psz );
-    }
-    else
-    {
-        return strlen( psz );
-    }
-#else
-    return wxMB2WC(buf, psz, n);
-#endif
-}
-
-size_t wxMBConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
-{
-#if IGNORE_LIBC
-    if (buf)
-    {
-        for (size_t i = 0; i < wxStrlen( psz )+1; i++)
-            buf[i] = (char) psz[i];
-        return wxStrlen( psz );
-    }
-    else
-    {
-        return wxStrlen( psz );
-    }
-#else
-    return wxWC2MB(buf, psz, n);
-#endif
+    // nothing to do here (necessary for Darwin linking probably)

 }
 
 const wxWCharBuffer wxMBConv::cMB2WC(const char *psz) const
 }
 
 const wxWCharBuffer wxMBConv::cMB2WC(const char *psz) const


@@ -244,9 +183,11 @@ const wxWCharBuffer wxMBConv::cMB2WC(const char *psz) const
         {
             // now do the actual conversion
             wxWCharBuffer buf(nLen);
         {
             // now do the actual conversion
             wxWCharBuffer buf(nLen);

-            MB2WC(buf.data(), psz, nLen + 1); // with the trailing NUL
-
-            return buf;
+            nLen = MB2WC(buf.data(), psz, nLen + 1); // with the trailing NULL
+            if ( nLen != (size_t)-1 )
+            {
+                return buf;
+            }

         }
     }
 
         }
     }
 


@@ -262,10 +203,12 @@ const wxCharBuffer wxMBConv::cWC2MB(const wchar_t *pwz) const
         size_t nLen = WC2MB(NULL, pwz, 0);
         if ( nLen != (size_t)-1 )
         {
         size_t nLen = WC2MB(NULL, pwz, 0);
         if ( nLen != (size_t)-1 )
         {

-            wxCharBuffer buf(nLen);
-            WC2MB(buf.data(), pwz, nLen + 1);
-
-            return buf;
+            wxCharBuffer buf(nLen+3);       // space for a wxUint32 trailing zero
+            nLen = WC2MB(buf.data(), pwz, nLen + 4);
+            if ( nLen != (size_t)-1 )
+            {
+                return buf;
+            }

         }
     }
 
         }
     }
 


@@ -275,84 +218,228 @@ const wxCharBuffer wxMBConv::cWC2MB(const wchar_t *pwz) const
 }
 
 // ----------------------------------------------------------------------------
 }
 
 // ----------------------------------------------------------------------------

-// standard gdk conversion
+// wxMBConvLibc

 // ----------------------------------------------------------------------------
 
 // ----------------------------------------------------------------------------
 

-#ifdef __WXGTK12__
+size_t wxMBConvLibc::MB2WC(wchar_t *buf, const char *psz, size_t n) const
+{
+    return wxMB2WC(buf, psz, n);
+}

 
 

-WXDLLEXPORT_DATA(wxMBConvGdk) wxConvGdk;
+size_t wxMBConvLibc::WC2MB(char *buf, const wchar_t *psz, size_t n) const
+{
+    return wxWC2MB(buf, psz, n);
+}
+// ----------------------------------------------------------------------------
+// UTF-7 
+// ----------------------------------------------------------------------------

 
 

-#include <gdk/gdk.h>
+// Implementation (C) 2004 Fredrik Roubert

 
 

-size_t wxMBConvGdk::MB2WC(wchar_t *buf, const char *psz, size_t n) const
+//
+// BASE64 decoding table
+//
+static const unsigned char utf7unb64[] =

 {
 {

-    if (buf)
-    {
-        return gdk_mbstowcs((GdkWChar *)buf, psz, n);
-    }
-    else
-    {
-        GdkWChar *nbuf = new GdkWChar[n=strlen(psz)];
-        size_t len = gdk_mbstowcs(nbuf, psz, n);
-        delete[] nbuf;
-        return len;
-    }
-}
+    0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
+    0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
+    0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
+    0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
+    0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
+    0xff, 0xff, 0xff, 0x3e, 0xff, 0xff, 0xff, 0x3f,
+    0x34, 0x35, 0x36, 0x37, 0x38, 0x39, 0x3a, 0x3b,
+    0x3c, 0x3d, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
+    0xff, 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06,
+    0x07, 0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e,
+    0x0f, 0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16,
+    0x17, 0x18, 0x19, 0xff, 0xff, 0xff, 0xff, 0xff,
+    0xff, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f, 0x20,
+    0x21, 0x22, 0x23, 0x24, 0x25, 0x26, 0x27, 0x28,
+    0x29, 0x2a, 0x2b, 0x2c, 0x2d, 0x2e, 0x2f, 0x30,
+    0x31, 0x32, 0x33, 0xff, 0xff, 0xff, 0xff, 0xff,
+    0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
+    0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
+    0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
+    0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
+    0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
+    0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
+    0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
+    0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
+    0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
+    0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
+    0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
+    0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
+    0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
+    0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
+    0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
+    0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff
+};

 
 

-size_t wxMBConvGdk::WC2MB(char *buf, const wchar_t *psz, size_t n) const
+size_t wxMBConvUTF7::MB2WC(wchar_t *buf, const char *psz, size_t n) const

 {
 {

-    char *mbstr = gdk_wcstombs((GdkWChar *)psz);
-    size_t len = mbstr ? strlen(mbstr) : 0;
-    if (buf)
+
+    size_t len = 0;
+
+    while (*psz && ((!buf) || (len < n)))

     {
     {

-        if (len > n)
-            len = n;
-        memcpy(buf, psz, len);
-        if (len < n)
-            buf[len] = 0;
+        unsigned char cc = *psz++;
+        if (cc != '+')
+        {
+            // plain ASCII char
+            if (buf)
+                *buf++ = cc;
+            len++;
+        }
+        else if (*psz == '-')
+        {
+            // encoded plus sign
+            if (buf)
+                *buf++ = cc;
+            len++;
+            psz++;
+        }
+        else
+        {
+            // BASE64 encoded string
+            bool lsb;
+            unsigned char c;
+            unsigned int d, l;
+            for (lsb = false, d = 0, l = 0;
+                (cc = utf7unb64[(unsigned char)*psz]) != 0xff; psz++)
+            {
+                d <<= 6;
+                d += cc;
+                for (l += 6; l >= 8; lsb = !lsb)
+                {
+                    c = (d >> (l -= 8)) % 256;
+                    if (lsb)
+                    {
+                        if (buf)
+                            *buf++ |= c;
+                        len ++;
+                    }
+                    else
+                        if (buf)
+                            *buf = c << 8;
+                }
+            }
+            if (*psz == '-')
+                psz++;
+        }

     }
     }

+    if (buf && (len < n))
+        *buf = 0;

     return len;
 }
 
     return len;
 }
 

-#endif // GTK > 1.0
+//
+// BASE64 encoding table
+//
+static const unsigned char utf7enb64[] =
+{
+    'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H',
+    'I', 'J', 'K', 'L', 'M', 'N', 'O', 'P',
+    'Q', 'R', 'S', 'T', 'U', 'V', 'W', 'X',
+    'Y', 'Z', 'a', 'b', 'c', 'd', 'e', 'f',
+    'g', 'h', 'i', 'j', 'k', 'l', 'm', 'n',
+    'o', 'p', 'q', 'r', 's', 't', 'u', 'v',
+    'w', 'x', 'y', 'z', '0', '1', '2', '3',
+    '4', '5', '6', '7', '8', '9', '+', '/'
+};

 
 

-// ----------------------------------------------------------------------------
-// UTF-7
-// ----------------------------------------------------------------------------
+//
+// UTF-7 encoding table
+//
+// 0 - Set D (directly encoded characters)
+// 1 - Set O (optional direct characters)
+// 2 - whitespace characters (optional)
+// 3 - special characters
+//
+static const unsigned char utf7encode[128] =
+{
+    3, 3, 3, 3, 3, 3, 3, 3, 3, 2, 2, 3, 3, 2, 3, 3,
+    3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,
+    2, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1, 3, 0, 0, 0, 3,
+    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 0,
+    1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 3, 1, 1, 1,
+    1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 3, 3
+};

 
 

-WXDLLEXPORT_DATA(wxMBConvUTF7) wxConvUTF7;
+size_t wxMBConvUTF7::WC2MB(char *buf, const wchar_t 
+*psz, size_t n) const
+{

 
 

-#if 0
-static char utf7_setD[]="ABCDEFGHIJKLMNOPQRSTUVWXYZ"
-                        "abcdefghijklmnopqrstuvwxyz"
-                        "0123456789'(),-./:?";
-static char utf7_setO[]="!\"#$%&*;<=>@[]^_`{|}";
-static char utf7_setB[]="ABCDEFGHIJKLMNOPQRSTUVWXYZ"
-                        "abcdefghijklmnopqrstuvwxyz"
-                        "0123456789+/";
-#endif

 
 

-// TODO: write actual implementations of UTF-7 here
-size_t wxMBConvUTF7::MB2WC(wchar_t * WXUNUSED(buf),
-                           const char * WXUNUSED(psz),
-                           size_t WXUNUSED(n)) const
-{
-  return 0;
-}
+    size_t len = 0;

 
 

-size_t wxMBConvUTF7::WC2MB(char * WXUNUSED(buf),
-                           const wchar_t * WXUNUSED(psz),
-                           size_t WXUNUSED(n)) const
-{
-  return 0;
+    while (*psz && ((!buf) || (len < n)))
+    {
+        wchar_t cc = *psz++;
+        if (cc < 0x80 && utf7encode[cc] < 1)
+        {
+            // plain ASCII char
+            if (buf)
+                *buf++ = (char)cc;
+            len++;
+        }
+#ifndef WC_UTF16
+        else if (cc > 0xffff)
+        {
+            // no surrogate pair generation (yet?)
+            return (size_t)-1;
+        }
+#endif
+        else
+        {
+            if (buf)
+                *buf++ = '+';
+            len++;
+            if (cc != '+')
+            {
+                // BASE64 encode string
+                unsigned int lsb, d, l;
+                for (d = 0, l = 0;; psz++)
+                {
+                    for (lsb = 0; lsb < 2; lsb ++)
+                    {
+                        d <<= 8;
+                        d += lsb ? cc & 0xff : (cc & 0xff00) >> 8;
+
+                        for (l += 8; l >= 6; )
+                        {
+                            l -= 6;
+                            if (buf)
+                                *buf++ = utf7enb64[(d >> l) % 64];
+                            len++;
+                        }
+                    }
+                    cc = *psz;
+                    if (!(cc) || (cc < 0x80 && utf7encode[cc] < 1))
+                        break;
+                }
+                if (l != 0)
+                {
+                    if (buf)
+                        *buf++ = utf7enb64[((d % 16) << (6 - l)) % 64];
+                    len++;
+                }
+            }
+            if (buf)
+                *buf++ = '-';
+            len++;
+        }
+    }
+    if (buf && (len < n))
+        *buf = 0;
+    return len;

 }
 
 // ----------------------------------------------------------------------------
 // UTF-8
 // ----------------------------------------------------------------------------
 
 }
 
 // ----------------------------------------------------------------------------
 // UTF-8
 // ----------------------------------------------------------------------------
 

-WXDLLEXPORT_DATA(wxMBConvUTF8) wxConvUTF8;
-

 static wxUint32 utf8_max[]=
     { 0x7f, 0x7ff, 0xffff, 0x1fffff, 0x3ffffff, 0x7fffffff, 0xffffffff };
 
 static wxUint32 utf8_max[]=
     { 0x7f, 0x7ff, 0xffff, 0x1fffff, 0x3ffffff, 0x7fffffff, 0xffffffff };
 


@@ -401,7 +488,8 @@ size_t wxMBConvUTF8::MB2WC(wchar_t *buf, const char *psz, size_t n) const
                     return (size_t)-1;
                 }
 #ifdef WC_UTF16
                     return (size_t)-1;
                 }
 #ifdef WC_UTF16

-                size_t pa = encode_utf16(res, buf);
+                // cast is ok because wchar_t == wxUuint16 if WC_UTF16
+                size_t pa = encode_utf16(res, (wxUint16 *)buf);

                 if (pa == (size_t)-1)
                   return (size_t)-1;
                 if (buf)
                 if (pa == (size_t)-1)
                   return (size_t)-1;
                 if (buf)


@@ -428,7 +516,8 @@ size_t wxMBConvUTF8::WC2MB(char *buf, const wchar_t *psz, size_t n) const
     {
         wxUint32 cc;
 #ifdef WC_UTF16
     {
         wxUint32 cc;
 #ifdef WC_UTF16

-        size_t pa = decode_utf16(psz, cc);
+        // cast is ok for WC_UTF16
+        size_t pa = decode_utf16((const wxUint16 *)psz, cc);

         psz += (pa == (size_t)-1) ? 1 : pa;
 #else
         cc=(*psz++) & 0x7fffffff;
         psz += (pa == (size_t)-1) ? 1 : pa;
 #else
         cc=(*psz++) & 0x7fffffff;


@@ -456,126 +545,550 @@ size_t wxMBConvUTF8::WC2MB(char *buf, const wchar_t *psz, size_t n) const
     }
 
     if (buf && (len<n)) *buf = 0;
     }
 
     if (buf && (len<n)) *buf = 0;

-    
+

     return len;
 }
 
     return len;
 }
 

-// ============================================================================
-// wxCharacterSet and derived classes
-// ============================================================================

 
 

-// ----------------------------------------------------------------------------
-// wxCharacterSet is the ABC for the classes below
-// ----------------------------------------------------------------------------

 
 

-class wxCharacterSet
-{
-public:
-    wxCharacterSet(const wxChar*name) : cname(name) {}
-    virtual ~wxCharacterSet() {}
-    virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n) = 0;
-    virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n) = 0;
-    virtual bool usable() const = 0;
-public:
-    const wxChar*cname;
-};

 
 // ----------------------------------------------------------------------------
 
 // ----------------------------------------------------------------------------

-// ID_CharSet: implementation of wxCharacterSet using an existing wxMBConv
+// UTF-16

 // ----------------------------------------------------------------------------
 
 // ----------------------------------------------------------------------------
 

-class ID_CharSet : public wxCharacterSet
+#ifdef WORDS_BIGENDIAN
+    #define wxMBConvUTF16straight wxMBConvUTF16BE
+    #define wxMBConvUTF16swap     wxMBConvUTF16LE
+#else
+    #define wxMBConvUTF16swap     wxMBConvUTF16BE
+    #define wxMBConvUTF16straight wxMBConvUTF16LE
+#endif
+
+
+#ifdef WC_UTF16
+
+// copy 16bit MB to 16bit String
+size_t wxMBConvUTF16straight::MB2WC(wchar_t *buf, const char *psz, size_t n) const

 {
 {

-public:
-    ID_CharSet(const wxChar *name, wxMBConv *cnv)
-        : wxCharacterSet(name), work(cnv) {}
+    size_t len=0;

 
 

-    size_t MB2WC(wchar_t *buf, const char *psz, size_t n)
-        { return work ? work->MB2WC(buf,psz,n) : (size_t)-1; }
+    while (*(wxUint16*)psz && (!buf || len < n))
+    {
+        if (buf)
+            *buf++ = *(wxUint16*)psz;
+        len++;

 
 

-    size_t WC2MB(char *buf, const wchar_t *psz, size_t n)
-        { return work ? work->WC2MB(buf,psz,n) : (size_t)-1; }
+        psz += sizeof(wxUint16);
+    }
+    if (buf && len<n)   *buf=0;

 
 

-    bool usable() const
-        { return work!=NULL; }
-public:
-    wxMBConv*work;
-};
+    return len;
+}

 
 
 
 

-// ============================================================================
-// The classes doing conversion using the iconv_xxx() functions
-// ============================================================================
+// copy 16bit String to 16bit MB
+size_t wxMBConvUTF16straight::WC2MB(char *buf, const wchar_t *psz, size_t n) const
+{
+    size_t len=0;

 
 

-#ifdef HAVE_ICONV
+    while (*psz && (!buf || len < n))
+    {
+        if (buf)
+        {
+            *(wxUint16*)buf = *psz;
+            buf += sizeof(wxUint16);
+        }
+        len += sizeof(wxUint16);
+        psz++;
+    }
+    if (buf && len<=n-sizeof(wxUint16))   *(wxUint16*)buf=0;

 
 

-// VS: glibc 2.1.3 is broken in that iconv() conversion to/from UCS4 fails with E2BIG
-//     if output buffer is _exactly_ as big as needed. Such case is (unless there's
-//     yet another bug in glibc) the only case when iconv() returns with (size_t)-1
-//     (which means error) and says there are 0 bytes left in the input buffer --
-//     when _real_ error occurs, bytes-left-in-input buffer is non-zero. Hence,
-//     this alternative test for iconv() failure.
-//     [This bug does not appear in glibc 2.2.]
-#if defined(__GLIBC__) && __GLIBC__ == 2 && __GLIBC_MINOR__ <= 1
-#define ICONV_FAILED(cres, bufLeft) ((cres == (size_t)-1) && \
-                                     (errno != E2BIG || bufLeft != 0))
-#else
-#define ICONV_FAILED(cres, bufLeft)  (cres == (size_t)-1)
-#endif
+    return len;
+}

 
 

-#define ICONV_CHAR_CAST(x)  ((ICONV_CONST char **)(x))

 
 

-// ----------------------------------------------------------------------------
-// IC_CharSet: encapsulates an iconv character set
-// ----------------------------------------------------------------------------
+// swap 16bit MB to 16bit String
+size_t wxMBConvUTF16swap::MB2WC(wchar_t *buf, const char *psz, size_t n) const
+{
+    size_t len=0;
+
+    while (*(wxUint16*)psz && (!buf || len < n))
+    {
+        if (buf)
+        {
+            ((char *)buf)[0] = psz[1];
+            ((char *)buf)[1] = psz[0];
+            buf++;
+        }
+        len++;
+        psz += sizeof(wxUint16);
+    }
+    if (buf && len<n)   *buf=0;
+
+    return len;
+}

 
 

-class IC_CharSet : public wxCharacterSet
+
+// swap 16bit MB to 16bit String
+size_t wxMBConvUTF16swap::WC2MB(char *buf, const wchar_t *psz, size_t n) const

 {
 {

-public:
-    IC_CharSet(const wxChar *name);
-    virtual ~IC_CharSet();
+    size_t len=0;

 
 

-    virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n);
-    virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n);
+    while (*psz && (!buf || len < n))
+    {
+        if (buf)
+        {
+            *buf++ = ((char*)psz)[1];
+            *buf++ = ((char*)psz)[0];
+        }
+        len += sizeof(wxUint16);
+        psz++;
+    }
+    if (buf && len<=n-sizeof(wxUint16))   *(wxUint16*)buf=0;

 
 

-    bool usable() const
-        { return (m2w != (iconv_t)-1) && (w2m != (iconv_t)-1); }
+    return len;
+}

 
 

-protected:
-    // the iconv handlers used to translate from multibyte to wide char and in
-    // the other direction
-    iconv_t m2w,
-            w2m;

 
 

-private:
-    // the name (for iconv_open()) of a wide char charset - if none is
-    // available on this machine, it will remain NULL
-    static const char *ms_wcCharsetName;
+#else // WC_UTF16

 
 

-    // true if the wide char encoding we use (i.e. ms_wcCharsetName) has
-    // different endian-ness than the native one
-    static bool ms_wcNeedsSwap;
-};

 
 

-const char *IC_CharSet::ms_wcCharsetName = NULL;
-bool IC_CharSet::ms_wcNeedsSwap = FALSE;
+// copy 16bit MB to 32bit String
+size_t wxMBConvUTF16straight::MB2WC(wchar_t *buf, const char *psz, size_t n) const
+{
+    size_t len=0;
+
+    while (*(wxUint16*)psz && (!buf || len < n))
+    {
+        wxUint32 cc;
+        size_t pa=decode_utf16((wxUint16*)psz, cc);
+        if (pa == (size_t)-1)
+            return pa;
+
+        if (buf)
+            *buf++ = cc;
+        len++;
+        psz += pa * sizeof(wxUint16);
+    }
+    if (buf && len<n)   *buf=0;
+
+    return len;
+}

 
 

-IC_CharSet::IC_CharSet(const wxChar *name)
-          : wxCharacterSet(name)
+
+// copy 32bit String to 16bit MB
+size_t wxMBConvUTF16straight::WC2MB(char *buf, const wchar_t *psz, size_t n) const

 {
 {

-    // Do it the hard way
-    char cname[100];
-    for (size_t i = 0; i < wxStrlen(name)+1; i++)
-        cname[i] = (char) name[i];
+    size_t len=0;

 
 

-    // check for charset that represents wchar_t:
-    if (ms_wcCharsetName == NULL)
+    while (*psz && (!buf || len < n))

     {
     {

-        ms_wcNeedsSwap = FALSE;
+        wxUint16 cc[2];
+        size_t pa=encode_utf16(*psz, cc);

 
 

-        // try charset with explicit bytesex info (e.g. "UCS-4LE"):
-        ms_wcCharsetName = WC_NAME_BEST;
-        m2w = iconv_open(ms_wcCharsetName, cname);
+        if (pa == (size_t)-1)
+            return pa;
+
+        if (buf)
+        {
+            *(wxUint16*)buf = cc[0];
+            buf += sizeof(wxUint16);
+            if (pa > 1)
+            {
+                *(wxUint16*)buf = cc[1];
+                buf += sizeof(wxUint16);
+            }
+        }
+
+        len += pa*sizeof(wxUint16);
+        psz++;
+    }
+    if (buf && len<=n-sizeof(wxUint16))   *(wxUint16*)buf=0;
+
+    return len;
+}
+
+
+// swap 16bit MB to 32bit String
+size_t wxMBConvUTF16swap::MB2WC(wchar_t *buf, const char *psz, size_t n) const
+{
+    size_t len=0;
+
+    while (*(wxUint16*)psz && (!buf || len < n))
+    {
+        wxUint32 cc;
+        char tmp[4];
+        tmp[0]=psz[1];  tmp[1]=psz[0];
+        tmp[2]=psz[3];  tmp[3]=psz[2];
+
+        size_t pa=decode_utf16((wxUint16*)tmp, cc);
+        if (pa == (size_t)-1)
+            return pa;
+
+        if (buf)
+            *buf++ = cc;
+
+        len++;
+        psz += pa * sizeof(wxUint16);
+    }
+    if (buf && len<n)   *buf=0;
+
+    return len;
+}
+
+
+// swap 32bit String to 16bit MB
+size_t wxMBConvUTF16swap::WC2MB(char *buf, const wchar_t *psz, size_t n) const
+{
+    size_t len=0;
+
+    while (*psz && (!buf || len < n))
+    {
+        wxUint16 cc[2];
+        size_t pa=encode_utf16(*psz, cc);
+
+        if (pa == (size_t)-1)
+            return pa;
+
+        if (buf)
+        {
+            *buf++ = ((char*)cc)[1];
+            *buf++ = ((char*)cc)[0];
+            if (pa > 1)
+            {
+                *buf++ = ((char*)cc)[3];
+                *buf++ = ((char*)cc)[2];
+            }
+        }
+
+        len += pa*sizeof(wxUint16);
+        psz++;
+    }
+    if (buf && len<=n-sizeof(wxUint16))   *(wxUint16*)buf=0;
+
+    return len;
+}
+
+#endif // WC_UTF16
+
+
+// ----------------------------------------------------------------------------
+// UTF-32
+// ----------------------------------------------------------------------------
+
+#ifdef WORDS_BIGENDIAN
+#define wxMBConvUTF32straight  wxMBConvUTF32BE
+#define wxMBConvUTF32swap      wxMBConvUTF32LE
+#else
+#define wxMBConvUTF32swap      wxMBConvUTF32BE
+#define wxMBConvUTF32straight  wxMBConvUTF32LE
+#endif
+
+
+WXDLLIMPEXP_DATA_BASE(wxMBConvUTF32LE) wxConvUTF32LE;
+WXDLLIMPEXP_DATA_BASE(wxMBConvUTF32BE) wxConvUTF32BE;
+
+
+#ifdef WC_UTF16
+
+// copy 32bit MB to 16bit String
+size_t wxMBConvUTF32straight::MB2WC(wchar_t *buf, const char *psz, size_t n) const
+{
+    size_t len=0;
+
+    while (*(wxUint32*)psz && (!buf || len < n))
+    {
+        wxUint16 cc[2];
+
+        size_t pa=encode_utf16(*(wxUint32*)psz, cc);
+        if (pa == (size_t)-1)
+            return pa;
+
+        if (buf)
+        {
+            *buf++ = cc[0];
+            if (pa > 1)
+                *buf++ = cc[1];
+        }
+        len += pa;
+        psz += sizeof(wxUint32);
+    }
+    if (buf && len<n)   *buf=0;
+
+    return len;
+}
+
+
+// copy 16bit String to 32bit MB
+size_t wxMBConvUTF32straight::WC2MB(char *buf, const wchar_t *psz, size_t n) const
+{
+    size_t len=0;
+
+    while (*psz && (!buf || len < n))
+    {
+        wxUint32 cc;
+
+        // cast is ok for WC_UTF16
+        size_t pa = decode_utf16((const wxUint16 *)psz, cc);
+        if (pa == (size_t)-1)
+            return pa;
+
+        if (buf)
+        {
+            *(wxUint32*)buf = cc;
+            buf += sizeof(wxUint32);
+        }
+        len += sizeof(wxUint32);
+        psz += pa;
+    }
+
+    if (buf && len<=n-sizeof(wxUint32))
+        *(wxUint32*)buf=0;
+
+    return len;
+}
+
+
+
+// swap 32bit MB to 16bit String
+size_t wxMBConvUTF32swap::MB2WC(wchar_t *buf, const char *psz, size_t n) const
+{
+    size_t len=0;
+
+    while (*(wxUint32*)psz && (!buf || len < n))
+    {
+        char tmp[4];
+        tmp[0] = psz[3];   tmp[1] = psz[2];
+        tmp[2] = psz[1];   tmp[3] = psz[0];
+
+
+        wxUint16 cc[2];
+
+        size_t pa=encode_utf16(*(wxUint32*)tmp, cc);
+        if (pa == (size_t)-1)
+            return pa;
+
+        if (buf)
+        {
+            *buf++ = cc[0];
+            if (pa > 1)
+                *buf++ = cc[1];
+        }
+        len += pa;
+        psz += sizeof(wxUint32);
+    }
+
+    if (buf && len<n)
+        *buf=0;
+
+    return len;
+}
+
+
+// swap 16bit String to 32bit MB
+size_t wxMBConvUTF32swap::WC2MB(char *buf, const wchar_t *psz, size_t n) const
+{
+    size_t len=0;
+
+    while (*psz && (!buf || len < n))
+    {
+        char cc[4];
+
+        // cast is ok for WC_UTF16
+        size_t pa=decode_utf16((const wxUint16 *)psz, *(wxUint32*)cc);
+        if (pa == (size_t)-1)
+            return pa;
+
+        if (buf)
+        {
+            *buf++ = cc[3];
+            *buf++ = cc[2];
+            *buf++ = cc[1];
+            *buf++ = cc[0];
+        }
+        len += sizeof(wxUint32);
+        psz += pa;
+    }
+
+    if (buf && len<=n-sizeof(wxUint32))
+        *(wxUint32*)buf=0;
+
+    return len;
+}
+
+#else // WC_UTF16
+
+
+// copy 32bit MB to 32bit String
+size_t wxMBConvUTF32straight::MB2WC(wchar_t *buf, const char *psz, size_t n) const
+{
+    size_t len=0;
+
+    while (*(wxUint32*)psz && (!buf || len < n))
+    {
+        if (buf)
+            *buf++ = *(wxUint32*)psz;
+        len++;
+        psz += sizeof(wxUint32);
+    }
+
+    if (buf && len<n)
+        *buf=0;
+
+    return len;
+}
+
+
+// copy 32bit String to 32bit MB
+size_t wxMBConvUTF32straight::WC2MB(char *buf, const wchar_t *psz, size_t n) const
+{
+    size_t len=0;
+
+    while (*psz && (!buf || len < n))
+    {
+        if (buf)
+        {
+            *(wxUint32*)buf = *psz;
+            buf += sizeof(wxUint32);
+        }
+
+        len += sizeof(wxUint32);
+        psz++;
+    }
+
+    if (buf && len<=n-sizeof(wxUint32))
+        *(wxUint32*)buf=0;
+
+    return len;
+}
+
+
+// swap 32bit MB to 32bit String
+size_t wxMBConvUTF32swap::MB2WC(wchar_t *buf, const char *psz, size_t n) const
+{
+    size_t len=0;
+
+    while (*(wxUint32*)psz && (!buf || len < n))
+    {
+        if (buf)
+        {
+            ((char *)buf)[0] = psz[3];
+            ((char *)buf)[1] = psz[2];
+            ((char *)buf)[2] = psz[1];
+            ((char *)buf)[3] = psz[0];
+            buf++;
+        }
+        len++;
+        psz += sizeof(wxUint32);
+    }
+
+    if (buf && len<n)
+        *buf=0;
+
+    return len;
+}
+
+
+// swap 32bit String to 32bit MB
+size_t wxMBConvUTF32swap::WC2MB(char *buf, const wchar_t *psz, size_t n) const
+{
+    size_t len=0;
+
+    while (*psz && (!buf || len < n))
+    {
+        if (buf)
+        {
+            *buf++ = ((char *)psz)[3];
+            *buf++ = ((char *)psz)[2];
+            *buf++ = ((char *)psz)[1];
+            *buf++ = ((char *)psz)[0];
+        }
+        len += sizeof(wxUint32);
+        psz++;
+    }
+
+    if (buf && len<=n-sizeof(wxUint32))
+        *(wxUint32*)buf=0;
+
+    return len;
+}
+
+
+#endif // WC_UTF16
+
+
+// ============================================================================
+// The classes doing conversion using the iconv_xxx() functions
+// ============================================================================
+
+#ifdef HAVE_ICONV
+
+// VS: glibc 2.1.3 is broken in that iconv() conversion to/from UCS4 fails with E2BIG
+//     if output buffer is _exactly_ as big as needed. Such case is (unless there's
+//     yet another bug in glibc) the only case when iconv() returns with (size_t)-1
+//     (which means error) and says there are 0 bytes left in the input buffer --
+//     when _real_ error occurs, bytes-left-in-input buffer is non-zero. Hence,
+//     this alternative test for iconv() failure.
+//     [This bug does not appear in glibc 2.2.]
+#if defined(__GLIBC__) && __GLIBC__ == 2 && __GLIBC_MINOR__ <= 1
+#define ICONV_FAILED(cres, bufLeft) ((cres == (size_t)-1) && \
+                                     (errno != E2BIG || bufLeft != 0))
+#else
+#define ICONV_FAILED(cres, bufLeft)  (cres == (size_t)-1)
+#endif
+
+#define ICONV_CHAR_CAST(x)  ((ICONV_CONST char **)(x))
+
+// ----------------------------------------------------------------------------
+// wxMBConv_iconv: encapsulates an iconv character set
+// ----------------------------------------------------------------------------
+
+class wxMBConv_iconv : public wxMBConv
+{
+public:
+    wxMBConv_iconv(const wxChar *name);
+    virtual ~wxMBConv_iconv();
+
+    virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n) const;
+    virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n) const;
+
+    bool IsOk() const
+        { return (m2w != (iconv_t)-1) && (w2m != (iconv_t)-1); }
+
+protected:
+    // the iconv handlers used to translate from multibyte to wide char and in
+    // the other direction
+    iconv_t m2w,
+            w2m;
+
+private:
+    // the name (for iconv_open()) of a wide char charset -- if none is
+    // available on this machine, it will remain NULL
+    static const char *ms_wcCharsetName;
+
+    // true if the wide char encoding we use (i.e. ms_wcCharsetName) has
+    // different endian-ness than the native one
+    static bool ms_wcNeedsSwap;
+};
+
+const char *wxMBConv_iconv::ms_wcCharsetName = NULL;
+bool wxMBConv_iconv::ms_wcNeedsSwap = false;
+
+wxMBConv_iconv::wxMBConv_iconv(const wxChar *name)
+{
+    // Do it the hard way
+    char cname[100];
+    for (size_t i = 0; i < wxStrlen(name)+1; i++)
+        cname[i] = (char) name[i];
+
+    // check for charset that represents wchar_t:
+    if (ms_wcCharsetName == NULL)
+    {
+        ms_wcNeedsSwap = false;
+
+        // try charset with explicit bytesex info (e.g. "UCS-4LE"):
+        ms_wcCharsetName = WC_NAME_BEST;
+        m2w = iconv_open(ms_wcCharsetName, cname);

 
         if (m2w == (iconv_t)-1)
         {
 
         if (m2w == (iconv_t)-1)
         {


@@ -624,7 +1137,7 @@ IC_CharSet::IC_CharSet(const wxChar *name)
             {
                 ms_wcCharsetName = NULL;
 
             {
                 ms_wcCharsetName = NULL;
 

-                // VS: we must not output an error here, since wxWindows will safely
+                // VS: we must not output an error here, since wxWidgets will safely

                 //     fall back to using wxEncodingConverter.
                 wxLogTrace(wxT("strconv"), wxT("Impossible to convert to/from charset '%s' with iconv, falling back to wxEncodingConverter."), name);
                 //wxLogError(
                 //     fall back to using wxEncodingConverter.
                 wxLogTrace(wxT("strconv"), wxT("Impossible to convert to/from charset '%s' with iconv, falling back to wxEncodingConverter."), name);
                 //wxLogError(


@@ -648,7 +1161,7 @@ IC_CharSet::IC_CharSet(const wxChar *name)
     }
 }
 
     }
 }
 

-IC_CharSet::~IC_CharSet()
+wxMBConv_iconv::~wxMBConv_iconv()

 {
     if ( m2w != (iconv_t)-1 )
         iconv_close(m2w);
 {
     if ( m2w != (iconv_t)-1 )
         iconv_close(m2w);


@@ -656,7 +1169,7 @@ IC_CharSet::~IC_CharSet()
         iconv_close(w2m);
 }
 
         iconv_close(w2m);
 }
 

-size_t IC_CharSet::MB2WC(wchar_t *buf, const char *psz, size_t n)
+size_t wxMBConv_iconv::MB2WC(wchar_t *buf, const char *psz, size_t n) const

 {
     size_t inbuf = strlen(psz);
     size_t outbuf = n * SIZEOF_WCHAR_T;
 {
     size_t inbuf = strlen(psz);
     size_t outbuf = n * SIZEOF_WCHAR_T;


@@ -678,7 +1191,7 @@ size_t IC_CharSet::MB2WC(wchar_t *buf, const char *psz, size_t n)
             // convert to native endianness
             WC_BSWAP(buf /* _not_ bufPtr */, res)
         }
             // convert to native endianness
             WC_BSWAP(buf /* _not_ bufPtr */, res)
         }

-        
+

         // NB: iconv was given only strlen(psz) characters on input, and so
         //     it couldn't convert the trailing zero. Let's do it ourselves
         //     if there's some room left for it in the output buffer.
         // NB: iconv was given only strlen(psz) characters on input, and so
         //     it couldn't convert the trailing zero. Let's do it ourselves
         //     if there's some room left for it in the output buffer.


@@ -713,7 +1226,7 @@ size_t IC_CharSet::MB2WC(wchar_t *buf, const char *psz, size_t n)
     return res;
 }
 
     return res;
 }
 

-size_t IC_CharSet::WC2MB(char *buf, const wchar_t *psz, size_t n)
+size_t wxMBConv_iconv::WC2MB(char *buf, const wchar_t *psz, size_t n) const

 {
     size_t inbuf = wxWcslen(psz) * SIZEOF_WCHAR_T;
     size_t outbuf = n;
 {
     size_t inbuf = wxWcslen(psz) * SIZEOF_WCHAR_T;
     size_t outbuf = n;


@@ -739,7 +1252,7 @@ size_t IC_CharSet::WC2MB(char *buf, const wchar_t *psz, size_t n)
         cres = iconv( w2m, ICONV_CHAR_CAST(&psz), &inbuf, &buf, &outbuf );
 
         res = n-outbuf;
         cres = iconv( w2m, ICONV_CHAR_CAST(&psz), &inbuf, &buf, &outbuf );
 
         res = n-outbuf;

-        
+

         // NB: iconv was given only wcslen(psz) characters on input, and so
         //     it couldn't convert the trailing zero. Let's do it ourselves
         //     if there's some room left for it in the output buffer.
         // NB: iconv was given only wcslen(psz) characters on input, and so
         //     it couldn't convert the trailing zero. Let's do it ourselves
         //     if there's some room left for it in the output buffer.


@@ -778,204 +1291,1116 @@ size_t IC_CharSet::WC2MB(char *buf, const wchar_t *psz, size_t n)
 
 #endif // HAVE_ICONV
 
 
 #endif // HAVE_ICONV
 

+

 // ============================================================================
 // Win32 conversion classes
 // ============================================================================
 
 // ============================================================================
 // Win32 conversion classes
 // ============================================================================
 

-#if defined(__WIN32__) && !defined(__WXMICROWIN__)
+#ifdef wxHAVE_WIN32_MB2WC

 
 

-#ifdef __WXWINE__
-    #define WINE_CAST (WCHAR *)
-#else
-    #define WINE_CAST
+// from utils.cpp
+#if wxUSE_FONTMAP
+extern WXDLLIMPEXP_BASE long wxCharsetToCodepage(const wxChar *charset);
+extern WXDLLIMPEXP_BASE long wxEncodingToCodepage(wxFontEncoding encoding);

 #endif
 
 #endif
 

-extern long wxCharsetToCodepage(const wxChar *charset); // from utils.cpp
-
-class CP_CharSet : public wxCharacterSet
+class wxMBConv_win32 : public wxMBConv

 {
 public:
 {
 public:

-    CP_CharSet(const wxChar* name)
-        : wxCharacterSet(name)
-        {
-            m_CodePage = wxCharsetToCodepage(name);
-        }
+    wxMBConv_win32()
+    {
+        m_CodePage = CP_ACP;
+    }
+
+#if wxUSE_FONTMAP
+    wxMBConv_win32(const wxChar* name)
+    {
+        m_CodePage = wxCharsetToCodepage(name);
+    }
+
+    wxMBConv_win32(wxFontEncoding encoding)
+    {
+        m_CodePage = wxEncodingToCodepage(encoding);
+    }
+#endif

 
 

-    size_t MB2WC(wchar_t *buf, const char *psz, size_t n)
+    size_t MB2WC(wchar_t *buf, const char *psz, size_t n) const

     {
     {

+        // note that we have to use MB_ERR_INVALID_CHARS flag as it without it
+        // the behaviour is not compatible with the Unix version (using iconv)
+        // and break the library itself, e.g. wxTextInputStream::NextChar()
+        // wouldn't work if reading an incomplete MB char didn't result in an
+        // error

         const size_t len = ::MultiByteToWideChar
                              (
                                 m_CodePage,     // code page
         const size_t len = ::MultiByteToWideChar
                              (
                                 m_CodePage,     // code page

-                                0,              // flags (none)
+                                MB_ERR_INVALID_CHARS, // flags: fall on error

                                 psz,            // input string
                                 -1,             // its length (NUL-terminated)
                                 psz,            // input string
                                 -1,             // its length (NUL-terminated)

-                                WINE_CAST buf,  // output string
+                                buf,            // output string

                                 buf ? n : 0     // size of output buffer
                              );
 
                                 buf ? n : 0     // size of output buffer
                              );
 

-        // note that it returns # of written chars for buf != NULL and *size*
-        // of the needed buffer for buf == NULL
-        return len ? (buf ? len : len - 1) : (size_t)-1;
+        // note that it returns count of written chars for buf != NULL and size
+        // of the needed buffer for buf == NULL so in either case the length of
+        // the string (which never includes the terminating NUL) is one less
+        return len ? len - 1 : (size_t)-1;

     }
 
     }
 

-    size_t WC2MB(char *buf, const wchar_t *psz, size_t n)
+    size_t WC2MB(char *buf, const wchar_t *pwz, size_t n) const

     {
     {

-        const size_t len = ::WideCharToMultiByte
-                             (
+        /*
+            we have a problem here: by default, WideCharToMultiByte() may
+            replace characters unrepresentable in the target code page with bad
+            quality approximations such as turning "1/2" symbol (U+00BD) into
+            "1" for the code pages which don't have it and we, obviously, want
+            to avoid this at any price
+
+            the trouble is that this function does it _silently_, i.e. it won't
+            even tell us whether it did or not... Win98/2000 and higher provide
+            WC_NO_BEST_FIT_CHARS but it doesn't work for the older systems and
+            we have to resort to a round trip, i.e. check that converting back
+            results in the same string -- this is, of course, expensive but
+            otherwise we simply can't be sure to not garble the data.
+         */
+
+        // determine if we can rely on WC_NO_BEST_FIT_CHARS: according to MSDN
+        // it doesn't work with CJK encodings (which we test for rather roughly
+        // here...) nor with UTF-7/8 nor, of course, with Windows versions not
+        // supporting it
+        BOOL usedDef wxDUMMY_INITIALIZE(false);
+        BOOL *pUsedDef;
+        int flags;
+        if ( CanUseNoBestFit() && m_CodePage < 50000 )
+        {
+            // it's our lucky day
+            flags = WC_NO_BEST_FIT_CHARS;
+            pUsedDef = &usedDef;
+        }
+        else // old system or unsupported encoding
+        {
+            flags = 0;
+            pUsedDef = NULL;
+        }
+
+        const size_t len = ::WideCharToMultiByte
+                             (

                                 m_CodePage,     // code page
                                 m_CodePage,     // code page

-                                0,              // flags (none)
-                                WINE_CAST psz,  // input string
+                                flags,          // either none or no best fit
+                                pwz,            // input string

                                 -1,             // it is (wide) NUL-terminated
                                 buf,            // output buffer
                                 buf ? n : 0,    // and its size
                                 NULL,           // default "replacement" char
                                 -1,             // it is (wide) NUL-terminated
                                 buf,            // output buffer
                                 buf ? n : 0,    // and its size
                                 NULL,           // default "replacement" char

-                                NULL            // [out] was it used?
+                                pUsedDef        // [out] was it used?

                              );
 
                              );
 

-        // see the comment above!
-        return len ? (buf ? len : len - 1) : (size_t)-1;
+        if ( !len )
+        {
+            // function totally failed
+            return (size_t)-1;
+        }
+
+        // if we were really converting, check if we succeeded
+        if ( buf )
+        {
+            if ( flags )
+            {
+                // check if the conversion failed, i.e. if any replacements
+                // were done
+                if ( usedDef )
+                    return (size_t)-1;
+            }
+            else // we must resort to double tripping...
+            {
+                wxWCharBuffer wcBuf(n);
+                if ( MB2WC(wcBuf.data(), buf, n) == (size_t)-1 ||
+                        wcscmp(wcBuf, pwz) != 0 )
+                {
+                    // we didn't obtain the same thing we started from, hence
+                    // the conversion was lossy and we consider that it failed
+                    return (size_t)-1;
+                }
+            }
+        }
+
+        // see the comment above for the reason of "len - 1"
+        return len - 1;

     }
 
     }
 

-    bool usable() const
-        { return m_CodePage != -1; }
+    bool IsOk() const { return m_CodePage != -1; }
+
+private:
+    static bool CanUseNoBestFit()
+    {
+        static int s_isWin98Or2k = -1;
+
+        if ( s_isWin98Or2k == -1 )
+        {
+            int verMaj, verMin;
+            switch ( wxGetOsVersion(&verMaj, &verMin) )
+            {
+                case wxWIN95:
+                    s_isWin98Or2k = verMaj >= 4 && verMin >= 10;
+                    break;
+
+                case wxWINDOWS_NT:
+                    s_isWin98Or2k = verMaj >= 5;
+                    break;
+
+                default:
+                    // unknown, be conseravtive by default
+                    s_isWin98Or2k = 0;
+            }
+
+            wxASSERT_MSG( s_isWin98Or2k != -1, _T("should be set above") );
+        }
+
+        return s_isWin98Or2k == 1;
+    }

 
 

-public:

     long m_CodePage;
 };
     long m_CodePage;
 };

-#endif // __WIN32__
+
+#endif // wxHAVE_WIN32_MB2WC

 
 // ============================================================================
 
 // ============================================================================

-// wxEncodingConverter based conversion classes
+// Cocoa conversion classes

 // ============================================================================
 
 // ============================================================================
 

-#if wxUSE_FONTMAP
+#if defined(__WXCOCOA__)
+
+// RN:  There is no UTF-32 support in either Core Foundation or
+// Cocoa.  Strangely enough, internally Core Foundation uses
+// UTF 32 internally quite a bit - its just not public (yet).
+
+#include <CoreFoundation/CFString.h>
+#include <CoreFoundation/CFStringEncodingExt.h>
+
+CFStringEncoding wxCFStringEncFromFontEnc(wxFontEncoding encoding)
+{
+    CFStringEncoding enc = 0 ;
+    if ( encoding == wxFONTENCODING_DEFAULT )
+    {
+#if wxUSE_GUI
+        encoding = wxFont::GetDefaultEncoding() ;
+#else
+        encoding = wxLocale::GetSystemEncoding() ;
+#endif
+    }
+    else switch( encoding)
+    {
+        case wxFONTENCODING_ISO8859_1 :
+            enc = kCFStringEncodingISOLatin1 ;
+            break ;
+        case wxFONTENCODING_ISO8859_2 :
+            enc = kCFStringEncodingISOLatin2;
+            break ;
+        case wxFONTENCODING_ISO8859_3 :
+            enc = kCFStringEncodingISOLatin3 ;
+            break ;
+        case wxFONTENCODING_ISO8859_4 :
+            enc = kCFStringEncodingISOLatin4;
+            break ;
+        case wxFONTENCODING_ISO8859_5 :
+            enc = kCFStringEncodingISOLatinCyrillic;
+            break ;
+        case wxFONTENCODING_ISO8859_6 :
+            enc = kCFStringEncodingISOLatinArabic;
+            break ;
+        case wxFONTENCODING_ISO8859_7 :
+            enc = kCFStringEncodingISOLatinGreek;
+            break ;
+        case wxFONTENCODING_ISO8859_8 :
+            enc = kCFStringEncodingISOLatinHebrew;
+            break ;
+        case wxFONTENCODING_ISO8859_9 :
+            enc = kCFStringEncodingISOLatin5;
+            break ;
+        case wxFONTENCODING_ISO8859_10 :
+            enc = kCFStringEncodingISOLatin6;
+            break ;
+        case wxFONTENCODING_ISO8859_11 :
+            enc = kCFStringEncodingISOLatinThai;
+            break ;
+        case wxFONTENCODING_ISO8859_13 :
+            enc = kCFStringEncodingISOLatin7;
+            break ;
+        case wxFONTENCODING_ISO8859_14 :
+            enc = kCFStringEncodingISOLatin8;
+            break ;
+        case wxFONTENCODING_ISO8859_15 :
+            enc = kCFStringEncodingISOLatin9;
+            break ;
+
+        case wxFONTENCODING_KOI8 :
+            enc = kCFStringEncodingKOI8_R;
+            break ;
+        case wxFONTENCODING_ALTERNATIVE : // MS-DOS CP866
+            enc = kCFStringEncodingDOSRussian;
+            break ;
+
+//      case wxFONTENCODING_BULGARIAN :
+//          enc = ;
+//          break ;
+
+        case wxFONTENCODING_CP437 :
+            enc =kCFStringEncodingDOSLatinUS ;
+            break ;
+        case wxFONTENCODING_CP850 :
+            enc = kCFStringEncodingDOSLatin1;
+            break ;
+        case wxFONTENCODING_CP852 :
+            enc = kCFStringEncodingDOSLatin2;
+            break ;
+        case wxFONTENCODING_CP855 :
+            enc = kCFStringEncodingDOSCyrillic;
+            break ;
+        case wxFONTENCODING_CP866 :
+            enc =kCFStringEncodingDOSRussian ;
+            break ;
+        case wxFONTENCODING_CP874 :
+            enc = kCFStringEncodingDOSThai;
+            break ;
+        case wxFONTENCODING_CP932 :
+            enc = kCFStringEncodingDOSJapanese;
+            break ;
+        case wxFONTENCODING_CP936 :
+            enc =kCFStringEncodingDOSChineseSimplif ;
+            break ;
+        case wxFONTENCODING_CP949 :
+            enc = kCFStringEncodingDOSKorean;
+            break ;
+        case wxFONTENCODING_CP950 :
+            enc = kCFStringEncodingDOSChineseTrad;
+            break ;
+
+        case wxFONTENCODING_CP1250 :
+            enc = kCFStringEncodingWindowsLatin2;
+            break ;
+        case wxFONTENCODING_CP1251 :
+            enc =kCFStringEncodingWindowsCyrillic ;
+            break ;
+        case wxFONTENCODING_CP1252 :
+            enc =kCFStringEncodingWindowsLatin1 ;
+            break ;
+        case wxFONTENCODING_CP1253 :
+            enc = kCFStringEncodingWindowsGreek;
+            break ;
+        case wxFONTENCODING_CP1254 :
+            enc = kCFStringEncodingWindowsLatin5;
+            break ;
+        case wxFONTENCODING_CP1255 :
+            enc =kCFStringEncodingWindowsHebrew ;
+            break ;
+        case wxFONTENCODING_CP1256 :
+            enc =kCFStringEncodingWindowsArabic ;
+            break ;
+        case wxFONTENCODING_CP1257 :
+            enc = kCFStringEncodingWindowsBalticRim;
+            break ;
+        case wxFONTENCODING_UTF7 :
+            enc = kCFStringEncodingNonLossyASCII ;
+            break ;
+        case wxFONTENCODING_UTF8 :
+            enc = kCFStringEncodingUTF8 ;
+            break ;
+        case wxFONTENCODING_EUC_JP :
+            enc = kCFStringEncodingEUC_JP;
+            break ;
+        case wxFONTENCODING_UTF16 :
+            enc = kCFStringEncodingUnicode ;
+            break ;
+        case wxFONTENCODING_MACROMAN :
+            enc = kCFStringEncodingMacRoman ;
+            break ;
+        case wxFONTENCODING_MACJAPANESE :
+            enc = kCFStringEncodingMacJapanese ;
+            break ;
+        case wxFONTENCODING_MACCHINESETRAD :
+            enc = kCFStringEncodingMacChineseTrad ;
+            break ;
+        case wxFONTENCODING_MACKOREAN :
+            enc = kCFStringEncodingMacKorean ;
+            break ;
+        case wxFONTENCODING_MACARABIC :
+            enc = kCFStringEncodingMacArabic ;
+            break ;
+        case wxFONTENCODING_MACHEBREW :
+            enc = kCFStringEncodingMacHebrew ;
+            break ;
+        case wxFONTENCODING_MACGREEK :
+            enc = kCFStringEncodingMacGreek ;
+            break ;
+        case wxFONTENCODING_MACCYRILLIC :
+            enc = kCFStringEncodingMacCyrillic ;
+            break ;
+        case wxFONTENCODING_MACDEVANAGARI :
+            enc = kCFStringEncodingMacDevanagari ;
+            break ;
+        case wxFONTENCODING_MACGURMUKHI :
+            enc = kCFStringEncodingMacGurmukhi ;
+            break ;
+        case wxFONTENCODING_MACGUJARATI :
+            enc = kCFStringEncodingMacGujarati ;
+            break ;
+        case wxFONTENCODING_MACORIYA :
+            enc = kCFStringEncodingMacOriya ;
+            break ;
+        case wxFONTENCODING_MACBENGALI :
+            enc = kCFStringEncodingMacBengali ;
+            break ;
+        case wxFONTENCODING_MACTAMIL :
+            enc = kCFStringEncodingMacTamil ;
+            break ;
+        case wxFONTENCODING_MACTELUGU :
+            enc = kCFStringEncodingMacTelugu ;
+            break ;
+        case wxFONTENCODING_MACKANNADA :
+            enc = kCFStringEncodingMacKannada ;
+            break ;
+        case wxFONTENCODING_MACMALAJALAM :
+            enc = kCFStringEncodingMacMalayalam ;
+            break ;
+        case wxFONTENCODING_MACSINHALESE :
+            enc = kCFStringEncodingMacSinhalese ;
+            break ;
+        case wxFONTENCODING_MACBURMESE :
+            enc = kCFStringEncodingMacBurmese ;
+            break ;
+        case wxFONTENCODING_MACKHMER :
+            enc = kCFStringEncodingMacKhmer ;
+            break ;
+        case wxFONTENCODING_MACTHAI :
+            enc = kCFStringEncodingMacThai ;
+            break ;
+        case wxFONTENCODING_MACLAOTIAN :
+            enc = kCFStringEncodingMacLaotian ;
+            break ;
+        case wxFONTENCODING_MACGEORGIAN :
+            enc = kCFStringEncodingMacGeorgian ;
+            break ;
+        case wxFONTENCODING_MACARMENIAN :
+            enc = kCFStringEncodingMacArmenian ;
+            break ;
+        case wxFONTENCODING_MACCHINESESIMP :
+            enc = kCFStringEncodingMacChineseSimp ;
+            break ;
+        case wxFONTENCODING_MACTIBETAN :
+            enc = kCFStringEncodingMacTibetan ;
+            break ;
+        case wxFONTENCODING_MACMONGOLIAN :
+            enc = kCFStringEncodingMacMongolian ;
+            break ;
+        case wxFONTENCODING_MACETHIOPIC :
+            enc = kCFStringEncodingMacEthiopic ;
+            break ;
+        case wxFONTENCODING_MACCENTRALEUR :
+            enc = kCFStringEncodingMacCentralEurRoman ;
+            break ;
+        case wxFONTENCODING_MACVIATNAMESE :
+            enc = kCFStringEncodingMacVietnamese ;
+            break ;
+        case wxFONTENCODING_MACARABICEXT :
+            enc = kCFStringEncodingMacExtArabic ;
+            break ;
+        case wxFONTENCODING_MACSYMBOL :
+            enc = kCFStringEncodingMacSymbol ;
+            break ;
+        case wxFONTENCODING_MACDINGBATS :
+            enc = kCFStringEncodingMacDingbats ;
+            break ;
+        case wxFONTENCODING_MACTURKISH :
+            enc = kCFStringEncodingMacTurkish ;
+            break ;
+        case wxFONTENCODING_MACCROATIAN :
+            enc = kCFStringEncodingMacCroatian ;
+            break ;
+        case wxFONTENCODING_MACICELANDIC :
+            enc = kCFStringEncodingMacIcelandic ;
+            break ;
+        case wxFONTENCODING_MACROMANIAN :
+            enc = kCFStringEncodingMacRomanian ;
+            break ;
+        case wxFONTENCODING_MACCELTIC :
+            enc = kCFStringEncodingMacCeltic ;
+            break ;
+        case wxFONTENCODING_MACGAELIC :
+            enc = kCFStringEncodingMacGaelic ;
+            break ;
+//      case wxFONTENCODING_MACKEYBOARD :
+//          enc = kCFStringEncodingMacKeyboardGlyphs ;
+//          break ;
+        default :
+            // because gcc is picky
+            break ;
+    } ;
+    return enc ;
+}
+
+wxFontEncoding wxFontEncFromCFStringEnc(CFStringEncoding encoding)
+{
+    wxFontEncoding enc = wxFONTENCODING_DEFAULT ;
+
+    switch( encoding)
+    {
+        case kCFStringEncodingISOLatin1  :
+            enc = wxFONTENCODING_ISO8859_1 ;
+            break ;
+        case kCFStringEncodingISOLatin2 :
+            enc = wxFONTENCODING_ISO8859_2;
+            break ;
+        case kCFStringEncodingISOLatin3 :
+            enc = wxFONTENCODING_ISO8859_3 ;
+            break ;
+        case kCFStringEncodingISOLatin4 :
+            enc = wxFONTENCODING_ISO8859_4;
+            break ;
+        case kCFStringEncodingISOLatinCyrillic :
+            enc = wxFONTENCODING_ISO8859_5;
+            break ;
+        case kCFStringEncodingISOLatinArabic :
+            enc = wxFONTENCODING_ISO8859_6;
+            break ;
+        case kCFStringEncodingISOLatinGreek :
+            enc = wxFONTENCODING_ISO8859_7;
+            break ;
+        case kCFStringEncodingISOLatinHebrew :
+            enc = wxFONTENCODING_ISO8859_8;
+            break ;
+        case kCFStringEncodingISOLatin5 :
+            enc = wxFONTENCODING_ISO8859_9;
+            break ;
+        case kCFStringEncodingISOLatin6 :
+            enc = wxFONTENCODING_ISO8859_10;
+            break ;
+        case kCFStringEncodingISOLatin7 :
+            enc = wxFONTENCODING_ISO8859_13;
+            break ;
+        case kCFStringEncodingISOLatin8 :
+            enc = wxFONTENCODING_ISO8859_14;
+            break ;
+        case kCFStringEncodingISOLatin9 :
+            enc =wxFONTENCODING_ISO8859_15 ;
+            break ;
+
+        case kCFStringEncodingKOI8_R :
+            enc = wxFONTENCODING_KOI8;
+            break ;
+
+//      case  :
+//          enc = wxFONTENCODING_BULGARIAN;
+//          break ;
+
+        case kCFStringEncodingDOSLatinUS :
+            enc = wxFONTENCODING_CP437;
+            break ;
+        case kCFStringEncodingDOSLatin1 :
+            enc = wxFONTENCODING_CP850;
+            break ;
+        case kCFStringEncodingDOSLatin2 :
+            enc =wxFONTENCODING_CP852 ;
+            break ;
+        case kCFStringEncodingDOSCyrillic :
+            enc = wxFONTENCODING_CP855;
+            break ;
+        case kCFStringEncodingDOSRussian :
+            enc = wxFONTENCODING_CP866;
+            break ;
+        case kCFStringEncodingDOSThai :
+            enc =wxFONTENCODING_CP874 ;
+            break ;
+        case kCFStringEncodingDOSJapanese :
+            enc = wxFONTENCODING_CP932;
+            break ;
+        case kCFStringEncodingDOSChineseSimplif :
+            enc = wxFONTENCODING_CP936;
+            break ;
+        case kCFStringEncodingDOSKorean :
+            enc = wxFONTENCODING_CP949;
+            break ;
+        case kCFStringEncodingDOSChineseTrad :
+            enc = wxFONTENCODING_CP950;
+            break ;
+
+        case kCFStringEncodingWindowsLatin2 :
+            enc = wxFONTENCODING_CP1250;
+            break ;
+        case kCFStringEncodingWindowsCyrillic :
+            enc = wxFONTENCODING_CP1251;
+            break ;
+        case kCFStringEncodingWindowsLatin1 :
+            enc = wxFONTENCODING_CP1252;
+            break ;
+        case kCFStringEncodingWindowsGreek :
+            enc = wxFONTENCODING_CP1253;
+            break ;
+        case kCFStringEncodingWindowsLatin5 :
+            enc = wxFONTENCODING_CP1254;
+            break ;
+        case kCFStringEncodingWindowsHebrew :
+            enc = wxFONTENCODING_CP1255;
+            break ;
+        case kCFStringEncodingWindowsArabic :
+            enc = wxFONTENCODING_CP1256;
+            break ;
+        case kCFStringEncodingWindowsBalticRim :
+            enc =wxFONTENCODING_CP1257 ;
+            break ;
+        case kCFStringEncodingEUC_JP :
+            enc = wxFONTENCODING_EUC_JP;
+            break ;
+        case kCFStringEncodingUnicode :
+            enc = wxFONTENCODING_UTF16;
+            break;
+        case kCFStringEncodingMacRoman :
+            enc = wxFONTENCODING_MACROMAN ;
+            break ;
+        case kCFStringEncodingMacJapanese :
+            enc = wxFONTENCODING_MACJAPANESE ;
+            break ;
+        case kCFStringEncodingMacChineseTrad :
+            enc = wxFONTENCODING_MACCHINESETRAD ;
+            break ;
+        case kCFStringEncodingMacKorean :
+            enc = wxFONTENCODING_MACKOREAN ;
+            break ;
+        case kCFStringEncodingMacArabic :
+            enc =wxFONTENCODING_MACARABIC ;
+            break ;
+        case kCFStringEncodingMacHebrew :
+            enc = wxFONTENCODING_MACHEBREW ;
+            break ;
+        case kCFStringEncodingMacGreek :
+            enc = wxFONTENCODING_MACGREEK ;
+            break ;
+        case kCFStringEncodingMacCyrillic :
+            enc = wxFONTENCODING_MACCYRILLIC ;
+            break ;
+        case kCFStringEncodingMacDevanagari :
+            enc = wxFONTENCODING_MACDEVANAGARI ;
+            break ;
+        case kCFStringEncodingMacGurmukhi :
+            enc = wxFONTENCODING_MACGURMUKHI ;
+            break ;
+        case kCFStringEncodingMacGujarati :
+            enc = wxFONTENCODING_MACGUJARATI ;
+            break ;
+        case kCFStringEncodingMacOriya :
+            enc =wxFONTENCODING_MACORIYA ;
+            break ;
+        case kCFStringEncodingMacBengali :
+            enc =wxFONTENCODING_MACBENGALI ;
+            break ;
+        case kCFStringEncodingMacTamil :
+            enc = wxFONTENCODING_MACTAMIL ;
+            break ;
+        case kCFStringEncodingMacTelugu :
+            enc = wxFONTENCODING_MACTELUGU ;
+            break ;
+        case kCFStringEncodingMacKannada :
+            enc = wxFONTENCODING_MACKANNADA ;
+            break ;
+        case kCFStringEncodingMacMalayalam :
+            enc = wxFONTENCODING_MACMALAJALAM ;
+            break ;
+        case kCFStringEncodingMacSinhalese :
+            enc = wxFONTENCODING_MACSINHALESE ;
+            break ;
+        case kCFStringEncodingMacBurmese :
+            enc = wxFONTENCODING_MACBURMESE ;
+            break ;
+        case kCFStringEncodingMacKhmer :
+            enc = wxFONTENCODING_MACKHMER ;
+            break ;
+        case kCFStringEncodingMacThai :
+            enc = wxFONTENCODING_MACTHAI ;
+            break ;
+        case kCFStringEncodingMacLaotian :
+            enc = wxFONTENCODING_MACLAOTIAN ;
+            break ;
+        case kCFStringEncodingMacGeorgian :
+            enc = wxFONTENCODING_MACGEORGIAN ;
+            break ;
+        case kCFStringEncodingMacArmenian :
+            enc = wxFONTENCODING_MACARMENIAN ;
+            break ;
+        case kCFStringEncodingMacChineseSimp :
+            enc = wxFONTENCODING_MACCHINESESIMP ;
+            break ;
+        case kCFStringEncodingMacTibetan :
+            enc = wxFONTENCODING_MACTIBETAN ;
+            break ;
+        case kCFStringEncodingMacMongolian :
+            enc = wxFONTENCODING_MACMONGOLIAN ;
+            break ;
+        case kCFStringEncodingMacEthiopic :
+            enc = wxFONTENCODING_MACETHIOPIC ;
+            break ;
+        case kCFStringEncodingMacCentralEurRoman:
+            enc = wxFONTENCODING_MACCENTRALEUR  ;
+            break ;
+        case kCFStringEncodingMacVietnamese:
+            enc = wxFONTENCODING_MACVIATNAMESE  ;
+            break ;
+        case kCFStringEncodingMacExtArabic :
+            enc = wxFONTENCODING_MACARABICEXT ;
+            break ;
+        case kCFStringEncodingMacSymbol :
+            enc = wxFONTENCODING_MACSYMBOL ;
+            break ;
+        case kCFStringEncodingMacDingbats :
+            enc = wxFONTENCODING_MACDINGBATS ;
+            break ;
+        case kCFStringEncodingMacTurkish :
+            enc = wxFONTENCODING_MACTURKISH ;
+            break ;
+        case kCFStringEncodingMacCroatian :
+            enc = wxFONTENCODING_MACCROATIAN ;
+            break ;
+        case kCFStringEncodingMacIcelandic :
+            enc = wxFONTENCODING_MACICELANDIC ;
+            break ;
+        case kCFStringEncodingMacRomanian :
+            enc = wxFONTENCODING_MACROMANIAN ;
+            break ;
+        case kCFStringEncodingMacCeltic :
+            enc = wxFONTENCODING_MACCELTIC ;
+            break ;
+        case kCFStringEncodingMacGaelic :
+            enc = wxFONTENCODING_MACGAELIC ;
+            break ;
+//        case kCFStringEncodingMacKeyboardGlyphs :
+//            enc = wxFONTENCODING_MACKEYBOARD ;
+//            break ;
+    } ;
+    return enc ;
+}

 
 

-class EC_CharSet : public wxCharacterSet
+class wxMBConv_cocoa : public wxMBConv

 {
 public:
 {
 public:

-    // temporarily just use wxEncodingConverter stuff,
-    // so that it works while a better implementation is built
-    EC_CharSet(const wxChar* name) : wxCharacterSet(name),
-                                     enc(wxFONTENCODING_SYSTEM)
+    wxMBConv_cocoa()

     {
     {

-        if (name)
-            enc = wxFontMapper::Get()->CharsetToEncoding(name, FALSE);
+        Init(CFStringGetSystemEncoding()) ;
+    }

 
 

-        m_ok = m2w.Init(enc, wxFONTENCODING_UNICODE) &&
-               w2m.Init(wxFONTENCODING_UNICODE, enc);
+    wxMBConv_cocoa(const wxChar* name)
+    {
+        Init( wxCFStringEncFromFontEnc(wxFontMapper::Get()->CharsetToEncoding(name, false) ) ) ;

     }
 
     }
 

-    size_t MB2WC(wchar_t *buf, const char *psz, size_t WXUNUSED(n))
+    wxMBConv_cocoa(wxFontEncoding encoding)

     {
     {

-        size_t inbuf = strlen(psz);
-        if (buf)
-            m2w.Convert(psz,buf);
-        return inbuf;
+        Init( wxCFStringEncFromFontEnc(encoding) );

     }
 
     }
 

-    size_t WC2MB(char *buf, const wchar_t *psz, size_t WXUNUSED(n))
+    ~wxMBConv_cocoa()

     {
     {

-        const size_t inbuf = wxWcslen(psz);
-        if (buf)
-            w2m.Convert(psz,buf);
+    }

 
 

-        return inbuf;
+    void Init( CFStringEncoding encoding)
+    {
+        m_char_encoding = encoding ;
+        m_unicode_encoding = kCFStringEncodingUnicode;

     }
 
     }
 

-    bool usable() const { return m_ok; }
+    size_t MB2WC(wchar_t * szOut, const char * szUnConv, size_t nOutSize) const
+    {
+        wxASSERT(szUnConv);

 
 

-public:
-    wxFontEncoding enc;
-    wxEncodingConverter m2w, w2m;
+        size_t nBufSize = strlen(szUnConv) + 1;
+        size_t nRealOutSize;

 
 

-    // were we initialized successfully?
-    bool m_ok;
+        UniChar* szUniCharBuffer    = (UniChar*) szOut;
+        wchar_t* szConvBuffer       = szOut;
+
+        if (szConvBuffer == NULL && nOutSize != 0)
+        {
+            szConvBuffer = new wchar_t[nOutSize] ;
+        }
+
+#if SIZEOF_WCHAR_T == 4
+        szUniCharBuffer = new UniChar[nOutSize];
+#endif
+
+        CFDataRef theData = CFDataCreateWithBytesNoCopy (
+                                            NULL,     //allocator
+                                            (const UInt8*)szUnConv,
+                                            nBufSize - 1,
+                                            NULL      //deallocator
+                                            );
+
+        wxASSERT(theData);
+
+        CFStringRef theString = CFStringCreateFromExternalRepresentation (
+                                                NULL,
+                                                theData,
+                                                m_char_encoding
+                                                );
+
+        wxASSERT(theString);
+
+        if (nOutSize == 0)
+        {
+            nRealOutSize = CFStringGetLength(theString) + 1;
+            CFRelease(theString);
+            return nRealOutSize - 1;
+        }
+
+        CFRange theRange = { 0, CFStringGetLength(theString) };
+
+        CFStringGetCharacters(theString, theRange, szUniCharBuffer);
+
+
+        nRealOutSize = (CFStringGetLength(theString) + 1);
+
+        CFRelease(theString);
+
+        szUniCharBuffer[nRealOutSize-1] = '\0' ;
+
+#if SIZEOF_WCHAR_T == 4
+        wxMBConvUTF16 converter ;
+        converter.MB2WC(szConvBuffer  , (const char*)szUniCharBuffer , nRealOutSize ) ;
+        delete[] szUniCharBuffer;
+#endif
+        if ( szOut == NULL )
+            delete [] szConvBuffer;
+
+        return nRealOutSize ;
+    }
+
+    size_t WC2MB(char *szOut, const wchar_t *szUnConv, size_t nOutSize) const
+    {
+        size_t nBufSize = wxWcslen(szUnConv) + 1;
+        size_t nRealOutSize;
+        char* szBuffer = szOut;
+        UniChar* szUniBuffer = (UniChar*) szUnConv;
+
+        if (szOut == NULL)
+        {
+            // worst case
+            nRealOutSize = wxString::WorstEncodingCase(nBufSize - 1, *this)+1 ;
+            szBuffer = new char[ nRealOutSize ] ;
+        }
+        else
+            nRealOutSize = nOutSize;
+
+#if SIZEOF_WCHAR_T == 4
+        wxMBConvUTF16BE converter ;
+        nBufSize = converter.WC2MB( NULL , szUnConv , 0 );
+        szUniBuffer = new UniChar[ (nBufSize / sizeof(UniChar)) + 1] ;
+        converter.WC2MB( (char*) szUniBuffer , szUnConv, nBufSize + sizeof(UniChar)) ;
+        nBufSize /= sizeof(UniChar);
+        ++nBufSize;
+#endif
+
+        CFStringRef theString = CFStringCreateWithCharactersNoCopy(
+                                NULL, //allocator
+                                szUniBuffer,
+                                nBufSize,
+                                NULL //deallocator
+                            );
+
+        wxASSERT(theString);
+
+        //Note that CER puts a BOM when converting to unicode
+        //so we may want to check and use getchars instead in that case
+        CFDataRef theData = CFStringCreateExternalRepresentation(
+                                NULL, //allocator
+                                theString,
+                                m_char_encoding,
+                                0 //what to put in characters that can't be converted -
+                                    //0 tells CFString to return NULL if it meets such a character
+                        );
+
+        if(!theData)
+            return (size_t)-1;
+
+        CFRelease(theString);
+
+        nRealOutSize = CFDataGetLength(theData);
+
+        if ( szOut == NULL )
+            delete[] szBuffer;
+
+        if(nOutSize == 0)
+        {
+//TODO: This gets flagged as a non-malloced address by the debugger...
+//#if SIZEOF_WCHAR_T == 4
+//        delete[] szUniBuffer;
+//#endif
+            CFRelease(theData);
+            return nRealOutSize - 1;
+        }
+
+        CFRange theRange = {0, CFDataGetLength(theData) };
+        CFDataGetBytes(theData, theRange, (UInt8*) szBuffer);
+
+        CFRelease(theData);
+
+//TODO: This gets flagged as a non-malloced address by the debugger...
+//#if SIZEOF_WCHAR_T == 4
+//        delete[] szUniBuffer;
+//#endif
+        return  nRealOutSize - 1;
+    }
+
+    bool IsOk() const
+    {
+        //TODO: check for invalid en/de/coding
+        return true;
+    }
+
+private:
+    CFStringEncoding m_char_encoding ;
+    CFStringEncoding m_unicode_encoding ;

 };
 
 };
 

-#endif // wxUSE_FONTMAP
+#endif // defined(__WXCOCOA__)

 
 

-// ----------------------------------------------------------------------------
-// the function creating the wxCharacterSet for the specified charset on the
-// current system, trying all possibilities
-// ----------------------------------------------------------------------------
+// ============================================================================
+// Mac conversion classes
+// ============================================================================
+
+#if defined(__WXMAC__) && defined(TARGET_CARBON)

 
 

-static wxCharacterSet *wxGetCharacterSet(const wxChar *name)
+class wxMBConv_mac : public wxMBConv

 {
 {

-    // check for the special case of ASCII charset
-#if wxUSE_FONTMAP
-    if ( wxFontMapper::Get()->CharsetToEncoding(name) == wxFONTENCODING_DEFAULT )
-#else // wxUSE_FONTMAP
-    if ( !name )
-#endif // wxUSE_FONTMAP/!wxUSE_FONTMAP
+public:
+    wxMBConv_mac()

     {
     {

-        // don't convert at all
-        return NULL;
+        Init(CFStringGetSystemEncoding()) ;

     }
 
     }
 

-    // the test above must have taken care of this case
-    wxCHECK_MSG( name, NULL, _T("NULL name must be wxFONTENCODING_DEFAULT") );
+    wxMBConv_mac(const wxChar* name)
+    {
+        Init( wxMacGetSystemEncFromFontEnc(wxFontMapper::Get()->CharsetToEncoding(name, false) ) ) ;
+    }

 
 

-    wxCharacterSet *cset;
+    wxMBConv_mac(wxFontEncoding encoding)
+    {
+        Init( wxMacGetSystemEncFromFontEnc(encoding) );
+    }

 
 

-    if ( wxStricmp(name, wxT("UTF8")) == 0 || wxStricmp(name, wxT("UTF-8")) == 0)
+    ~wxMBConv_mac()

     {
     {

-        cset = new ID_CharSet(name, &wxConvUTF8);
+        OSStatus status = noErr ;
+        status = TECDisposeConverter(m_MB2WC_converter);
+        status = TECDisposeConverter(m_WC2MB_converter);

     }
     }

-    else
+
+
+    void Init( TextEncodingBase encoding)

     {
     {

-#ifdef HAVE_ICONV
-        cset = new IC_CharSet(name);
-#else // !HAVE_ICONV
-        cset = NULL;
-#endif // HAVE_ICONV/!HAVE_ICONV
+        OSStatus status = noErr ;
+        m_char_encoding = encoding ;
+        m_unicode_encoding = CreateTextEncoding(kTextEncodingUnicodeDefault,0,kUnicode16BitFormat) ;
+
+        status = TECCreateConverter(&m_MB2WC_converter,
+                                    m_char_encoding,
+                                    m_unicode_encoding);
+        status = TECCreateConverter(&m_WC2MB_converter,
+                                    m_unicode_encoding,
+                                    m_char_encoding);

     }
 
     }
 

-    // it can only be NULL in this case
-#ifndef HAVE_ICONV
-    if ( cset )
-#endif // !HAVE_ICONV
+    size_t MB2WC(wchar_t *buf, const char *psz, size_t n) const

     {
     {

-        if ( cset->usable() )
-            return cset;
+        OSStatus status = noErr ;
+        ByteCount byteOutLen ;
+        ByteCount byteInLen = strlen(psz) ;
+        wchar_t *tbuf = NULL ;
+        UniChar* ubuf = NULL ;
+        size_t res = 0 ;
+
+        if (buf == NULL)
+        {
+            n = byteInLen ;
+            tbuf = (wchar_t*) malloc( n * SIZEOF_WCHAR_T) ;
+        }
+        ByteCount byteBufferLen = n * sizeof( UniChar ) ;
+#if SIZEOF_WCHAR_T == 4
+        ubuf = (UniChar*) malloc( byteBufferLen + 2 ) ;
+#else
+        ubuf = (UniChar*) (buf ? buf : tbuf) ;
+#endif
+        status = TECConvertText(m_MB2WC_converter, (ConstTextPtr) psz , byteInLen, &byteInLen,
+          (TextPtr) ubuf , byteBufferLen, &byteOutLen);
+#if SIZEOF_WCHAR_T == 4
+        // we have to terminate here, because n might be larger for the trailing zero, and if UniChar
+        // is not properly terminated we get random characters at the end
+        ubuf[byteOutLen / sizeof( UniChar ) ] = 0 ;
+        wxMBConvUTF16BE converter ;
+        res = converter.MB2WC( (buf ? buf : tbuf) , (const char*)ubuf , n ) ;
+        free( ubuf ) ;
+#else
+        res = byteOutLen / sizeof( UniChar ) ;
+#endif
+        if ( buf == NULL )
+             free(tbuf) ;

 
 

-        delete cset;
-        cset = NULL;
+        if ( buf  && res < n)
+            buf[res] = 0;
+
+        return res ;

     }
 
     }
 

-#if defined(__WIN32__) && !defined(__WXMICROWIN__)
-    cset = new CP_CharSet(name);
-    if ( cset->usable() )
-        return cset;
+    size_t WC2MB(char *buf, const wchar_t *psz, size_t n) const
+    {
+        OSStatus status = noErr ;
+        ByteCount byteOutLen ;
+        ByteCount byteInLen = wxWcslen(psz) * SIZEOF_WCHAR_T ;
+
+        char *tbuf = NULL ;
+
+        if (buf == NULL)
+        {
+            // worst case
+            n = wxString::WorstEncodingCase(byteInLen / SIZEOF_WCHAR_T, *this) + SIZEOF_WCHAR_T;
+            tbuf = (char*) malloc( n ) ;
+        }
+
+        ByteCount byteBufferLen = n ;
+        UniChar* ubuf = NULL ;
+#if SIZEOF_WCHAR_T == 4
+        wxMBConvUTF16BE converter ;
+        size_t unicharlen = converter.WC2MB( NULL , psz , 0 ) ;
+        byteInLen = unicharlen ;
+        ubuf = (UniChar*) malloc( byteInLen + 2 ) ;
+        converter.WC2MB( (char*) ubuf , psz, unicharlen + 2 ) ;
+#else
+        ubuf = (UniChar*) psz ;
+#endif
+        status = TECConvertText(m_WC2MB_converter, (ConstTextPtr) ubuf , byteInLen, &byteInLen,
+            (TextPtr) (buf ? buf : tbuf) , byteBufferLen, &byteOutLen);
+#if SIZEOF_WCHAR_T == 4
+        free( ubuf ) ;
+#endif
+        if ( buf == NULL )
+            free(tbuf) ;
+
+        size_t res = byteOutLen ;
+        if ( buf  && res < n)
+            buf[res] = 0;
+
+        return res ;
+    }

 
 

-    delete cset;
-    cset = NULL;
-#endif // __WIN32__
+    bool IsOk() const
+        { return m_MB2WC_converter !=  NULL && m_WC2MB_converter != NULL  ; }
+
+private:
+    TECObjectRef m_MB2WC_converter ;
+    TECObjectRef m_WC2MB_converter ;
+
+    TextEncodingBase m_char_encoding ;
+    TextEncodingBase m_unicode_encoding ;
+};
+
+#endif // defined(__WXMAC__) && defined(TARGET_CARBON)
+
+// ============================================================================
+// wxEncodingConverter based conversion classes
+// ============================================================================

 
 #if wxUSE_FONTMAP
 
 #if wxUSE_FONTMAP

-    cset = new EC_CharSet(name);
-    if ( cset->usable() )
-        return cset;

 
 

-    delete cset;
-    cset = NULL;
-#endif // wxUSE_FONTMAP
+class wxMBConv_wxwin : public wxMBConv
+{
+private:
+    void Init()
+    {
+        m_ok = m2w.Init(m_enc, wxFONTENCODING_UNICODE) &&
+               w2m.Init(wxFONTENCODING_UNICODE, m_enc);
+    }
+
+public:
+    // temporarily just use wxEncodingConverter stuff,
+    // so that it works while a better implementation is built
+    wxMBConv_wxwin(const wxChar* name)
+    {
+        if (name)
+            m_enc = wxFontMapper::Get()->CharsetToEncoding(name, false);
+        else
+            m_enc = wxFONTENCODING_SYSTEM;

 
 

-    wxLogError(_("Cannot convert from encoding '%s'!"), name);
+        Init();
+    }

 
 

-    return NULL;
-}
+    wxMBConv_wxwin(wxFontEncoding enc)
+    {
+        m_enc = enc;
+
+        Init();
+    }
+
+    size_t MB2WC(wchar_t *buf, const char *psz, size_t WXUNUSED(n)) const
+    {
+        size_t inbuf = strlen(psz);
+        if (buf)
+            m2w.Convert(psz,buf);
+        return inbuf;
+    }
+
+    size_t WC2MB(char *buf, const wchar_t *psz, size_t WXUNUSED(n)) const
+    {
+        const size_t inbuf = wxWcslen(psz);
+        if (buf)
+            w2m.Convert(psz,buf);
+
+        return inbuf;
+    }
+
+    bool IsOk() const { return m_ok; }
+
+public:
+    wxFontEncoding m_enc;
+    wxEncodingConverter m2w, w2m;
+
+    // were we initialized successfully?
+    bool m_ok;
+
+    DECLARE_NO_COPY_CLASS(wxMBConv_wxwin)
+};
+
+#endif // wxUSE_FONTMAP

 
 // ============================================================================
 // wxCSConv implementation
 // ============================================================================
 
 
 // ============================================================================
 // wxCSConv implementation
 // ============================================================================
 

+void wxCSConv::Init()
+{
+    m_name = NULL;
+    m_convReal =  NULL;
+    m_deferred = true;
+}
+

 wxCSConv::wxCSConv(const wxChar *charset)
 {
 wxCSConv::wxCSConv(const wxChar *charset)
 {

-    m_name = (wxChar *)NULL;
-    m_cset = (wxCharacterSet *) NULL;
-    m_deferred = TRUE;
+    Init();
+
+    if ( charset )
+    {
+        SetName(charset);
+    }
+
+    m_encoding = wxFONTENCODING_SYSTEM;
+}
+
+wxCSConv::wxCSConv(wxFontEncoding encoding)
+{
+    if ( encoding == wxFONTENCODING_MAX || encoding == wxFONTENCODING_DEFAULT )
+    {
+        wxFAIL_MSG( _T("invalid encoding value in wxCSConv ctor") );

 
 

-    SetName(charset);
+        encoding = wxFONTENCODING_SYSTEM;
+    }
+
+    Init();
+
+    m_encoding = encoding;

 }
 
 wxCSConv::~wxCSConv()
 }
 
 wxCSConv::~wxCSConv()


@@ -984,27 +2409,31 @@ wxCSConv::~wxCSConv()
 }
 
 wxCSConv::wxCSConv(const wxCSConv& conv)
 }
 
 wxCSConv::wxCSConv(const wxCSConv& conv)

-    : wxMBConv()
+        : wxMBConv()

 {
 {

-    Clear();
+    Init();
+

     SetName(conv.m_name);
     SetName(conv.m_name);

+    m_encoding = conv.m_encoding;

 }
 
 wxCSConv& wxCSConv::operator=(const wxCSConv& conv)
 {
     Clear();
 }
 
 wxCSConv& wxCSConv::operator=(const wxCSConv& conv)
 {
     Clear();

+

     SetName(conv.m_name);
     SetName(conv.m_name);

+    m_encoding = conv.m_encoding;
+

     return *this;
 }
 
 void wxCSConv::Clear()
 {
     return *this;
 }
 
 void wxCSConv::Clear()
 {

-    if (m_name)
-        free(m_name);
-    if (m_cset)
-        delete m_cset;
+    free(m_name);
+    delete m_convReal;
+

     m_name = NULL;
     m_name = NULL;

-    m_cset = NULL;
+    m_convReal = NULL;

 }
 
 void wxCSConv::SetName(const wxChar *charset)
 }
 
 void wxCSConv::SetName(const wxChar *charset)


@@ -1012,35 +2441,193 @@ void wxCSConv::SetName(const wxChar *charset)
     if (charset)
     {
         m_name = wxStrdup(charset);
     if (charset)
     {
         m_name = wxStrdup(charset);

-        m_deferred = TRUE;
+        m_deferred = true;

     }
 }
 
     }
 }
 

-void wxCSConv::LoadNow()
+wxMBConv *wxCSConv::DoCreate() const

 {
 {

-    if (m_deferred)
+    // check for the special case of ASCII or ISO8859-1 charset: as we have
+    // special knowledge of it anyhow, we don't need to create a special
+    // conversion object
+    if ( m_encoding == wxFONTENCODING_ISO8859_1 )
+    {
+        // don't convert at all
+        return NULL;
+    }
+
+    // we trust OS to do conversion better than we can so try external
+    // conversion methods first
+    //
+    // the full order is:
+    //      1. OS conversion (iconv() under Unix or Win32 API)
+    //      2. hard coded conversions for UTF
+    //      3. wxEncodingConverter as fall back
+
+    // step (1)
+#ifdef HAVE_ICONV
+#if !wxUSE_FONTMAP
+    if ( m_name )
+#endif // !wxUSE_FONTMAP

     {
     {

-        if ( !m_name )
+        wxString name(m_name);
+
+#if wxUSE_FONTMAP
+        if ( name.empty() )
+            name = wxFontMapper::Get()->GetEncodingName(m_encoding);
+#endif // wxUSE_FONTMAP
+
+        wxMBConv_iconv *conv = new wxMBConv_iconv(name);
+        if ( conv->IsOk() )
+            return conv;
+
+        delete conv;
+    }
+#endif // HAVE_ICONV
+
+#ifdef wxHAVE_WIN32_MB2WC
+    {
+#if wxUSE_FONTMAP
+        wxMBConv_win32 *conv = m_name ? new wxMBConv_win32(m_name)
+                                      : new wxMBConv_win32(m_encoding);
+        if ( conv->IsOk() )
+            return conv;
+
+        delete conv;
+#else
+        return NULL;
+#endif
+    }
+#endif // wxHAVE_WIN32_MB2WC
+#if defined(__WXMAC__)
+    {
+        if ( m_name || ( m_encoding < wxFONTENCODING_UTF16BE ) )

         {
         {

-            wxString name = wxLocale::GetSystemEncodingName();
-            if ( !name.empty() )
-            {
-                SetName(name);
-            }
+
+            wxMBConv_mac *conv = m_name ? new wxMBConv_mac(m_name)
+                                        : new wxMBConv_mac(m_encoding);
+            if ( conv->IsOk() )
+                 return conv;
+
+            delete conv;
+        }
+    }
+#endif
+#if defined(__WXCOCOA__)
+    {
+        if ( m_name || ( m_encoding <= wxFONTENCODING_UTF16 ) )
+        {
+
+            wxMBConv_cocoa *conv = m_name ? new wxMBConv_cocoa(m_name)
+                                          : new wxMBConv_cocoa(m_encoding);
+            if ( conv->IsOk() )
+                 return conv;
+
+            delete conv;

         }
         }

+    }
+#endif
+    // step (2)
+    wxFontEncoding enc = m_encoding;
+#if wxUSE_FONTMAP
+    if ( enc == wxFONTENCODING_SYSTEM && m_name )
+    {
+        // use "false" to suppress interactive dialogs -- we can be called from
+        // anywhere and popping up a dialog from here is the last thing we want to
+        // do
+        enc = wxFontMapper::Get()->CharsetToEncoding(m_name, false);
+    }
+#endif // wxUSE_FONTMAP

 
 

-        // wxGetCharacterSet() complains about NULL name
-        m_cset = m_name ? wxGetCharacterSet(m_name) : NULL;
-        m_deferred = FALSE;
+    switch ( enc )
+    {
+        case wxFONTENCODING_UTF7:
+             return new wxMBConvUTF7;
+
+        case wxFONTENCODING_UTF8:
+             return new wxMBConvUTF8;
+
+        case wxFONTENCODING_UTF16BE:
+             return new wxMBConvUTF16BE;
+
+        case wxFONTENCODING_UTF16LE:
+             return new wxMBConvUTF16LE;
+
+        case wxFONTENCODING_UTF32BE:
+             return new wxMBConvUTF32BE;
+
+        case wxFONTENCODING_UTF32LE:
+             return new wxMBConvUTF32LE;
+
+        default:
+             // nothing to do but put here to suppress gcc warnings
+             ;
+    }
+
+    // step (3)
+#if wxUSE_FONTMAP
+    {
+        wxMBConv_wxwin *conv = m_name ? new wxMBConv_wxwin(m_name)
+                                      : new wxMBConv_wxwin(m_encoding);
+        if ( conv->IsOk() )
+            return conv;
+
+        delete conv;
+    }
+#endif // wxUSE_FONTMAP
+
+    // NB: This is a hack to prevent deadlock. What could otherwise happen
+    //     in Unicode build: wxConvLocal creation ends up being here
+    //     because of some failure and logs the error. But wxLog will try to
+    //     attach timestamp, for which it will need wxConvLocal (to convert
+    //     time to char* and then wchar_t*), but that fails, tries to log
+    //     error, but wxLog has a (already locked) critical section that
+    //     guards static buffer.
+    static bool alreadyLoggingError = false;
+    if (!alreadyLoggingError)
+    {
+        alreadyLoggingError = true;
+        wxLogError(_("Cannot convert from the charset '%s'!"),
+                   m_name ? m_name
+                      :
+#if wxUSE_FONTMAP
+                         wxFontMapper::GetEncodingDescription(m_encoding).c_str()
+#else // !wxUSE_FONTMAP
+                         wxString::Format(_("encoding %s"), m_encoding).c_str()
+#endif // wxUSE_FONTMAP/!wxUSE_FONTMAP
+              );
+        alreadyLoggingError = false;
+    }
+
+    return NULL;
+}
+
+void wxCSConv::CreateConvIfNeeded() const
+{
+    if ( m_deferred )
+    {
+        wxCSConv *self = (wxCSConv *)this; // const_cast
+
+#if wxUSE_INTL
+        // if we don't have neither the name nor the encoding, use the default
+        // encoding for this system
+        if ( !m_name && m_encoding == wxFONTENCODING_SYSTEM )
+        {
+            self->m_name = wxStrdup(wxLocale::GetSystemEncodingName());
+        }
+#endif // wxUSE_INTL
+
+        self->m_convReal = DoCreate();
+        self->m_deferred = false;

     }
 }
 
 size_t wxCSConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
 {
     }
 }
 
 size_t wxCSConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
 {

-    ((wxCSConv *)this)->LoadNow(); // discard constness
+    CreateConvIfNeeded();

 
 

-    if (m_cset)
-        return m_cset->MB2WC(buf, psz, n);
+    if (m_convReal)
+        return m_convReal->MB2WC(buf, psz, n);

 
     // latin-1 (direct)
     size_t len = strlen(psz);
 
     // latin-1 (direct)
     size_t len = strlen(psz);


@@ -1056,22 +2643,67 @@ size_t wxCSConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
 
 size_t wxCSConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
 {
 
 size_t wxCSConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
 {

-    ((wxCSConv *)this)->LoadNow(); // discard constness
+    CreateConvIfNeeded();

 
 

-    if (m_cset)
-        return m_cset->WC2MB(buf, psz, n);
+    if (m_convReal)
+        return m_convReal->WC2MB(buf, psz, n);

 
     // latin-1 (direct)
     const size_t len = wxWcslen(psz);
     if (buf)
     {
         for (size_t c = 0; c <= len; c++)
 
     // latin-1 (direct)
     const size_t len = wxWcslen(psz);
     if (buf)
     {
         for (size_t c = 0; c <= len; c++)

-            buf[c] = (psz[c] > 0xff) ? '?' : psz[c];
+        {
+            if (psz[c] > 0xFF)
+                return (size_t)-1;
+            buf[c] = (char)psz[c];
+        }
+    }
+    else
+    {
+        for (size_t c = 0; c <= len; c++)
+        {
+            if (psz[c] > 0xFF)
+                return (size_t)-1;
+        }

     }
 
     return len;
 }
 
     }
 
     return len;
 }
 

-#endif // wxUSE_WCHAR_T
+// ----------------------------------------------------------------------------
+// globals
+// ----------------------------------------------------------------------------
+
+#ifdef __WINDOWS__
+    static wxMBConv_win32 wxConvLibcObj;
+#elif defined(__WXMAC__) && !defined(__MACH__)
+    static wxMBConv_mac wxConvLibcObj ;
+#else
+    static wxMBConvLibc wxConvLibcObj;
+#endif
+
+static wxCSConv wxConvLocalObj(wxFONTENCODING_SYSTEM);
+static wxCSConv wxConvISO8859_1Obj(wxFONTENCODING_ISO8859_1);
+static wxMBConvUTF7 wxConvUTF7Obj;
+static wxMBConvUTF8 wxConvUTF8Obj;
+
+
+WXDLLIMPEXP_DATA_BASE(wxMBConv&) wxConvLibc = wxConvLibcObj;
+WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvLocal = wxConvLocalObj;
+WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvISO8859_1 = wxConvISO8859_1Obj;
+WXDLLIMPEXP_DATA_BASE(wxMBConvUTF7&) wxConvUTF7 = wxConvUTF7Obj;
+WXDLLIMPEXP_DATA_BASE(wxMBConvUTF8&) wxConvUTF8 = wxConvUTF8Obj;
+WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent = &wxConvLibcObj;
+
+#else // !wxUSE_WCHAR_T
+
+// stand-ins in absence of wchar_t
+WXDLLIMPEXP_DATA_BASE(wxMBConv) wxConvLibc,
+                                wxConvISO8859_1,
+                                wxConvLocal,
+                                wxConvUTF8;
+
+#endif // wxUSE_WCHAR_T/!wxUSE_WCHAR_T