fixing iterator comparison

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

diff --git a/src/common/strconv.cpp b/src/common/strconv.cpp
index 7f2c37a398868787a1db31016cd8578ee5ce2f44..bf0b0923c76c3c152d9cbea9016054fca0007da0 100644 (file)
--- a/src/common/strconv.cpp
+++ b/src/common/strconv.cpp
@@ -28,8 +28,6 @@
 
 #include "wx/strconv.h"
 
 
 #include "wx/strconv.h"
 

-#if wxUSE_WCHAR_T
-

 #ifndef __WXWINCE__
 #include <errno.h>
 #endif
 #ifndef __WXWINCE__
 #include <errno.h>
 #endif


@@ -53,11 +51,11 @@
 #include "wx/fontmap.h"
 
 #ifdef __DARWIN__
 #include "wx/fontmap.h"
 
 #ifdef __DARWIN__

-#include "wx/mac/corefoundation/private/strconv_cf.h"
+#include "wx/osx/core/private/strconv_cf.h"

 #endif //def __DARWIN__
 
 
 #endif //def __DARWIN__
 
 

-#define TRACE_STRCONV _T("strconv")
+#define TRACE_STRCONV wxT("strconv")

 
 // WC_UTF16 is defined only if sizeof(wchar_t) == 2, otherwise it's supposed to
 // be 4 bytes
 
 // WC_UTF16 is defined only if sizeof(wchar_t) == 2, otherwise it's supposed to
 // be 4 bytes


@@ -142,7 +140,7 @@ static wxUint32 wxDecodeSurrogate(const wxDecodeSurrogate_t **pSrc)
 {
     wxUint32 out;
     const size_t
 {
     wxUint32 out;
     const size_t

-        n = decode_utf16(wx_reinterpret_cast(const wxUint16 *, *pSrc), out);
+        n = decode_utf16(reinterpret_cast<const wxUint16 *>(*pSrc), out);

     if ( n == wxCONV_FAILED )
         *pSrc = NULL;
     else
     if ( n == wxCONV_FAILED )
         *pSrc = NULL;
     else


@@ -160,11 +158,15 @@ wxMBConv::ToWChar(wchar_t *dst, size_t dstLen,
                   const char *src, size_t srcLen) const
 {
     // although new conversion classes are supposed to implement this function
                   const char *src, size_t srcLen) const
 {
     // although new conversion classes are supposed to implement this function

-    // directly, the existins ones only implement the old MB2WC() and so, to
+    // directly, the existing ones only implement the old MB2WC() and so, to

     // avoid to have to rewrite all conversion classes at once, we provide a
     // default (but not efficient) implementation of this one in terms of the
     // old function by copying the input to ensure that it's NUL-terminated and
     // then using MB2WC() to convert it
     // avoid to have to rewrite all conversion classes at once, we provide a
     // default (but not efficient) implementation of this one in terms of the
     // old function by copying the input to ensure that it's NUL-terminated and
     // then using MB2WC() to convert it

+    //
+    // moreover, some conversion classes simply can't implement ToWChar()
+    // directly, the primary example is wxConvLibc: mbstowcs() only handles
+    // NUL-terminated strings

 
     // the number of chars [which would be] written to dst [if it were not NULL]
     size_t dstWritten = 0;
 
     // the number of chars [which would be] written to dst [if it were not NULL]
     size_t dstWritten = 0;


@@ -205,6 +207,21 @@ wxMBConv::ToWChar(wchar_t *dst, size_t dstLen,
         srcEnd = NULL;
     }
 
         srcEnd = NULL;
     }
 

+    // the idea of this code is straightforward: it converts a NUL-terminated
+    // chunk of the string during each iteration and updates the output buffer
+    // with the result
+    //
+    // all the complication come from the fact that this function, for
+    // historical reasons, must behave in 2 subtly different ways when it's
+    // called with a fixed number of characters and when it's called for the
+    // entire NUL-terminated string: in the former case (srcEnd != NULL) we
+    // must count all characters we convert, NUL or not; but in the latter we
+    // do not count the trailing NUL -- but still count all the NULs inside the
+    // string
+    //
+    // so for the (simple) former case we just always count the trailing NUL,
+    // but for the latter we need to wait until we see if there is going to be
+    // another loop iteration and only count it then

     for ( ;; )
     {
         // try to convert the current chunk
     for ( ;; )
     {
         // try to convert the current chunk


@@ -212,11 +229,11 @@ wxMBConv::ToWChar(wchar_t *dst, size_t dstLen,
         if ( lenChunk == wxCONV_FAILED )
             return wxCONV_FAILED;
 
         if ( lenChunk == wxCONV_FAILED )
             return wxCONV_FAILED;
 

-        lenChunk++; // for the L'\0' at the end of this chunk
-

         dstWritten += lenChunk;
         dstWritten += lenChunk;

+        if ( !srcEnd )
+            dstWritten++;

 
 

-        if ( lenChunk == 1 )
+        if ( !lenChunk )

         {
             // nothing left in the input string, conversion succeeded
             break;
         {
             // nothing left in the input string, conversion succeeded
             break;


@@ -227,20 +244,25 @@ wxMBConv::ToWChar(wchar_t *dst, size_t dstLen,
             if ( dstWritten > dstLen )
                 return wxCONV_FAILED;
 
             if ( dstWritten > dstLen )
                 return wxCONV_FAILED;
 

-            if ( MB2WC(dst, src, lenChunk) == wxCONV_FAILED )
+            // +1 is for trailing NUL
+            if ( MB2WC(dst, src, lenChunk + 1) == wxCONV_FAILED )

                 return wxCONV_FAILED;
 
             dst += lenChunk;
                 return wxCONV_FAILED;
 
             dst += lenChunk;

+            if ( !srcEnd )
+                dst++;

         }
 
         if ( !srcEnd )
         {
             // we convert just one chunk in this case as this is the entire
         }
 
         if ( !srcEnd )
         {
             // we convert just one chunk in this case as this is the entire

-            // string anyhow
+            // string anyhow (and we don't count the trailing NUL in this case)

             break;
         }
 
             break;
         }
 

-        // advance the input pointer past the end of this chunk
+        // advance the input pointer past the end of this chunk: notice that we
+        // will always stop before srcEnd because we know that the chunk is
+        // always properly NUL-terminated

         while ( NotAllNULs(src, nulLen) )
         {
             // notice that we must skip over multiple bytes here as we suppose
         while ( NotAllNULs(src, nulLen) )
         {
             // notice that we must skip over multiple bytes here as we suppose


@@ -250,11 +272,18 @@ wxMBConv::ToWChar(wchar_t *dst, size_t dstLen,
             src += nulLen;
         }
 
             src += nulLen;
         }
 

-        src += nulLen; // skipping over its terminator as well
+        // if the buffer ends before this NUL, we shouldn't count it in our
+        // output so skip the code below
+        if ( src == srcEnd )
+            break;
+
+        // do count this terminator as it's inside the buffer we convert
+        dstWritten++;
+        if ( dst )
+            dst++;
+
+        src += nulLen; // skip the terminator itself

 
 

-        // note that ">=" (and not just "==") is needed here as the terminator
-        // we skipped just above could be inside or just after the buffer
-        // delimited by inEnd

         if ( src >= srcEnd )
             break;
     }
         if ( src >= srcEnd )
             break;
     }


@@ -269,13 +298,15 @@ wxMBConv::FromWChar(char *dst, size_t dstLen,
     // the number of chars [which would be] written to dst [if it were not NULL]
     size_t dstWritten = 0;
 
     // the number of chars [which would be] written to dst [if it were not NULL]
     size_t dstWritten = 0;
 

+    // if we don't know its length we have no choice but to assume that it is
+    // NUL-terminated (notice that it can still be NUL-terminated even if
+    // explicit length is given but it doesn't change our return value)
+    const bool isNulTerminated = srcLen == wxNO_LEN;
+

     // make a copy of the input string unless it is already properly
     // NUL-terminated
     // make a copy of the input string unless it is already properly
     // NUL-terminated

-    //
-    // if we don't know its length we have no choice but to assume that it is,
-    // indeed, properly terminated

     wxWCharBuffer bufTmp;
     wxWCharBuffer bufTmp;

-    if ( srcLen == wxNO_LEN )
+    if ( isNulTerminated )

     {
         srcLen = wxWcslen(src) + 1;
     }
     {
         srcLen = wxWcslen(src) + 1;
     }


@@ -290,27 +321,65 @@ wxMBConv::FromWChar(char *dst, size_t dstLen,
     const size_t lenNul = GetMBNulLen();
     for ( const wchar_t * const srcEnd = src + srcLen;
           src < srcEnd;
     const size_t lenNul = GetMBNulLen();
     for ( const wchar_t * const srcEnd = src + srcLen;
           src < srcEnd;

-          src += wxWcslen(src) + 1 /* skip L'\0' too */ )
+          src++ /* skip L'\0' too */ )

     {
         // try to convert the current chunk
         size_t lenChunk = WC2MB(NULL, src, 0);
     {
         // try to convert the current chunk
         size_t lenChunk = WC2MB(NULL, src, 0);

-

         if ( lenChunk == wxCONV_FAILED )
             return wxCONV_FAILED;
 
         if ( lenChunk == wxCONV_FAILED )
             return wxCONV_FAILED;
 

-        lenChunk += lenNul;

         dstWritten += lenChunk;
 
         dstWritten += lenChunk;
 

+        const wchar_t * const
+            chunkEnd = isNulTerminated ? srcEnd - 1 : src + wxWcslen(src);
+
+        // our return value accounts for the trailing NUL(s), unlike that of
+        // WC2MB(), however don't do it for the last NUL we artificially added
+        // ourselves above
+        if ( chunkEnd < srcEnd )
+            dstWritten += lenNul;
+

         if ( dst )
         {
             if ( dstWritten > dstLen )
                 return wxCONV_FAILED;
 
         if ( dst )
         {
             if ( dstWritten > dstLen )
                 return wxCONV_FAILED;
 

-            if ( WC2MB(dst, src, lenChunk) == wxCONV_FAILED )
+            // if we know that there is enough space in the destination buffer
+            // (because we accounted for lenNul in dstWritten above), we can
+            // convert directly in place -- but otherwise we need another
+            // temporary buffer to ensure that we don't overwrite the output
+            wxCharBuffer dstBuf;
+            char *dstTmp;
+            if ( chunkEnd == srcEnd )
+            {
+                dstBuf = wxCharBuffer(lenChunk + lenNul - 1);
+                dstTmp = dstBuf.data();
+            }
+            else
+            {
+                dstTmp = dst;
+            }
+
+            if ( WC2MB(dstTmp, src, lenChunk + lenNul) == wxCONV_FAILED )

                 return wxCONV_FAILED;
 
                 return wxCONV_FAILED;
 

+            if ( dstTmp != dst )
+            {
+                // copy everything up to but excluding the terminating NUL(s)
+                // into the real output buffer
+                memcpy(dst, dstTmp, lenChunk);
+
+                // micro-optimization: if dstTmp != dst it means that chunkEnd
+                // == srcEnd and so we're done, no need to update anything below
+                break;
+            }
+

             dst += lenChunk;
             dst += lenChunk;

+            if ( chunkEnd < srcEnd )
+                dst += lenNul;

         }
         }

+
+        src = chunkEnd;

     }
 
     return dstWritten;
     }
 
     return dstWritten;


@@ -391,13 +460,19 @@ wxMBConv::cMB2WC(const char *inBuff, size_t inLen, size_t *outLen) const
         // because we want the buffer to always be NUL-terminated, even if the
         // input isn't (as otherwise the caller has no way to know its length)
         wxWCharBuffer wbuf(dstLen);
         // because we want the buffer to always be NUL-terminated, even if the
         // input isn't (as otherwise the caller has no way to know its length)
         wxWCharBuffer wbuf(dstLen);

-        wbuf.data()[dstLen - 1] = L'\0';
+        wbuf.data()[dstLen] = L'\0';

         if ( ToWChar(wbuf.data(), dstLen, inBuff, inLen) != wxCONV_FAILED )
         {
             if ( outLen )
             {
                 *outLen = dstLen;
         if ( ToWChar(wbuf.data(), dstLen, inBuff, inLen) != wxCONV_FAILED )
         {
             if ( outLen )
             {
                 *outLen = dstLen;

-                if ( wbuf[dstLen - 1] == L'\0' )
+
+                // we also need to handle NUL-terminated input strings
+                // specially: for them the output is the length of the string
+                // excluding the trailing NUL, however if we're asked to
+                // convert a specific number of characters we return the length
+                // of the resulting output even if it's NUL-terminated
+                if ( inLen == wxNO_LEN )

                     (*outLen)--;
             }
 
                     (*outLen)--;
             }
 


@@ -429,11 +504,10 @@ wxMBConv::cWC2MB(const wchar_t *inBuff, size_t inLen, size_t *outLen) const
             {
                 *outLen = dstLen;
 
             {
                 *outLen = dstLen;
 

-                if ( dstLen >= nulLen &&
-                        !NotAllNULs(buf.data() + dstLen - nulLen, nulLen) )
+                if ( inLen == wxNO_LEN )

                 {
                 {

-                    // in this case the output is NUL-terminated and we're not
-                    // supposed to count NUL
+                    // in this case both input and output are NUL-terminated
+                    // and we're not supposed to count NUL

                     *outLen -= nulLen;
                 }
             }
                     *outLen -= nulLen;
                 }
             }


@@ -448,6 +522,42 @@ wxMBConv::cWC2MB(const wchar_t *inBuff, size_t inLen, size_t *outLen) const
     return wxCharBuffer();
 }
 
     return wxCharBuffer();
 }
 

+const wxWCharBuffer wxMBConv::cMB2WC(const wxScopedCharBuffer& buf) const
+{
+    const size_t srcLen = buf.length();
+    if ( srcLen )
+    {
+        const size_t dstLen = ToWChar(NULL, 0, buf, srcLen);
+        if ( dstLen != wxCONV_FAILED )
+        {
+            wxWCharBuffer wbuf(dstLen);
+            wbuf.data()[dstLen] = L'\0';
+            if ( ToWChar(wbuf.data(), dstLen, buf, srcLen) != wxCONV_FAILED )
+                return wbuf;
+        }
+    }
+
+    return wxScopedWCharBuffer::CreateNonOwned(L"", 0);
+}
+
+const wxCharBuffer wxMBConv::cWC2MB(const wxScopedWCharBuffer& wbuf) const
+{
+    const size_t srcLen = wbuf.length();
+    if ( srcLen )
+    {
+        const size_t dstLen = FromWChar(NULL, 0, wbuf, srcLen);
+        if ( dstLen != wxCONV_FAILED )
+        {
+            wxCharBuffer buf(dstLen);
+            buf.data()[dstLen] = '\0';
+            if ( FromWChar(buf.data(), dstLen, wbuf, srcLen) != wxCONV_FAILED )
+                return buf;
+        }
+    }
+
+    return wxScopedCharBuffer::CreateNonOwned("", 0);
+}
+

 // ----------------------------------------------------------------------------
 // wxMBConvLibc
 // ----------------------------------------------------------------------------
 // ----------------------------------------------------------------------------
 // wxMBConvLibc
 // ----------------------------------------------------------------------------


@@ -470,8 +580,8 @@ size_t wxMBConvLibc::WC2MB(char *buf, const wchar_t *psz, size_t n) const
 
 wxConvBrokenFileNames::wxConvBrokenFileNames(const wxString& charset)
 {
 
 wxConvBrokenFileNames::wxConvBrokenFileNames(const wxString& charset)
 {

-    if ( wxStricmp(charset, _T("UTF-8")) == 0 ||
-         wxStricmp(charset, _T("UTF8")) == 0  )
+    if ( wxStricmp(charset, wxT("UTF-8")) == 0 ||
+         wxStricmp(charset, wxT("UTF8")) == 0  )

         m_conv = new wxMBConvUTF8(wxMBConvUTF8::MAP_INVALID_UTF8_TO_PUA);
     else
         m_conv = new wxCSConv(charset);
         m_conv = new wxMBConvUTF8(wxMBConvUTF8::MAP_INVALID_UTF8_TO_PUA);
     else
         m_conv = new wxCSConv(charset);


@@ -484,6 +594,8 @@ wxConvBrokenFileNames::wxConvBrokenFileNames(const wxString& charset)
 // ----------------------------------------------------------------------------
 
 // Implementation (C) 2004 Fredrik Roubert
 // ----------------------------------------------------------------------------
 
 // Implementation (C) 2004 Fredrik Roubert

+//
+// Changes to work in streaming mode (C) 2008 Vadim Zeitlin

 
 //
 // BASE64 decoding table
 
 //
 // BASE64 decoding table


@@ -524,70 +636,149 @@ static const unsigned char utf7unb64[] =
     0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff
 };
 
     0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff
 };
 

-size_t wxMBConvUTF7::MB2WC(wchar_t *buf, const char *psz, size_t n) const
+size_t wxMBConvUTF7::ToWChar(wchar_t *dst, size_t dstLen,
+                             const char *src, size_t srcLen) const

 {
 {

+    DecoderState stateOrig,
+                *statePtr;
+    if ( srcLen == wxNO_LEN )
+    {
+        // convert the entire string, up to and including the trailing NUL
+        srcLen = strlen(src) + 1;
+
+        // when working on the entire strings we don't update nor use the shift
+        // state from the previous call
+        statePtr = &stateOrig;
+    }
+    else // when working with partial strings we do use the shift state
+    {
+        statePtr = const_cast<DecoderState *>(&m_stateDecoder);
+
+        // also save the old state to be able to rollback to it on error
+        stateOrig = m_stateDecoder;
+    }
+
+    // but to simplify the code below we use this variable in both cases
+    DecoderState& state = *statePtr;
+
+
+    // number of characters [which would have been] written to dst [if it were
+    // not NULL]

     size_t len = 0;
 
     size_t len = 0;
 

-    while ( *psz && (!buf || (len < n)) )
+    const char * const srcEnd = src + srcLen;
+
+    while ( (src < srcEnd) && (!dst || (len < dstLen)) )

     {
     {

-        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 // start of BASE64 encoded string
+        const unsigned char cc = *src++;
+
+        if ( state.IsShifted() )

         {
         {

-            bool lsb, ok;
-            unsigned int d, l;
-            for ( ok = lsb = false, d = 0, l = 0;
-                  (cc = utf7unb64[(unsigned char)*psz]) != 0xff;
-                  psz++ )
+            const unsigned char dc = utf7unb64[cc];
+            if ( dc == 0xff )

             {
             {

-                d <<= 6;
-                d += cc;
-                for (l += 6; l >= 8; lsb = !lsb)
+                // end of encoded part, check that nothing was left: there can
+                // be up to 4 bits of 0 padding but nothing else (we also need
+                // to check isLSB as we count bits modulo 8 while a valid UTF-7
+                // encoded sequence must contain an integral number of UTF-16
+                // characters)
+                if ( state.isLSB || state.bit > 4 ||
+                        (state.accum & ((1 << state.bit) - 1)) )

                 {
                 {

-                    unsigned char c = (unsigned char)((d >> (l -= 8)) % 256);
-                    if (lsb)
+                    if ( !len )
+                        state = stateOrig;
+
+                    return wxCONV_FAILED;
+                }
+
+                state.ToDirect();
+
+                // re-parse this character normally below unless it's '-' which
+                // is consumed by the decoder
+                if ( cc == '-' )
+                    continue;
+            }
+            else // valid encoded character
+            {
+                // mini base64 decoder: each character is 6 bits
+                state.bit += 6;
+                state.accum <<= 6;
+                state.accum += dc;
+
+                if ( state.bit >= 8 )
+                {
+                    // got the full byte, consume it
+                    state.bit -= 8;
+                    unsigned char b = (state.accum >> state.bit) & 0x00ff;
+
+                    if ( state.isLSB )

                     {
                     {

-                        if (buf)
-                            *buf++ |= c;
-                        len ++;
+                        // we've got the full word, output it
+                        if ( dst )
+                            *dst++ = (state.msb << 8) | b;
+                        len++;
+                        state.isLSB = false;

                     }
                     }

-                    else
+                    else // MSB

                     {
                     {

-                        if (buf)
-                            *buf = (wchar_t)(c << 8);
+                        // just store it while we wait for LSB
+                        state.msb = b;
+                        state.isLSB = true;

                     }
                     }

-
-                    ok = true;

                 }
             }
                 }
             }

+        }

 
 

-            if ( !ok )
+        if ( state.IsDirect() )
+        {
+            // start of an encoded segment?
+            if ( cc == '+' )

             {
             {

-                // in valid UTF7 we should have valid characters after '+'
-                return wxCONV_FAILED;
+                if ( *src == '-' )
+                {
+                    // just the encoded plus sign, don't switch to shifted mode
+                    if ( dst )
+                        *dst++ = '+';
+                    len++;
+                    src++;
+                }
+                else if ( utf7unb64[(unsigned)*src] == 0xff )
+                {
+                    // empty encoded chunks are not allowed
+                    if ( !len )
+                        state = stateOrig;
+
+                    return wxCONV_FAILED;
+                }
+                else // base-64 encoded chunk follows
+                {
+                    state.ToShifted();
+                }

             }
             }

+            else // not '+'
+            {
+                // only printable 7 bit ASCII characters (with the exception of
+                // NUL, TAB, CR and LF) can be used directly
+                if ( cc >= 0x7f || (cc < ' ' &&
+                      !(cc == '\0' || cc == '\t' || cc == '\r' || cc == '\n')) )
+                    return wxCONV_FAILED;

 
 

-            if (*psz == '-')
-                psz++;
+                if ( dst )
+                    *dst++ = cc;
+                len++;
+            }

         }
     }
 
         }
     }
 

-    if ( buf && (len < n) )
-        *buf = '\0';
+    if ( !len )
+    {
+        // as we didn't read any characters we should be called with the same
+        // data (followed by some more new data) again later so don't save our
+        // state
+        state = stateOrig;
+
+        return wxCONV_FAILED;
+    }

 
     return len;
 }
 
     return len;
 }


@@ -617,7 +808,7 @@ static const unsigned char utf7enb64[] =
 //
 static const unsigned char utf7encode[128] =
 {
 //
 static const unsigned char utf7encode[128] =
 {

-    3, 3, 3, 3, 3, 3, 3, 3, 3, 2, 2, 3, 3, 2, 3, 3,
+    0, 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,
     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,


@@ -627,21 +818,72 @@ static const unsigned char utf7encode[128] =
     0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 3, 3
 };
 
     0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 3, 3
 };
 

-size_t wxMBConvUTF7::WC2MB(char *buf, const wchar_t *psz, size_t n) const
+static inline bool wxIsUTF7Direct(wchar_t wc)
+{
+    return wc < 0x80 && utf7encode[wc] < 1;
+}
+
+size_t wxMBConvUTF7::FromWChar(char *dst, size_t dstLen,
+                               const wchar_t *src, size_t srcLen) const

 {
 {

+    EncoderState stateOrig,
+                *statePtr;
+    if ( srcLen == wxNO_LEN )
+    {
+        // we don't apply the stored state when operating on entire strings at
+        // once
+        statePtr = &stateOrig;
+
+        srcLen = wxWcslen(src) + 1;
+    }
+    else // do use the mode we left the output in previously
+    {
+        stateOrig = m_stateEncoder;
+        statePtr = const_cast<EncoderState *>(&m_stateEncoder);
+    }
+
+    EncoderState& state = *statePtr;
+
+

     size_t len = 0;
 
     size_t len = 0;
 

-    while (*psz && ((!buf) || (len < n)))
+    const wchar_t * const srcEnd = src + srcLen;
+    while ( src < srcEnd && (!dst || len < dstLen) )

     {
     {

-        wchar_t cc = *psz++;
-        if (cc < 0x80 && utf7encode[cc] < 1)
+        wchar_t cc = *src++;
+        if ( wxIsUTF7Direct(cc) )

         {
         {

-            // plain ASCII char
-            if (buf)
-                *buf++ = (char)cc;
+            if ( state.IsShifted() )
+            {
+                // pad with zeros the last encoded block if necessary
+                if ( state.bit )
+                {
+                    if ( dst )
+                        *dst++ = utf7enb64[((state.accum % 16) << (6 - state.bit)) % 64];
+                    len++;
+                }
+
+                state.ToDirect();
+
+                if ( dst )
+                    *dst++ = '-';
+                len++;
+            }

 
 

+            if ( dst )
+                *dst++ = (char)cc;

             len++;
         }
             len++;
         }

+        else if ( cc == '+' && state.IsDirect() )
+        {
+            if ( dst )
+            {
+                *dst++ = '+';
+                *dst++ = '-';
+            }
+
+            len += 2;
+        }

 #ifndef WC_UTF16
         else if (((wxUint32)cc) > 0xffff)
         {
 #ifndef WC_UTF16
         else if (((wxUint32)cc) > 0xffff)
         {


@@ -651,52 +893,45 @@ size_t wxMBConvUTF7::WC2MB(char *buf, const wchar_t *psz, size_t n) const
 #endif
         else
         {
 #endif
         else
         {

-            if (buf)
-                *buf++ = '+';
+            if ( state.IsDirect() )
+            {
+                state.ToShifted();

 
 

-            len++;
-            if (cc != '+')
+                if ( dst )
+                    *dst++ = '+';
+                len++;
+            }
+
+            // BASE64 encode string
+            for ( ;; )

             {
             {

-                // BASE64 encode string
-                unsigned int lsb, d, l;
-                for (d = 0, l = 0; /*nothing*/; psz++)
+                for ( unsigned lsb = 0; lsb < 2; lsb++ )

                 {
                 {

-                    for (lsb = 0; lsb < 2; lsb ++)
-                    {
-                        d <<= 8;
-                        d += lsb ? cc & 0xff : (cc & 0xff00) >> 8;
+                    state.accum <<= 8;
+                    state.accum += lsb ? cc & 0xff : (cc & 0xff00) >> 8;

 
 

-                        for (l += 8; l >= 6; )
-                        {
-                            l -= 6;
-                            if (buf)
-                                *buf++ = utf7enb64[(d >> l) % 64];
-                            len++;
-                        }
+                    for (state.bit += 8; state.bit >= 6; )
+                    {
+                        state.bit -= 6;
+                        if ( dst )
+                            *dst++ = utf7enb64[(state.accum >> state.bit) % 64];
+                        len++;

                     }
                     }

-
-                    cc = *psz;
-                    if (!(cc) || (cc < 0x80 && utf7encode[cc] < 1))
-                        break;

                 }
 
                 }
 

-                if (l != 0)
-                {
-                    if (buf)
-                        *buf++ = utf7enb64[((d % 16) << (6 - l)) % 64];
+                if ( src == srcEnd || wxIsUTF7Direct(cc = *src) )
+                    break;

 
 

-                    len++;
-                }
+                src++;

             }
             }

-
-            if (buf)
-                *buf++ = '-';
-            len++;

         }
     }
 
         }
     }
 

-    if (buf && (len < n))
-        *buf = 0;
+    // we need to restore the original encoder state if we were called just to
+    // calculate the amount of space needed as we will presumably be called
+    // again to really convert the data now
+    if ( !dst )
+        state = stateOrig;

 
     return len;
 }
 
     return len;
 }


@@ -880,7 +1115,7 @@ wxMBConvStrictUTF8::FromWChar(char *dst, size_t dstLen,
 
     for ( const wchar_t *wp = src; ; wp++ )
     {
 
     for ( const wchar_t *wp = src; ; wp++ )
     {

-        if ( !(srcLen == wxNO_LEN ? *wp : srcLen--) )
+        if ( !(srcLen == wxNO_LEN ? *wp : srcLen) )

         {
             // all done successfully, just add the trailing NULL if we are not
             // using explicit length
         {
             // all done successfully, just add the trailing NULL if we are not
             // using explicit length


@@ -900,6 +1135,8 @@ wxMBConvStrictUTF8::FromWChar(char *dst, size_t dstLen,
             return written;
         }
 
             return written;
         }
 

+        if ( srcLen != wxNO_LEN )
+            srcLen--;

 
         wxUint32 code;
 #ifdef WC_UTF16
 
         wxUint32 code;
 #ifdef WC_UTF16


@@ -968,7 +1205,7 @@ wxMBConvStrictUTF8::FromWChar(char *dst, size_t dstLen,
         }
         else
         {
         }
         else
         {

-            wxFAIL_MSG( _T("trying to encode undefined Unicode character") );
+            wxFAIL_MSG( wxT("trying to encode undefined Unicode character") );

             break;
         }
 
             break;
         }
 


@@ -1238,7 +1475,7 @@ size_t wxMBConvUTF16Base::GetLength(const char *src, size_t srcLen)
     if ( srcLen == wxNO_LEN )
     {
         // count the number of bytes in input, including the trailing NULs
     if ( srcLen == wxNO_LEN )
     {
         // count the number of bytes in input, including the trailing NULs

-        const wxUint16 *inBuff = wx_reinterpret_cast(const wxUint16 *, src);
+        const wxUint16 *inBuff = reinterpret_cast<const wxUint16 *>(src);

         for ( srcLen = 1; *inBuff++; srcLen++ )
             ;
 
         for ( srcLen = 1; *inBuff++; srcLen++ )
             ;
 


@@ -1322,7 +1559,7 @@ wxMBConvUTF16swap::ToWChar(wchar_t *dst, size_t dstLen,
         if ( dstLen < srcLen )
             return wxCONV_FAILED;
 
         if ( dstLen < srcLen )
             return wxCONV_FAILED;
 

-        const wxUint16 *inBuff = wx_reinterpret_cast(const wxUint16 *, src);
+        const wxUint16 *inBuff = reinterpret_cast<const wxUint16 *>(src);

         for ( size_t n = 0; n < srcLen; n++, inBuff++ )
         {
             *dst++ = wxUINT16_SWAP_ALWAYS(*inBuff);
         for ( size_t n = 0; n < srcLen; n++, inBuff++ )
         {
             *dst++ = wxUINT16_SWAP_ALWAYS(*inBuff);


@@ -1346,7 +1583,7 @@ wxMBConvUTF16swap::FromWChar(char *dst, size_t dstLen,
         if ( dstLen < srcLen )
             return wxCONV_FAILED;
 
         if ( dstLen < srcLen )
             return wxCONV_FAILED;
 

-        wxUint16 *outBuff = wx_reinterpret_cast(wxUint16 *, dst);
+        wxUint16 *outBuff = reinterpret_cast<wxUint16 *>(dst);

         for ( size_t n = 0; n < srcLen; n += BYTES_PER_CHAR, src++ )
         {
             *outBuff++ = wxUINT16_SWAP_ALWAYS(*src);
         for ( size_t n = 0; n < srcLen; n += BYTES_PER_CHAR, src++ )
         {
             *outBuff++ = wxUINT16_SWAP_ALWAYS(*src);


@@ -1380,7 +1617,7 @@ wxMBConvUTF16straight::ToWChar(wchar_t *dst, size_t dstLen,
     }
 
     size_t outLen = 0;
     }
 
     size_t outLen = 0;

-    const wxUint16 *inBuff = wx_reinterpret_cast(const wxUint16 *, src);
+    const wxUint16 *inBuff = reinterpret_cast<const wxUint16 *>(src);

     for ( const wxUint16 * const inEnd = inBuff + inLen; inBuff < inEnd; )
     {
         const wxUint32 ch = wxDecodeSurrogate(&inBuff);
     for ( const wxUint16 * const inEnd = inBuff + inLen; inBuff < inEnd; )
     {
         const wxUint32 ch = wxDecodeSurrogate(&inBuff);


@@ -1405,7 +1642,7 @@ wxMBConvUTF16straight::FromWChar(char *dst, size_t dstLen,
         srcLen = wxWcslen(src) + 1;
 
     size_t outLen = 0;
         srcLen = wxWcslen(src) + 1;
 
     size_t outLen = 0;

-    wxUint16 *outBuff = wx_reinterpret_cast(wxUint16 *, dst);
+    wxUint16 *outBuff = reinterpret_cast<wxUint16 *>(dst);

     for ( size_t n = 0; n < srcLen; n++ )
     {
         wxUint16 cc[2];
     for ( size_t n = 0; n < srcLen; n++ )
     {
         wxUint16 cc[2];


@@ -1453,7 +1690,7 @@ wxMBConvUTF16swap::ToWChar(wchar_t *dst, size_t dstLen,
     }
 
     size_t outLen = 0;
     }
 
     size_t outLen = 0;

-    const wxUint16 *inBuff = wx_reinterpret_cast(const wxUint16 *, src);
+    const wxUint16 *inBuff = reinterpret_cast<const wxUint16 *>(src);

     for ( const wxUint16 * const inEnd = inBuff + inLen; inBuff < inEnd; )
     {
         wxUint32 ch;
     for ( const wxUint16 * const inEnd = inBuff + inLen; inBuff < inEnd; )
     {
         wxUint32 ch;


@@ -1488,7 +1725,7 @@ wxMBConvUTF16swap::FromWChar(char *dst, size_t dstLen,
         srcLen = wxWcslen(src) + 1;
 
     size_t outLen = 0;
         srcLen = wxWcslen(src) + 1;
 
     size_t outLen = 0;

-    wxUint16 *outBuff = wx_reinterpret_cast(wxUint16 *, dst);
+    wxUint16 *outBuff = reinterpret_cast<wxUint16 *>(dst);

     for ( const wchar_t *srcEnd = src + srcLen; src < srcEnd; src++ )
     {
         wxUint16 cc[2];
     for ( const wchar_t *srcEnd = src + srcLen; src < srcEnd; src++ )
     {
         wxUint16 cc[2];


@@ -1539,7 +1776,7 @@ size_t wxMBConvUTF32Base::GetLength(const char *src, size_t srcLen)
     if ( srcLen == wxNO_LEN )
     {
         // count the number of bytes in input, including the trailing NULs
     if ( srcLen == wxNO_LEN )
     {
         // count the number of bytes in input, including the trailing NULs

-        const wxUint32 *inBuff = wx_reinterpret_cast(const wxUint32 *, src);
+        const wxUint32 *inBuff = reinterpret_cast<const wxUint32 *>(src);

         for ( srcLen = 1; *inBuff++; srcLen++ )
             ;
 
         for ( srcLen = 1; *inBuff++; srcLen++ )
             ;
 


@@ -1570,7 +1807,7 @@ wxMBConvUTF32straight::ToWChar(wchar_t *dst, size_t dstLen,
     if ( srcLen == wxNO_LEN )
         return wxCONV_FAILED;
 
     if ( srcLen == wxNO_LEN )
         return wxCONV_FAILED;
 

-    const wxUint32 *inBuff = wx_reinterpret_cast(const wxUint32 *, src);
+    const wxUint32 *inBuff = reinterpret_cast<const wxUint32 *>(src);

     const size_t inLen = srcLen / BYTES_PER_CHAR;
     size_t outLen = 0;
     for ( size_t n = 0; n < inLen; n++ )
     const size_t inLen = srcLen / BYTES_PER_CHAR;
     size_t outLen = 0;
     for ( size_t n = 0; n < inLen; n++ )


@@ -1617,7 +1854,7 @@ wxMBConvUTF32straight::FromWChar(char *dst, size_t dstLen,
         return srcLen * BYTES_PER_CHAR;
     }
 
         return srcLen * BYTES_PER_CHAR;
     }
 

-    wxUint32 *outBuff = wx_reinterpret_cast(wxUint32 *, dst);
+    wxUint32 *outBuff = reinterpret_cast<wxUint32 *>(dst);

     size_t outLen = 0;
     for ( const wchar_t * const srcEnd = src + srcLen; src < srcEnd; )
     {
     size_t outLen = 0;
     for ( const wchar_t * const srcEnd = src + srcLen; src < srcEnd; )
     {


@@ -1648,7 +1885,7 @@ wxMBConvUTF32swap::ToWChar(wchar_t *dst, size_t dstLen,
     if ( srcLen == wxNO_LEN )
         return wxCONV_FAILED;
 
     if ( srcLen == wxNO_LEN )
         return wxCONV_FAILED;
 

-    const wxUint32 *inBuff = wx_reinterpret_cast(const wxUint32 *, src);
+    const wxUint32 *inBuff = reinterpret_cast<const wxUint32 *>(src);

     const size_t inLen = srcLen / BYTES_PER_CHAR;
     size_t outLen = 0;
     for ( size_t n = 0; n < inLen; n++, inBuff++ )
     const size_t inLen = srcLen / BYTES_PER_CHAR;
     size_t outLen = 0;
     for ( size_t n = 0; n < inLen; n++, inBuff++ )


@@ -1695,7 +1932,7 @@ wxMBConvUTF32swap::FromWChar(char *dst, size_t dstLen,
         return srcLen*BYTES_PER_CHAR;
     }
 
         return srcLen*BYTES_PER_CHAR;
     }
 

-    wxUint32 *outBuff = wx_reinterpret_cast(wxUint32 *, dst);
+    wxUint32 *outBuff = reinterpret_cast<wxUint32 *>(dst);

     size_t outLen = 0;
     for ( const wchar_t * const srcEnd = src + srcLen; src < srcEnd; )
     {
     size_t outLen = 0;
     for ( const wchar_t * const srcEnd = src + srcLen; src < srcEnd; )
     {


@@ -1780,7 +2017,7 @@ wxMBConvUTF32swap::ToWChar(wchar_t *dst, size_t dstLen,
         if ( dstLen < srcLen )
             return wxCONV_FAILED;
 
         if ( dstLen < srcLen )
             return wxCONV_FAILED;
 

-        const wxUint32 *inBuff = wx_reinterpret_cast(const wxUint32 *, src);
+        const wxUint32 *inBuff = reinterpret_cast<const wxUint32 *>(src);

         for ( size_t n = 0; n < srcLen; n++, inBuff++ )
         {
             *dst++ = wxUINT32_SWAP_ALWAYS(*inBuff);
         for ( size_t n = 0; n < srcLen; n++, inBuff++ )
         {
             *dst++ = wxUINT32_SWAP_ALWAYS(*inBuff);


@@ -1804,7 +2041,7 @@ wxMBConvUTF32swap::FromWChar(char *dst, size_t dstLen,
         if ( dstLen < srcLen )
             return wxCONV_FAILED;
 
         if ( dstLen < srcLen )
             return wxCONV_FAILED;
 

-        wxUint32 *outBuff = wx_reinterpret_cast(wxUint32 *, dst);
+        wxUint32 *outBuff = reinterpret_cast<wxUint32 *>(dst);

         for ( size_t n = 0; n < srcLen; n += BYTES_PER_CHAR, src++ )
         {
             *outBuff++ = wxUINT32_SWAP_ALWAYS(*src);
         for ( size_t n = 0; n < srcLen; n += BYTES_PER_CHAR, src++ )
         {
             *outBuff++ = wxUINT32_SWAP_ALWAYS(*src);


@@ -1876,7 +2113,7 @@ public:
 
     virtual wxMBConv *Clone() const
     {
 
     virtual wxMBConv *Clone() const
     {

-        wxMBConv_iconv *p = new wxMBConv_iconv(m_name.ToAscii());
+        wxMBConv_iconv *p = new wxMBConv_iconv(m_name);

         p->m_minMBCharWidth = m_minMBCharWidth;
         return p;
     }
         p->m_minMBCharWidth = m_minMBCharWidth;
         return p;
     }


@@ -1906,7 +2143,7 @@ private:
 
 
     // name of the encoding handled by this conversion
 
 
     // name of the encoding handled by this conversion

-    wxString m_name;
+    const char *m_name;

 
     // cached result of GetMBNulLen(); set to 0 meaning "unknown"
     // initially
 
     // cached result of GetMBNulLen(); set to 0 meaning "unknown"
     // initially


@@ -1930,28 +2167,28 @@ wxString wxMBConv_iconv::ms_wcCharsetName;
 bool wxMBConv_iconv::ms_wcNeedsSwap = false;
 
 wxMBConv_iconv::wxMBConv_iconv(const char *name)
 bool wxMBConv_iconv::ms_wcNeedsSwap = false;
 
 wxMBConv_iconv::wxMBConv_iconv(const char *name)

-              : m_name(name)
+              : m_name(wxStrdup(name))

 {
     m_minMBCharWidth = 0;
 
     // check for charset that represents wchar_t:
     if ( ms_wcCharsetName.empty() )
     {
 {
     m_minMBCharWidth = 0;
 
     // check for charset that represents wchar_t:
     if ( ms_wcCharsetName.empty() )
     {

-        wxLogTrace(TRACE_STRCONV, _T("Looking for wide char codeset:"));
+        wxLogTrace(TRACE_STRCONV, wxT("Looking for wide char codeset:"));

 
 #if wxUSE_FONTMAP
 
 #if wxUSE_FONTMAP

-        const wxChar **names = wxFontMapperBase::GetAllEncodingNames(WC_ENC);
+        const wxChar *const *names = wxFontMapperBase::GetAllEncodingNames(WC_ENC);

 #else // !wxUSE_FONTMAP
 #else // !wxUSE_FONTMAP

-        static const wxChar *names_static[] =
+        static const wxChar *const names_static[] =

         {
 #if SIZEOF_WCHAR_T == 4
         {
 #if SIZEOF_WCHAR_T == 4

-            _T("UCS-4"),
+            wxT("UCS-4"),

 #elif SIZEOF_WCHAR_T = 2
 #elif SIZEOF_WCHAR_T = 2

-            _T("UCS-2"),
+            wxT("UCS-2"),

 #endif
             NULL
         };
 #endif
             NULL
         };

-        const wxChar **names = names_static;
+        const wxChar *const *names = names_static;

 #endif // wxUSE_FONTMAP/!wxUSE_FONTMAP
 
         for ( ; *names && ms_wcCharsetName.empty(); ++names )
 #endif // wxUSE_FONTMAP/!wxUSE_FONTMAP
 
         for ( ; *names && ms_wcCharsetName.empty(); ++names )


@@ -1962,19 +2199,19 @@ wxMBConv_iconv::wxMBConv_iconv(const char *name)
             wxString nameXE(nameCS);
 
 #ifdef WORDS_BIGENDIAN
             wxString nameXE(nameCS);
 
 #ifdef WORDS_BIGENDIAN

-                nameXE += _T("BE");
+                nameXE += wxT("BE");

 #else // little endian
 #else // little endian

-                nameXE += _T("LE");
+                nameXE += wxT("LE");

 #endif
 
 #endif
 

-            wxLogTrace(TRACE_STRCONV, _T("  trying charset \"%s\""),
+            wxLogTrace(TRACE_STRCONV, wxT("  trying charset \"%s\""),

                        nameXE.c_str());
 
             m2w = iconv_open(nameXE.ToAscii(), name);
             if ( m2w == ICONV_T_INVALID )
             {
                 // try charset w/o bytesex info (e.g. "UCS4")
                        nameXE.c_str());
 
             m2w = iconv_open(nameXE.ToAscii(), name);
             if ( m2w == ICONV_T_INVALID )
             {
                 // try charset w/o bytesex info (e.g. "UCS4")

-                wxLogTrace(TRACE_STRCONV, _T("  trying charset \"%s\""),
+                wxLogTrace(TRACE_STRCONV, wxT("  trying charset \"%s\""),

                            nameCS.c_str());
                 m2w = iconv_open(nameCS.ToAscii(), name);
 
                            nameCS.c_str());
                 m2w = iconv_open(nameCS.ToAscii(), name);
 


@@ -2021,8 +2258,8 @@ wxMBConv_iconv::wxMBConv_iconv(const char *name)
                    wxT("iconv wchar_t charset is \"%s\"%s"),
                    ms_wcCharsetName.empty() ? wxString("<none>")
                                             : ms_wcCharsetName,
                    wxT("iconv wchar_t charset is \"%s\"%s"),
                    ms_wcCharsetName.empty() ? wxString("<none>")
                                             : ms_wcCharsetName,

-                   ms_wcNeedsSwap ? _T(" (needs swap)")
-                                  : _T(""));
+                   ms_wcNeedsSwap ? wxT(" (needs swap)")
+                                  : wxT(""));

     }
     else // we already have ms_wcCharsetName
     {
     }
     else // we already have ms_wcCharsetName
     {


@@ -2047,6 +2284,8 @@ wxMBConv_iconv::wxMBConv_iconv(const char *name)
 
 wxMBConv_iconv::~wxMBConv_iconv()
 {
 
 wxMBConv_iconv::~wxMBConv_iconv()
 {

+    free(const_cast<char *>(m_name));
+

     if ( m2w != ICONV_T_INVALID )
         iconv_close(m2w);
     if ( w2m != ICONV_T_INVALID )
     if ( m2w != ICONV_T_INVALID )
         iconv_close(m2w);
     if ( w2m != ICONV_T_INVALID )


@@ -2083,6 +2322,11 @@ wxMBConv_iconv::ToWChar(wchar_t *dst, size_t dstLen,
                 srcLen = p - src;
                 break;
         }
                 srcLen = p - src;
                 break;
         }

+
+        // when we're determining the length of the string ourselves we count
+        // the terminating NUL(s) as part of it and always NUL-terminate the
+        // output
+        srcLen += nulLen;

     }
 
     // we express length in the number of (wide) characters but iconv always
     }
 
     // we express length in the number of (wide) characters but iconv always


@@ -2107,26 +2351,26 @@ wxMBConv_iconv::ToWChar(wchar_t *dst, size_t dstLen,
         char* bufPtr = (char*)dst;
 
         // have destination buffer, convert there
         char* bufPtr = (char*)dst;
 
         // have destination buffer, convert there

+        size_t dstLenOrig = dstLen;

         cres = iconv(m2w,
                      ICONV_CHAR_CAST(&pszPtr), &srcLen,
                      &bufPtr, &dstLen);
         cres = iconv(m2w,
                      ICONV_CHAR_CAST(&pszPtr), &srcLen,
                      &bufPtr, &dstLen);

-        res = dstLen - (dstLen / SIZEOF_WCHAR_T);
+
+        // convert the number of bytes converted as returned by iconv to the
+        // number of (wide) characters converted that we need
+        res = (dstLenOrig - dstLen) / SIZEOF_WCHAR_T;

 
         if (ms_wcNeedsSwap)
         {
             // convert to native endianness
             for ( unsigned i = 0; i < res; i++ )
 
         if (ms_wcNeedsSwap)
         {
             // convert to native endianness
             for ( unsigned i = 0; i < res; i++ )

-                dst[dstLen] = WC_BSWAP(dst[i]);
+                dst[i] = WC_BSWAP(dst[i]);

         }
         }

-
-        // NUL-terminate the string if there is any space left
-        if (res < dstLen)
-            dst[res] = 0;

     }
     else // no destination buffer
     {
         // convert using temp buffer to calculate the size of the buffer needed
     }
     else // no destination buffer
     {
         // convert using temp buffer to calculate the size of the buffer needed

-        wchar_t tbuf[8];
+        wchar_t tbuf[256];

         res = 0;
 
         do
         res = 0;
 
         do


@@ -2162,7 +2406,7 @@ size_t wxMBConv_iconv::FromWChar(char *dst, size_t dstLen,
 #endif
 
     if ( srcLen == wxNO_LEN )
 #endif
 
     if ( srcLen == wxNO_LEN )

-        srcLen = wxWcslen(src);
+        srcLen = wxWcslen(src) + 1;

 
     size_t inbuflen = srcLen * SIZEOF_WCHAR_T;
     size_t outbuflen = dstLen;
 
     size_t inbuflen = srcLen * SIZEOF_WCHAR_T;
     size_t outbuflen = dstLen;


@@ -2173,13 +2417,12 @@ size_t wxMBConv_iconv::FromWChar(char *dst, size_t dstLen,
     if (ms_wcNeedsSwap)
     {
         // need to copy to temp buffer to switch endianness
     if (ms_wcNeedsSwap)
     {
         // need to copy to temp buffer to switch endianness

-        // (doing WC_BSWAP twice on the original buffer won't help, as it
+        // (doing WC_BSWAP twice on the original buffer won't work, as it

         //  could be in read-only memory, or be accessed in some other thread)
         //  could be in read-only memory, or be accessed in some other thread)

-        tmpbuf = (wchar_t *)malloc(inbuflen + SIZEOF_WCHAR_T);
+        tmpbuf = (wchar_t *)malloc(inbuflen);

         for ( size_t i = 0; i < srcLen; i++ )
             tmpbuf[i] = WC_BSWAP(src[i]);
 
         for ( size_t i = 0; i < srcLen; i++ )
             tmpbuf[i] = WC_BSWAP(src[i]);
 

-        tmpbuf[srcLen] = L'\0';

         src = tmpbuf;
     }
 
         src = tmpbuf;
     }
 


@@ -2190,26 +2433,20 @@ size_t wxMBConv_iconv::FromWChar(char *dst, size_t dstLen,
         cres = iconv(w2m, ICONV_CHAR_CAST(&inbuf), &inbuflen, &dst, &outbuflen);
 
         res = dstLen - outbuflen;
         cres = iconv(w2m, ICONV_CHAR_CAST(&inbuf), &inbuflen, &dst, &outbuflen);
 
         res = dstLen - outbuflen;

-
-        // NB: iconv was given only wcslen(src) 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.
-        if (res < dstLen)
-            dst[0] = 0;

     }
     else // no destination buffer
     {
         // convert using temp buffer to calculate the size of the buffer needed
     }
     else // no destination buffer
     {
         // convert using temp buffer to calculate the size of the buffer needed

-        char tbuf[16];
+        char tbuf[256];

         res = 0;
         do
         {
             dst = tbuf;
         res = 0;
         do
         {
             dst = tbuf;

-            outbuflen = 16;
+            outbuflen = WXSIZEOF(tbuf);

 
             cres = iconv(w2m, ICONV_CHAR_CAST(&inbuf), &inbuflen, &dst, &outbuflen);
 
 
             cres = iconv(w2m, ICONV_CHAR_CAST(&inbuf), &inbuflen, &dst, &outbuflen);
 

-            res += 16 - outbuflen;
+            res += WXSIZEOF(tbuf) - outbuflen;

         }
         while ((cres == (size_t)-1) && (errno == E2BIG));
     }
         }
         while ((cres == (size_t)-1) && (errno == E2BIG));
     }


@@ -2500,7 +2737,7 @@ public:
             switch ( len )
             {
                 default:
             switch ( len )
             {
                 default:

-                    wxLogDebug(_T("Unexpected NUL length %d"), len);
+                    wxLogDebug(wxT("Unexpected NUL length %d"), len);

                     self->m_minMBCharWidth = (size_t)-1;
                     break;
 
                     self->m_minMBCharWidth = (size_t)-1;
                     break;
 


@@ -2547,7 +2784,7 @@ private:
                     break;
             }
 
                     break;
             }
 

-            wxASSERT_MSG( s_isWin98Or2k != -1, _T("should be set above") );
+            wxASSERT_MSG( s_isWin98Or2k != -1, wxT("should be set above") );

         }
 
         return s_isWin98Or2k == 1;
         }
 
         return s_isWin98Or2k == 1;


@@ -2682,7 +2919,7 @@ private:
     // were we initialized successfully?
     bool m_ok;
 
     // were we initialized successfully?
     bool m_ok;
 

-    DECLARE_NO_COPY_CLASS(wxMBConv_wxwin)
+    wxDECLARE_NO_COPY_CLASS(wxMBConv_wxwin);

 };
 
 // make the constructors available for unit testing
 };
 
 // make the constructors available for unit testing


@@ -2708,7 +2945,41 @@ void wxCSConv::Init()
 {
     m_name = NULL;
     m_convReal =  NULL;
 {
     m_name = NULL;
     m_convReal =  NULL;

-    m_deferred = true;
+}
+
+void wxCSConv::SetEncoding(wxFontEncoding encoding)
+{
+    switch ( encoding )
+    {
+        case wxFONTENCODING_MAX:
+        case wxFONTENCODING_SYSTEM:
+            if ( m_name )
+            {
+                // It's ok to not have encoding value if we have a name for it.
+                m_encoding = wxFONTENCODING_SYSTEM;
+            }
+            else // No name neither.
+            {
+                // Fall back to the system default encoding in this case (not
+                // sure how much sense does this make but this is how the old
+                // code used to behave).
+#if wxUSE_INTL
+                m_encoding = wxLocale::GetSystemEncoding();
+                if ( m_encoding == wxFONTENCODING_SYSTEM )
+#endif // wxUSE_INTL
+                    m_encoding = wxFONTENCODING_ISO8859_1;
+            }
+            break;
+
+        case wxFONTENCODING_DEFAULT:
+            // wxFONTENCODING_DEFAULT is same as US-ASCII in this context
+            m_encoding = wxFONTENCODING_ISO8859_1;
+            break;
+
+        default:
+            // Just use the provided encoding.
+            m_encoding = encoding;
+    }

 }
 
 wxCSConv::wxCSConv(const wxString& charset)
 }
 
 wxCSConv::wxCSConv(const wxString& charset)


@@ -2721,24 +2992,28 @@ wxCSConv::wxCSConv(const wxString& charset)
     }
 
 #if wxUSE_FONTMAP
     }
 
 #if wxUSE_FONTMAP

-    m_encoding = wxFontMapperBase::GetEncodingFromName(charset);
+    SetEncoding(wxFontMapperBase::GetEncodingFromName(charset));

 #else
 #else

-    m_encoding = wxFONTENCODING_SYSTEM;
+    SetEncoding(wxFONTENCODING_SYSTEM);

 #endif
 #endif

+
+    m_convReal = DoCreate();

 }
 
 wxCSConv::wxCSConv(wxFontEncoding encoding)
 {
     if ( encoding == wxFONTENCODING_MAX || encoding == wxFONTENCODING_DEFAULT )
     {
 }
 
 wxCSConv::wxCSConv(wxFontEncoding encoding)
 {
     if ( encoding == wxFONTENCODING_MAX || encoding == wxFONTENCODING_DEFAULT )
     {

-        wxFAIL_MSG( _T("invalid encoding value in wxCSConv ctor") );
+        wxFAIL_MSG( wxT("invalid encoding value in wxCSConv ctor") );

 
         encoding = wxFONTENCODING_SYSTEM;
     }
 
     Init();
 
 
         encoding = wxFONTENCODING_SYSTEM;
     }
 
     Init();
 

-    m_encoding = encoding;
+    SetEncoding(encoding);
+
+    m_convReal = DoCreate();

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


@@ -2752,7 +3027,9 @@ wxCSConv::wxCSConv(const wxCSConv& conv)
     Init();
 
     SetName(conv.m_name);
     Init();
 
     SetName(conv.m_name);

-    m_encoding = conv.m_encoding;
+    SetEncoding(conv.m_encoding);
+
+    m_convReal = DoCreate();

 }
 
 wxCSConv& wxCSConv::operator=(const wxCSConv& conv)
 }
 
 wxCSConv& wxCSConv::operator=(const wxCSConv& conv)


@@ -2760,7 +3037,9 @@ wxCSConv& wxCSConv::operator=(const wxCSConv& conv)
     Clear();
 
     SetName(conv.m_name);
     Clear();
 
     SetName(conv.m_name);

-    m_encoding = conv.m_encoding;
+    SetEncoding(conv.m_encoding);
+
+    m_convReal = DoCreate();

 
     return *this;
 }
 
     return *this;
 }


@@ -2768,19 +3047,15 @@ wxCSConv& wxCSConv::operator=(const wxCSConv& conv)
 void wxCSConv::Clear()
 {
     free(m_name);
 void wxCSConv::Clear()
 {
     free(m_name);

-    delete m_convReal;
-

     m_name = NULL;
     m_name = NULL;

-    m_convReal = NULL;
+
+    wxDELETE(m_convReal);

 }
 
 void wxCSConv::SetName(const char *charset)
 {
 }
 
 void wxCSConv::SetName(const char *charset)
 {

-    if (charset)
-    {
+    if ( charset )

         m_name = wxStrdup(charset);
         m_name = wxStrdup(charset);

-        m_deferred = true;
-    }

 }
 
 #if wxUSE_FONTMAP
 }
 
 #if wxUSE_FONTMAP


@@ -2803,8 +3078,7 @@ wxMBConv *wxCSConv::DoCreate() const
     // 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
     // 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 ||
-            m_encoding == wxFONTENCODING_DEFAULT )
+    if ( m_encoding == wxFONTENCODING_ISO8859_1 )

     {
         // don't convert at all
         return NULL;
     {
         // don't convert at all
         return NULL;


@@ -2856,7 +3130,7 @@ wxMBConv *wxCSConv::DoCreate() const
                 delete conv;
             }
 
                 delete conv;
             }
 

-            const wxChar** names = wxFontMapperBase::GetAllEncodingNames(encoding);
+            const wxChar* const* names = wxFontMapperBase::GetAllEncodingNames(encoding);

             // CS : in case this does not return valid names (eg for MacRoman)
             // encoding got a 'failure' entry in the cache all the same,
             // although it just has to be created using a different method, so
             // CS : in case this does not return valid names (eg for MacRoman)
             // encoding got a 'failure' entry in the cache all the same,
             // although it just has to be created using a different method, so


@@ -2879,7 +3153,7 @@ wxMBConv *wxCSConv::DoCreate() const
                     delete conv;
                 }
 
                     delete conv;
                 }
 

-                gs_nameCache[encoding] = _T(""); // cache the failure
+                gs_nameCache[encoding] = wxT(""); // cache the failure

             }
         }
 #endif // wxUSE_FONTMAP
             }
         }
 #endif // wxUSE_FONTMAP


@@ -2969,62 +3243,18 @@ wxMBConv *wxCSConv::DoCreate() const
 
         delete 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 a timestamp, for which it will need wxConvLocal (to convert
-    //     time to char* and then wchar_t*), but that fails, tries to log the
-    //     error, but wxLog has an (already locked) critical section that
-    //     guards the static buffer.
-    static bool alreadyLoggingError = false;
-    if (!alreadyLoggingError)
-    {
-        alreadyLoggingError = true;
-        wxLogError(_("Cannot convert from the charset '%s'!"),
-                   m_name ? m_name
-                      :
-#if wxUSE_FONTMAP
-                         (const char*)wxFontMapperBase::GetEncodingDescription(m_encoding).ToAscii()
-#else // !wxUSE_FONTMAP
-                         (const char*)wxString::Format(_("encoding %i"), m_encoding).ToAscii()
-#endif // wxUSE_FONTMAP/!wxUSE_FONTMAP
-              );

 
 

-        alreadyLoggingError = false;
-    }
+    wxLogTrace(TRACE_STRCONV,
+               wxT("encoding \"%s\" is not supported by this system"),
+               (m_name ? wxString(m_name)
+                       : wxFontMapperBase::GetEncodingName(m_encoding)));
+#endif // wxUSE_FONTMAP

 
     return NULL;
 }
 
 
     return NULL;
 }
 

-void wxCSConv::CreateConvIfNeeded() const
-{
-    if ( m_deferred )
-    {
-        wxCSConv *self = (wxCSConv *)this; // const_cast
-
-        // 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 )
-        {
-#if wxUSE_INTL
-            self->m_encoding = wxLocale::GetSystemEncoding();
-#else
-            // fallback to some reasonable default:
-            self->m_encoding = wxFONTENCODING_ISO8859_1;
-#endif // wxUSE_INTL
-        }
-
-        self->m_convReal = DoCreate();
-        self->m_deferred = false;
-    }
-}
-

 bool wxCSConv::IsOk() const
 {
 bool wxCSConv::IsOk() const
 {

-    CreateConvIfNeeded();
-

     // special case: no convReal created for wxFONTENCODING_ISO8859_1
     if ( m_encoding == wxFONTENCODING_ISO8859_1 )
         return true; // always ok as we do it ourselves
     // special case: no convReal created for wxFONTENCODING_ISO8859_1
     if ( m_encoding == wxFONTENCODING_ISO8859_1 )
         return true; // always ok as we do it ourselves


@@ -3037,85 +3267,65 @@ bool wxCSConv::IsOk() const
 size_t wxCSConv::ToWChar(wchar_t *dst, size_t dstLen,
                          const char *src, size_t srcLen) const
 {
 size_t wxCSConv::ToWChar(wchar_t *dst, size_t dstLen,
                          const char *src, size_t srcLen) const
 {

-    CreateConvIfNeeded();
-

     if (m_convReal)
         return m_convReal->ToWChar(dst, dstLen, src, srcLen);
 
     // latin-1 (direct)
     if (m_convReal)
         return m_convReal->ToWChar(dst, dstLen, src, srcLen);
 
     // latin-1 (direct)

-    return wxMBConv::ToWChar(dst, dstLen, src, srcLen);
-}
+    if ( srcLen == wxNO_LEN )
+        srcLen = strlen(src) + 1; // take trailing NUL too

 
 

-size_t wxCSConv::FromWChar(char *dst, size_t dstLen,
-                           const wchar_t *src, size_t srcLen) const
-{
-    CreateConvIfNeeded();
+    if ( dst )
+    {
+        if ( dstLen < srcLen )
+            return wxCONV_FAILED;

 
 

-    if (m_convReal)
-        return m_convReal->FromWChar(dst, dstLen, src, srcLen);
+        for ( size_t n = 0; n < srcLen; n++ )
+            dst[n] = (unsigned char)(src[n]);
+    }

 
 

-    // latin-1 (direct)
-    return wxMBConv::FromWChar(dst, dstLen, src, srcLen);
+    return srcLen;

 }
 
 }
 

-size_t wxCSConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
+size_t wxCSConv::FromWChar(char *dst, size_t dstLen,
+                           const wchar_t *src, size_t srcLen) const

 {
 {

-    CreateConvIfNeeded();
-

     if (m_convReal)
     if (m_convReal)

-        return m_convReal->MB2WC(buf, psz, n);
+        return m_convReal->FromWChar(dst, dstLen, src, srcLen);

 
     // latin-1 (direct)
 
     // latin-1 (direct)

-    size_t len = strlen(psz);
+    if ( srcLen == wxNO_LEN )
+        srcLen = wxWcslen(src) + 1;

 
 

-    if (buf)
+    if ( dst )

     {
     {

-        for (size_t c = 0; c <= len; c++)
-            buf[c] = (unsigned char)(psz[c]);
-    }
-
-    return len;
-}
-
-size_t wxCSConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
-{
-    CreateConvIfNeeded();
-
-    if (m_convReal)
-        return m_convReal->WC2MB(buf, psz, n);
+        if ( dstLen < srcLen )
+            return wxCONV_FAILED;

 
 

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

         {
         {

-            if (psz[c] > 0xFF)
+            if ( src[n] > 0xFF )

                 return wxCONV_FAILED;
 
                 return wxCONV_FAILED;
 

-            buf[c] = (char)psz[c];
+            dst[n] = (char)src[n];

         }
         }

+

     }
     }

-    else
+    else // still need to check the input validity

     {
     {

-        for (size_t c = 0; c <= len; c++)
+        for ( size_t n = 0; n < srcLen; n++ )

         {
         {

-            if (psz[c] > 0xFF)
+            if ( src[n] > 0xFF )

                 return wxCONV_FAILED;
         }
     }
 
                 return wxCONV_FAILED;
         }
     }
 

-    return len;
+    return srcLen;

 }
 
 size_t wxCSConv::GetMBNulLen() const
 {
 }
 
 size_t wxCSConv::GetMBNulLen() const
 {

-    CreateConvIfNeeded();
-

     if ( m_convReal )
     if ( m_convReal )

-    {

         return m_convReal->GetMBNulLen();
         return m_convReal->GetMBNulLen();

-    }

 
     // otherwise, we are ISO-8859-1
     return 1;
 
     // otherwise, we are ISO-8859-1
     return 1;


@@ -3124,12 +3334,8 @@ size_t wxCSConv::GetMBNulLen() const
 #if wxUSE_UNICODE_UTF8
 bool wxCSConv::IsUTF8() const
 {
 #if wxUSE_UNICODE_UTF8
 bool wxCSConv::IsUTF8() const
 {

-    CreateConvIfNeeded();
-

     if ( m_convReal )
     if ( m_convReal )

-    {

         return m_convReal->IsUTF8();
         return m_convReal->IsUTF8();

-    }

 
     // otherwise, we are ISO-8859-1
     return false;
 
     // otherwise, we are ISO-8859-1
     return false;


@@ -3199,8 +3405,15 @@ wxCharBuffer wxSafeConvertWX2MB(const wchar_t *ws)
 #define WX_DEFINE_GLOBAL_CONV(klass, name, ctor_args) \
     WX_DEFINE_GLOBAL_CONV2(klass, klass, name, ctor_args)
 
 #define WX_DEFINE_GLOBAL_CONV(klass, name, ctor_args) \
     WX_DEFINE_GLOBAL_CONV2(klass, klass, name, ctor_args)
 

+#ifdef __INTELC__
+    // disable warning "variable 'xxx' was declared but never referenced"
+    #pragma warning(disable: 177)
+#endif // Intel C++
+

 #ifdef __WINDOWS__
     WX_DEFINE_GLOBAL_CONV2(wxMBConv, wxMBConv_win32, wxConvLibc, wxEMPTY_PARAMETER_VALUE);
 #ifdef __WINDOWS__
     WX_DEFINE_GLOBAL_CONV2(wxMBConv, wxMBConv_win32, wxConvLibc, wxEMPTY_PARAMETER_VALUE);

+#elif 0 // defined(__WXOSX__)
+    WX_DEFINE_GLOBAL_CONV2(wxMBConv, wxMBConv_cf, wxConvLibc,  (wxFONTENCODING_UTF8));

 #else
     WX_DEFINE_GLOBAL_CONV2(wxMBConv, wxMBConvLibc, wxConvLibc, wxEMPTY_PARAMETER_VALUE);
 #endif
 #else
     WX_DEFINE_GLOBAL_CONV2(wxMBConv, wxMBConvLibc, wxConvLibc, wxEMPTY_PARAMETER_VALUE);
 #endif


@@ -3221,8 +3434,9 @@ WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent = wxGet_wxConvLibcPtr();
 WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvUI = wxGet_wxConvLocalPtr();
 
 #ifdef __DARWIN__
 WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvUI = wxGet_wxConvLocalPtr();
 
 #ifdef __DARWIN__

-// The xnu kernel always communicates file paths in decomposed UTF-8.
-// WARNING: Are we sure that CFString's conversion will cause decomposition?
+// It is important to use this conversion object under Darwin as it ensures
+// that Unicode strings are (re)composed correctly even though xnu kernel uses
+// decomposed form internally (at least for the file names).

 static wxMBConv_cf wxConvMacUTF8DObj(wxFONTENCODING_UTF8);
 #endif
 
 static wxMBConv_cf wxConvMacUTF8DObj(wxFONTENCODING_UTF8);
 #endif
 


@@ -3232,14 +3446,3 @@ WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvFileName =
 #else // !__DARWIN__
                                     wxGet_wxConvLibcPtr();
 #endif // __DARWIN__/!__DARWIN__
 #else // !__DARWIN__
                                     wxGet_wxConvLibcPtr();
 #endif // __DARWIN__/!__DARWIN__

-
-#else // !wxUSE_WCHAR_T
-
-// FIXME-UTF8: remove this, wxUSE_WCHAR_T is required now
-// stand-ins in absence of wchar_t
-WXDLLIMPEXP_DATA_BASE(wxMBConv) wxConvLibc,
-                                wxConvISO8859_1,
-                                wxConvLocal,
-                                wxConvUTF8;
-
-#endif // wxUSE_WCHAR_T/!wxUSE_WCHAR_T