no real changes, just some cleanup

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

diff --git a/src/common/string.cpp b/src/common/string.cpp
index 70fdb635e1785ea8a705afe5808b679c184983af..d36f85bd5a2cb4259e40df65fdd87a9cc3bfb7f8 100644 (file)
--- a/src/common/string.cpp
+++ b/src/common/string.cpp
@@ -1,25 +1,15 @@
 /////////////////////////////////////////////////////////////////////////////
 /////////////////////////////////////////////////////////////////////////////

-// Name:        string.cpp
+// Name:        src/common/string.cpp

 // Purpose:     wxString class
 // Purpose:     wxString class

-// Author:      Vadim Zeitlin
+// Author:      Vadim Zeitlin, Ryan Norton

 // Modified by:
 // Created:     29/01/98
 // RCS-ID:      $Id$
 // Copyright:   (c) 1998 Vadim Zeitlin <zeitlin@dptmaths.ens-cachan.fr>
 // Modified by:
 // Created:     29/01/98
 // RCS-ID:      $Id$
 // Copyright:   (c) 1998 Vadim Zeitlin <zeitlin@dptmaths.ens-cachan.fr>

-// Licence:     wxWindows license
+//              (c) 2004 Ryan Norton <wxprojects@comcast.net>
+// Licence:     wxWindows licence

 /////////////////////////////////////////////////////////////////////////////
 
 /////////////////////////////////////////////////////////////////////////////
 

-#ifdef __GNUG__
-#pragma implementation "string.h"
-#endif
-
-/*
- * About ref counting:
- *  1) all empty strings use g_strEmpty, nRefs = -1 (set in Init())
- *  2) AllocBuffer() sets nRefs to 1, Lock() increments it by one
- *  3) Unlock() decrements nRefs and frees memory if it goes to 0
- */
-

 // ===========================================================================
 // headers, declarations, constants
 // ===========================================================================
 // ===========================================================================
 // headers, declarations, constants
 // ===========================================================================


@@ -28,612 +18,1121 @@
 #include "wx/wxprec.h"
 
 #ifdef __BORLANDC__
 #include "wx/wxprec.h"
 
 #ifdef __BORLANDC__

-#pragma hdrstop
+    #pragma hdrstop

 #endif
 
 #ifndef WX_PRECOMP
 #endif
 
 #ifndef WX_PRECOMP

-  #include "wx/defs.h"
-  #include "wx/string.h"
-  #include "wx/intl.h"
+    #include "wx/string.h"
+    #include "wx/wxcrtvararg.h"

 #endif
 
 #include <ctype.h>
 #endif
 
 #include <ctype.h>

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

 #include <string.h>
 #include <stdlib.h>
 
 #include <string.h>
 #include <stdlib.h>
 

-#ifdef wxUSE_WCSRTOMBS
-  #include <wchar.h>    // for wcsrtombs(), see comments where it's used
-#endif // GNU
+#ifdef __SALFORDC__
+    #include <clib.h>
+#endif

 
 

-#ifdef  WXSTRING_IS_WXOBJECT
-  IMPLEMENT_DYNAMIC_CLASS(wxString, wxObject)
-#endif  //WXSTRING_IS_WXOBJECT
+#include "wx/hashmap.h"
+
+// string handling functions used by wxString:
+#if wxUSE_UNICODE_UTF8
+    #define wxStringMemcpy   memcpy
+    #define wxStringMemcmp   memcmp
+    #define wxStringMemchr   memchr
+    #define wxStringStrlen   strlen
+#else
+    #define wxStringMemcpy   wxTmemcpy
+    #define wxStringMemcmp   wxTmemcmp
+    #define wxStringMemchr   wxTmemchr
+    #define wxStringStrlen   wxStrlen
+#endif

 
 

-// allocating extra space for each string consumes more memory but speeds up
-// the concatenation operations (nLen is the current string's length)
-// NB: EXTRA_ALLOC must be >= 0!
-#define EXTRA_ALLOC       (19 - nLen % 16)

 
 // ---------------------------------------------------------------------------
 // static class variables definition
 // ---------------------------------------------------------------------------
 
 
 // ---------------------------------------------------------------------------
 // static class variables definition
 // ---------------------------------------------------------------------------
 

-#ifdef  STD_STRING_COMPATIBILITY
-  const size_t wxString::npos = STRING_MAXLEN;
-#endif
+//According to STL _must_ be a -1 size_t
+const size_t wxString::npos = (size_t) -1;

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

-// static data
+// global functions

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

-// for an empty string, GetStringData() will return this address: this
-// structure has the same layout as wxStringData and it's data() method will
-// return the empty string (dummy pointer)
-static const struct
+#if wxUSE_STD_IOSTREAM
+
+#include <iostream>
+
+wxSTD ostream& operator<<(wxSTD ostream& os, const wxCStrData& str)
+{
+// FIXME-UTF8: always, not only if wxUSE_UNICODE
+#if wxUSE_UNICODE && !defined(__BORLANDC__)
+    return os << (const wchar_t*)str.AsWCharBuf();
+#else
+    return os << (const char*)str.AsCharBuf();
+#endif
+}
+
+wxSTD ostream& operator<<(wxSTD ostream& os, const wxString& str)

 {
 {

-  wxStringData data;
-  char dummy;
-} g_strEmpty = { {-1, 0, 0}, '\0' };
+    return os << str.c_str();
+}
+
+wxSTD ostream& operator<<(wxSTD ostream& os, const wxCharBuffer& str)
+{
+    return os << str.data();
+}
+
+#ifndef __BORLANDC__
+wxSTD ostream& operator<<(wxSTD ostream& os, const wxWCharBuffer& str)
+{
+    return os << str.data();
+}
+#endif
+
+#endif // wxUSE_STD_IOSTREAM
+
+// ===========================================================================
+// wxString class core
+// ===========================================================================
+
+#if wxUSE_UNICODE_UTF8
+
+void wxString::PosLenToImpl(size_t pos, size_t len,
+                            size_t *implPos, size_t *implLen) const
+{
+    if ( pos == npos )
+        *implPos = npos;
+    else
+    {
+        const_iterator i = begin() + pos;
+        *implPos = wxStringImpl::const_iterator(i.impl()) - m_impl.begin();
+        if ( len == npos )
+            *implLen = npos;
+        else
+        {
+            // too large length is interpreted as "to the end of the string"
+            // FIXME-UTF8: verify this is the case in std::string, assert
+            // otherwise
+            if ( pos + len > length() )
+                len = length() - pos;
+
+            *implLen = (i + len).impl() - i.impl();
+        }
+    }
+}

 
 

-// empty C style string: points to 'string data' byte of g_strEmpty
-extern const char *g_szNul = &g_strEmpty.dummy;
+#endif // wxUSE_UNICODE_UTF8

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

-// global functions
+// wxCStrData converted strings caching

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

-#ifdef  STD_STRING_COMPATIBILITY
+// FIXME-UTF8: temporarily disabled because it doesn't work with global
+//             string objects; re-enable after fixing this bug and benchmarking
+//             performance to see if using a hash is a good idea at all
+#if 0

 
 

-// MS Visual C++ version 5.0 provides the new STL headers as well as the old
-// iostream ones.
+// For backward compatibility reasons, it must be possible to assign the value
+// returned by wxString::c_str() to a char* or wchar_t* variable and work with
+// it. Returning wxCharBuffer from (const char*)c_str() wouldn't do the trick,
+// because the memory would be freed immediately, but it has to be valid as long
+// as the string is not modified, so that code like this still works:

 //
 //

-// ATTN: you can _not_ use both of these in the same program!
-#if wxUSE_IOSTREAMH
-#include <iostream.h>
-#define   NAMESPACE
-#else
-#include <iostream>
-#  ifdef _MSC_VER
-      using namespace std;
-#  endif
-// for msvc (bcc50+ also) you don't need these NAMESPACE defines,
-// using namespace std; takes care of that.
-#define   NAMESPACE   std::
-#endif
+// const wxChar *s = str.c_str();
+// while ( s ) { ... }
+
+// FIXME-UTF8: not thread safe!
+// FIXME-UTF8: we currently clear the cached conversion only when the string is
+//             destroyed, but we should do it when the string is modified, to
+//             keep memory usage down
+// FIXME-UTF8: we do the conversion every time As[W]Char() is called, but if we
+//             invalidated the cache on every change, we could keep the previous
+//             conversion
+// FIXME-UTF8: add tracing of usage of these two methods - new code is supposed
+//             to use mb_str() or wc_str() instead of (const [w]char*)c_str()
+
+template<typename T>
+static inline void DeleteStringFromConversionCache(T& hash, const wxString *s)
+{
+    typename T::iterator i = hash.find(wxConstCast(s, wxString));
+    if ( i != hash.end() )
+    {
+        free(i->second);
+        hash.erase(i);
+    }
+}

 
 

+#if wxUSE_UNICODE
+// NB: non-STL implementation doesn't compile with "const wxString*" key type,
+//     so we have to use wxString* here and const-cast when used
+WX_DECLARE_HASH_MAP(wxString*, char*, wxPointerHash, wxPointerEqual,
+                    wxStringCharConversionCache);
+static wxStringCharConversionCache gs_stringsCharCache;

 
 

-NAMESPACE istream& operator>>(NAMESPACE istream& is, wxString& WXUNUSED(str))
+const char* wxCStrData::AsChar() const

 {
 {

-#if 0
-  int w = is.width(0);
-  if ( is.ipfx(0) ) {
-    NAMESPACE streambuf *sb = is.rdbuf();
-    str.erase();
-    while ( true ) {
-      int ch = sb->sbumpc ();
-      if ( ch == EOF ) {
-        is.setstate(NAMESPACE ios::eofbit);
-        break;
-      }
-      else if ( isspace(ch) ) {
-        sb->sungetc();
-        break;
-      }
+    // remove previously cache value, if any (see FIXMEs above):
+    DeleteStringFromConversionCache(gs_stringsCharCache, m_str);

 
 

-      str += ch;
-      if ( --w == 1 )
-        break;
-    }
-  }
+    // convert the string and keep it:
+    const char *s = gs_stringsCharCache[wxConstCast(m_str, wxString)] =
+        m_str->mb_str().release();
+
+    return s + m_offset;
+}
+#endif // wxUSE_UNICODE
+
+#if !wxUSE_UNICODE_WCHAR
+WX_DECLARE_HASH_MAP(wxString*, wchar_t*, wxPointerHash, wxPointerEqual,
+                    wxStringWCharConversionCache);
+static wxStringWCharConversionCache gs_stringsWCharCache;
+
+const wchar_t* wxCStrData::AsWChar() const
+{
+    // remove previously cache value, if any (see FIXMEs above):
+    DeleteStringFromConversionCache(gs_stringsWCharCache, m_str);
+
+    // convert the string and keep it:
+    const wchar_t *s = gs_stringsWCharCache[wxConstCast(m_str, wxString)] =
+        m_str->wc_str().release();
+
+    return s + m_offset;
+}
+#endif // !wxUSE_UNICODE_WCHAR

 
 

-  is.isfx();
-  if ( str.length() == 0 )
-    is.setstate(NAMESPACE ios::failbit);
+wxString::~wxString()
+{
+#if wxUSE_UNICODE
+    // FIXME-UTF8: do this only if locale is not UTF8 if wxUSE_UNICODE_UTF8
+    DeleteStringFromConversionCache(gs_stringsCharCache, this);
+#endif
+#if !wxUSE_UNICODE_WCHAR
+    DeleteStringFromConversionCache(gs_stringsWCharCache, this);

 #endif
 #endif

-  return is;

 }
 }

+#endif

 
 

-#endif  //std::string compatibility
+#if wxUSE_UNICODE && !wxUSE_UTF8_LOCALE_ONLY
+const char* wxCStrData::AsChar() const
+{
+#if wxUSE_UNICODE_UTF8
+    if ( wxLocaleIsUtf8 )
+        return AsInternal();
+#endif
+    // under non-UTF8 locales, we have to convert the internal UTF-8
+    // representation using wxConvLibc and cache the result

 
 

-// ----------------------------------------------------------------------------
-// private classes
-// ----------------------------------------------------------------------------
+    wxString *str = wxConstCast(m_str, wxString);

 
 

-// this small class is used to gather statistics for performance tuning
-//#define WXSTRING_STATISTICS
-#ifdef  WXSTRING_STATISTICS
-  class Averager
-  {
-  public:
-    Averager(const char *sz) { m_sz = sz; m_nTotal = m_nCount = 0; }
-   ~Averager()
-   { printf("wxString: average %s = %f\n", m_sz, ((float)m_nTotal)/m_nCount); }
-
-    void Add(size_t n) { m_nTotal += n; m_nCount++; }
-
-  private:
-    size_t m_nCount, m_nTotal;
-    const char *m_sz;
-  } g_averageLength("allocation size"),
-    g_averageSummandLength("summand length"),
-    g_averageConcatHit("hit probability in concat"),
-    g_averageInitialLength("initial string length");
-
-  #define STATISTICS_ADD(av, val) g_average##av.Add(val)
-#else
-  #define STATISTICS_ADD(av, val)
-#endif // WXSTRING_STATISTICS
+    // convert the string:
+    wxCharBuffer buf(str->mb_str());
+
+    // FIXME-UTF8: do the conversion in-place in the existing buffer
+    if ( str->m_convertedToChar &&
+         strlen(buf) == strlen(str->m_convertedToChar) )
+    {
+        // keep the same buffer for as long as possible, so that several calls
+        // to c_str() in a row still work:
+        strcpy(str->m_convertedToChar, buf);
+    }
+    else
+    {
+        str->m_convertedToChar = buf.release();
+    }
+
+    // and keep it:
+    return str->m_convertedToChar + m_offset;
+}
+#endif // wxUSE_UNICODE && !wxUSE_UTF8_LOCALE_ONLY
+
+#if !wxUSE_UNICODE_WCHAR
+const wchar_t* wxCStrData::AsWChar() const
+{
+    wxString *str = wxConstCast(m_str, wxString);
+
+    // convert the string:
+    wxWCharBuffer buf(str->wc_str());
+
+    // FIXME-UTF8: do the conversion in-place in the existing buffer
+    if ( str->m_convertedToWChar &&
+         wxWcslen(buf) == wxWcslen(str->m_convertedToWChar) )
+    {
+        // keep the same buffer for as long as possible, so that several calls
+        // to c_str() in a row still work:
+        memcpy(str->m_convertedToWChar, buf, sizeof(wchar_t) * wxWcslen(buf));
+    }
+    else
+    {
+        str->m_convertedToWChar = buf.release();
+    }
+
+    // and keep it:
+    return str->m_convertedToWChar + m_offset;
+}
+#endif // !wxUSE_UNICODE_WCHAR

 
 // ===========================================================================
 // wxString class core
 // ===========================================================================
 
 // ---------------------------------------------------------------------------
 
 // ===========================================================================
 // wxString class core
 // ===========================================================================
 
 // ---------------------------------------------------------------------------

-// construction
+// construction and conversion

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

-// constructs string of <nLength> copies of character <ch>
-wxString::wxString(char ch, size_t nLength)
+#if wxUSE_UNICODE_WCHAR
+/* static */
+wxString::SubstrBufFromMB wxString::ConvertStr(const char *psz, size_t nLength,
+                                               const wxMBConv& conv)

 {
 {

-  Init();
+    // anything to do?
+    if ( !psz || nLength == 0 )
+        return SubstrBufFromMB(L"", 0);

 
 

-  if ( nLength > 0 ) {
-    AllocBuffer(nLength);
+    if ( nLength == npos )
+        nLength = wxNO_LEN;

 
 

-    wxASSERT( sizeof(char) == 1 );  // can't use memset if not
+    size_t wcLen;
+    wxWCharBuffer wcBuf(conv.cMB2WC(psz, nLength, &wcLen));
+    if ( !wcLen )
+        return SubstrBufFromMB(_T(""), 0);
+    else
+        return SubstrBufFromMB(wcBuf, wcLen);
+}
+#endif // wxUSE_UNICODE_WCHAR

 
 

-    memset(m_pchData, ch, nLength);
-  }
+#if wxUSE_UNICODE_UTF8
+/* static */
+wxString::SubstrBufFromMB wxString::ConvertStr(const char *psz, size_t nLength,
+                                               const wxMBConv& conv)
+{
+    // anything to do?
+    if ( !psz || nLength == 0 )
+        return SubstrBufFromMB("", 0);
+
+    // if psz is already in UTF-8, we don't have to do the roundtrip to
+    // wchar_t* and back:
+    if ( conv.IsUTF8() )
+    {
+        // we need to validate the input because UTF8 iterators assume valid
+        // UTF-8 sequence and psz may be invalid:
+        if ( wxStringOperations::IsValidUtf8String(psz, nLength) )
+        {
+            return SubstrBufFromMB(wxCharBuffer::CreateNonOwned(psz), nLength);
+        }
+        // else: do the roundtrip through wchar_t*
+    }
+
+    if ( nLength == npos )
+        nLength = wxNO_LEN;
+
+    // first convert to wide string:
+    size_t wcLen;
+    wxWCharBuffer wcBuf(conv.cMB2WC(psz, nLength, &wcLen));
+    if ( !wcLen )
+        return SubstrBufFromMB("", 0);
+
+    // and then to UTF-8:
+    SubstrBufFromMB buf(ConvertStr(wcBuf, wcLen, wxMBConvStrictUTF8()));
+    // widechar -> UTF-8 conversion isn't supposed to ever fail:
+    wxASSERT_MSG( buf.data, _T("conversion to UTF-8 failed") );
+
+    return buf;

 }
 }

+#endif // wxUSE_UNICODE_UTF8

 
 

-// takes nLength elements of psz starting at nPos
-void wxString::InitWith(const char *psz, size_t nPos, size_t nLength)
+#if wxUSE_UNICODE_UTF8 || !wxUSE_UNICODE
+/* static */
+wxString::SubstrBufFromWC wxString::ConvertStr(const wchar_t *pwz, size_t nLength,
+                                               const wxMBConv& conv)

 {
 {

-  Init();
+    // anything to do?
+    if ( !pwz || nLength == 0 )
+        return SubstrBufFromWC("", 0);

 
 

-  wxASSERT( nPos <= Strlen(psz) );
+    if ( nLength == npos )
+        nLength = wxNO_LEN;

 
 

-  if ( nLength == STRING_MAXLEN )
-    nLength = Strlen(psz + nPos);
+    size_t mbLen;
+    wxCharBuffer mbBuf(conv.cWC2MB(pwz, nLength, &mbLen));
+    if ( !mbLen )
+        return SubstrBufFromWC("", 0);
+    else
+        return SubstrBufFromWC(mbBuf, mbLen);
+}
+#endif // wxUSE_UNICODE_UTF8 || !wxUSE_UNICODE

 
 

-  STATISTICS_ADD(InitialLength, nLength);

 
 

-  if ( nLength > 0 ) {
-    // trailing '\0' is written in AllocBuffer()
-    AllocBuffer(nLength);
-    memcpy(m_pchData, psz + nPos, nLength*sizeof(char));
-  }
+#if wxUSE_UNICODE_WCHAR
+
+//Convert wxString in Unicode mode to a multi-byte string
+const wxCharBuffer wxString::mb_str(const wxMBConv& conv) const
+{
+    return conv.cWC2MB(wx_str(), length() + 1 /* size, not length */, NULL);

 }
 
 }
 

-// the same as previous constructor, but for compilers using unsigned char
-wxString::wxString(const unsigned char* psz, size_t nLength)
+#elif wxUSE_UNICODE_UTF8
+
+const wxWCharBuffer wxString::wc_str() const

 {
 {

-  InitWith((const char *)psz, 0, nLength);
+    return wxMBConvStrictUTF8().cMB2WC
+                                (
+                                    m_impl.c_str(),
+                                    m_impl.length() + 1, // size, not length
+                                    NULL
+                                );

 }
 
 }
 

-#ifdef  STD_STRING_COMPATIBILITY
+const wxCharBuffer wxString::mb_str(const wxMBConv& conv) const
+{
+    if ( conv.IsUTF8() )
+        return wxCharBuffer::CreateNonOwned(m_impl.c_str());
+
+    // FIXME-UTF8: use wc_str() here once we have buffers with length
+
+    size_t wcLen;
+    wxWCharBuffer wcBuf(wxMBConvStrictUTF8().cMB2WC
+                                             (
+                                                m_impl.c_str(),
+                                                m_impl.length() + 1, // size
+                                                &wcLen
+                                             ));
+    if ( !wcLen )
+        return wxCharBuffer("");

 
 

-// poor man's iterators are "void *" pointers
-wxString::wxString(const void *pStart, const void *pEnd)
+    return conv.cWC2MB(wcBuf, wcLen, NULL);
+}
+
+#else // ANSI
+
+//Converts this string to a wide character string if unicode
+//mode is not enabled and wxUSE_WCHAR_T is enabled
+const wxWCharBuffer wxString::wc_str(const wxMBConv& conv) const

 {
 {

-  InitWith((const char *)pStart, 0,
-           (const char *)pEnd - (const char *)pStart);
+    return conv.cMB2WC(wx_str(), length() + 1 /* size, not length */, NULL);

 }
 
 }
 

-#endif  //std::string compatibility
+#endif // Unicode/ANSI
+
+// shrink to minimal size (releasing extra memory)
+bool wxString::Shrink()
+{
+  wxString tmp(begin(), end());
+  swap(tmp);
+  return tmp.length() == length();
+}

 
 

-// from wide string
-wxString::wxString(const wchar_t *pwz)
+// deprecated compatibility code:
+#if WXWIN_COMPATIBILITY_2_8 && !wxUSE_STL_BASED_WXSTRING && !wxUSE_UNICODE_UTF8
+wxStringCharType *wxString::GetWriteBuf(size_t nLen)

 {
 {

-  // first get necessary size
+    return DoGetWriteBuf(nLen);
+}

 
 

-  // NB: GNU libc5 wcstombs() is completely broken, don't use it (it doesn't
-  //     honor the 3rd parameter, thus it will happily crash here).
-#ifdef wxUSE_WCSRTOMBS
-  // don't know if it's really needed (or if we can pass NULL), but better safe
-  // than quick
-  mbstate_t mbstate;  
-  size_t nLen = wcsrtombs((char *) NULL, &pwz, 0, &mbstate);
-#else  // !GNU libc
-  size_t nLen = wcstombs((char *) NULL, pwz, 0);
-#endif // GNU
+void wxString::UngetWriteBuf()
+{
+    DoUngetWriteBuf();
+}

 
 

-  // empty?
-  if ( nLen != 0 ) {
-    AllocBuffer(nLen);
-    wcstombs(m_pchData, pwz, nLen);
-  }
-  else {
-    Init();
-  }
+void wxString::UngetWriteBuf(size_t nLen)
+{
+    DoUngetWriteBuf(nLen);

 }
 }

+#endif // WXWIN_COMPATIBILITY_2_8 && !wxUSE_STL_BASED_WXSTRING && !wxUSE_UNICODE_UTF8
+
+
+// ---------------------------------------------------------------------------
+// data access
+// ---------------------------------------------------------------------------
+
+// all functions are inline in string.h

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

-// memory allocation
+// concatenation operators

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

-// allocates memory needed to store a C string of length nLen
-void wxString::AllocBuffer(size_t nLen)
+/*
+ * concatenation functions come in 5 flavours:
+ *  string + string
+ *  char   + string      and      string + char
+ *  C str  + string      and      string + C str
+ */
+
+wxString operator+(const wxString& str1, const wxString& str2)

 {
 {

-  wxASSERT( nLen >  0         );    //
-  wxASSERT( nLen <= INT_MAX-1 );    // max size (enough room for 1 extra)
+#if !wxUSE_STL_BASED_WXSTRING
+    wxASSERT( str1.IsValid() );
+    wxASSERT( str2.IsValid() );
+#endif

 
 

-  STATISTICS_ADD(Length, nLen);
+    wxString s = str1;
+    s += str2;

 
 

-  // allocate memory:
-  // 1) one extra character for '\0' termination
-  // 2) sizeof(wxStringData) for housekeeping info
-  wxStringData* pData = (wxStringData*)
-    malloc(sizeof(wxStringData) + (nLen + EXTRA_ALLOC + 1)*sizeof(char));
-  pData->nRefs        = 1;
-  pData->nDataLength  = nLen;
-  pData->nAllocLength = nLen + EXTRA_ALLOC;
-  m_pchData           = pData->data();  // data starts after wxStringData
-  m_pchData[nLen]     = '\0';
+    return s;

 }
 
 }
 

-// must be called before changing this string
-void wxString::CopyBeforeWrite()
+wxString operator+(const wxString& str, wxUniChar ch)

 {
 {

-  wxStringData* pData = GetStringData();
+#if !wxUSE_STL_BASED_WXSTRING
+    wxASSERT( str.IsValid() );
+#endif

 
 

-  if ( pData->IsShared() ) {
-    pData->Unlock();                // memory not freed because shared
-    size_t nLen = pData->nDataLength;
-    AllocBuffer(nLen);
-    memcpy(m_pchData, pData->data(), nLen*sizeof(char));
-  }
+    wxString s = str;
+    s += ch;

 
 

-  wxASSERT( !GetStringData()->IsShared() );  // we must be the only owner
+    return s;

 }
 
 }
 

-// must be called before replacing contents of this string
-void wxString::AllocBeforeWrite(size_t nLen)
+wxString operator+(wxUniChar ch, const wxString& str)

 {
 {

-  wxASSERT( nLen != 0 );  // doesn't make any sense
+#if !wxUSE_STL_BASED_WXSTRING
+    wxASSERT( str.IsValid() );
+#endif

 
 

-  // must not share string and must have enough space
-  wxStringData* pData = GetStringData();
-  if ( pData->IsShared() || (nLen > pData->nAllocLength) ) {
-    // can't work with old buffer, get new one
-    pData->Unlock();
-    AllocBuffer(nLen);
-  }
+    wxString s = ch;
+    s += str;

 
 

-  wxASSERT( !GetStringData()->IsShared() );  // we must be the only owner
+    return s;

 }
 
 }
 

-// allocate enough memory for nLen characters
-void wxString::Alloc(size_t nLen)
+wxString operator+(const wxString& str, const char *psz)

 {
 {

-  wxStringData *pData = GetStringData();
-  if ( pData->nAllocLength <= nLen ) {
-    if ( pData->IsEmpty() ) {
-      nLen += EXTRA_ALLOC;
+#if !wxUSE_STL_BASED_WXSTRING
+    wxASSERT( str.IsValid() );
+#endif

 
 

-      wxStringData* pData = (wxStringData*)
-        malloc(sizeof(wxStringData) + (nLen + 1)*sizeof(char));
-      pData->nRefs = 1;
-      pData->nDataLength = 0;
-      pData->nAllocLength = nLen;
-      m_pchData = pData->data();  // data starts after wxStringData
-      m_pchData[0u] = '\0';
-    }
-    else if ( pData->IsShared() ) {
-      pData->Unlock();                // memory not freed because shared
-      size_t nOldLen = pData->nDataLength;
-      AllocBuffer(nLen);
-      memcpy(m_pchData, pData->data(), nOldLen*sizeof(char));
+    wxString s;
+    if ( !s.Alloc(strlen(psz) + str.length()) ) {
+        wxFAIL_MSG( _T("out of memory in wxString::operator+") );

     }
     }

-    else {
-      nLen += EXTRA_ALLOC;
+    s += str;
+    s += psz;

 
 

-      wxStringData *p = (wxStringData *)
-        realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(char));
+    return s;
+}

 
 

-      if ( p == NULL ) {
-        // @@@ what to do on memory error?
-        return;
-      }
+wxString operator+(const wxString& str, const wchar_t *pwz)
+{
+#if !wxUSE_STL_BASED_WXSTRING
+    wxASSERT( str.IsValid() );
+#endif

 
 

-      // it's not important if the pointer changed or not (the check for this
-      // is not faster than assigning to m_pchData in all cases)
-      p->nAllocLength = nLen;
-      m_pchData = p->data();
+    wxString s;
+    if ( !s.Alloc(wxWcslen(pwz) + str.length()) ) {
+        wxFAIL_MSG( _T("out of memory in wxString::operator+") );

     }
     }

-  }
-  //else: we've already got enough
+    s += str;
+    s += pwz;
+
+    return s;

 }
 
 }
 

-// shrink to minimal size (releasing extra memory)
-void wxString::Shrink()
+wxString operator+(const char *psz, const wxString& str)

 {
 {

-  wxStringData *pData = GetStringData();
+#if !wxUSE_STL_BASED_WXSTRING
+    wxASSERT( str.IsValid() );
+#endif
+
+    wxString s;
+    if ( !s.Alloc(strlen(psz) + str.length()) ) {
+        wxFAIL_MSG( _T("out of memory in wxString::operator+") );
+    }
+    s = psz;
+    s += str;

 
 

-  // this variable is unused in release build, so avoid the compiler warning by
-  // just not declaring it
-#ifdef __WXDEBUG__
-  void *p =
+    return s;
+}
+
+wxString operator+(const wchar_t *pwz, const wxString& str)
+{
+#if !wxUSE_STL_BASED_WXSTRING
+    wxASSERT( str.IsValid() );

 #endif
 #endif

-  realloc(pData, sizeof(wxStringData) + (pData->nDataLength + 1)*sizeof(char));

 
 

-  wxASSERT( p != NULL );  // can't free memory?
-  wxASSERT( p == pData ); // we're decrementing the size - block shouldn't move!
+    wxString s;
+    if ( !s.Alloc(wxWcslen(pwz) + str.length()) ) {
+        wxFAIL_MSG( _T("out of memory in wxString::operator+") );
+    }
+    s = pwz;
+    s += str;
+
+    return s;

 }
 
 }
 

-// get the pointer to writable buffer of (at least) nLen bytes
-char *wxString::GetWriteBuf(size_t nLen)
+// ---------------------------------------------------------------------------
+// string comparison
+// ---------------------------------------------------------------------------
+
+bool wxString::IsSameAs(wxUniChar c, bool compareWithCase) const

 {
 {

-  AllocBeforeWrite(nLen);
+    return (length() == 1) && (compareWithCase ? GetChar(0u) == c
+                               : wxToupper(GetChar(0u)) == wxToupper(c));
+}

 
 

-  wxASSERT( GetStringData()->nRefs == 1 );
-  GetStringData()->Validate(FALSE);
+#ifdef HAVE_STD_STRING_COMPARE

 
 

-  return m_pchData;
+// NB: Comparison code (both if HAVE_STD_STRING_COMPARE and if not) works with
+//     UTF-8 encoded strings too, thanks to UTF-8's design which allows us to
+//     sort strings in characters code point order by sorting the byte sequence
+//     in byte values order (i.e. what strcmp() and memcmp() do).
+
+int wxString::compare(const wxString& str) const
+{
+    return m_impl.compare(str.m_impl);

 }
 
 }
 

-// put string back in a reasonable state after GetWriteBuf
-void wxString::UngetWriteBuf()
+int wxString::compare(size_t nStart, size_t nLen,
+                      const wxString& str) const

 {
 {

-  GetStringData()->nDataLength = strlen(m_pchData);
-  GetStringData()->Validate(TRUE);
+    size_t pos, len;
+    PosLenToImpl(nStart, nLen, &pos, &len);
+    return m_impl.compare(pos, len, str.m_impl);

 }
 
 }
 

-// ---------------------------------------------------------------------------
-// data access
-// ---------------------------------------------------------------------------
+int wxString::compare(size_t nStart, size_t nLen,
+                      const wxString& str,
+                      size_t nStart2, size_t nLen2) const
+{
+    size_t pos, len;
+    PosLenToImpl(nStart, nLen, &pos, &len);

 
 

-// all functions are inline in string.h
+    size_t pos2, len2;
+    str.PosLenToImpl(nStart2, nLen2, &pos2, &len2);

 
 

-// ---------------------------------------------------------------------------
-// assignment operators
-// ---------------------------------------------------------------------------
+    return m_impl.compare(pos, len, str.m_impl, pos2, len2);
+}

 
 

-// helper function: does real copy
-void wxString::AssignCopy(size_t nSrcLen, const char *pszSrcData)
+int wxString::compare(const char* sz) const

 {
 {

-  if ( nSrcLen == 0 ) {
-    Reinit();
-  }
-  else {
-    AllocBeforeWrite(nSrcLen);
-    memcpy(m_pchData, pszSrcData, nSrcLen*sizeof(char));
-    GetStringData()->nDataLength = nSrcLen;
-    m_pchData[nSrcLen] = '\0';
-  }
+    return m_impl.compare(ImplStr(sz));
+}
+
+int wxString::compare(const wchar_t* sz) const
+{
+    return m_impl.compare(ImplStr(sz));

 }
 
 }
 

-// assigns one string to another
-wxString& wxString::operator=(const wxString& stringSrc)
+int wxString::compare(size_t nStart, size_t nLen,
+                      const char* sz, size_t nCount) const

 {
 {

-  wxASSERT( stringSrc.GetStringData()->IsValid() );
+    size_t pos, len;
+    PosLenToImpl(nStart, nLen, &pos, &len);
+
+    SubstrBufFromMB str(ImplStr(sz, nCount));
+
+    return m_impl.compare(pos, len, str.data, str.len);
+}
+
+int wxString::compare(size_t nStart, size_t nLen,
+                      const wchar_t* sz, size_t nCount) const
+{
+    size_t pos, len;
+    PosLenToImpl(nStart, nLen, &pos, &len);
+
+    SubstrBufFromWC str(ImplStr(sz, nCount));

 
 

-  // don't copy string over itself
-  if ( m_pchData != stringSrc.m_pchData ) {
-    if ( stringSrc.GetStringData()->IsEmpty() ) {
-      Reinit();
+    return m_impl.compare(pos, len, str.data, str.len);
+}
+
+#else // !HAVE_STD_STRING_COMPARE
+
+static inline int wxDoCmp(const wxStringCharType* s1, size_t l1,
+                          const wxStringCharType* s2, size_t l2)
+{
+    if( l1 == l2 )
+        return wxStringMemcmp(s1, s2, l1);
+    else if( l1 < l2 )
+    {
+        int ret = wxStringMemcmp(s1, s2, l1);
+        return ret == 0 ? -1 : ret;

     }
     }

-    else {
-      // adjust references
-      GetStringData()->Unlock();
-      m_pchData = stringSrc.m_pchData;
-      GetStringData()->Lock();
+    else
+    {
+        int ret = wxStringMemcmp(s1, s2, l2);
+        return ret == 0 ? +1 : ret;

     }
     }

-  }
+}

 
 

-  return *this;
+int wxString::compare(const wxString& str) const
+{
+    return ::wxDoCmp(m_impl.data(), m_impl.length(),
+                     str.m_impl.data(), str.m_impl.length());

 }
 
 }
 

-// assigns a single character
-wxString& wxString::operator=(char ch)
+int wxString::compare(size_t nStart, size_t nLen,
+                      const wxString& str) const

 {
 {

-  AssignCopy(1, &ch);
-  return *this;
+    wxASSERT(nStart <= length());
+    size_type strLen = length() - nStart;
+    nLen = strLen < nLen ? strLen : nLen;
+
+    size_t pos, len;
+    PosLenToImpl(nStart, nLen, &pos, &len);
+
+    return ::wxDoCmp(m_impl.data() + pos,  len,
+                     str.m_impl.data(), str.m_impl.length());

 }
 
 }
 

-// assigns C string
-wxString& wxString::operator=(const char *psz)
+int wxString::compare(size_t nStart, size_t nLen,
+                      const wxString& str,
+                      size_t nStart2, size_t nLen2) const

 {
 {

-  AssignCopy(Strlen(psz), psz);
-  return *this;
+    wxASSERT(nStart <= length());
+    wxASSERT(nStart2 <= str.length());
+    size_type strLen  =     length() - nStart,
+              strLen2 = str.length() - nStart2;
+    nLen  = strLen  < nLen  ? strLen  : nLen;
+    nLen2 = strLen2 < nLen2 ? strLen2 : nLen2;
+
+    size_t pos, len;
+    PosLenToImpl(nStart, nLen, &pos, &len);
+    size_t pos2, len2;
+    str.PosLenToImpl(nStart2, nLen2, &pos2, &len2);
+
+    return ::wxDoCmp(m_impl.data() + pos, len,
+                     str.m_impl.data() + pos2, len2);

 }
 
 }
 

-// same as 'signed char' variant
-wxString& wxString::operator=(const unsigned char* psz)
+int wxString::compare(const char* sz) const

 {
 {

-  *this = (const char *)psz;
-  return *this;
+    SubstrBufFromMB str(ImplStr(sz, npos));
+    if ( str.len == npos )
+        str.len = wxStringStrlen(str.data);
+    return ::wxDoCmp(m_impl.data(), m_impl.length(), str.data, str.len);

 }
 
 }
 

-wxString& wxString::operator=(const wchar_t *pwz)
+int wxString::compare(const wchar_t* sz) const

 {
 {

-  wxString str(pwz);
-  *this = str;
-  return *this;
+    SubstrBufFromWC str(ImplStr(sz, npos));
+    if ( str.len == npos )
+        str.len = wxStringStrlen(str.data);
+    return ::wxDoCmp(m_impl.data(), m_impl.length(), str.data, str.len);
+}
+
+int wxString::compare(size_t nStart, size_t nLen,
+                      const char* sz, size_t nCount) const
+{
+    wxASSERT(nStart <= length());
+    size_type strLen = length() - nStart;
+    nLen = strLen < nLen ? strLen : nLen;
+
+    size_t pos, len;
+    PosLenToImpl(nStart, nLen, &pos, &len);
+
+    SubstrBufFromMB str(ImplStr(sz, nCount));
+    if ( str.len == npos )
+        str.len = wxStringStrlen(str.data);
+
+    return ::wxDoCmp(m_impl.data() + pos, len, str.data, str.len);

 }
 
 }
 

+int wxString::compare(size_t nStart, size_t nLen,
+                      const wchar_t* sz, size_t nCount) const
+{
+    wxASSERT(nStart <= length());
+    size_type strLen = length() - nStart;
+    nLen = strLen < nLen ? strLen : nLen;
+
+    size_t pos, len;
+    PosLenToImpl(nStart, nLen, &pos, &len);
+
+    SubstrBufFromWC str(ImplStr(sz, nCount));
+    if ( str.len == npos )
+        str.len = wxStringStrlen(str.data);
+
+    return ::wxDoCmp(m_impl.data() + pos, len, str.data, str.len);
+}
+
+#endif // HAVE_STD_STRING_COMPARE/!HAVE_STD_STRING_COMPARE
+
+

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

-// string concatenation
+// find_{first,last}_[not]_of functions

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

-// add something to this string
-void wxString::ConcatSelf(int nSrcLen, const char *pszSrcData)
-{
-  STATISTICS_ADD(SummandLength, nSrcLen);
+#if !wxUSE_STL_BASED_WXSTRING || wxUSE_UNICODE_UTF8

 
 

-  // concatenating an empty string is a NOP
-  if ( nSrcLen > 0 ) {
-    wxStringData *pData = GetStringData();
-    size_t nLen = pData->nDataLength;
-    size_t nNewLen = nLen + nSrcLen;
+// NB: All these functions are implemented  with the argument being wxChar*,
+//     i.e. widechar string in any Unicode build, even though native string
+//     representation is char* in the UTF-8 build. This is because we couldn't
+//     use memchr() to determine if a character is in a set encoded as UTF-8.

 
 

-    // alloc new buffer if current is too small
-    if ( pData->IsShared() ) {
-      STATISTICS_ADD(ConcatHit, 0);
+size_t wxString::find_first_of(const wxChar* sz, size_t nStart) const
+{
+    return find_first_of(sz, nStart, wxStrlen(sz));
+}

 
 

-      // we have to allocate another buffer
-      wxStringData* pOldData = GetStringData();
-      AllocBuffer(nNewLen);
-      memcpy(m_pchData, pOldData->data(), nLen*sizeof(char));
-      pOldData->Unlock();
-    }
-    else if ( nNewLen > pData->nAllocLength ) {
-      STATISTICS_ADD(ConcatHit, 0);
+size_t wxString::find_first_not_of(const wxChar* sz, size_t nStart) const
+{
+    return find_first_not_of(sz, nStart, wxStrlen(sz));
+}

 
 

-      // we have to grow the buffer
-      Alloc(nNewLen);
-    }
-    else {
-      STATISTICS_ADD(ConcatHit, 1);
+size_t wxString::find_first_of(const wxChar* sz, size_t nStart, size_t n) const
+{
+    wxASSERT_MSG( nStart <= length(),  _T("invalid index") );

 
 

-      // the buffer is already big enough
+    size_t idx = nStart;
+    for ( const_iterator i = begin() + nStart; i != end(); ++idx, ++i )
+    {
+        if ( wxTmemchr(sz, *i, n) )
+            return idx;

     }
 
     }
 

-    // should be enough space
-    wxASSERT( nNewLen <= GetStringData()->nAllocLength );
+    return npos;
+}
+
+size_t wxString::find_first_not_of(const wxChar* sz, size_t nStart, size_t n) const
+{
+    wxASSERT_MSG( nStart <= length(),  _T("invalid index") );

 
 

-    // fast concatenation - all is done in our buffer
-    memcpy(m_pchData + nLen, pszSrcData, nSrcLen*sizeof(char));
+    size_t idx = nStart;
+    for ( const_iterator i = begin() + nStart; i != end(); ++idx, ++i )
+    {
+        if ( !wxTmemchr(sz, *i, n) )
+            return idx;
+    }

 
 

-    m_pchData[nNewLen] = '\0';              // put terminating '\0'
-    GetStringData()->nDataLength = nNewLen; // and fix the length
-  }
-  //else: the string to append was empty
+    return npos;

 }
 
 }
 

-/*
- * concatenation functions come in 5 flavours:
- *  string + string
- *  char   + string      and      string + char
- *  C str  + string      and      string + C str
- */

 
 

-wxString operator+(const wxString& string1, const wxString& string2)
+size_t wxString::find_last_of(const wxChar* sz, size_t nStart) const

 {
 {

-  wxASSERT( string1.GetStringData()->IsValid() );
-  wxASSERT( string2.GetStringData()->IsValid() );
-
-  wxString s = string1;
-  s += string2;
+    return find_last_of(sz, nStart, wxStrlen(sz));
+}

 
 

-  return s;
+size_t wxString::find_last_not_of(const wxChar* sz, size_t nStart) const
+{
+    return find_last_not_of(sz, nStart, wxStrlen(sz));

 }
 
 }
 

-wxString operator+(const wxString& string, char ch)
+size_t wxString::find_last_of(const wxChar* sz, size_t nStart, size_t n) const

 {
 {

-  wxASSERT( string.GetStringData()->IsValid() );
+    size_t len = length();

 
 

-  wxString s = string;
-  s += ch;
+    if ( nStart == npos )
+    {
+        nStart = len - 1;
+    }
+    else
+    {
+        wxASSERT_MSG( nStart <= len, _T("invalid index") );
+    }

 
 

-  return s;
+    size_t idx = nStart;
+    for ( const_reverse_iterator i = rbegin() + (len - nStart - 1);
+          i != rend(); --idx, ++i )
+    {
+        if ( wxTmemchr(sz, *i, n) )
+            return idx;
+    }
+
+    return npos;

 }
 
 }
 

-wxString operator+(char ch, const wxString& string)
+size_t wxString::find_last_not_of(const wxChar* sz, size_t nStart, size_t n) const

 {
 {

-  wxASSERT( string.GetStringData()->IsValid() );
+    size_t len = length();
+
+    if ( nStart == npos )
+    {
+        nStart = len - 1;
+    }
+    else
+    {
+        wxASSERT_MSG( nStart <= len, _T("invalid index") );
+    }

 
 

-  wxString s = ch;
-  s += string;
+    size_t idx = nStart;
+    for ( const_reverse_iterator i = rbegin() + (len - nStart - 1);
+          i != rend(); --idx, ++i )
+    {
+        if ( !wxTmemchr(sz, *i, n) )
+            return idx;
+    }

 
 

-  return s;
+    return npos;

 }
 
 }
 

-wxString operator+(const wxString& string, const char *psz)
+size_t wxString::find_first_not_of(wxUniChar ch, size_t nStart) const

 {
 {

-  wxASSERT( string.GetStringData()->IsValid() );
+    wxASSERT_MSG( nStart <= length(),  _T("invalid index") );

 
 

-  wxString s;
-  s.Alloc(Strlen(psz) + string.Len());
-  s = string;
-  s += psz;
+    size_t idx = nStart;
+    for ( const_iterator i = begin() + nStart; i != end(); ++idx, ++i )
+    {
+        if ( *i != ch )
+            return idx;
+    }

 
 

-  return s;
+    return npos;

 }
 
 }
 

-wxString operator+(const char *psz, const wxString& string)
+size_t wxString::find_last_not_of(wxUniChar ch, size_t nStart) const

 {
 {

-  wxASSERT( string.GetStringData()->IsValid() );
+    size_t len = length();
+
+    if ( nStart == npos )
+    {
+        nStart = len - 1;
+    }
+    else
+    {
+        wxASSERT_MSG( nStart <= len, _T("invalid index") );
+    }

 
 

-  wxString s;
-  s.Alloc(Strlen(psz) + string.Len());
-  s = psz;
-  s += string;
+    size_t idx = nStart;
+    for ( const_reverse_iterator i = rbegin() + (len - nStart - 1);
+          i != rend(); --idx, ++i )
+    {
+        if ( *i != ch )
+            return idx;
+    }

 
 

-  return s;
+    return npos;

 }
 
 }
 

+// the functions above were implemented for wchar_t* arguments in Unicode
+// build and char* in ANSI build; below are implementations for the other
+// version:
+#if wxUSE_UNICODE
+    #define wxOtherCharType char
+    #define STRCONV         (const wxChar*)wxConvLibc.cMB2WC
+#else
+    #define wxOtherCharType wchar_t
+    #define STRCONV         (const wxChar*)wxConvLibc.cWC2MB
+#endif
+
+size_t wxString::find_first_of(const wxOtherCharType* sz, size_t nStart) const
+    { return find_first_of(STRCONV(sz), nStart); }
+
+size_t wxString::find_first_of(const wxOtherCharType* sz, size_t nStart,
+                               size_t n) const
+    { return find_first_of(STRCONV(sz, n, NULL), nStart, n); }
+size_t wxString::find_last_of(const wxOtherCharType* sz, size_t nStart) const
+    { return find_last_of(STRCONV(sz), nStart); }
+size_t wxString::find_last_of(const wxOtherCharType* sz, size_t nStart,
+                              size_t n) const
+    { return find_last_of(STRCONV(sz, n, NULL), nStart, n); }
+size_t wxString::find_first_not_of(const wxOtherCharType* sz, size_t nStart) const
+    { return find_first_not_of(STRCONV(sz), nStart); }
+size_t wxString::find_first_not_of(const wxOtherCharType* sz, size_t nStart,
+                                   size_t n) const
+    { return find_first_not_of(STRCONV(sz, n, NULL), nStart, n); }
+size_t wxString::find_last_not_of(const wxOtherCharType* sz, size_t nStart) const
+    { return find_last_not_of(STRCONV(sz), nStart); }
+size_t wxString::find_last_not_of(const wxOtherCharType* sz, size_t nStart,
+                                  size_t n) const
+    { return find_last_not_of(STRCONV(sz, n, NULL), nStart, n); }
+
+#undef wxOtherCharType
+#undef STRCONV
+
+#endif // !wxUSE_STL_BASED_WXSTRING || wxUSE_UNICODE_UTF8
+

 // ===========================================================================
 // other common string functions
 // ===========================================================================
 
 // ===========================================================================
 // other common string functions
 // ===========================================================================
 

-// ---------------------------------------------------------------------------
-// simple sub-string extraction
-// ---------------------------------------------------------------------------
+int wxString::CmpNoCase(const wxString& s) const
+{
+    // FIXME-UTF8: use wxUniChar::ToLower/ToUpper once added
+
+    size_t idx = 0;
+    const_iterator i1 = begin();
+    const_iterator end1 = end();
+    const_iterator i2 = s.begin();
+    const_iterator end2 = s.end();
+
+    for ( ; i1 != end1 && i2 != end2; ++idx, ++i1, ++i2 )
+    {
+        wxUniChar lower1 = (wxChar)wxTolower(*i1);
+        wxUniChar lower2 = (wxChar)wxTolower(*i2);
+        if ( lower1 != lower2 )
+            return lower1 < lower2 ? -1 : 1;
+    }
+
+    size_t len1 = length();
+    size_t len2 = s.length();
+
+    if ( len1 < len2 )
+        return -1;
+    else if ( len1 > len2 )
+        return 1;
+    return 0;
+}
+
+
+#if wxUSE_UNICODE

 
 

-// helper function: clone the data attached to this string
-void wxString::AllocCopy(wxString& dest, int nCopyLen, int nCopyIndex) const
+#ifdef __MWERKS__
+#ifndef __SCHAR_MAX__
+#define __SCHAR_MAX__ 127
+#endif
+#endif
+
+wxString wxString::FromAscii(const char *ascii, size_t len)

 {
 {

-  if ( nCopyLen == 0 ) {
-    dest.Init();
-  }
-  else {
-    dest.AllocBuffer(nCopyLen);
-    memcpy(dest.m_pchData, m_pchData + nCopyIndex, nCopyLen*sizeof(char));
-  }
+    if (!ascii || len == 0)
+       return wxEmptyString;
+
+    wxString res;
+
+    {
+        wxImplStringBuffer buf(res, len);
+        wxStringCharType *dest = buf;
+
+        for ( ; len > 0; --len )
+        {
+            unsigned char c = (unsigned char)*ascii++;
+            wxASSERT_MSG( c < 0x80,
+                          _T("Non-ASCII value passed to FromAscii().") );
+
+            *dest++ = (wchar_t)c;
+        }
+    }
+
+    return res;
+}
+
+wxString wxString::FromAscii(const char *ascii)
+{
+    return FromAscii(ascii, wxStrlen(ascii));
+}
+
+wxString wxString::FromAscii(const char ascii)
+{
+    // What do we do with '\0' ?
+
+    unsigned char c = (unsigned char)ascii;
+
+    wxASSERT_MSG( c < 0x80, _T("Non-ASCII value passed to FromAscii().") );
+
+    // NB: the cast to wchar_t causes interpretation of 'ascii' as Latin1 value
+    return wxString(wxUniChar((wchar_t)c));

 }
 
 }
 

+const wxCharBuffer wxString::ToAscii() const
+{
+    // this will allocate enough space for the terminating NUL too
+    wxCharBuffer buffer(length());
+    char *dest = buffer.data();
+
+    for ( const_iterator i = begin(); i != end(); ++i )
+    {
+        wxUniChar c(*i);
+        // FIXME-UTF8: unify substituted char ('_') with wxUniChar ('?')
+        *dest++ = c.IsAscii() ? (char)c : '_';
+
+        // the output string can't have embedded NULs anyhow, so we can safely
+        // stop at first of them even if we do have any
+        if ( !c )
+            break;
+    }
+
+    return buffer;
+}
+
+#endif // wxUSE_UNICODE
+

 // extract string of length nCount starting at nFirst
 wxString wxString::Mid(size_t nFirst, size_t nCount) const
 {
 // extract string of length nCount starting at nFirst
 wxString wxString::Mid(size_t nFirst, size_t nCount) const
 {

-  wxStringData *pData = GetStringData();
-  size_t nLen = pData->nDataLength;
+    size_t nLen = length();

 
 

-  // default value of nCount is STRING_MAXLEN and means "till the end"
-  if ( nCount == STRING_MAXLEN )
-  {
-    nCount = nLen - nFirst;
-  }
+    // default value of nCount is npos and means "till the end"
+    if ( nCount == npos )
+    {
+        nCount = nLen - nFirst;
+    }

 
 

-  // out-of-bounds requests return sensible things
-  if ( nFirst + nCount > nLen )
-  {
-    nCount = nLen - nFirst;
-  }
+    // out-of-bounds requests return sensible things
+    if ( nFirst + nCount > nLen )
+    {
+        nCount = nLen - nFirst;
+    }
+
+    if ( nFirst > nLen )
+    {
+        // AllocCopy() will return empty string
+        return wxEmptyString;
+    }
+
+    wxString dest(*this, nFirst, nCount);
+    if ( dest.length() != nCount )
+    {
+        wxFAIL_MSG( _T("out of memory in wxString::Mid") );
+    }
+
+    return dest;
+}
+
+// check that the string starts with prefix and return the rest of the string
+// in the provided pointer if it is not NULL, otherwise return false
+bool wxString::StartsWith(const wxString& prefix, wxString *rest) const
+{
+    if ( compare(0, prefix.length(), prefix) != 0 )
+        return false;
+
+    if ( rest )
+    {
+        // put the rest of the string into provided pointer
+        rest->assign(*this, prefix.length(), npos);
+    }
+
+    return true;
+}
+
+
+// check that the string ends with suffix and return the rest of it in the
+// provided pointer if it is not NULL, otherwise return false
+bool wxString::EndsWith(const wxString& suffix, wxString *rest) const
+{
+    int start = length() - suffix.length();

 
 

-  if ( nFirst > nLen )
-  {
-    // AllocCopy() will return empty string
-    nCount = 0;
-  }
+    if ( start < 0 || compare(start, npos, suffix) != 0 )
+        return false;

 
 

-  wxString dest;
-  AllocCopy(dest, nCount, nFirst);
+    if ( rest )
+    {
+        // put the rest of the string into provided pointer
+        rest->assign(*this, 0, start);
+    }

 
 

-  return dest;
+    return true;

 }
 
 }
 

+

 // extract nCount last (rightmost) characters
 wxString wxString::Right(size_t nCount) const
 {
 // extract nCount last (rightmost) characters
 wxString wxString::Right(size_t nCount) const
 {

-  if ( nCount > (size_t)GetStringData()->nDataLength )
-    nCount = GetStringData()->nDataLength;
+  if ( nCount > length() )
+    nCount = length();

 
 

-  wxString dest;
-  AllocCopy(dest, nCount, GetStringData()->nDataLength - nCount);
+  wxString dest(*this, length() - nCount, nCount);
+  if ( dest.length() != nCount ) {
+    wxFAIL_MSG( _T("out of memory in wxString::Right") );
+  }

   return dest;
 }
 
 // get all characters after the last occurence of ch
 // (returns the whole string if ch not found)
   return dest;
 }
 
 // get all characters after the last occurence of ch
 // (returns the whole string if ch not found)

-wxString wxString::Right(char ch) const
+wxString wxString::AfterLast(wxUniChar ch) const

 {
   wxString str;
 {
   wxString str;

-  int iPos = Find(ch, TRUE);
-  if ( iPos == NOT_FOUND )
+  int iPos = Find(ch, true);
+  if ( iPos == wxNOT_FOUND )

     str = *this;
   else
     str = *this;
   else

-    str = c_str() + iPos + 1;
+    str = wx_str() + iPos + 1;

 
   return str;
 }
 
   return str;
 }


@@ -641,32 +1140,32 @@ wxString wxString::Right(char ch) const
 // extract nCount first (leftmost) characters
 wxString wxString::Left(size_t nCount) const
 {
 // extract nCount first (leftmost) characters
 wxString wxString::Left(size_t nCount) const
 {

-  if ( nCount > (size_t)GetStringData()->nDataLength )
-    nCount = GetStringData()->nDataLength;
+  if ( nCount > length() )
+    nCount = length();

 
 

-  wxString dest;
-  AllocCopy(dest, nCount, 0);
+  wxString dest(*this, 0, nCount);
+  if ( dest.length() != nCount ) {
+    wxFAIL_MSG( _T("out of memory in wxString::Left") );
+  }

   return dest;
 }
 
 // get all characters before the first occurence of ch
 // (returns the whole string if ch not found)
   return dest;
 }
 
 // get all characters before the first occurence of ch
 // (returns the whole string if ch not found)

-wxString wxString::Left(char ch) const
+wxString wxString::BeforeFirst(wxUniChar ch) const

 {
 {

-  wxString str;
-  for ( const char *pc = m_pchData; *pc != '\0' && *pc != ch; pc++ )
-    str += *pc;
-
-  return str;
+  int iPos = Find(ch);
+  if ( iPos == wxNOT_FOUND ) iPos = length();
+  return wxString(*this, 0, iPos);

 }
 
 /// get all characters before the last occurence of ch
 /// (returns empty string if ch not found)
 }
 
 /// get all characters before the last occurence of ch
 /// (returns empty string if ch not found)

-wxString wxString::Before(char ch) const
+wxString wxString::BeforeLast(wxUniChar ch) const

 {
   wxString str;
 {
   wxString str;

-  int iPos = Find(ch, TRUE);
-  if ( iPos != NOT_FOUND && iPos != 0 )
+  int iPos = Find(ch, true);
+  if ( iPos != wxNOT_FOUND && iPos != 0 )

     str = wxString(c_str(), iPos);
 
   return str;
     str = wxString(c_str(), iPos);
 
   return str;


@@ -674,93 +1173,105 @@ wxString wxString::Before(char ch) const
 
 /// get all characters after the first occurence of ch
 /// (returns empty string if ch not found)
 
 /// get all characters after the first occurence of ch
 /// (returns empty string if ch not found)

-wxString wxString::After(char ch) const
+wxString wxString::AfterFirst(wxUniChar ch) const

 {
   wxString str;
   int iPos = Find(ch);
 {
   wxString str;
   int iPos = Find(ch);

-  if ( iPos != NOT_FOUND )
-    str = c_str() + iPos + 1;
+  if ( iPos != wxNOT_FOUND )
+    str = wx_str() + iPos + 1;

 
   return str;
 }
 
 // replace first (or all) occurences of some substring with another one
 
   return str;
 }
 
 // replace first (or all) occurences of some substring with another one

-size_t wxString::Replace(const char *szOld, const char *szNew, bool bReplaceAll)
+size_t wxString::Replace(const wxString& strOld,
+                         const wxString& strNew, bool bReplaceAll)

 {
 {

-  size_t uiCount = 0;   // count of replacements made
+    // if we tried to replace an empty string we'd enter an infinite loop below
+    wxCHECK_MSG( !strOld.empty(), 0,
+                 _T("wxString::Replace(): invalid parameter") );

 
 

-  size_t uiOldLen = Strlen(szOld);
+    size_t uiCount = 0;   // count of replacements made

 
 

-  wxString strTemp;
-  const char *pCurrent = m_pchData;
-  const char *pSubstr;
-  while ( *pCurrent != '\0' ) {
-    pSubstr = strstr(pCurrent, szOld);
-    if ( pSubstr == NULL ) {
-      // strTemp is unused if no replacements were made, so avoid the copy
-      if ( uiCount == 0 )
-        return 0;
+    size_t uiOldLen = strOld.length();
+    size_t uiNewLen = strNew.length();

 
 

-      strTemp += pCurrent;    // copy the rest
-      break;                  // exit the loop
-    }
-    else {
-      // take chars before match
-      strTemp.ConcatSelf(pSubstr - pCurrent, pCurrent);
-      strTemp += szNew;
-      pCurrent = pSubstr + uiOldLen;  // restart after match
-
-      uiCount++;
-
-      // stop now?
-      if ( !bReplaceAll ) {
-        strTemp += pCurrent;    // copy the rest
-        break;                  // exit the loop
-      }
-    }
-  }
+    size_t dwPos = 0;
+
+    while ( (*this)[dwPos] != wxT('\0') )
+    {
+        //DO NOT USE STRSTR HERE
+        //this string can contain embedded null characters,
+        //so strstr will function incorrectly
+        dwPos = find(strOld, dwPos);
+        if ( dwPos == npos )
+            break;                  // exit the loop
+        else
+        {
+            //replace this occurance of the old string with the new one
+            replace(dwPos, uiOldLen, strNew, uiNewLen);
+
+            //move up pos past the string that was replaced
+            dwPos += uiNewLen;

 
 

-  // only done if there were replacements, otherwise would have returned above
-  *this = strTemp;
+            //increase replace count
+            ++uiCount;
+
+            // stop now?
+            if ( !bReplaceAll )
+                break;                  // exit the loop
+        }
+    }

 
 

-  return uiCount;
+    return uiCount;

 }
 
 bool wxString::IsAscii() const
 {
 }
 
 bool wxString::IsAscii() const
 {

-  const char *s = (const char*) *this;
-  while(*s){
-    if(!isascii(*s)) return(FALSE);
-    s++;
-  }
-  return(TRUE);
+    for ( const_iterator i = begin(); i != end(); ++i )
+    {
+        if ( !(*i).IsAscii() )
+            return false;
+    }
+
+    return true;

 }
 
 bool wxString::IsWord() const
 {
 }
 
 bool wxString::IsWord() const
 {

-  const char *s = (const char*) *this;
-  while(*s){
-    if(!isalpha(*s)) return(FALSE);
-    s++;
-  }
-  return(TRUE);
+    for ( const_iterator i = begin(); i != end(); ++i )
+    {
+        if ( !wxIsalpha(*i) )
+            return false;
+    }
+
+    return true;

 }
 
 bool wxString::IsNumber() const
 {
 }
 
 bool wxString::IsNumber() const
 {

-  const char *s = (const char*) *this;
-  while(*s){
-    if(!isdigit(*s)) return(FALSE);
-    s++;
-  }
-  return(TRUE);
+    if ( empty() )
+        return true;
+
+    const_iterator i = begin();
+
+    if ( *i == _T('-') || *i == _T('+') )
+        ++i;
+
+    for ( ; i != end(); ++i )
+    {
+        if ( !wxIsdigit(*i) )
+            return false;
+    }
+
+    return true;

 }
 
 wxString wxString::Strip(stripType w) const
 {
     wxString s = *this;
 }
 
 wxString wxString::Strip(stripType w) const
 {
     wxString s = *this;

-    if ( w & leading ) s.Trim(FALSE);
-    if ( w & trailing ) s.Trim(TRUE);
+    if ( w & leading ) s.Trim(false);
+    if ( w & trailing ) s.Trim(true);

     return s;
 }
 
     return s;
 }
 


@@ -770,20 +1281,16 @@ wxString wxString::Strip(stripType w) const
 
 wxString& wxString::MakeUpper()
 {
 
 wxString& wxString::MakeUpper()
 {

-  CopyBeforeWrite();
-
-  for ( char *p = m_pchData; *p; p++ )
-    *p = (char)toupper(*p);
+  for ( iterator it = begin(), en = end(); it != en; ++it )
+    *it = (wxChar)wxToupper(*it);

 
   return *this;
 }
 
 wxString& wxString::MakeLower()
 {
 
   return *this;
 }
 
 wxString& wxString::MakeLower()
 {

-  CopyBeforeWrite();
-
-  for ( char *p = m_pchData; *p; p++ )
-    *p = (char)tolower(*p);
+  for ( iterator it = begin(), en = end(); it != en; ++it )
+    *it = (wxChar)wxTolower(*it);

 
   return *this;
 }
 
   return *this;
 }


@@ -792,536 +1299,536 @@ wxString& wxString::MakeLower()
 // trimming and padding
 // ---------------------------------------------------------------------------
 
 // trimming and padding
 // ---------------------------------------------------------------------------
 

+// some compilers (VC++ 6.0 not to name them) return true for a call to
+// isspace('ê') in the C locale which seems to be broken to me, but we have to
+// live with this by checking that the character is a 7 bit one - even if this
+// may fail to detect some spaces (I don't know if Unicode doesn't have
+// space-like symbols somewhere except in the first 128 chars), it is arguably
+// still better than trimming away accented letters
+inline int wxSafeIsspace(wxChar ch) { return (ch < 127) && wxIsspace(ch); }
+

 // trims spaces (in the sense of isspace) from left or right side
 wxString& wxString::Trim(bool bFromRight)
 {
 // trims spaces (in the sense of isspace) from left or right side
 wxString& wxString::Trim(bool bFromRight)
 {

-  // first check if we're going to modify the string at all
-  if ( !IsEmpty() && 
-       ( 
-        (bFromRight && isspace(GetChar(Len() - 1))) ||
-        (!bFromRight && isspace(GetChar(0u)))
+    // first check if we're going to modify the string at all
+    if ( !empty() &&
+         (
+          (bFromRight && wxSafeIsspace(GetChar(length() - 1))) ||
+          (!bFromRight && wxSafeIsspace(GetChar(0u)))
+         )

        )
        )

-     )
-  {
-    // ok, there is at least one space to trim
-    CopyBeforeWrite();
-
-    if ( bFromRight )
-    {
-      // find last non-space character
-      char *psz = m_pchData + GetStringData()->nDataLength - 1;
-      while ( isspace(*psz) && (psz >= m_pchData) )
-        psz--;
-
-      // truncate at trailing space start
-      *++psz = '\0';
-      GetStringData()->nDataLength = psz - m_pchData;
-    }
-    else

     {
     {

-      // find first non-space character
-      const char *psz = m_pchData;
-      while ( isspace(*psz) )
-        psz++;
-
-      // fix up data and length
-      int nDataLength = GetStringData()->nDataLength - (psz - m_pchData);
-      memmove(m_pchData, psz, (nDataLength + 1)*sizeof(char));
-      GetStringData()->nDataLength = nDataLength;
+        if ( bFromRight )
+        {
+            // find last non-space character
+            reverse_iterator psz = rbegin();
+            while ( (psz != rend()) && wxSafeIsspace(*psz) )
+                psz++;
+
+            // truncate at trailing space start
+            erase(psz.base(), end());
+        }
+        else
+        {
+            // find first non-space character
+            iterator psz = begin();
+            while ( (psz != end()) && wxSafeIsspace(*psz) )
+                psz++;
+
+            // fix up data and length
+            erase(begin(), psz);
+        }

     }
     }

-  }

 
 

-  return *this;
+    return *this;

 }
 
 // adds nCount characters chPad to the string from either side
 }
 
 // adds nCount characters chPad to the string from either side

-wxString& wxString::Pad(size_t nCount, char chPad, bool bFromRight)
+wxString& wxString::Pad(size_t nCount, wxUniChar chPad, bool bFromRight)

 {
 {

-  wxString s(chPad, nCount);
+    wxString s(chPad, nCount);

 
 

-  if ( bFromRight )
-    *this += s;
-  else
-  {
-    s += *this;
-    *this = s;
-  }
+    if ( bFromRight )
+        *this += s;
+    else
+    {
+        s += *this;
+        swap(s);
+    }

 
 

-  return *this;
+    return *this;

 }
 
 // truncate the string
 wxString& wxString::Truncate(size_t uiLen)
 {
 }
 
 // truncate the string
 wxString& wxString::Truncate(size_t uiLen)
 {

-  *(m_pchData + uiLen) = '\0';
-  GetStringData()->nDataLength = uiLen;
+    if ( uiLen < length() )
+    {
+        erase(begin() + uiLen, end());
+    }
+    //else: nothing to do, string is already short enough

 
 

-  return *this;
+    return *this;

 }
 
 // ---------------------------------------------------------------------------
 }
 
 // ---------------------------------------------------------------------------

-// finding (return NOT_FOUND if not found and index otherwise)
+// finding (return wxNOT_FOUND if not found and index otherwise)

 // ---------------------------------------------------------------------------
 
 // find a character
 // ---------------------------------------------------------------------------
 
 // find a character

-int wxString::Find(char ch, bool bFromEnd) const
+int wxString::Find(wxUniChar ch, bool bFromEnd) const

 {
 {

-  const char *psz = bFromEnd ? strrchr(m_pchData, ch) : strchr(m_pchData, ch);
+    size_type idx = bFromEnd ? find_last_of(ch) : find_first_of(ch);

 
 

-  return (psz == NULL) ? NOT_FOUND : psz - m_pchData;
+    return (idx == npos) ? wxNOT_FOUND : (int)idx;

 }
 
 }
 

-// find a sub-string (like strstr)
-int wxString::Find(const char *pszSub) const
-{
-  const char *psz = strstr(m_pchData, pszSub);
+// ----------------------------------------------------------------------------
+// conversion to numbers
+// ----------------------------------------------------------------------------

 
 

-  return (psz == NULL) ? NOT_FOUND : psz - m_pchData;
-}
+// The implementation of all the functions below is exactly the same so factor
+// it out. Note that number extraction works correctly on UTF-8 strings, so
+// we can use wxStringCharType and wx_str() for maximum efficiency.

 
 

-// ---------------------------------------------------------------------------
-// formatted output
-// ---------------------------------------------------------------------------
-int wxString::Printf(const char *pszFormat, ...)
+template <typename T>
+bool wxStringToIntType(const wxStringCharType *start,
+                       T *val,
+                       int base,
+                       T (*func)(const wxStringCharType*, wxStringCharType**, int))

 {
 {

-  va_list argptr;
-  va_start(argptr, pszFormat);
+    wxCHECK_MSG( val, false, _T("NULL output pointer") );
+    wxASSERT_MSG( !base || (base > 1 && base <= 36), _T("invalid base") );

 
 

-  int iLen = PrintfV(pszFormat, argptr);
+#ifndef __WXWINCE__
+    errno = 0;
+#endif

 
 

-  va_end(argptr);
+    wxStringCharType *end;
+    *val = (*func)(start, &end, base);

 
 

-  return iLen;
+    // return true only if scan was stopped by the terminating NUL and if the
+    // string was not empty to start with and no under/overflow occurred
+    return !*end && (end != start)
+#ifndef __WXWINCE__
+        && (errno != ERANGE)
+#endif
+    ;

 }
 
 }
 

-int wxString::PrintfV(const char* pszFormat, va_list argptr)
+bool wxString::ToLong(long *val, int base) const

 {
 {

-  static char s_szScratch[1024];
-
-  int iLen = vsprintf(s_szScratch, pszFormat, argptr);
-  AllocBeforeWrite(iLen);
-  strcpy(m_pchData, s_szScratch);
-
-  return iLen;
+    return wxStringToIntType(wx_str(), val, base, wxStrtol);

 }
 
 }
 

-// ----------------------------------------------------------------------------
-// misc other operations
-// ----------------------------------------------------------------------------
-bool wxString::Matches(const char *pszMask) const
+bool wxString::ToULong(unsigned long *val, int base) const

 {
 {

-  // check char by char
-  const char *pszTxt;
-  for ( pszTxt = c_str(); *pszMask != '\0'; pszMask++, pszTxt++ ) {
-    switch ( *pszMask ) {
-      case '?':
-        if ( *pszTxt == '\0' )
-          return FALSE;
-
-        pszTxt++;
-        pszMask++;
-        break;
-
-      case '*':
-        {
-          // ignore special chars immediately following this one
-          while ( *pszMask == '*' || *pszMask == '?' )
-            pszMask++;
-
-          // if there is nothing more, match
-          if ( *pszMask == '\0' )
-            return TRUE;
+    return wxStringToIntType(wx_str(), val, base, wxStrtoul);
+}

 
 

-          // are there any other metacharacters in the mask?
-          size_t uiLenMask;
-          const char *pEndMask = strpbrk(pszMask, "*?");
+bool wxString::ToLongLong(wxLongLong_t *val, int base) const
+{
+    return wxStringToIntType(wx_str(), val, base, wxStrtoll);
+}

 
 

-          if ( pEndMask != NULL ) {
-            // we have to match the string between two metachars
-            uiLenMask = pEndMask - pszMask;
-          }
-          else {
-            // we have to match the remainder of the string
-            uiLenMask = strlen(pszMask);
-          }
+bool wxString::ToULongLong(wxULongLong_t *val, int base) const
+{
+    return wxStringToIntType(wx_str(), val, base, wxStrtoull);
+}

 
 

-          wxString strToMatch(pszMask, uiLenMask);
-          const char* pMatch = strstr(pszTxt, strToMatch);
-          if ( pMatch == NULL )
-            return FALSE;
+bool wxString::ToDouble(double *val) const
+{
+    wxCHECK_MSG( val, false, _T("NULL pointer in wxString::ToDouble") );

 
 

-          // -1 to compensate "++" in the loop
-          pszTxt = pMatch + uiLenMask - 1;
-          pszMask += uiLenMask - 1;
-        }
-        break;
+#ifndef __WXWINCE__
+    errno = 0;
+#endif

 
 

-      default:
-        if ( *pszMask != *pszTxt )
-          return FALSE;
-        break;
-    }
-  }
+    const wxChar *start = c_str();
+    wxChar *end;
+    *val = wxStrtod(start, &end);

 
 

-  // match only if nothing left
-  return *pszTxt == '\0';
+    // return true only if scan was stopped by the terminating NUL and if the
+    // string was not empty to start with and no under/overflow occurred
+    return !*end && (end != start)
+#ifndef __WXWINCE__
+        && (errno != ERANGE)
+#endif
+    ;

 }
 
 // ---------------------------------------------------------------------------
 }
 
 // ---------------------------------------------------------------------------

-// standard C++ library string functions
+// formatted output

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

-#ifdef  STD_STRING_COMPATIBILITY
-
-wxString& wxString::insert(size_t nPos, const wxString& str)
-{
-  wxASSERT( str.GetStringData()->IsValid() );
-  wxASSERT( nPos <= Len() );
-
-  if ( !str.IsEmpty() ) {
-    wxString strTmp;
-    char *pc = strTmp.GetWriteBuf(Len() + str.Len());
-    strncpy(pc, c_str(), nPos);
-    strcpy(pc + nPos, str);
-    strcpy(pc + nPos + str.Len(), c_str() + nPos);
-    strTmp.UngetWriteBuf();
-    *this = strTmp;
-  }
-
-  return *this;
-}

 
 

-size_t wxString::find(const wxString& str, size_t nStart) const
+#if !wxUSE_UTF8_LOCALE_ONLY
+/* static */
+#ifdef wxNEEDS_WXSTRING_PRINTF_MIXIN
+wxString wxStringPrintfMixinBase::DoFormatWchar(const wxChar *format, ...)
+#else
+wxString wxString::DoFormatWchar(const wxChar *format, ...)
+#endif

 {
 {

-  wxASSERT( str.GetStringData()->IsValid() );
-  wxASSERT( nStart <= Len() );
+    va_list argptr;
+    va_start(argptr, format);

 
 

-  const char *p = strstr(c_str() + nStart, str);
+    wxString s;
+    s.PrintfV(format, argptr);

 
 

-  return p == NULL ? npos : p - c_str();
-}
+    va_end(argptr);

 
 

-// VC++ 1.5 can't cope with the default argument in the header.
-#if ! (defined(_MSC_VER) && !defined(__WIN32__))
-size_t wxString::find(const char* sz, size_t nStart, size_t n) const
-{
-  return find(wxString(sz, n == npos ? 0 : n), nStart);
+    return s;

 }
 }

-#endif
+#endif // !wxUSE_UTF8_LOCALE_ONLY

 
 

-size_t wxString::find(char ch, size_t nStart) const
+#if wxUSE_UNICODE_UTF8
+/* static */
+wxString wxString::DoFormatUtf8(const char *format, ...)

 {
 {

-  wxASSERT( nStart <= Len() );
-
-  const char *p = strchr(c_str() + nStart, ch);
-
-  return p == NULL ? npos : p - c_str();
-}
+    va_list argptr;
+    va_start(argptr, format);

 
 

-size_t wxString::rfind(const wxString& str, size_t nStart) const
-{
-  wxASSERT( str.GetStringData()->IsValid() );
-  wxASSERT( nStart <= Len() );
+    wxString s;
+    s.PrintfV(format, argptr);

 
 

-  // # could be quicker than that
-  const char *p = c_str() + (nStart == npos ? Len() : nStart);
-  while ( p >= c_str() + str.Len() ) {
-    if ( strncmp(p - str.Len(), str, str.Len()) == 0 )
-      return p - str.Len() - c_str();
-    p--;
-  }
+    va_end(argptr);

 
 

-  return npos;
+    return s;

 }
 }

+#endif // wxUSE_UNICODE_UTF8

 
 

-// VC++ 1.5 can't cope with the default argument in the header.
-#if ! (defined(_MSC_VER) && !defined(__WIN32__))
-size_t wxString::rfind(const char* sz, size_t nStart, size_t n) const
+/* static */
+wxString wxString::FormatV(const wxString& format, va_list argptr)

 {
 {

-  return rfind(wxString(sz, n == npos ? 0 : n), nStart);
+    wxString s;
+    s.PrintfV(format, argptr);
+    return s;

 }
 
 }
 

-size_t wxString::rfind(char ch, size_t nStart) const
-{
-  wxASSERT( nStart <= Len() );
-
-  const char *p = strrchr(c_str() + nStart, ch);
-
-  return p == NULL ? npos : p - c_str();
-}
+#if !wxUSE_UTF8_LOCALE_ONLY
+#ifdef wxNEEDS_WXSTRING_PRINTF_MIXIN
+int wxStringPrintfMixinBase::DoPrintfWchar(const wxChar *format, ...)
+#else
+int wxString::DoPrintfWchar(const wxChar *format, ...)

 #endif
 #endif

-
-wxString wxString::substr(size_t nStart, size_t nLen) const

 {
 {

-  // npos means 'take all'
-  if ( nLen == npos )
-    nLen = 0;
-
-  wxASSERT( nStart + nLen <= Len() );
+    va_list argptr;
+    va_start(argptr, format);

 
 

-  return wxString(c_str() + nStart, nLen == npos ? 0 : nLen);
-}
+#ifdef wxNEEDS_WXSTRING_PRINTF_MIXIN
+    // get a pointer to the wxString instance; we have to use dynamic_cast<>
+    // because it's the only cast that works safely for downcasting when
+    // multiple inheritance is used:
+    wxString *str = static_cast<wxString*>(this);
+#else
+    wxString *str = this;
+#endif

 
 

-wxString& wxString::erase(size_t nStart, size_t nLen)
-{
-  wxString strTmp(c_str(), nStart);
-  if ( nLen != npos ) {
-    wxASSERT( nStart + nLen <= Len() );
+    int iLen = str->PrintfV(format, argptr);

 
 

-    strTmp.append(c_str() + nStart + nLen);
-  }
+    va_end(argptr);

 
 

-  *this = strTmp;
-  return *this;
+    return iLen;

 }
 }

+#endif // !wxUSE_UTF8_LOCALE_ONLY

 
 

-wxString& wxString::replace(size_t nStart, size_t nLen, const char *sz)
+#if wxUSE_UNICODE_UTF8
+int wxString::DoPrintfUtf8(const char *format, ...)

 {
 {

-  wxASSERT( nStart + nLen <= Strlen(sz) );
+    va_list argptr;
+    va_start(argptr, format);

 
 

-  wxString strTmp;
-  if ( nStart != 0 )
-    strTmp.append(c_str(), nStart);
-  strTmp += sz;
-  strTmp.append(c_str() + nStart + nLen);
+    int iLen = PrintfV(format, argptr);

 
 

-  *this = strTmp;
-  return *this;
-}
+    va_end(argptr);

 
 

-wxString& wxString::replace(size_t nStart, size_t nLen, size_t nCount, char ch)
-{
-  return replace(nStart, nLen, wxString(ch, nCount));
+    return iLen;

 }
 }

+#endif // wxUSE_UNICODE_UTF8

 
 

-wxString& wxString::replace(size_t nStart, size_t nLen,
-                            const wxString& str, size_t nStart2, size_t nLen2)
+#if wxUSE_UNICODE_UTF8
+template<typename BufferType>
+#else
+// we only need one version in non-UTF8 builds and at least two Windows
+// compilers have problems with this function template, so use just one
+// normal function here
+#endif
+static int DoStringPrintfV(wxString& str,
+                           const wxString& format, va_list argptr)

 {
 {

-  return replace(nStart, nLen, str.substr(nStart2, nLen2));
-}
+    int size = 1024;

 
 

-wxString& wxString::replace(size_t nStart, size_t nLen,
-                        const char* sz, size_t nCount)
-{
-  return replace(nStart, nLen, wxString(sz, nCount));
-}
+    for ( ;; )
+    {
+#if wxUSE_UNICODE_UTF8
+        BufferType tmp(str, size + 1);
+        typename BufferType::CharType *buf = tmp;
+#else
+        wxStringBuffer tmp(str, size + 1);
+        wxChar *buf = tmp;
+#endif

 
 

-#endif  //std::string compatibility
+        if ( !buf )
+        {
+            // out of memory

 
 

-// ============================================================================
-// ArrayString
-// ============================================================================
+            // in UTF-8 build, leaving uninitialized junk in the buffer
+            // could result in invalid non-empty UTF-8 string, so just
+            // reset the string to empty on failure:
+            buf[0] = '\0';
+            return -1;
+        }

 
 

-// size increment = max(50% of current size, ARRAY_MAXSIZE_INCREMENT)
-#define   ARRAY_MAXSIZE_INCREMENT       4096
-#ifndef   ARRAY_DEFAULT_INITIAL_SIZE    // also defined in dynarray.h
-  #define   ARRAY_DEFAULT_INITIAL_SIZE    (16)
+        // wxVsnprintf() may modify the original arg pointer, so pass it
+        // only a copy
+        va_list argptrcopy;
+        wxVaCopy(argptrcopy, argptr);
+        int len = wxVsnprintf(buf, size, format, argptrcopy);
+        va_end(argptrcopy);
+
+        // some implementations of vsnprintf() don't NUL terminate
+        // the string if there is not enough space for it so
+        // always do it manually
+        buf[size] = _T('\0');
+
+        // vsnprintf() may return either -1 (traditional Unix behaviour) or the
+        // total number of characters which would have been written if the
+        // buffer were large enough (newer standards such as Unix98)
+        if ( len < 0 )
+        {
+            // NB: wxVsnprintf() may call either wxCRT_VsnprintfW or
+            //     wxCRT_VsnprintfA in UTF-8 build; wxUSE_WXVSNPRINTF
+            //     is true if *both* of them use our own implementation,
+            //     otherwise we can't be sure
+#if wxUSE_WXVSNPRINTF
+            // we know that our own implementation of wxVsnprintf() returns -1
+            // only for a format error - thus there's something wrong with
+            // the user's format string
+            buf[0] = '\0';
+            return -1;
+#else // possibly using system version
+            // assume it only returns error if there is not enough space, but
+            // as we don't know how much we need, double the current size of
+            // the buffer
+            size *= 2;
+#endif // wxUSE_WXVSNPRINTF/!wxUSE_WXVSNPRINTF
+        }
+        else if ( len >= size )
+        {
+#if wxUSE_WXVSNPRINTF
+            // we know that our own implementation of wxVsnprintf() returns 
+            // size+1 when there's not enough space but that's not the size
+            // of the required buffer!
+            size *= 2;      // so we just double the current size of the buffer
+#else
+            // some vsnprintf() implementations NUL-terminate the buffer and
+            // some don't in len == size case, to be safe always add 1
+            size = len + 1;

 #endif
 #endif

+        }
+        else // ok, there was enough space
+        {
+            break;
+        }
+    }

 
 

-#define   STRING(p)   ((wxString *)(&(p)))
+    // we could have overshot
+    str.Shrink();

 
 

-// ctor
-wxArrayString::wxArrayString()
-{
-  m_nSize  =
-  m_nCount = 0;
-  m_pItems = (char **) NULL;
+    return str.length();

 }
 
 }
 

-// copy ctor
-wxArrayString::wxArrayString(const wxArrayString& src)
+int wxString::PrintfV(const wxString& format, va_list argptr)

 {
 {

-  m_nSize  =
-  m_nCount = 0;
-  m_pItems = (char **) NULL;
+#if wxUSE_UNICODE_UTF8
+    #if wxUSE_STL_BASED_WXSTRING
+        typedef wxStringTypeBuffer<char> Utf8Buffer;
+    #else
+        typedef wxImplStringBuffer Utf8Buffer;
+    #endif
+#endif

 
 

-  *this = src;
+#if wxUSE_UTF8_LOCALE_ONLY
+    return DoStringPrintfV<Utf8Buffer>(*this, format, argptr);
+#else
+    #if wxUSE_UNICODE_UTF8
+    if ( wxLocaleIsUtf8 )
+        return DoStringPrintfV<Utf8Buffer>(*this, format, argptr);
+    else
+        // wxChar* version
+        return DoStringPrintfV<wxStringBuffer>(*this, format, argptr);
+    #else
+        return DoStringPrintfV(*this, format, argptr);
+    #endif // UTF8/WCHAR
+#endif

 }
 
 }
 

-// assignment operator
-wxArrayString& wxArrayString::operator=(const wxArrayString& src)
-{
-  if ( m_nSize > 0 )
-    Clear();
-
-  if ( src.m_nCount > ARRAY_DEFAULT_INITIAL_SIZE )
-    Alloc(src.m_nCount);
-
-  // we can't just copy the pointers here because otherwise we would share
-  // the strings with another array
-  for ( size_t n = 0; n < src.m_nCount; n++ )
-    Add(src[n]);
-
-  if ( m_nCount != 0 )
-    memcpy(m_pItems, src.m_pItems, m_nCount*sizeof(char *));
+// ----------------------------------------------------------------------------
+// misc other operations
+// ----------------------------------------------------------------------------

 
 

-  return *this;
-}
+// returns true if the string matches the pattern which may contain '*' and
+// '?' metacharacters (as usual, '?' matches any character and '*' any number
+// of them)
+bool wxString::Matches(const wxString& mask) const
+{
+    // I disable this code as it doesn't seem to be faster (in fact, it seems
+    // to be much slower) than the old, hand-written code below and using it
+    // here requires always linking with libregex even if the user code doesn't
+    // use it
+#if 0 // wxUSE_REGEX
+    // first translate the shell-like mask into a regex
+    wxString pattern;
+    pattern.reserve(wxStrlen(pszMask));
+
+    pattern += _T('^');
+    while ( *pszMask )
+    {
+        switch ( *pszMask )
+        {
+            case _T('?'):
+                pattern += _T('.');
+                break;
+
+            case _T('*'):
+                pattern += _T(".*");
+                break;
+
+            case _T('^'):
+            case _T('.'):
+            case _T('$'):
+            case _T('('):
+            case _T(')'):
+            case _T('|'):
+            case _T('+'):
+            case _T('\\'):
+                // these characters are special in a RE, quote them
+                // (however note that we don't quote '[' and ']' to allow
+                // using them for Unix shell like matching)
+                pattern += _T('\\');
+                // fall through
+
+            default:
+                pattern += *pszMask;
+        }

 
 

-// grow the array
-void wxArrayString::Grow()
-{
-  // only do it if no more place
-  if( m_nCount == m_nSize ) {
-    if( m_nSize == 0 ) {
-      // was empty, alloc some memory
-      m_nSize = ARRAY_DEFAULT_INITIAL_SIZE;
-      m_pItems = new char *[m_nSize];
-    }
-    else {
-      // otherwise when it's called for the first time, nIncrement would be 0
-      // and the array would never be expanded
-      wxASSERT( ARRAY_DEFAULT_INITIAL_SIZE != 0 );
-
-      // add 50% but not too much
-      size_t nIncrement = m_nSize < ARRAY_DEFAULT_INITIAL_SIZE
-                          ? ARRAY_DEFAULT_INITIAL_SIZE : m_nSize >> 1;
-      if ( nIncrement > ARRAY_MAXSIZE_INCREMENT )
-        nIncrement = ARRAY_MAXSIZE_INCREMENT;
-      m_nSize += nIncrement;
-      char **pNew = new char *[m_nSize];
-
-      // copy data to new location
-      memcpy(pNew, m_pItems, m_nCount*sizeof(char *));
-
-      // delete old memory (but do not release the strings!)
-      wxDELETEA(m_pItems);
-
-      m_pItems = pNew;
+        pszMask++;

     }
     }

-  }
-}
+    pattern += _T('$');
+
+    // and now use it
+    return wxRegEx(pattern, wxRE_NOSUB | wxRE_EXTENDED).Matches(c_str());
+#else // !wxUSE_REGEX
+  // TODO: this is, of course, awfully inefficient...
+
+  // FIXME-UTF8: implement using iterators, remove #if
+#if wxUSE_UNICODE_UTF8
+  wxWCharBuffer maskBuf = mask.wc_str();
+  wxWCharBuffer txtBuf = wc_str();
+  const wxChar *pszMask = maskBuf.data();
+  const wxChar *pszTxt = txtBuf.data();
+#else
+  const wxChar *pszMask = mask.wx_str();
+  // the char currently being checked
+  const wxChar *pszTxt = wx_str();
+#endif

 
 

-void wxArrayString::Free()
-{
-  for ( size_t n = 0; n < m_nCount; n++ ) {
-    STRING(m_pItems[n])->GetStringData()->Unlock();
-  }
-}
+  // the last location where '*' matched
+  const wxChar *pszLastStarInText = NULL;
+  const wxChar *pszLastStarInMask = NULL;

 
 

-// deletes all the strings from the list
-void wxArrayString::Empty()
-{
-  Free();
+match:
+  for ( ; *pszMask != wxT('\0'); pszMask++, pszTxt++ ) {
+    switch ( *pszMask ) {
+      case wxT('?'):
+        if ( *pszTxt == wxT('\0') )
+          return false;

 
 

-  m_nCount = 0;
-}
+        // pszTxt and pszMask will be incremented in the loop statement

 
 

-// as Empty, but also frees memory
-void wxArrayString::Clear()
-{
-  Free();
+        break;

 
 

-  m_nSize  =
-  m_nCount = 0;
+      case wxT('*'):
+        {
+          // remember where we started to be able to backtrack later
+          pszLastStarInText = pszTxt;
+          pszLastStarInMask = pszMask;

 
 

-  wxDELETEA(m_pItems);
-}
+          // ignore special chars immediately following this one
+          // (should this be an error?)
+          while ( *pszMask == wxT('*') || *pszMask == wxT('?') )
+            pszMask++;

 
 

-// dtor
-wxArrayString::~wxArrayString()
-{
-  Free();
+          // if there is nothing more, match
+          if ( *pszMask == wxT('\0') )
+            return true;

 
 

-  wxDELETEA(m_pItems);
-}
+          // are there any other metacharacters in the mask?
+          size_t uiLenMask;
+          const wxChar *pEndMask = wxStrpbrk(pszMask, wxT("*?"));

 
 

-// pre-allocates memory (frees the previous data!)
-void wxArrayString::Alloc(size_t nSize)
-{
-  wxASSERT( nSize > 0 );
+          if ( pEndMask != NULL ) {
+            // we have to match the string between two metachars
+            uiLenMask = pEndMask - pszMask;
+          }
+          else {
+            // we have to match the remainder of the string
+            uiLenMask = wxStrlen(pszMask);
+          }

 
 

-  // only if old buffer was not big enough
-  if ( nSize > m_nSize ) {
-    Free();
-    wxDELETEA(m_pItems);
-    m_pItems = new char *[nSize];
-    m_nSize  = nSize;
-  }
+          wxString strToMatch(pszMask, uiLenMask);
+          const wxChar* pMatch = wxStrstr(pszTxt, strToMatch);
+          if ( pMatch == NULL )
+            return false;

 
 

-  m_nCount = 0;
-}
+          // -1 to compensate "++" in the loop
+          pszTxt = pMatch + uiLenMask - 1;
+          pszMask += uiLenMask - 1;
+        }
+        break;

 
 

-// searches the array for an item (forward or backwards)
-int wxArrayString::Index(const char *sz, bool bCase, bool bFromEnd) const
-{
-  if ( bFromEnd ) {
-    if ( m_nCount > 0 ) {
-      size_t ui = m_nCount;
-      do {
-        if ( STRING(m_pItems[--ui])->IsSameAs(sz, bCase) )
-          return ui;
-      }
-      while ( ui != 0 );
-    }
-  }
-  else {
-    for( size_t ui = 0; ui < m_nCount; ui++ ) {
-      if( STRING(m_pItems[ui])->IsSameAs(sz, bCase) )
-        return ui;
+      default:
+        if ( *pszMask != *pszTxt )
+          return false;
+        break;

     }
   }
 
     }
   }
 

-  return NOT_FOUND;
-}
-
-// add item at the end
-void wxArrayString::Add(const wxString& str)
-{
-  wxASSERT( str.GetStringData()->IsValid() );
-
-  Grow();
-
-  // the string data must not be deleted!
-  str.GetStringData()->Lock();
-  m_pItems[m_nCount++] = (char *)str.c_str();
-}
-
-// add item at the given position
-void wxArrayString::Insert(const wxString& str, size_t nIndex)
-{
-  wxASSERT( str.GetStringData()->IsValid() );
-
-  wxCHECK_RET( nIndex <= m_nCount, ("bad index in wxArrayString::Insert") );
-
-  Grow();
-
-  memmove(&m_pItems[nIndex + 1], &m_pItems[nIndex],
-          (m_nCount - nIndex)*sizeof(char *));
+  // match only if nothing left
+  if ( *pszTxt == wxT('\0') )
+    return true;

 
 

-  str.GetStringData()->Lock();
-  m_pItems[nIndex] = (char *)str.c_str();
+  // if we failed to match, backtrack if we can
+  if ( pszLastStarInText ) {
+    pszTxt = pszLastStarInText + 1;
+    pszMask = pszLastStarInMask;

 
 

-  m_nCount++;
-}
+    pszLastStarInText = NULL;

 
 

-// removes item from array (by index)
-void wxArrayString::Remove(size_t nIndex)
-{
-  wxCHECK_RET( nIndex <= m_nCount, _("bad index in wxArrayString::Remove") );
+    // don't bother resetting pszLastStarInMask, it's unnecessary

 
 

-  // release our lock
-  Item(nIndex).GetStringData()->Unlock();
+    goto match;
+  }

 
 

-  memmove(&m_pItems[nIndex], &m_pItems[nIndex + 1],
-          (m_nCount - nIndex - 1)*sizeof(char *));
-  m_nCount--;
+  return false;
+#endif // wxUSE_REGEX/!wxUSE_REGEX

 }
 
 }
 

-// removes item from array (by value)
-void wxArrayString::Remove(const char *sz)
+// Count the number of chars
+int wxString::Freq(wxUniChar ch) const

 {
 {

-  int iIndex = Index(sz);
-
-  wxCHECK_RET( iIndex != NOT_FOUND,
-               _("removing inexistent element in wxArrayString::Remove") );
-
-  Remove(iIndex);
+    int count = 0;
+    for ( const_iterator i = begin(); i != end(); ++i )
+    {
+        if ( *i == ch )
+            count ++;
+    }
+    return count;

 }
 
 }
 

-// sort array elements using passed comparaison function
+// convert to upper case, return the copy of the string
+wxString wxString::Upper() const
+{ wxString s(*this); return s.MakeUpper(); }

 
 

-void wxArrayString::Sort(bool WXUNUSED(bCase), bool WXUNUSED(bReverse) )
-{
-  //@@@@ TO DO
-  //qsort(m_pItems, m_nCount, sizeof(char *), fCmp);
-}
+// convert to lower case, return the copy of the string
+wxString wxString::Lower() const { wxString s(*this); return s.MakeLower(); }