X-Git-Url: https://git.saurik.com/wxWidgets.git/blobdiff_plain/e542e141a3c26522c053ee3cfc72314e6109a227..b50d93d1dae385443f0ca4a57736cb5a57a3e7fe:/include/wx/string.h?ds=sidebyside

diff --git a/include/wx/string.h b/include/wx/string.h
index 1dc72edeb7..9ab7f3f10a 100644
--- a/include/wx/string.h
+++ b/include/wx/string.h
@@ -209,7 +209,7 @@ inline int Stricmp(const char *psz1, const char *psz2)
 
 // Lightweight object returned by wxString::c_str() and implicitly convertible
 // to either const char* or const wchar_t*.
-class WXDLLIMPEXP_BASE wxCStrData
+class wxCStrData
 {
 private:
     // Ctors; for internal use by wxString and wxCStrData only
@@ -246,8 +246,15 @@ public:
     operator const void*() const { return AsChar(); }
 
     // returns buffers that are valid as long as the associated wxString exists
-    inline const wxScopedCharBuffer AsCharBuf() const;
-    inline const wxScopedWCharBuffer AsWCharBuf() const;
+    const wxScopedCharBuffer AsCharBuf() const
+    {
+        return wxScopedCharBuffer::CreateNonOwned(AsChar());
+    }
+
+    const wxScopedWCharBuffer AsWCharBuf() const
+    {
+        return wxScopedWCharBuffer::CreateNonOwned(AsWChar());
+    }
 
     inline wxString AsString() const;
 
@@ -281,7 +288,7 @@ public:
     wxCStrData operator-(ptrdiff_t n) const
     {
         wxASSERT_MSG( n <= (ptrdiff_t)m_offset,
-                      _T("attempt to construct address before the beginning of the string") );
+                      wxT("attempt to construct address before the beginning of the string") );
         return wxCStrData(m_str, m_offset - n, m_owned);
     }
 
@@ -401,6 +408,7 @@ protected:
 #ifdef wxNEEDS_WXSTRING_PRINTF_MIXIN
     // "non dll-interface class 'wxStringPrintfMixin' used as base interface
     // for dll-interface class 'wxString'" -- this is OK in our case
+    #pragma warning (push)
     #pragma warning (disable:4275)
 #endif
 
@@ -997,8 +1005,8 @@ public:
         { return iterator(str(), wxStringOperations::AddToIter(m_cur, -n)); }
 
   private:
-      iterator(wxString *str, underlying_iterator ptr)
-          : m_cur(ptr), m_node(str, &m_cur) {}
+      iterator(wxString *wxstr, underlying_iterator ptr)
+          : m_cur(ptr), m_node(wxstr, &m_cur) {}
 
       wxString* str() const { return const_cast<wxString*>(m_node.m_str); }
 
@@ -1042,8 +1050,8 @@ public:
 
   private:
       // for internal wxString use only:
-      const_iterator(const wxString *str, underlying_iterator ptr)
-          : m_cur(ptr), m_node(str, &m_cur) {}
+      const_iterator(const wxString *wxstr, underlying_iterator ptr)
+          : m_cur(ptr), m_node(wxstr, &m_cur) {}
 
       const wxString* str() const { return m_node.m_str; }
 
@@ -1277,9 +1285,9 @@ public:
     { assign(pwz, nLength); }
 
   wxString(const wxScopedCharBuffer& buf)
-    { assign(buf.data()); } // FIXME-UTF8: fix for embedded NUL and buffer length
+    { assign(buf.data(), buf.length()); }
   wxString(const wxScopedWCharBuffer& buf)
-    { assign(buf.data()); } // FIXME-UTF8: fix for embedded NUL and buffer length
+    { assign(buf.data(), buf.length()); }
 
     // NB: this version uses m_impl.c_str() to force making a copy of the
     //     string, so that "wxString(str.c_str())" idiom for passing strings
@@ -1341,8 +1349,10 @@ public:
   #else
     // wxStringImpl is either not std::string or needs conversion
     operator wxStdWideString() const
-        // FIXME-UTF8: broken for embedded NULs
-        { return wxStdWideString(wc_str()); }
+    {
+        wxScopedWCharBuffer buf(wc_str());
+        return wxStdWideString(buf.data(), buf.length());
+    }
   #endif
 
   #if (!wxUSE_UNICODE || wxUSE_UTF8_LOCALE_ONLY) && wxUSE_STL_BASED_WXSTRING
@@ -1351,8 +1361,10 @@ public:
   #else
     // wxStringImpl is either not std::string or needs conversion
     operator std::string() const
-        // FIXME-UTF8: broken for embedded NULs
-        { return std::string(mb_str()); }
+    {
+        wxScopedCharBuffer buf(mb_str());
+        return std::string(buf.data(), buf.length());
+    }
   #endif
 #endif // wxUSE_STL
 
@@ -1501,14 +1513,14 @@ public:
     // get last character
     wxUniChar Last() const
     {
-      wxASSERT_MSG( !empty(), _T("wxString: index out of bounds") );
+      wxASSERT_MSG( !empty(), wxT("wxString: index out of bounds") );
       return *rbegin();
     }
 
     // get writable last character
     wxUniCharRef Last()
     {
-      wxASSERT_MSG( !empty(), _T("wxString: index out of bounds") );
+      wxASSERT_MSG( !empty(), wxT("wxString: index out of bounds") );
       return *rbegin();
     }
 
@@ -1706,8 +1718,8 @@ public:
         return FromImpl(wxStringImpl(utf8, len));
     }
 
-    const char* utf8_str() const { return wx_str(); }
-    const char* ToUTF8() const { return wx_str(); }
+    const wxScopedCharBuffer utf8_str() const
+        { return wxCharBuffer::CreateNonOwned(m_impl.c_str(), m_impl.length()); }
 
     // this function exists in UTF-8 build only and returns the length of the
     // internal UTF-8 representation
@@ -1723,7 +1735,6 @@ public:
         return s;
     }
     const wxScopedCharBuffer utf8_str() const { return mb_str(wxMBConvUTF8()); }
-    const wxScopedCharBuffer ToUTF8() const { return utf8_str(); }
 #else // ANSI
     static wxString FromUTF8(const char *utf8)
       { return wxString(wxMBConvUTF8().cMB2WC(utf8)); }
@@ -1752,9 +1763,10 @@ public:
     }
     const wxScopedCharBuffer utf8_str() const
       { return wxMBConvUTF8().cWC2MB(wc_str()); }
-    const wxScopedCharBuffer ToUTF8() const { return utf8_str(); }
 #endif
 
+    const wxScopedCharBuffer ToUTF8() const { return utf8_str(); }
+
     // functions for storing binary data in wxString:
 #if wxUSE_UNICODE
     static wxString From8BitData(const char *data, size_t len)
@@ -1782,21 +1794,34 @@ public:
     // accepting the file names. The return value is always the same, but the
     // type differs because a function may either return pointer to the buffer
     // directly or have to use intermediate buffer for translation.
+
 #if wxUSE_UNICODE
 
+    // this is an optimization: even though using mb_str(wxConvLibc) does the
+    // same thing (i.e. returns pointer to internal representation as locale is
+    // always an UTF-8 one) in wxUSE_UTF8_LOCALE_ONLY case, we can avoid the
+    // extra checks and the temporary buffer construction by providing a
+    // separate mb_str() overload
 #if wxUSE_UTF8_LOCALE_ONLY
     const char* mb_str() const { return wx_str(); }
-    const wxScopedCharBuffer mb_str(const wxMBConv& conv) const;
-#else
-    const wxScopedCharBuffer mb_str(const wxMBConv& conv = wxConvLibc) const;
-#endif
+    const wxScopedCharBuffer mb_str(const wxMBConv& conv) const
+    {
+        return AsCharBuf(conv);
+    }
+#else // !wxUSE_UTF8_LOCALE_ONLY
+    const wxScopedCharBuffer mb_str(const wxMBConv& conv = wxConvLibc) const
+    {
+        return AsCharBuf(conv);
+    }
+#endif // wxUSE_UTF8_LOCALE_ONLY/!wxUSE_UTF8_LOCALE_ONLY
 
     const wxWX2MBbuf mbc_str() const { return mb_str(*wxConvCurrent); }
 
 #if wxUSE_UNICODE_WCHAR
     const wchar_t* wc_str() const { return wx_str(); }
 #elif wxUSE_UNICODE_UTF8
-    const wxScopedWCharBuffer wc_str() const;
+    const wxScopedWCharBuffer wc_str() const
+        { return AsWCharBuf(wxMBConvStrictUTF8()); }
 #endif
     // for compatibility with !wxUSE_UNICODE version
     const wxWX2WCbuf wc_str(const wxMBConv& WXUNUSED(conv)) const
@@ -1809,16 +1834,16 @@ public:
 #endif // wxMBFILES/!wxMBFILES
 
 #else // ANSI
-    const wxChar* mb_str() const { return wx_str(); }
+    const char* mb_str() const { return wx_str(); }
 
     // for compatibility with wxUSE_UNICODE version
     const char* mb_str(const wxMBConv& WXUNUSED(conv)) const { return wx_str(); }
 
     const wxWX2MBbuf mbc_str() const { return mb_str(); }
 
-#if wxUSE_WCHAR_T
-    const wxScopedWCharBuffer wc_str(const wxMBConv& conv = wxConvLibc) const;
-#endif // wxUSE_WCHAR_T
+    const wxScopedWCharBuffer wc_str(const wxMBConv& conv = wxConvLibc) const
+        { return AsWCharBuf(conv); }
+
     const wxScopedCharBuffer fn_str() const
         { return wxConvFile.cWC2WX( wc_str( wxConvLibc ) ); }
 #endif // Unicode/ANSI
@@ -1921,10 +1946,10 @@ public:
 
     // from wxScopedWCharBuffer
   wxString& operator=(const wxScopedWCharBuffer& s)
-    { return operator=(s.data()); } // FIXME-UTF8: fix for embedded NULs
+    { return assign(s); }
     // from wxScopedCharBuffer
   wxString& operator=(const wxScopedCharBuffer& s)
-    { return operator=(s.data()); } // FIXME-UTF8: fix for embedded NULs
+    { return assign(s); }
 
   // string concatenation
     // in place concatenation
@@ -1938,7 +1963,7 @@ public:
   {
 #if WXWIN_COMPATIBILITY_2_8 && !wxUSE_STL_BASED_WXSTRING && !wxUSE_UNICODE_UTF8
     wxASSERT_MSG( s.IsValid(),
-                  _T("did you forget to call UngetWriteBuf()?") );
+                  wxT("did you forget to call UngetWriteBuf()?") );
 #endif
 
     append(s);
@@ -1960,9 +1985,9 @@ public:
 
       // string += buffer (i.e. from wxGetString)
   wxString& operator<<(const wxScopedWCharBuffer& s)
-    { return operator<<((const wchar_t *)s); }
+    { return append(s); }
   wxString& operator<<(const wxScopedCharBuffer& s)
-    { return operator<<((const char *)s); }
+    { return append(s); }
 
     // string += C string
   wxString& Append(const wxString& s)
@@ -2032,36 +2057,36 @@ public:
   // stream-like functions
       // insert an int into string
   wxString& operator<<(int i)
-    { return (*this) << Format(_T("%d"), i); }
+    { return (*this) << Format(wxT("%d"), i); }
       // insert an unsigned int into string
   wxString& operator<<(unsigned int ui)
-    { return (*this) << Format(_T("%u"), ui); }
+    { return (*this) << Format(wxT("%u"), ui); }
       // insert a long into string
   wxString& operator<<(long l)
-    { return (*this) << Format(_T("%ld"), l); }
+    { return (*this) << Format(wxT("%ld"), l); }
       // insert an unsigned long into string
   wxString& operator<<(unsigned long ul)
-    { return (*this) << Format(_T("%lu"), ul); }
+    { return (*this) << Format(wxT("%lu"), ul); }
 #if defined wxLongLong_t && !defined wxLongLongIsLong
       // insert a long long if they exist and aren't longs
   wxString& operator<<(wxLongLong_t ll)
     {
-      const wxChar *fmt = _T("%") wxLongLongFmtSpec _T("d");
+      const wxChar *fmt = wxT("%") wxLongLongFmtSpec wxT("d");
       return (*this) << Format(fmt, ll);
     }
       // insert an unsigned long long
   wxString& operator<<(wxULongLong_t ull)
     {
-      const wxChar *fmt = _T("%") wxLongLongFmtSpec _T("u");
+      const wxChar *fmt = wxT("%") wxLongLongFmtSpec wxT("u");
       return (*this) << Format(fmt , ull);
     }
 #endif // wxLongLong_t && !wxLongLongIsLong
       // insert a float into string
   wxString& operator<<(float f)
-    { return (*this) << Format(_T("%f"), f); }
+    { return (*this) << Format(wxT("%f"), f); }
       // insert a double into string
   wxString& operator<<(double d)
-    { return (*this) << Format(_T("%g"), d); }
+    { return (*this) << Format(wxT("%g"), d); }
 
   // string comparison
     // case-sensitive comparison (returns a value < 0, = 0 or > 0)
@@ -2248,7 +2273,7 @@ public:
       // convert to a double
   bool ToCDouble(double *val) const;
 #endif
-  
+
 #ifndef wxNEEDS_WXSTRING_PRINTF_MIXIN
   // formatted input/output
     // as sprintf(), returns the number of characters written or < 0 on error
@@ -2403,7 +2428,7 @@ public:
                CreateConstIterator(last).impl())
   {
       wxASSERT_MSG( first.m_str == last.m_str,
-                    _T("pointers must be into the same string") );
+                    wxT("pointers must be into the same string") );
   }
 #endif // WXWIN_COMPATIBILITY_STRING_PTR_AS_ITER
 
@@ -2464,9 +2489,9 @@ public:
   wxString& append(const wxCStrData& str)
     { return append(str.AsString()); }
   wxString& append(const wxScopedCharBuffer& str)
-    { return append(str.data()); }
+    { return append(str.data(), str.length()); }
   wxString& append(const wxScopedWCharBuffer& str)
-    { return append(str.data()); }
+    { return append(str.data(), str.length()); }
   wxString& append(const wxCStrData& str, size_t n)
     { return append(str.AsString(), 0, n); }
   wxString& append(const wxScopedCharBuffer& str, size_t n)
@@ -2596,9 +2621,9 @@ public:
   wxString& assign(const wxCStrData& str)
     { return assign(str.AsString()); }
   wxString& assign(const wxScopedCharBuffer& str)
-    { return assign(str.data()); }
+    { return assign(str.data(), str.length()); }
   wxString& assign(const wxScopedWCharBuffer& str)
-    { return assign(str.data()); }
+    { return assign(str.data(), str.length()); }
   wxString& assign(const wxCStrData& str, size_t len)
     { return assign(str.AsString(), len); }
   wxString& assign(const wxScopedCharBuffer& str, size_t len)
@@ -3352,9 +3377,9 @@ public:
       return *this;
   }
   wxString& operator+=(const wxScopedCharBuffer& s)
-    { return operator+=(s.data()); }
+    { return append(s); }
   wxString& operator+=(const wxScopedWCharBuffer& s)
-    { return operator+=(s.data()); }
+    { return append(s); }
       // string += char
   wxString& operator+=(wxUniChar ch)
   {
@@ -3417,36 +3442,117 @@ private:
   wxStringImpl m_impl;
 
   // buffers for compatibility conversion from (char*)c_str() and
-  // (wchar_t*)c_str():
-  // FIXME-UTF8: bechmark various approaches to keeping compatibility buffers
+  // (wchar_t*)c_str(): the pointers returned by these functions should remain
+  // valid until the string itself is modified for compatibility with the
+  // existing code and consistency with std::string::c_str() so returning a
+  // temporary buffer won't do and we need to cache the conversion results
+
+  // TODO-UTF8: benchmark various approaches to keeping compatibility buffers
   template<typename T>
   struct ConvertedBuffer
   {
-      ConvertedBuffer() : m_buf(NULL) {}
+      // notice that there is no need to initialize m_len here as it's unused
+      // as long as m_str is NULL
+      ConvertedBuffer() : m_str(NULL) {}
       ~ConvertedBuffer()
-          { free(m_buf); }
+          { free(m_str); }
 
-      operator T*() const { return m_buf; }
+      bool Extend(size_t len)
+      {
+          // add extra 1 for the trailing NUL
+          void * const str = realloc(m_str, sizeof(T)*(len + 1));
+          if ( !str )
+              return false;
+
+          m_str = static_cast<T *>(str);
+          m_len = len;
 
-      ConvertedBuffer& operator=(T *str)
+          return true;
+      }
+
+      const wxScopedCharTypeBuffer<T> AsScopedBuffer() const
       {
-          free(m_buf);
-          m_buf = str;
-          return *this;
+          return wxScopedCharTypeBuffer<T>::CreateNonOwned(m_str, m_len);
       }
 
-      T *m_buf;
+      T *m_str;     // pointer to the string data
+      size_t m_len; // length, not size, i.e. in chars and without last NUL
   };
-#if wxUSE_UNICODE && !wxUSE_UTF8_LOCALE_ONLY
+
+
+#if wxUSE_UNICODE
+  // common mb_str() and wxCStrData::AsChar() helper: performs the conversion
+  // and returns either m_convertedToChar.m_str (in which case its m_len is
+  // also updated) or NULL if it failed
+  //
+  // there is an important exception: in wxUSE_UNICODE_UTF8 build if conv is a
+  // UTF-8 one, we return m_impl.c_str() directly, without doing any conversion
+  // as optimization and so the caller needs to check for this before using
+  // m_convertedToChar
+  //
+  // NB: AsChar() returns char* in any build, unlike mb_str()
+  const char *AsChar(const wxMBConv& conv) const;
+
+  // mb_str() implementation helper
+  wxScopedCharBuffer AsCharBuf(const wxMBConv& conv) const
+  {
+#if wxUSE_UNICODE_UTF8
+      // avoid conversion if we can
+      if ( conv.IsUTF8() )
+      {
+          return wxScopedCharBuffer::CreateNonOwned(m_impl.c_str(),
+                  m_impl.length());
+      }
+#endif // wxUSE_UNICODE_UTF8
+
+      // call this solely in order to fill in m_convertedToChar as AsChar()
+      // updates it as a side effect: this is a bit ugly but it's a completely
+      // internal function so the users of this class shouldn't care or know
+      // about it and doing it like this, i.e. having a separate AsChar(),
+      // allows us to avoid the creation and destruction of a temporary buffer
+      // when using wxCStrData without duplicating any code
+      if ( !AsChar(conv) )
+      {
+          // although it would be probably more correct to return NULL buffer
+          // from here if the conversion fails, a lot of existing code doesn't
+          // expect mb_str() (or wc_str()) to ever return NULL so return an
+          // empty string otherwise to avoid crashes in it
+          //
+          // also, some existing code does check for the conversion success and
+          // so asserting here would be bad too -- even if it does mean that
+          // silently losing data is possible for badly written code
+          return wxScopedCharBuffer::CreateNonOwned("", 0);
+      }
+
+      return m_convertedToChar.AsScopedBuffer();
+  }
+
   ConvertedBuffer<char> m_convertedToChar;
-#endif
+#endif // !wxUSE_UNICODE
+
 #if !wxUSE_UNICODE_WCHAR
+  // common wc_str() and wxCStrData::AsWChar() helper for both UTF-8 and ANSI
+  // builds: converts the string contents into m_convertedToWChar and returns
+  // NULL if the conversion failed (this can only happen in ANSI build)
+  //
+  // NB: AsWChar() returns wchar_t* in any build, unlike wc_str()
+  const wchar_t *AsWChar(const wxMBConv& conv) const;
+
+  // wc_str() implementation helper
+  wxScopedWCharBuffer AsWCharBuf(const wxMBConv& conv) const
+  {
+      if ( !AsWChar(conv) )
+          return wxScopedWCharBuffer::CreateNonOwned(L"", 0);
+
+      return m_convertedToWChar.AsScopedBuffer();
+  }
+
   ConvertedBuffer<wchar_t> m_convertedToWChar;
-#endif
+#endif // !wxUSE_UNICODE_WCHAR
 
 #if wxUSE_UNICODE_UTF8
   // FIXME-UTF8: (try to) move this elsewhere (TLS) or solve differently
-  //             assigning to character pointer to by wxString::interator may
+  //             assigning to character pointer to by wxString::iterator may
   //             change the underlying wxStringImpl iterator, so we have to
   //             keep track of all iterators and update them as necessary:
   struct wxStringIteratorNodeHead
@@ -3471,7 +3577,7 @@ private:
 };
 
 #ifdef wxNEEDS_WXSTRING_PRINTF_MIXIN
-    #pragma warning (default:4275)
+    #pragma warning (pop)
 #endif
 
 // string iterator operators that satisfy STL Random Access Iterator
@@ -3990,45 +4096,53 @@ inline wxCStrData::~wxCStrData()
         delete const_cast<wxString*>(m_str); // cast to silence warnings
 }
 
-// simple cases for AsChar() and AsWChar(), the complicated ones are
-// in string.cpp
-#if wxUSE_UNICODE_WCHAR
+// AsChar() and AsWChar() implementations simply forward to wxString methods
+
 inline const wchar_t* wxCStrData::AsWChar() const
 {
-    return m_str->wx_str() + m_offset;
-}
-#endif // wxUSE_UNICODE_WCHAR
+    const wchar_t * const p =
+#if wxUSE_UNICODE_WCHAR
+        m_str->wc_str();
+#elif wxUSE_UNICODE_UTF8
+        m_str->AsWChar(wxMBConvStrictUTF8());
+#else
+        m_str->AsWChar(wxConvLibc);
+#endif
 
+    // in Unicode build the string always has a valid Unicode representation
+    // and even if a conversion is needed (as in UTF8 case) it can't fail
+    //
+    // but in ANSI build the string contents might be not convertible to
+    // Unicode using the current locale encoding so we do need to check for
+    // errors
 #if !wxUSE_UNICODE
-inline const char* wxCStrData::AsChar() const
-{
-    return m_str->wx_str() + m_offset;
-}
+    if ( !p )
+    {
+        // if conversion fails, return empty string and not NULL to avoid
+        // crashes in code written with either wxWidgets 2 wxString or
+        // std::string behaviour in mind: neither of them ever returns NULL
+        // from its c_str() and so we shouldn't neither
+        //
+        // notice that the same is done in AsChar() below and
+        // wxString::wc_str() and mb_str() for the same reasons
+        return L"";
+    }
 #endif // !wxUSE_UNICODE
 
-#if wxUSE_UTF8_LOCALE_ONLY
-inline const char* wxCStrData::AsChar() const
-{
-    return wxStringOperations::AddToIter(m_str->wx_str(), m_offset);
-}
-#endif // wxUSE_UTF8_LOCALE_ONLY
-
-inline const wxScopedCharBuffer wxCStrData::AsCharBuf() const
-{
-#if !wxUSE_UNICODE || wxUSE_UTF8_LOCALE_ONLY
-    return wxScopedCharBuffer::CreateNonOwned(AsChar());
-#else
-    return AsString().mb_str();
-#endif
+    return p + m_offset;
 }
 
-inline const wxScopedWCharBuffer wxCStrData::AsWCharBuf() const
+inline const char* wxCStrData::AsChar() const
 {
-#if wxUSE_UNICODE_WCHAR
-    return wxScopedWCharBuffer::CreateNonOwned(AsWChar());
-#else
-    return AsString().wc_str();
-#endif
+#if wxUSE_UNICODE && !wxUSE_UTF8_LOCALE_ONLY
+    const char * const p = m_str->AsChar(wxConvLibc);
+    if ( !p )
+        return "";
+#else // !wxUSE_UNICODE || wxUSE_UTF8_LOCALE_ONLY
+    const char * const p = m_str->mb_str();
+#endif // wxUSE_UNICODE && !wxUSE_UTF8_LOCALE_ONLY
+
+    return p + m_offset;
 }
 
 inline wxString wxCStrData::AsString() const
@@ -4051,7 +4165,7 @@ inline const wxStringCharType *wxCStrData::AsInternal() const
 inline wxUniChar wxCStrData::operator*() const
 {
     if ( m_str->empty() )
-        return wxUniChar(_T('\0'));
+        return wxUniChar(wxT('\0'));
     else
         return (*m_str)[m_offset];
 }