Reimplement wxUTF8StringBuffer correctly and more efficiently

This buffer class can avoid copying strings entirely in UTF-8 build, as
it can write directly to the storage provided by the underlying
std::string, and it also needs to keep the contents in UTF-8, instead of
converting it from the current locale encoding, which was at best
useless, and ensure that it is correct, which is necessary with at least
MSVC, as its CRT can return invalid UTF-8 strings even when using this
encoding.

This finally fixes "PrintfError" unit test under MSW.

include/wx/string.h

@@ -3557,6 +3557,8 @@ private:
 
   friend class WXDLLIMPEXP_FWD_BASE wxStringIteratorNode;
   friend class WXDLLIMPEXP_FWD_BASE wxUniCharRef;
+  friend class wxUTF8StringBuffer;
+  friend class wxUTF8StringBufferLength;
 #endif // wxUSE_UNICODE_UTF8
 
   friend class WXDLLIMPEXP_FWD_BASE wxCStrData;
@@ -3802,8 +3804,89 @@ typedef wxStringInternalBufferLength          wxStringBufferLength;
 #endif // wxUSE_UNICODE_UTF8/wxUSE_UNICODE_WCHAR
 
 #if wxUSE_UNICODE_UTF8
-typedef wxStringInternalBuffer                wxUTF8StringBuffer;
-typedef wxStringInternalBufferLength          wxUTF8StringBufferLength;
+
+// Special implementation of buffer classes for UTF-8 build which exploit the
+// fact that we can write directly to std::string used by wxString, avoiding an
+// extra copy which could be significant for long strings.
+
+namespace wxPrivate
+{
+
+class wxUTF8StringBufferBase
+{
+public:
+    using CharType = char;
+
+    ~wxUTF8StringBufferBase()
+    {
+        // This class works only with UTF-8 strings, so we need to check if the
+        // string has valid contents. Note that it isn't an error if it
+        // doesn't, as it can happen that the function we use this buffer with
+        // (e.g. vsnprintf()) writes something invalid into the provided buffer
+        // in some cases.
+        if ( !wxStringOperations::IsValidUtf8String(m_str.c_str()) )
+            m_str.clear();
+    }
+
+    operator char*() const { return const_cast<char*>(m_str.c_str()); }
+
+protected:
+    explicit wxUTF8StringBufferBase(std::string& str, size_t size)
+        : m_str{str}
+    {
+        m_str.resize(size);
+    }
+
+    std::string& m_str;
+
+    wxDECLARE_NO_COPY_CLASS(wxUTF8StringBufferBase);
+};
+
+} // wxPrivate
+
+class wxUTF8StringBuffer : public wxPrivate::wxUTF8StringBufferBase
+{
+public:
+    wxUTF8StringBuffer(wxString& str, size_t size)
+        : wxPrivate::wxUTF8StringBufferBase{str.m_impl, size}
+    {
+    }
+
+    ~wxUTF8StringBuffer()
+    {
+        // This class works only with NUL-terminated strings, so we need to
+        // resize the string to have the correct length.
+        m_str.resize(strlen(m_str.c_str()));
+    }
+
+private:
+    wxDECLARE_NO_COPY_CLASS(wxUTF8StringBuffer);
+};
+
+class wxUTF8StringBufferLength : public wxPrivate::wxUTF8StringBufferBase
+{
+public:
+    wxUTF8StringBufferLength(wxString& str, size_t size)
+        : wxPrivate::wxUTF8StringBufferBase{str.m_impl, size}
+    {
+    }
+
+    ~wxUTF8StringBufferLength()
+    {
+        wxASSERT_MSG( m_lenSet, "forgot to call SetLength()" );
+
+        m_str.resize(m_len);
+    }
+
+    void SetLength(size_t length) { m_len = length; m_lenSet = true; }
+
+protected:
+    size_t m_len = 0;
+    bool m_lenSet = false;
+
+    wxDECLARE_NO_COPY_CLASS(wxUTF8StringBufferLength);
+};
+
 #else // wxUSE_UNICODE_WCHAR
 
 // Note about inlined dtors in the classes below: this is done not for

src/common/string.cpp

@@ -1899,16 +1899,12 @@ static int DoStringPrintfV(wxString& str,
 
 int wxString::PrintfV(const wxString& format, va_list argptr)
 {
-#if wxUSE_UNICODE_UTF8
-    typedef wxStringTypeBuffer<char> Utf8Buffer;
-#endif
-
 #if wxUSE_UTF8_LOCALE_ONLY
-    return DoStringPrintfV<Utf8Buffer>(*this, format, argptr);
+    return DoStringPrintfV<wxUTF8StringBuffer>(*this, format, argptr);
 #else
     #if wxUSE_UNICODE_UTF8
     if ( wxLocaleIsUtf8 )
-        return DoStringPrintfV<Utf8Buffer>(*this, format, argptr);
+        return DoStringPrintfV<wxUTF8StringBuffer>(*this, format, argptr);
     else
         // wxChar* version
         return DoStringPrintfV<wxStringBuffer>(*this, format, argptr);