try to deal more gracefully (than simply not showing anything at all) with invalid UTF-8 strings (e.g. ISO-8859-1 strings inserted in controls with default (UTF-8) font

git-svn-id: https://svn.wxwidgets.org/svn/wx/wxWidgets/trunk@40800 c3d73ce0-8a6f-49c7-b76d-6d57e0e08775

src/gtk/textctrl.cpp

@@ -846,7 +846,9 @@ void wxTextCtrl::SetValue( const wxString &value )
         const wxCharBuffer buffer(wxGTK_CONV_ENC(value, GetTextEncoding()));
         if ( !buffer )
         {
-            // what else can we do? at least don't crash...
+            // see comment in WriteText() as to why we must warn the user about
+            // this
+            wxLogWarning(_("Failed to set text in the text control."));
             return;
         }
 

src/gtk/utilsgtk.cpp

@@ -172,12 +172,35 @@ wxWindow* wxFindWindowAtPoint(const wxPoint& pt)
 
 wxCharBuffer wxConvertToGTK(const wxString& s, wxFontEncoding enc)
 {
+    wxCharBuffer buf;
     if ( enc == wxFONTENCODING_UTF8 )
     {
-        // no need for conversion at all
+        // no need for conversion at all, but do check that we have a valid
-        return wxCharBuffer(s);
+        // UTF-8 string because passing invalid UTF-8 to GTK+ is going to
-    }
+        // result in a GTK+ error message and, especially, loss of data which
+        // was supposed to be shown in the GUI
+        if ( wxConvUTF8.ToWChar(NULL, 0, s, s.length()) == wxCONV_FAILED )
+        {
+            // warn the programmer that something is probably wrong in his code
+            //
+            // NB: don't include the string in output because chances are that
+            //     this invalid UTF-8 string could result in more errors itself
+            //     if the application shows logs in the GUI and so we get into
+            //     an infinite loop
+            wxLogDebug(_T("Invalid UTF-8 string in wxConvertToGTK()"));
 
+            // but still try to show at least something on the screen
+            wxMBConvUTF8 utf8permissive(wxMBConvUTF8::MAP_INVALID_UTF8_TO_OCTAL);
+            wxWCharBuffer wbuf(utf8permissive.cMB2WC(s));
+            buf = wxConvUTF8.cWC2MB(wbuf);
+        }
+        else // valid UTF-8 string, no need to convert
+        {
+            buf = wxCharBuffer(s);
+        }
+    }
+    else // !UTF-8
+    {
         wxWCharBuffer wbuf;
         if ( enc == wxFONTENCODING_SYSTEM || enc == wxFONTENCODING_DEFAULT )
         {
@@ -188,9 +211,9 @@ wxCharBuffer wxConvertToGTK(const wxString& s, wxFontEncoding enc)
             wbuf = wxCSConv(enc).cMB2WC(s);
         }
 
-    wxCharBuffer buf;
         if ( wbuf )
             buf = wxConvUTF8.cWC2MB(wbuf);
+    }
 
     return buf;
 }