Firefox Multiple Write Problem using custom Response Filter - Invalid Viewstate

I implemented a whitespace filter based on the following article: I've discovered a problem with Firefox when the response size requires multiple calls to the Filter's Write method. On the rendered page, Firefox inserts a non-standard character between the two Write() outputs. In particular, this often happens in the middle of the Viewstate variable which causes the dreaded System.Web.HttpException: Invalid_Viewstate error. Here's an excerpt from such a problem viewstate:


except that the "_" between the C and Z is a single character. If I paste it in here directly, it renders as the following: &#0 or what may be an ascii null. In firefox, it appears as the diamond question mark to indicate a bad charset character.

This only happens in Firefox. Can someone explain to me why the null character is being inserted and how to correct for this problem?

Thanks in advance.

From WhitespaceFilter Class:
   Public Overrides Sub Write( _
         ByVal MyBuffer() As Byte, _
         ByVal offset As Integer, _
         ByVal count As Integer)
      Static bFirstPass As Boolean = True
      Static bHTML As Boolean = True
      ' See if Filter is Enabled
      Dim sFilter As Object = HttpRuntime.Cache("AppFilterWhiteSpace")
      If sFilter Is Nothing Then
         Dim sSQL As String = _
               "SELECT GR0_Value FROM GenericRef WHERE GR0_Type = 'System' AND GR0_Key = 'FilterWhiteSpace'"
         sFilter = CNull(DBInfo.ExecuteScalar(sSQL), "False")
         HttpRuntime.Cache.Insert("AppFilterWhiteSpace", sFilter, Nothing, _
               DateTime.Now.AddMinutes(15), TimeSpan.Zero)
      End If
      Dim data(count) As Byte
      Buffer.BlockCopy(MyBuffer, offset, data, 0, count)
      ' Process Filter 
      If sFilter <> "True" Then
         ' Use Output As Is 
         _sink.Write(data, 0, data.GetLength(0))
         ' Filter Leading Whitespace
         ' Don't use ASCII encoding here.  The .NET IDE replaces
         ' some characters, such as ®
         ' with a UTF-8 entity.  If you use ASCII encoding,
         ' you'll get B. instead of the registered
         ' trademark symbol.
         Dim s As String = System.Text.Encoding.UTF8.GetString(data)
         If bFirstPass Then
            bFirstPass = False
            bHTML = s.ToUpper.IndexOf("<HTML>") > -1
         End If
         If Not bHTML Then
            ' Use Output As Is 
            _sink.Write(data, 0, data.GetLength(0))
            ' Remove Whitespace
            ' Break string into separate lines
            ' - each line will end with prior line's carriage return
            ' - each line will begin with leading whitespace that we want to remove
            Dim aLine() As String = s.Split(vbLf)
            For iLine As Integer = 0 To aLine.GetUpperBound(0)
               aLine(iLine) = aLine(iLine).TrimStart()
            ' Rejoin the lines
            s = String.Join(vbLf, aLine)
            ' Finally, we spit out what we have done.
            Dim outdata() As Byte = System.Text.Encoding.UTF8.GetBytes(s)
            _sink.Write(outdata, 0, outdata.GetLength(0))
        End If
      End If
   End Sub
From Global.asax.vb
    Sub Application_BeginRequest(ByVal sender As Object, ByVal e As EventArgs)
      Response.Filter = New WhitespaceFilter(Response.Filter)
    End Sub

Open in new window

Who is Participating?
alexpercsiConnect With a Mentor Commented:
Just a stab in the dark:

When calling _sink.Write(outdata, 0, outdata.GetLength(0)), you might actually be getting the length wtih one unit too big, causing an ASCII null to be inserted.


_sink.Write(outdata, 0, outdata.GetLength(0)-1) and see if you are missing any characters or if it solves your problem.

Best Regards,
Alex Percsi.
ZekeLAAuthor Commented:
Thank you Alex. For a stab in the dark, it seems to be the correct solution. I still need to do more testing and won't get to do that for a few days but it certainly looks like you solved my problem. I'll update this thread once I complete the additional testing.
ZekeLAAuthor Commented:
I've tested the solution and plan to move it into Production. Apparently, GetBytes includes a final null character [bytes(nLastChar) is nothing] which is not part of the original string. IE ignores the null but Firefox doesn't. Thanks again.
Question has a verified solution.

Are you are experiencing a similar issue? Get a personalized answer when you ask a related question.

Have a better answer? Share it in a comment.

All Courses

From novice to tech pro — start learning today.