Question

Overview

I have a Microsoft Word Add-In, written in VBA (Visual Basic for Applications), that compresses a document and all of it’s related contents (embedded media) into a zip archive. After creating the zip archive it then turns the file into a byte array and posts it to an ASMX web service. This mostly works.

Issues

The main issue I have is transferring large files to the web site. I can successfully upload a file that is around 40MB, but not one that is 140MB (timeout/general failure).

A secondary issue is that building the byte array in the VBScript Word Add-In can fail by running out of memory on the client machine if the zip archive is too large.

Potential Solutions

I am considering the following options and am looking for feedback on either option or any other suggestions.

Option One

Opening a file stream on the client (MS Word VBA) and reading one “chunk” at a time and transmitting to ASMX web service which assembles the “chunks” into a file on the server.

This has the benefit of not adding any additional dependencies or components to the application, I would only be modifying existing functionality. (Fewer dependencies is better as this solution should work in a variety of server environments and be relatively easy to set up.)

Question:

• Are there examples of doing this or any recommended techniques (either on the client in VBA or in the web service in C#/VB.NET)?

Option Two

I understand WCF may provide a solution to the issue of transferring large files by “chunking” or streaming data. However, I am not very familiar with WCF, and am not sure what exactly it is capable of or if I can communicate with a WCF service from VBA. This has the downside of adding another dependency (.NET 3.0). But if using WCF is definitely a better solution I may not mind taking that dependency.

Questions:

• Does WCF reliably support large file transfers of this nature? If so, what does this involve? Any resources or examples?
• Are you able to call a WCF service from VBA? Any examples?

Answer 1

I ended up implementing option one referenced in the original question.

I “chunk” the file in VBA and transfer each “chunk” to the web service. I based the VBA portion of the solution on the code found here: Copy Large File by Chunk with Progress Notification. Instead of copying to the file system, however, I send it to the server.

The Code: VBA Land

Here is the (fugly) VBA code that creates the file chunks:

Function CopyFileByChunk(fileName As String, sSource As String) As Boolean

   Dim FileSize As Long, OddSize As Long, SoFar As Long
   Dim Buffer() As Byte, f1 As Integer, ChunkSize As Long

   On Error GoTo CopyFileByChunk_Error

   f1 = FreeFile: Open sSource For Binary Access Read As #f1
   FileSize = LOF(f1)
   If FileSize = 0 Then GoTo Exit_CopyFileByChunk ' -- done!

   ChunkSize = 5505024 '5.25MB
   OddSize = FileSize Mod ChunkSize

   Dim index As Integer
   index = 0

   If OddSize Then
      ReDim Buffer(1 To OddSize)
      Get #f1, , Buffer

      index = index + 1
      SoFar = OddSize

      If UploadFileViaWebService(Buffer, fileName, index, SoFar = FileSize) Then
            g_frmProgress.lblProgress = "Percent uploaded: " & Format(SoFar / FileSize, "0.0%")
            Debug.Print SoFar, Format(SoFar / FileSize, "0.0%")
            DoEvents
         Else
            GoTo CopyFileByChunk_Error
         End If
   End If

   If ChunkSize Then
      ReDim Buffer(1 To ChunkSize)
      Do While SoFar < FileSize
         Get #f1, , Buffer

         index = index + 1
         SoFar = SoFar + ChunkSize

         If UploadFileViaWebService(Buffer, fileName, index, SoFar = FileSize) Then
            g_frmProgress.lblProgress = "Percent uploaded: " & Format(SoFar / FileSize, "0.0%")
            Debug.Print SoFar, Format(SoFar / FileSize, "0.0%")
            DoEvents
         Else
            GoTo CopyFileByChunk_Error
         End If
      Loop
   End If

   CopyFileByChunk = True

Exit_CopyFileByChunk:
   Close #f1
   Exit Function

CopyFileByChunk_Error:
   CopyFileByChunk = False
   Resume Exit_CopyFileByChunk
End Function

Here is the referenced VBA method that uploads the chunks to the server:

Public Function UploadFileViaWebService(dataChunk() As Byte, fileName As String, index As Integer, lastChunk As Boolean) As Boolean

    On Error GoTo ErrHand
    Dim blnResult As Boolean
    blnResult = False

        'mdlConvert.SetProgressInfo "Connecting to the web server:" & vbNewLine & _
            DQUOT & server_title() & DQUOT
        If InternetAttemptConnect(0) = 0 Then
            On Error Resume Next

            Dim strSoapAction As String
            Dim strXml As String
            strXml = "<?xml version=""1.0"" encoding=""utf-8""?>" & _
            "<soap:Envelope xmlns:xsi=""http://www.w3.org/2001/XMLSchema-instance"" xmlns:xsd=""http://www.w3.org/2001/XMLSchema"" xmlns:soap=""http://schemas.xmlsoap.org/soap/envelope/"">" & _
            "<soap:Body>" & _
            "<UploadZipFile xmlns=""http://something.com/"">" & _
            "<zipBytes></zipBytes>" & _
            "<index>" & index & "</index>" & _
            "<isLastChunk>" & IIf(lastChunk, 1, 0) & "</isLastChunk>" & _
            "</UploadZipFile>" & _
            "</soap:Body>" & _
            "</soap:Envelope>"

            Dim objXmlhttp As Object
            Dim objDom As Object
            Set objXmlhttp = New MSXML2.xmlhttp

            ' Load XML
            Set objDom = CreateObject("MSXML2.DOMDocument")
            objDom.LoadXML strXml

            'insert data chunk into XML doc
            objDom.SelectSingleNode("//zipBytes").dataType = "bin.base64"
            objDom.SelectSingleNode("//zipBytes").nodeTypedValue = dataChunk

            ' Open the webservice
            objXmlhttp.Open "POST", webServiceUrl, False

            ' Create headings
            strSoapAction = "http://something.com/UploadZipFile"
            objXmlhttp.setRequestHeader "Content-Type", "text/xml; charset=utf-8"
            objXmlhttp.setRequestHeader "SOAPAction", strSoapAction

            ' Send XML command
            objXmlhttp.send objDom.XML

            ' Get all response text from webservice
            Dim strRet
            strRet = objXmlhttp.responseText

            ' Close object
            Set objXmlhttp = Nothing
            Set objDom = Nothing

            'get the error if any
            Set objDom = CreateObject("MSXML2.DOMDocument")
            objDom.LoadXML strRet
            Dim isSoapResponse As Boolean
            isSoapResponse = Not (objDom.SelectSingleNode("//soap:Envelope") Is Nothing)
            Dim error As String
            If Not isSoapResponse Then
                error = "Woops"
            Else
                error = objDom.SelectSingleNode("//soap:Envelope/soap:Body/soap:Fault/faultstring").text
            End If
            If error <> "" Then
                ShowServerError error, True
                blnResult = False
            Else
                Err.Clear 'clear the error caused in the XPath query above
                blnResult = True
            End If
            'close dom object
            Set objDom = Nothing


         Else
             GetErrorInfo "UploadFileViaWebService:InternetCheckConnection"
        End If

ErrHand:
    If Err.Number <> 0 Then
        ShowError Err, "UploadFileViaWebService"
        blnResult = False
    End If

    UploadFileViaWebService = blnResult
End Function

The Code: C# ASMX Web Service

Now, on the server side, the web service method accepts a few important parameters.

1. string fileName: The name of the
   file (each chunk has the same file
   name)
2. byte[] zipBytes: The contents
   of each chunk
3. int index: The index
   (used in conjunction with fileName
   to provide unique ordered partial
   files on the file system)
4. bool isLastChunk: This is the “i’m done –
   go ahead and merge all the “chunks”
   and clean up after yourself” flag.

int index and bool isLastChunk. With this context provided from the VBA world, I know enough to save each of these file chunks and then combine them when the isLastChunk flag is true.

   /// <summary>
    /// Accepts a chunk of a zip file.  Once all chunks have been received,  combines the chunks into a zip file that is processed.
    /// </summary>
    /// <param name="fileName">Name of the file.</param>
    /// <param name="zipBytes">The collection of bytes in this chunk.</param>
    /// <param name="index">The index of this chunk.</param>
    /// <param name="isLastChunk">if set to <c>true</c> this is the last chunk.</param>
    /// <returns>Whether the file was successfully read and parsed</returns>
    /// <exception cref="ParserException">An error occurred while trying to upload your file. The details have been written to the system log.</exception>
    [WebMethod]
    public bool UploadZipFile(string fileName, byte[] zipBytes, int index, bool isLastChunk)
    {
        try
        {
            const string ModuleRootUrl = "/Somewhere/";
            string folderName = HostingEnvironment.MapPath("~" + ModuleRootUrl);
            string fullDirectoryName = Path.Combine(folderName, Path.GetFileNameWithoutExtension(fileName));

            try
            {
                if (!Directory.Exists(fullDirectoryName))
                {
                    Directory.CreateDirectory(fullDirectoryName);
                }

                string pathAndFileName = Path.Combine(fullDirectoryName, AddIndexToFileName(fileName, index));
                using (var stream = new MemoryStream(zipBytes))
                {
                    WriteStreamToFile(stream, pathAndFileName);
                }

                if (isLastChunk)
                {
                    try
                    {
                        MergeFiles(fullDirectoryName, fileName, index);

                        // file transfer is done.
                        // extract the zip file
                        // and do whatever you need to do with its contents
                        // we'll assume that it works - but your "parsing" should return true or false
                        return true;
                    }
                    finally
                    {
                        DeleteDirectoryAndAllContents(fullDirectoryName);
                    }
                }
            }
            catch
            {
                DeleteDirectoryAndAllContents(fullDirectoryName);
                throw;
            }
        }
        return false;
    }

Here is the C# code that writes each incoming chunk the the file system:

/// <summary>
/// Writes the contents of the given <paramref name="stream"/> into a file at <paramref name="newFilePath"/>.
/// </summary>
/// <param name="stream">The stream to write to the given file</param>
/// <param name="newFilePath">The full path to the new file which should contain the contents of the <paramref name="stream"/></param>
public static void WriteStreamToFile(Stream stream, string newFilePath)
{
    using (FileStream fs = File.OpenWrite(newFilePath))
    {
        const int BlockSize = 1024;
        var buffer = new byte[BlockSize];
        int numBytes;
        while ((numBytes = stream.Read(buffer, 0, BlockSize)) > 0)
        {
            fs.Write(buffer, 0, numBytes);
        }
    }
}

Here is the C# code to merge all of the zip file “chunks”:

/// <summary>
/// Merges each file chunk into one complete zip archive.
/// </summary>
/// <param name="directoryPath">The full path to the directory.</param>
/// <param name="fileName">Name of the file.</param>
/// <param name="finalChunkIndex">The index of the last file chunk.</param>
private static void MergeFiles(string directoryPath, string fileName, int finalChunkIndex)
{
    var fullNewFilePath = Path.Combine(directoryPath, fileName);

    using (var newFileStream = File.Create(fullNewFilePath))
    {
        for (int i = 1; i <= finalChunkIndex; i++)
        {
            using (var chunkFileStream = new FileStream(AddIndexToFileName(fullNewFilePath, i), FileMode.Open))
            {
                var buffer = new byte[chunkFileStream.Length];
                chunkFileStream.Read(buffer, 0, (int)chunkFileStream.Length);
                newFileStream.Write(buffer, 0, (int)chunkFileStream.Length);
            }
        }
    }
}