C# HttpListener: storing posted file is always 2 byte larger than original

Question

Problem description:

I have an HttpListener within my C# program that takes a POST request with a posted file (jpg, doc, zip or alike). I am sending and receiving the file on the same Windows machine.

I took the code from https://stackoverflow.com/a/8468520/241475 (coming from the Java world, I'm fairly new to C#/.Net), which is said to just be a proof of concept, and the upload and storing of data on the server side almost works fine, except that the stored files is always 2 bytes larger than the original.

This does not matter for jpg, doc or single zip files - I can still open them and they display fine - but when I upload a zip file that is part of a multi-volume zip, unzipping usually fails for one file inside the zip, while the rest gets unzipped fine.

Example:

In this example, you can see the original file (import_filesystem.kmumitkst.z01.ORIGINAL), which was 10485760 bytes large, which I posted to my HttpListener and which stored the version named import_filesystem.kmumitkst.z01, which is 10485762 bytes large.

When I then unzip it via WinRar, it tells me that the CRC/checksum isn't correct and one file in it is corrupt due to these two extra bytes.

Code:

This is the code storing the received file:

while (_listener.IsListening)
{
    ThreadPool.QueueUserWorkItem((c) =>
    {
        var ctx = c as HttpListenerContext;  
        SaveFile(ctx.Request.ContentEncoding, GetBoundary(ctx.Request.ContentType), 
            ctx.Request.InputStream, targetFilePath);
        ...
    }
}

// =====================================


private static void SaveFile(Encoding enc, String boundary, Stream input, string targetPath)
{
    Byte[] boundaryBytes = enc.GetBytes(boundary);
    Int32 boundaryLen = boundaryBytes.Length;

    using (FileStream output = new FileStream(targetPath, FileMode.Create, FileAccess.Write))
    {
        Byte[] buffer = new Byte[1024];
        Int32 len = input.Read(buffer, 0, 1024);
        Int32 startPos = -1;

        // Find start boundary
        while (true)
        {
            if (len == 0)
            {
                throw new Exception("Start Boundaray Not Found");
            }

            startPos = IndexOf(buffer, len, boundaryBytes);
            if (startPos >= 0)
            {
                break;
            }
            else
            {
                Array.Copy(buffer, len - boundaryLen, buffer, 0, boundaryLen);
                len = input.Read(buffer, boundaryLen, 1024 - boundaryLen);
            }
        }

        // Skip four lines (Boundary, Content-Disposition, Content-Type, and a blank)
        for (Int32 i = 0; i < 4; i++)
        {
            while (true)
            {
                if (len == 0)
                {
                    throw new Exception("Preamble not Found.");
                }

                startPos = Array.IndexOf(buffer, enc.GetBytes("\n")[0], startPos);
                if (startPos >= 0)
                {
                    startPos++;
                    break;
                }
                else
                {
                    len = input.Read(buffer, 0, 1024);
                }
            }
        }

        Array.Copy(buffer, startPos, buffer, 0, len - startPos);
        len = len - startPos;

        while (true)
        {
            Int32 endPos = IndexOf(buffer, len, boundaryBytes);
            if (endPos >= 0)
            {
                if (endPos > 0) output.Write(buffer, 0, endPos);
                break;
            }
            else if (len <= boundaryLen)
            {
                throw new Exception("End Boundaray Not Found");
            }
            else
            {
                output.Write(buffer, 0, len - boundaryLen);
                Array.Copy(buffer, len - boundaryLen, buffer, 0, boundaryLen);
                len = input.Read(buffer, boundaryLen, 1024 - boundaryLen) + boundaryLen;
            }
        }
    }
}

private static String GetBoundary(String ctype)
{
    return "--" + ctype.Split(';')[1].Split('=')[1];
}

private static Int32 IndexOf(Byte[] buffer, Int32 len, Byte[] boundaryBytes)
{
    for (Int32 i = 0; i <= len - boundaryBytes.Length; i++)
    {
        Boolean match = true;
        for (Int32 j = 0; j < boundaryBytes.Length && match; j++)
        {
            match = buffer[i + j] == boundaryBytes[j];
        }

        if (match)
        {
            return i;
        }
    }
    return -1;
}

Question:

Where do these two extra bytes come from? Regardless of what kind of file or file size I sent, it's always exactly 2 bytes too much.

Is there any other way to save a received file through the HttpListener? Thanks for any hint.

Answer 1

Found the solution, had to change

while (true)
{
    Int32 endPos = IndexOf(buffer, len, boundaryBytes);
    if (endPos >= 0)
    {
        if (endPos > 0) output.Write(buffer, 0, endPos);
        break;
    }

to

while (true)
{
    Int32 endPos = IndexOf(buffer, len, boundaryBytes);
    if (endPos >= 0)
    {
        if (endPos > 0) output.Write(buffer, 0, endPos-2);
        break;
    }

adding a -2 to the endPos, to avoid extra line break at the end.