Is it possible to override MultipartFormDataStreamProvider so that is doesn't save uploads to the file system?

Question

I have an ASP.Net Web API application that allows clients (html pages and iPhone apps) to upload images to. I am using an async upload task as described in this article.

Everything works great when I want to save to the file system because that's what this code does automatically, behind the scenes it seems. But, I don't want to save the uploaded files to the file system. Instead, I want to take the uploaded stream and pass it through to an Amazon S3 bucket using the AWS SDK for .Net.

I have the code set up to send the stream up to AWS. The problem I can't figure out is how to get the uploaded content stream from the Web API method instead of having it automatically save to disk.

I was hoping there would be a virtual method I could override in MultipartFormDataStreamProvider which would allow me to do something else with the uploaded content other than save to disk, but there doesn't seem to be.

Any suggestions?

Answer 1

You could override MultipartFormDataStreamProvider's GetStream method to return a stream which is not a file stream but your AWS stream, but there are some issues doing so(which I will not elaborate here). Instead you could create a provider deriving from the abstract base class MultipartStreamProvider. Following sample is heavily based on the actual source code of MultipartFormDataStreamProvider and MultipartFileStreamProvider. You can check here and here for more details. Sample below:

public class CustomMultipartFormDataStreamProvider : MultipartStreamProvider
{
    private NameValueCollection _formData = new NameValueCollection(StringComparer.OrdinalIgnoreCase);

    private Collection<bool> _isFormData = new Collection<bool>();

    private Collection<MyMultipartFileData> _fileData = new Collection<MyMultipartFileData>();

    public NameValueCollection FormData
    {
        get { return _formData; }
    }

    public Collection<MultipartFileData> FileData
    {
        get { return _fileData; }
    }

    public override Stream GetStream(HttpContent parent, HttpContentHeaders headers)
    {
        // For form data, Content-Disposition header is a requirement
        ContentDispositionHeaderValue contentDisposition = headers.ContentDisposition;
        if (contentDisposition != null)
        {
            // If we have a file name then write contents out to AWS stream. Otherwise just write to MemoryStream
            if (!String.IsNullOrEmpty(contentDisposition.FileName))
            {
                // We won't post process files as form data
                _isFormData.Add(false);

                 MyMultipartFileData fileData = new MyMultipartFileData(headers, your-aws-filelocation-url-maybe);
                 _fileData.Add(fileData);

                return myAWSStream;//**return you AWS stream here**
            }

            // We will post process this as form data
            _isFormData.Add(true);

            // If no filename parameter was found in the Content-Disposition header then return a memory stream.
            return new MemoryStream();
        }

        throw new InvalidOperationException("Did not find required 'Content-Disposition' header field in MIME multipart body part..");
    }

    /// <summary>
    /// Read the non-file contents as form data.
    /// </summary>
    /// <returns></returns>
    public override async Task ExecutePostProcessingAsync()
    {
        // Find instances of HttpContent for which we created a memory stream and read them asynchronously
        // to get the string content and then add that as form data
        for (int index = 0; index < Contents.Count; index++)
        {
            if (_isFormData[index])
            {
                HttpContent formContent = Contents[index];
                // Extract name from Content-Disposition header. We know from earlier that the header is present.
                ContentDispositionHeaderValue contentDisposition = formContent.Headers.ContentDisposition;
                string formFieldName = UnquoteToken(contentDisposition.Name) ?? String.Empty;

                // Read the contents as string data and add to form data
                string formFieldValue = await formContent.ReadAsStringAsync();
                FormData.Add(formFieldName, formFieldValue);
            }
        }
    }

    /// <summary>
    /// Remove bounding quotes on a token if present
    /// </summary>
    /// <param name="token">Token to unquote.</param>
    /// <returns>Unquoted token.</returns>
    private static string UnquoteToken(string token)
    {
        if (String.IsNullOrWhiteSpace(token))
        {
            return token;
        }

        if (token.StartsWith("\"", StringComparison.Ordinal) && token.EndsWith("\"", StringComparison.Ordinal) && token.Length > 1)
        {
            return token.Substring(1, token.Length - 2);
        }

        return token;
    }
}

public class MyMultipartFileData
{
    public MultipartFileData(HttpContentHeaders headers, string awsFileUrl)
    {
        Headers = headers;
        AwsFileUrl = awsFileUrl;
    }

    public HttpContentHeaders Headers { get; private set; }

    public string AwsFileUrl { get; private set; }
}

Answer 2

Since @KiranChalla posted their answer, a new abstract class MultipartFormDataRemoteStreamProvider was introduced in Fix 1760: Make MultipartFormDataStreamProvider easier to work with non-FileStreams. to make this easier.

The summary of the class does a good job at explaining how to use it:

A MultipartStreamProvider implementation suited for use with HTML file uploads for writing file content to a remote storage Stream. The stream provider looks at the Content-Disposition header field and determines an output remote Stream based on the presence of a filename parameter. If a filename parameter is present in the Content-Disposition header field, then the body part is written to a remote Stream provided by GetRemoteStream. Otherwise it is written to a MemoryStream.