FileUpload Task

[andrueastman]

This PR closes microsoftgraph/msgraph-sdk-dotnet#467

Changes proposed in the pull request

• This PR proposes the adding of a LargeFileUploadTask to support upload of large files in drive or as message attachments.
• Task uses the ReadOnSubstream proposed in this PR.
  Essentially, one may perform a large upload by following the steps below.

Using the task

1. create an upload session (This example doesn't use the service library request builders)

// Create upload session 
// POST /v1.0/drive/items/01KGPRHTV6Y2GOVW7725BZO354PWSELRRZ:/SWEBOKv3.pdf:/microsoft.graph.createUploadSession
string uri = $"https://graph.microsoft.com/v1.0/drive/items/{itemId}:/SWEBOKv3.pdf:/microsoft.graph.createUploadSession";

HttpRequestMessage httpRequestMessage = new HttpRequestMessage(HttpMethod.Post, uri);
await graphClient.AuthenticationProvider.AuthenticateRequestAsync(httpRequestMessage);

// Read the session info from the response
var httpResponseMessage = await graphClient.HttpProvider.SendAsync(httpRequestMessage);
var content = await httpResponseMessage.Content.ReadAsStringAsync();
var uploadSession = graphClient.HttpProvider.Serializer.DeserializeObject<UploadSession>(content);

2. Create the task using the session.

// Create task
var maxSliceSize = 320 * 1024; // 320 KB - Change this to your chunk size. 4MB is the default.
LargeFileUploadTask<DriveItem> largeFileUploadTask = new LargeFileUploadTask<DriveItem>(uploadSession, stream, maxSliceSize);

3. Create an upload monitor

// Setup the progress monitoring
IProgress<long> progress = new Progress<long>(progress =>
{
    Console.WriteLine($"Uploaded {progress} bytes of {stream.Length} bytes");
});

4. Upload the file

UploadResult<DriveItem> uploadResult = null;
try
{
    uploadResult = await largeFileUploadTask.UploadAsync(progress);
                    
    if (uploadResult.UploadSucceeded)
    {
        Console.WriteLine($"File Uploaded {uploadResult.ItemResponse.Id}");//Sucessful Upload
    }
}
catch (ServiceException e)
{
    Console.WriteLine(e.Message);
}

FileAttachment support

The same steps can be followed for the FileAttachement upload by first creating an upload session and handing it over to the task as illustrated earlier.

// Create task
var maxSliceSize = 320 * 1024; // 320 KB - Change this to your chunk size. 4MB is the default.
LargeFileUploadTask<FileAttachment> largeFileUploadTask = new LargeFileUploadTask<FileAttachment>(uploadSession, stream, maxSliceSize);

However, the service only returns location URI which can be read off from the result object as follows. follows in step 4 as follows.

UploadResult<FileAttachment> uploadResult = null;
try
{
    uploadResult = await largeFileUploadTask.UploadAsync(progress);
    if (uploadResult.UploadSucceeded)
    {
        Console.WriteLine(uploadResult.Location);//the location of the object
    }
}
catch (ServiceException e)
{
    Console.WriteLine(e.Message);
}

Other links

• Chunked upload enhancements msgraph-sdk-dotnet#592 (related)
• LargeFileUpload task  msgraph-sdk-dotnet#584 (related)
• https://github.com/microsoftgraph/msgraph-sdk-design/blob/master/tasks/FileUploadTask.md (spec)

Notes

This PR can also be tested using the sample at the repository in the following link. It uses this PR as a git submodule.
https://github.com/andrueastman/UploadTest

[darrelmiller]

Is there a reason why we want to expose GetUploadSliceRequests() ? I don't see an advantage of exposing this. /cc @MIchaelMainer

[andrueastman]

I have updated the PR to :

• no longer expose GetUploadSliceRequests()
• update the UploadResponseHandler to use the DeserializeObject<T>(Stream stream) overload after the fix made in Use common serializer in Batch requests and responses #66.

cc @darrelmiller @peombwa @MIchaelMainer

[MIchaelMainer]

I'm sorry for the late review.