Can I abort a large FileStreamResult download on server side? - asp.net-mvc

Question

I have an action on which I simply download a file. Sometimes the user wants the download to be aborted and not to wait to be finished if the file was large.
private IActionResult Download(string path)
{
var length = new FileInfo(path).Length;
Response.Headers.Add("size", length.ToString());
Stream str = System.IO.File.OpenRead(path);
return File(str, "application/x-zip-compressed");
}
What I would like to know is if on backend we can abort that process, maybe storing on ram some task to be cancelled on another process.
EDITED for better understanding: the file downloads into a different device which calls the download method after I order, from my browser, that this device has to download the file. So the task need to be aborted on server side by calling an action from my browser.

Answer 1

On the server side, you can get notification when the request is aborted via HttpContext.RequestAborted, which is a CancellationToken that is cancelled when the underlying connection for the request is aborted. So you could pass it down into asynchronous methods that you call if they support cancellation, or you could hook a callback to process cancellation logic via CancellationToken.Register().
However, in your code sample, this is not necessary. All you are doing is creating a FileStream without reading the contents of the file. The actual reading of the file content is performed by the framework, when it processes the FileStreamResult which is created by the call to File(str, "application/x-zip-compressed"). As you can see from the source code for FileResultExecutorBase, the framework will automatically cancel the file reading when HttpContext.RequestAborted is cancelled.

Answer 2

Well, I've finally solved this issue by handling a global collection of HttpContexts. I store the download request HttpContext when my download action is called and the download starts. Also, I have a different action/method/whatelse where I read the context and I call .Abort() method.
private void CancelDownload(string key) //action, method...
{
string uniqueKeyDownload = externalDevice.ID + "#" + fileID;
log.Information("Cancelling download 1/3..." + uniqueKeyDownload );
if (Startup.MyDictionaryContexts.ContainsKey(uniqueKeyDownload ))
{
log.Information("Cancelling download 2/3..." + uniqueKeyDownload );
if (!Startup.MyListDownloadsCancelled.Contains(uniqueKeyDownload ))
{
log.Information("Cancelling download 3/3..." + uniqueKeyDownload );
Startup.MyListDownloadsCancelled.Add(uniqueKeyDownload );
Startup.MyDictionaryContexts[uniqueKeyDownload ].Abort();
}
}
}
public IActionResult Download(string path)
{
string uniqueKeyDownload = externalDevice.ID + "#" + fileID;
if (!Startup.MyDictionaryContexts.ContainsKey(uniqueKeyDownload ))
{
Startup.MyDictionaryContexts.Add(uniqueKeyDownload, HttpContext);
}
HttpContext.Response.OnCompleted(async () =>
{
//remove context from dictionary
if (Startup.MyDictionaryContexts.ContainsKey(uniqueKeyDownload))
Startup.MyDictionaryContexts.Remove(uniqueKeyDownload);
//Log if event was fired on completion or abort call
if (Startup.MyListDownloadsCancelled.Contains(uniqueKeyDownload))
{
Startup.MyListDownloadsCancelled.Remove(uniqueKeyDownload);
log.Information("DOWNLOAD CANCELLED!!!");
}
else
log.Information("DOWNLOAD COMPLETED!!!");
});
//if you are downloading on browser and make recalls, manage this block
// with conditions and return NoContent() or similar
var length = new FileInfo(path).Length;
Response.Headers.Add("size", length.ToString());
Stream str = System.IO.File.OpenRead(path);
log.Information("DOWNLOAD STARTED..." + uniqueKeyDownload);
return File(str, "application/x-zip-compressed");
}
I've realized that, depending on the client that is downloading the file, there is different behaviors. If the client is a http client from Java (my case), there will be a socket exception at client and that's all. But in case you are downloading on a browser (tested with Chrome), after calling .Abort() the browser will request one more time the download Action and, in this case, you will have to manage some flags to return NoContent() or similar instead of return File(..) again.