I am working on an ASP.NET MVC web service. In a web page, when a user clicks on a button, this triggers a complex method that takes a bit of time to finish. I want to redirect the user to a waiting page and then, when the process is finished, to redirect the user to a new page.
When the process is done it raises an event, which I can listen to from the controller. But I cannot make the last step to work (the controller redirecting to the new page upon receiving the event).
Here is my very naïve attempt at doing it (with simpler names):
public MyController()
{
EventsControllerClass.ProcessComplete += new EventHandler<MyArgsClass>(OnEventReceived);
}
private void OnEventReceived(object sender, MyArgsClass eventArguments)
{
RedirectToPage();
}
private ActionResult RedirectToPage()
{
return RedirectToAction("PageName");
}
After many days working on this, I have a viable solution. It may not be pretty, but it works, and maybe some ideas can be useful for other people, so here it goes:
I will explain the solution to my particular problem: I need a button to redirect to a "waiting" page while a longer process runs in the background and raises an event when it is finished. When this event is received, we want to redirect the user (automatically) to a final page.
First, I created a class to listen to the event. I tried doing this directly in the controller, but you need to be careful about signing and unsigning, because apparently controllers get created and destroyed at each request. In this "listener class" I have a bool property that is set to "true" when the event is received.
When the first action is triggered, the controller normally redirects to the "wait" page, where I have this simple java script redirecting to the new action:
<script type="text/javascript">
window.location = "#Url.Action("WaitThenRedirect", "AuxiliaryControllerName")";
</script>
This sets in motion the long process (through another event). The key is that I do this with an asynchronous action (this controller inherits from AsyncController). (Note I used an auxiliary controller. This is to keep all asynchronous stuff apart.) This is how this looks (more info here):
public static event EventHandler<AuxiliaryEventsArgs> ProcessReady;
public void WaitThenRedirectAsync()
{
AsyncManager.OutstandingOperations.Increment();
ProcessReady += (sender, e) =>
{
AsyncManager.Parameters["success"] = e.success;
AsyncManager.OutstandingOperations.Decrement();
};
WaitForEvent();
}
public ActionResult WaitThenRedirectCompleted(bool success)
{
if (success)
{
return RedirectToAction("RedirectToView", "ControllerName");
}
else
{
return RedirectToAction("UnexpectedError", "ControllerName");
}
}
private void WaitForEvent()
{
bool isWaitSuccessful = true;
int waitingLoops = 0;
int waitingThreshold = 200;
int sleepPeriod = 100; // (milliseconds)
while (!EventsListener.IsTheThingReady())
{
System.Threading.Thread.Sleep(sleepPeriod);
++waitingLoops;
if (waitingLoops > waitingThreshold)
{
System.Diagnostics.Debug.WriteLine("Waiting timed out!");
isWaitSuccessful = false;
break;
}
}
isWaitSuccessful = true;
if (null != ProcessReady)
{
AuxiliaryEventsArgs arguments = new AuxiliaryEventsArgs();
arguments.success = isWaitSuccessful;
try
{
ProcessReady(null, arguments);
}
catch (Exception ex)
{
System.Diagnostics.Debug.WriteLine("Error in event ProcessReady" + ex);
}
}
}
I believe it is possible to use ajax syntax for alternative solutions, but this is what I have and it works nicely. I believe this is not a very common need, but hopefully someone will benefit!
Related
I am using a DevExpress MVC Pivot Grid and trying to work out some problems with the loading and saving of layouts. So far I have the following:
I have set my CustomActionRouteValues in the PivotGridSettings as follows:
CustomActionRouteValues = new { Controller = "Home", Action = "PivotGridCustomCallback" },
Which points to the following:
public ActionResult PivotGridCustomCallback(string action, string reportName)
{
if (string.IsNullOrEmpty(reportName))
{
reportName = "Report 1";
}
var settings = PivotGridLayoutHelper.DefaultPivotGridSettings;
if (action == "Save")
{
// TODO: Find a better solution than this. At the moment, if Save is called once, it is then called again every time the user changes the layout.. which is why we have the 'saved' variable here.
bool saved = false;
settings.AfterPerformCallback = (sender, e) =>
{
if (saved)
{
return;
}
SaveLayout(((MVCxPivotGrid)sender).SaveLayoutToString(), reportName);
saved = true;
};
}
else if (action == "Load")
{
// TODO: Find a better solution than this. At the moment, if Load is called once, it is then called again every time the user changes the layout.. which is why we have the 'loaded' variable here.
bool loaded = false;
string layoutString = LoadLayout(reportName);
if (!string.IsNullOrEmpty(layoutString))
{
settings.BeforeGetCallbackResult = (sender, e) =>
{
if (loaded)
{
return;
}
((MVCxPivotGrid)sender).LoadLayoutFromString(layoutString, PivotGridWebOptionsLayout.DefaultLayout);
loaded = true;
};
}
}
ViewBag.PivotSettings = settings;
return PartialView("PivotPartial");
}
The problem, as you can see in the code comments, is that after performing an action just one time, it then gets called EVERY time I make any sort of change. So, for example... say I load a report.. that's fine.. but then when I try expand something or add a field.. or do ANYTHING, nothing seems to happen on the UI.. and I figured out that's because immediately, this code gets called again:
settings.BeforeGetCallbackResult = (sender, e) =>
{
((MVCxPivotGrid)sender).LoadLayoutFromString(layoutString, PivotGridWebOptionsLayout.DefaultLayout);
};
That just keeps resetting the values to the saved layout, which means the UI looks like it's unresponsive when trying to change anything.
This is why I now have the boolean variable called loaded to check if it's already loaded. That works.. but it's an ugly hack.. because it's making unnecessary trips to the server each and every time the user does anything on the pivot grid.
Surely there must be a way to prevent these actions from firing all the time?
I want to call method which will do something in background, but I don't want to change the current view. This is the method:
public ActionResult BayesTraining(string s,string path)
{
XmlParse xp = new XmlParse();
using (StreamWriter sw = System.IO.File.AppendText(path))
{
sw.WriteLine("d:/xml/"+xp.stripS(s)+".xml");
sw.Close();
}
return RedirectToAction("Index");
}
As you can see, I'm currently using RedirectToAction, that just reloads the page after the method is done working. Having in mind that method doesn't effect UI, I don't want to refresh web page every time I've used it. It's job should be done in background. So, how could I call it, without the need to redirect the view?
If you want something you can fire and forget use an ajax call. For instance if you change your action method to
public JsonResult BayesTraining(string s,string path)
{
XmlParse xp = new XmlParse();
using (StreamWriter sw = System.IO.File.AppendText(path))
{
sw.WriteLine("d:/xml/"+xp.stripS(s)+".xml");
sw.Close();
}
return Json("Success");
}
Then in your view bind to the UI event you need to via jQuery, for instance to bind to a button with id of BayesTraining do the following
$("#BayesTraining").click(function(){
$.post('#Url.Action( "BayesTraining" , "ControllerNameHere" , new { s = "stringcontent", path="//thepath//tothe//xmlfile//here//} )', function(data) {
//swallow success here.
});
}
DISCLAIMER: above code is not tested.
Hopefully it'll point you in the right direction.
If the method doesn't affect the UI, does it need to return an ActionResult? Couldn't it return void instead?
public void BayesTraining(string s,string path)
{
XmlParse xp = new XmlParse();
using (StreamWriter sw = System.IO.File.AppendText(path))
{
sw.WriteLine("d:/xml/"+xp.stripS(s)+".xml");
sw.Close();
}
}
I have a system whereby users can upload sometimes large(100-200 MB) files from within an MVC3 application. I would like to not block the UI while the file is uploading, and after some research, it looked like the new AsyncController might let me do what I'm trying to do. Problem is - every example I have seen isn't really doing the same thing, so I seem to be missing one crucial piece. After much futzing and fiddling, here's my current code:
public void CreateAsync(int CompanyId, FormCollection fc)
{
UserProfile up = new UserRepository().GetUserProfile(User.Identity.Name);
int companyId = CompanyId;
// make sure we got a file..
if (Request.Files.Count < 1)
{
RedirectToAction("Create");
}
HttpPostedFileBase hpf = Request.Files[0] as HttpPostedFileBase;
if (hpf.ContentLength > 0)
{
AsyncManager.OutstandingOperations.Increment();
BackgroundWorker worker = new BackgroundWorker();
worker.DoWork += (o, e) =>
{
string fileName = hpf.FileName;
AsyncManager.Parameters["recipientId"] = up.id;
AsyncManager.Parameters["fileName"] = fileName;
};
worker.RunWorkerCompleted += (o, e) => { AsyncManager.OutstandingOperations.Decrement(); };
worker.RunWorkerAsync();
}
RedirectToAction("Uploading");
}
public void CreateCompleted(int recipientId, string fileName)
{
SystemMessage msg = new SystemMessage();
msg.IsRead = false;
msg.Message = "Your file " + fileName + " has finished uploading.";
msg.MessageTypeId = 1;
msg.RecipientId = recipientId;
msg.SendDate = DateTime.Now;
SystemMessageRepository.AddMessage(msg);
}
public ActionResult Uploading()
{
return View();
}
Now the idea here is to have the user submit the file, call the background process which will do a bunch of things (for testing purposes is just pulling the filename for now), while directing them to the Uploading view which simply says "your file is uploading...carry on and we'll notify you when it's ready". The CreateCompleted method is handling that notification by inserting a message into the users's message queue.
So the problem is, I never get the Uploading view. Instead I get a blank Create view. I can't figure out why. Is it because the CreateCompleted method is getting called which shows the Create view? Why would it do that if it's returning void? I just want it to execute silently in the background, insert a message and stop.
So is this the right approach to take at ALL? my whole reason for doing it is with some network speeds, it can take 30 minutes to upload a file and in its current version, it blocks the entire application until it's complete. I'd rather not use something like a popup window if I can avoid it, since that gets into a bunch of support issues with popup-blocking scripts, etc.
Anyway - I am out of ideas. Suggestions? Help? Alternate methods I might consider?
Thanks in advance.
You are doing it all wrong here. Assume that your action name is Create.
CreateAsync will catch the request and should be a void method and returns nothing. If you have attributes, you should apply them to this method.
CreateCompleted is your method which you should treat as a standard controller action method and you should return your ActionResult inside this method.
Here is a simple example for you:
[HttpPost]
public void CreateAsync(int id) {
AsyncManager.OutstandingOperations.Increment();
var task = Task<double>.Factory.StartNew(() => {
double foo = 0;
for(var i = 0;i < 1000; i++) {
foo += Math.Sqrt(i);
}
return foo;
}).ContinueWith(t => {
if (!t.IsFaulted) {
AsyncManager.Parameters["headers1"] = t.Result;
}
else if (t.IsFaulted && t.Exception != null) {
AsyncManager.Parameters["error"] = t.Exception;
}
AsyncManager.OutstandingOperations.Decrement();
});
}
public ActionResult CreateCompleted(double headers1, Exception error) {
if(error != null)
throw error;
//Do what you need to do here
return RedirectToAction("Index");
}
Also keep in mind that this method will still block the till the operation is completed. This is not a "fire and forget" type async operation.
For more info, have a look:
Using an Asynchronous Controller in ASP.NET MVC
Edit
What you want here is something like the below code. Forget about all the AsyncController stuff and this is your create action post method:
[HttpPost]
public ActionResult About() {
Task.Factory.StartNew(() => {
System.Threading.Thread.Sleep(10000);
if (!System.IO.Directory.Exists(Server.MapPath("~/FooBar")))
System.IO.Directory.CreateDirectory(Server.MapPath("~/FooBar"));
System.IO.File.Create(Server.MapPath("~/FooBar/foo.txt"));
});
return RedirectToAction("Index");
}
Notice that I waited 10 seconds there in order to make it real. After you make the post, you will see the it will return immediately without waiting. Then, open up the root folder of you app and watch. You will notice that a folder and file will be created after 10 seconds.
But (a big one), here, there is no exception handling, a logic how to notify user, etc.
If I were you, I would look at a different approach here or make the user suffer and wait.
I'm using the mini-profiler on my asp.net MVC 3 application. I've implemented the profiler using the mvc nuget package. Everything works ok for standard page requests i get profile information sql everything.
However ajax requests don't seem to display initially. Just to let me confirm the request are completed without error. I've debugged this and they complete they also return http 200 responses in fiddler.
There is no request by the mini profiler that accompanies the ajax request. When i then navigate to another page i.e. a standard page request all the ajax request that were made on the last page are now displayed.
This is my mini profiler configuration page in App_Start
public static class MiniProfilerPackage
{
public static void PreStart()
{
//Setup sql formatter
MiniProfiler.Settings.SqlFormatter = new OracleFormatter();
//Make sure the MiniProfiler handles BeginRequest and EndRequest
DynamicModuleUtility.RegisterModule(typeof(MiniProfilerStartupModule));
//Setup profiler for Controllers via a Global ActionFilter
GlobalFilters.Filters.Add(new ProfilingActionFilter());
//Settings
MiniProfiler.Settings.PopupShowTimeWithChildren = true;
MiniProfiler.Settings.PopupShowTrivial = false;
//Ignore glimpse details in miniprofiler
var ignored = MiniProfiler.Settings.IgnoredPaths.ToList();
ignored.Add("Glimpse.axd");
MiniProfiler.Settings.IgnoredPaths = ignored.ToArray();
}
public static void PostStart()
{
// Intercept ViewEngines to profile all partial views and regular views.
// If you prefer to insert your profiling blocks manually you can comment this out
var copy = ViewEngines.Engines.ToList();
ViewEngines.Engines.Clear();
foreach (var item in copy)
{
ViewEngines.Engines.Add(new ProfilingViewEngine(item));
}
}
}
public class MiniProfilerStartupModule : IHttpModule
{
public void Init(HttpApplication context)
{
context.BeginRequest += (sender, e) =>
{
var request = ((HttpApplication)sender).Request;
MiniProfiler.Start();
};
//Profiling abadened if user is not in the admin role
context.PostAuthorizeRequest += (sender, e) =>
{
if (!context.User.IsInRole("Admin"))
MiniProfiler.Stop(discardResults: true);
};
context.EndRequest += (sender, e) =>
{
MiniProfiler.Stop();
};
}
public void Dispose() { }
}
Is there some configuration that is required or potential issues I should be aware of?
In the end i figured out that the Miniprofiler JavaScript block that is added to the page must be below the pages jquery reference.
My JavaScript reference are all added at the end of the page for best practice performance reasons.
But I'd left the #MvcMiniProfiler.MiniProfiler.RenderIncludes() in the page header. Moving it below the jquery script ref at the bottom of the page fixed my problem.
Just found out that the same situation repeat if
$.ajaxSetup({
global: false
});
We have an internal ASP.NET MVC application that requires a logon. Log on works great and does what's expected. We have a session expiration of 15 minutes. After sitting on a single page for that period of time, the user has lost the session. If they attempt to refresh the current page or browse to another, they will get a log on page. We keep their request stored so once they've logged in they can continue on to the page that they've requested. This works great.
However, my issue is that on some pages there are AJAX calls. For example, they may fill out part of a form, wander off and let their session expire. When they come back, the screen is still displayed. If they simply fill in a box (which will make an AJAX call) the AJAX call will return the Logon page (inside of whatever div the AJAX should have simply returned the actual results). This looks horrible.
I think that the solution is to make the page itself expire (so that when a session is terminated, they automatically are returned to the logon screen without any action by them). However, I'm wondering if there are opinions/ideas on how best to implement this specifically in regards to best practices in ASP.NET MVC.
Update:
So I went ahead and implemented this in my OnActionExecuting (per Keltex's suggestion)
if (!filterContext.HttpContext.User.Identity.IsAuthenticated)
{
if (filterContext.HttpContext.Request.IsAjaxRequest())
{
filterContext.HttpContext.Response.Write("Invalid session -- please login!");
filterContext.HttpContext.Response.End();
}
else
{
...
}
}
This definitely makes things better -- now even if they have two tabs (one with some AJAX calls that they can trigger) and they log out explicitly in the second tab, they will immediately get something that makes more sense rather than a bunch of screwed up AJAX data.
I still think I will implement the Javascript countdown as well that womp suggested.
Specifically, I don't know that there are any best practices regarding it, but I'm doing this right now for our app. We've opted for a client-side solution where we output the Session timeout value into some javascript in the master page, and calculate when the session will expire.
5 minutes before-hand, we pop up a modal dialog box saying "Are you still there?" with a countdown timer. Once the timer hits 0:00, we redirect the browser to the login page.
It's implemented with a minimal amount of javascript to do the time and timer calculations, and a simple .ashx handler that will refresh the session if the user clicks "I'm back!" on the dialog box before the session expires. That way if they return in time, they can refresh the session without any navigation.
I asked similar question yesterday. Here is my solution:
Modified Authorize attribute:
public class OptionalAuthorizeAttribute : AuthorizeAttribute
{
private class Http403Result : ActionResult
{
public override void ExecuteResult(ControllerContext context)
{
// Set the response code to 403.
context.HttpContext.Response.StatusCode = 403;
context.HttpContext.Response.Write(CTRes.AuthorizationLostPleaseLogOutAndLogInAgainToContinue);
}
}
private readonly bool _authorize;
public OptionalAuthorizeAttribute()
{
_authorize = true;
}
//OptionalAuthorize is turned on on base controller class, so it has to be turned off on some controller.
//That is why parameter is introduced.
public OptionalAuthorizeAttribute(bool authorize)
{
_authorize = authorize;
}
protected override bool AuthorizeCore(HttpContextBase httpContext)
{
//When authorize parameter is set to false, not authorization should be performed.
if (!_authorize)
return true;
var result = base.AuthorizeCore(httpContext);
return result;
}
protected override void HandleUnauthorizedRequest(AuthorizationContext filterContext)
{
if (filterContext.RequestContext.HttpContext.Request.IsAjaxRequest())
{
//Ajax request doesn't return to login page, it just returns 403 error.
filterContext.Result = new Http403Result();
}
else
base.HandleUnauthorizedRequest(filterContext);
}
}
HandleUnauthorizedRequest is overridden, so it returns Http403Result when using Ajax. Http403Result changes StatusCode to 403 and returns message to the user in response. There is some additional logic in attribute (authorize parameter), because I turn on [Authorize] in base controller and disable it in some pages.
Other important part is global handling of this response on client side. This is what I placed in Site.Master:
<script type="text/javascript">
$(document).ready(
function() {
$("body").ajaxError(
function(e,request) {
if (request.status == 403) {
alert(request.responseText);
window.location = '/Logout';
}
}
);
}
);
</script>
I place GLOBAL ajax error handler and when evert $.post fails with 403 error, response message is alerted and user is redirected to logout page. Now I don't have to handle error in every $.post request, because it is handled globally.
Why 403, not 401? 401 is handled internally by MVC framework (that is why redirection to login page is done after failed authorization).
What do you think about it?
EDIT:
About resigning from [Authorize] attribute: [Authorize] is not only about checking Identity.IsAuthenticated. It also handles page caching (so you don't cache material that requires authentication) and redirection. There is no need to copy this code.
You might look into the AjaxOptions that can be set in Ajax.BeginForm(). There is an OnBegin setting that you can associate with a javascript function, which could call a Controller method to confirm that the session is still valid, and if not, redirect to the login page using window.location.
Part of the problem appears to be that you're letting the framework do everything. I wouldn't decorate your AJAX method with the [Authorize] attribute. Instead check User.Identity.IsAuthenticated and if it returns false, create sensible error message.
My solution uses one meta-tag on login form and a bit of Javascript/jQuery.
LogOn.cshtml
<html>
<head>
<meta data-name="__loginform__" content="true" />
...
</head>
...
</html>
Common.js
var Common = {
IsLoginForm: function (data) {
var res = false;
if (data.indexOf("__loginform__") > 0) {
// Do a meta-test for login form
var temp =
$("<div>")
.html(data)
.find("meta[data-name='__loginform__']")
.attr("content");
res = !!temp;
}
return res;
}
};
AJAX code
$.get(myUrl, myData, function (serverData) {
if (Common.IsLoginForm(serverData)) {
location.reload();
return;
}
// Proceed with filling your placeholder or whatever you do with serverData response
// ...
});
Here's how I did it...
In my base controller
protected override void OnActionExecuting(ActionExecutingContext filterContext)
{
if (!filterContext.HttpContext.User.Identity.IsAuthenticated)
{
if (filterContext.HttpContext.Request.IsAjaxRequest())
{
filterContext.HttpContext.Response.StatusCode = 403;
filterContext.HttpContext.Response.Write(SessionTimeout);
filterContext.HttpContext.Response.End();
}
}
}
Then in my global .js file
$.ajaxSetup({
error: function (x, status, error) {
if (x.status == 403) {
alert("Sorry, your session has expired. Please login again to continue");
window.location.href = "/Account/Login";
}
else {
alert("An error occurred: " + status + "nError: " + error);
}
}
});
The SessionTimeout variable is a noty string. I omitted the implementation for brevity.