Making TPL Async calls from mvc controller on click of submit button - asp.net

Question

Basically I want to implement simple search functionality, whenever user enters some keyword in the text box on view and clicks submit button I want to make ASYNC calls to predefined website urls using TPL Async mechanism. When I do the same with console application it works like a charm but not with ASP.NET MVC3.
I couldn't find the reason
public ActionResult Index()
{
ViewBag.Message = "Welcome to ASP.NET MVC!";
return View();
}
public ActionResult About()
{
return View();
}
[HttpPost]
public ActionResult Index(string text)
{
string[] url = { "http://www.msnbc.com", "http://www.yahoo.com",
"http://www.nytimes.com", "http://www.washingtonpost.com",
"http://www.latimes.com", "http://www.newsday.com" };
Task<string[]> webTask = this.GetWordCounts(url, text);
string[] results = null;
try
{
results = webTask.Result;
}
catch (AggregateException e)
{
}
return View("Index", results);
}
//Taken from MSDN
Task<string[]> GetWordCounts(string[] urls, string name)
{
TaskCompletionSource<string[]> tcs = new TaskCompletionSource<string[]>();
WebClient[] webClients = new WebClient[urls.Length];
object m_lock = new object();
int count = 0;
List<string> results = new List<string>();
for (int i = 0; i < urls.Length; i++)
{
webClients[i] = new WebClient();
#region callback
// Specify the callback for the DownloadStringCompleted
// event that will be raised by this WebClient instance.
webClients[i].DownloadStringCompleted += (obj, args) =>
{
if (args.Cancelled == true)
{
tcs.TrySetCanceled();
return;
}
else if (args.Error != null)
{
// Pass through to the underlying Task
// any exceptions thrown by the WebClient
// during the asynchronous operation.
tcs.TrySetException(args.Error);
return;
}
else
{
// Split the string into an array of words,
// then count the number of elements that match
// the search term.
string[] words = null;
words = args.Result.Split(' ');
string NAME = name.ToUpper();
int nameCount = (from word in words.AsParallel()
where word.ToUpper().Contains(NAME)
select word)
.Count();
// Associate the results with the url, and add new string to the array that
// the underlying Task object will return in its Result property.
results.Add(String.Format("{0} has {1} instances of {2}", args.UserState, nameCount, name));
}
// If this is the last async operation to complete,
// then set the Result property on the underlying Task.
lock (m_lock)
{
count++;
if (count == urls.Length)
{
tcs.TrySetResult(results.ToArray());
}
}
};
#endregion
// Call DownloadStringAsync for each URL.
Uri address = null;
try
{
address = new Uri(urls[i]);
// Pass the address, and also use it for the userToken
// to identify the page when the delegate is invoked.
webClients[i].DownloadStringAsync(address, address);
}
catch (UriFormatException ex)
{
// Abandon the entire operation if one url is malformed.
// Other actions are possible here.
tcs.TrySetException(ex);
return tcs.Task;
}
}
// Return the underlying Task. The client code
// waits on the Result property, and handles exceptions
// in the try-catch block there.
return tcs.Task;
}
this is my view - for now I have hard coded keyword as microsoft
#using (Html.BeginForm("Index", "Home", new { text = "Microsoft" }))
{
<input type="submit" />
}
Update: It stays forever and inside the try block of Index Post method

Answer 1

I would recommend you using an AsyncController for this task to avoid jeopardizing ASP.NET worker threads which is one the worst thing that might happen to an ASP.NET application => running out of worker threads. It's like running out of fuel in the middle of the desert. You most certainly die.
So let's start by writing an extension method that will allow us converting the legacy WebClient event based pattern into the new task based pattern:
public static class TaskExtensions
{
public static Task<string> DownloadStringAsTask(this string url)
{
var tcs = new TaskCompletionSource<string>(url);
var client = new WebClient();
client.DownloadStringCompleted += (sender, args) =>
{
if (args.Error != null)
{
tcs.SetException(args.Error);
}
else
{
tcs.SetResult(args.Result);
}
};
client.DownloadStringAsync(new Uri(url));
return tcs.Task;
}
}
Armed with this extension method in hand we could now define a view model that will basically reflect the requirements of our view:
public class DownloadResultViewModel
{
public string Url { get; set; }
public int WordCount { get; set; }
public string Error { get; set; }
}
Then we move on to an asyncrhonous controller that will contain 2 actions: a standard synchronous Index action that will render the search form and an asynchronous Search action that will perform the actual work:
public class HomeController : AsyncController
{
public ActionResult Index()
{
return View();
}
[AsyncTimeout(600000)]
[HttpPost]
public void SearchAsync(string searchText)
{
AsyncManager.Parameters["searchText"] = searchText;
string[] urls =
{
"http://www.msnbc.com",
"http://www.yahoo.com",
"http://www.nytimes.com",
"http://www.washingtonpost.com",
"http://www.latimes.com",
"http://www.unexistentdomainthatwillcrash.com",
"http://www.newsday.com"
};
var tasks = urls.Select(url => url.DownloadStringAsTask());
AsyncManager.OutstandingOperations.Increment(urls.Length);
Task.Factory.ContinueWhenAll(tasks.ToArray(), allTasks =>
{
var results =
from task in allTasks
let error = task.IsFaulted ? task.Exception.Message : null
let result = !task.IsFaulted ? task.Result : string.Empty
select new DownloadResultViewModel
{
Url = (string)task.AsyncState,
Error = error,
WordCount = result.Split(' ')
.Where(x => string.Equals(x, searchText, StringComparison.OrdinalIgnoreCase))
.Count()
};
AsyncManager.Parameters["results"] = results;
AsyncManager.OutstandingOperations.Decrement(urls.Length);
});
}
public ActionResult SearchCompleted(IEnumerable<DownloadResultViewModel> results)
{
return View("index", results);
}
}
Now we define an ~/Views/Home/Index.cshtml view that will contain the search logic as well as the results:
#model IEnumerable<DownloadResultViewModel>
#using (Html.BeginForm("search", null, new { searchText = "politics" }))
{
<button type="submit">Search</button>
}
#if (Model != null)
{
<h3>Search results</h3>
<table>
<thead>
<tr>
<th>Url</th>
<th>Word count</th>
</tr>
</thead>
<tbody>
#Html.DisplayForModel()
</tbody>
</table>
}
And of course the corresponding display template that will be rendered automatically for each element of our model (~/Views/Shared/DisplayTemplates/DownloadResultViewModel.cshtml):
#model DownloadResultViewModel
<tr>
<td>#Html.DisplayFor(x => x.Url)</td>
<td>
#if (Model.Error != null)
{
#Html.DisplayFor(x => x.Error)
}
else
{
#Html.DisplayFor(x => x.WordCount)
}
</td>
</tr>
Now, since the search operation could take quite a long time your users could quickly get bored without being able to use some of the other hundredths of functionalities that your webpage has to offer them.
In this case it is absolutely trivial to invoke the Search controller action using an AJAX request and showing a spinner to inform the users that their search is in progress but without freezing the webpage allowing them to do other things (without navigating away from the page obviously).
So let's do that, shall we?
We start by externalizing the results into a partial (~/Views/Home/_Results.cshtml) without touching at the display template:
#model IEnumerable<DownloadResultViewModel>
#if (Model != null)
{
<h3>Search results</h3>
<table>
<thead>
<tr>
<th>Url</th>
<th>Word count</th>
</tr>
</thead>
<tbody>
#Html.DisplayForModel()
</tbody>
</table>
}
and we adapt our ~/Views/Home/Index.cshtml view to use this partial:
#model IEnumerable<DownloadResultViewModel>
#using (Html.BeginForm("search", null, new { searchText = "politics" }))
{
<button type="submit">Search</button>
}
<div id="results">
#Html.Partial("_Results")
</div>
and of course the SearchCompleted controller action that must now return only the partial result:
public ActionResult SearchCompleted(IEnumerable<DownloadResultViewModel> results)
{
return PartialView("_Results", results);
}
Now all that's left is to write a simple javascript that will AJAXify our search form. So this could happen into a separate js that will reference in our layout:
$(function () {
$('form').submit(function () {
$.ajax({
url: this.action,
type: this.method,
success: function (results) {
$('#results').html(results);
}
});
return false;
});
});
Depending on whether you referenced this script in the <head> section or at the end of the body you might not need to wrap it in a document.ready. If the script is at the end you could remove the wrapping document.ready function from my example.
And the last part is to give some visual indication to the user that the site is actually performing a search. This could be done using a global ajax event handler that we might subscribe to:
$(function () {
$(document).ajaxStart(function () {
$('#results').html('searching ...');
});
});