I am creating a stream like class, that inherits from Stream. When I use a StreamWriter class to write the information asynchronously:
using (var sw = new StreamWriter(messageWriter, Encoding.UTF8))
await sw.WriteAsync(msg);
I can see how StreamWriter goes straight to void Stream.Write(byte[] buffer, int offset, int count) rather than Task Stream.WriteAsync(byte[] buffer, int offset, int count, CancellationToken cancellationToken).
Is this a bug or am I missing something?
I spotted the source of confusion.
StreamWriter flushes synchronously by default unless you specifically flush asynchronously.
Stream.Flush is called when closing, and of course, flushing.
This will invoke Write rather than WriteAsync in the messageWriter stream:
using (var sw = new StreamWriter(messageWriter, Encoding.UTF8))
await sw.WriteAsync("Hi");
The StreamWriter will buffer the "Hi" message, and it will be released to the underlying stream on the Flush invocation synchronously when closing the StreamWriter.
So the asynchronous call to StreamWriter.WriteAsync could have no effect on the underlying stream (because the buffering), and when calling StreamWriter.Close, StreamWriter.Flush is called synchronously, and therefore the Flush method on the underlying stream.
This code also calls Write synchronously:
using (var sw = new StreamWriter(messageWriter, Encoding.UTF8))
{
sw.AutoFlush = true; // will call Write here to send the UTF8 BOM
await sw.WriteAsync("Hi"); // Write the actual data
}
obviously also this this:
using (var sw = new StreamWriter(messageWriter, Encoding.UTF8))
{
await sw.WriteAsync("Hi");
sw.Flush();
}
This will call WriteAsync:
using (var sw = new StreamWriter(messageWriter, Encoding.UTF8))
{
await sw.WriteAsync("Hi");
await sw.FlushAsync()
}
I have created a simple class to track how the methods are called:
public class MSWrite : MemoryStream
{
public override bool CanWrite { get { return true; } }
public override void Write(byte[] buffer, int offset, int count)
{
Console.WriteLine("Write");
}
public override Task WriteAsync(byte[] buffer, int offset, int count, CancellationToken cancellationToken)
{
Console.WriteLine("WriteAsync");
return Task.Run(() => { });
}
}
And this test:
Console.WriteLine("Simple");
using(var ms = new MSWrite())
using (StreamWriter sw = new StreamWriter(ms, Encoding.UTF8))
{
await sw.WriteAsync("Hi");
}
Console.WriteLine();
Console.WriteLine("AutoFlush");
using (var ms = new MSWrite())
using (StreamWriter sw = new StreamWriter(ms, Encoding.UTF8))
{
sw.AutoFlush = true;
await sw.WriteAsync("Hi");
}
Console.WriteLine();
Console.WriteLine("Flush async");
using (var ms = new MSWrite())
using (StreamWriter sw = new StreamWriter(ms, Encoding.UTF8))
{
await sw.WriteAsync("Hi");
await sw.FlushAsync();
}
Yields:
Simple
Write
Write
AutoFlush
Write
WriteAsync
Flush async
WriteAsync
WriteAsync
The first write is the UTF8 BOM, the second is the actual data.
No, StreamWriter.WriteAsync does in fact call Stream.WriteAsync (assuming you are on .NET 4.5). If you are seeing otherwise, please post a stack trace.
Note that the Stream base class by default will implement WriteAsync as a Write done on a thread pool thread. You'll need to override WriteAsync in your custom stream to change this behavior.
Related
I am working on a xamarin mobile application, upon making an async call to the exposed api, i do not get any error, however when i execute the .Result on the task the call never proceeds and it stuck forever.
Click here to see stringResourceResponse details
The same .Result call from a separate project (windows service) in the same solution works.
Any idea if .NET standard is causing limitation in executing async tasks, any advice would be helpful, thanks
Code added below:
//This is code from app.xaml.cs
var stringResourceApi = new StringResourceApiTask();
Task.Run(() =>
{
a = controller.CallStringResourceApi(stringResourceApi);
}).Wait();
public class MobileController
{
public string CallStringResourceApi(StringResourceApiTask stringResourceApiTask)
{
return stringResourceApiTask.Start(StringResourceUrl);
}
}
public override string Start(string URL)
{
var stringResourceResponse = SendRequest(url, "", HttpMethod.Get);
var result = stringResourceResponse.Result;
return result;
}
protected async Task < string > SendRequest(string url, string uri, HttpMethod method, int attempt = 1, int maxAttempts = 5)
{
return await SendRequest(
url, uri, Key, Secret, method, string.Empty, attempt, maxAttempts)
.ConfigureAwait(false);
}
protected async Task<string> SendRequest(string url, string uri, string key, string secret, HttpMethod method,
string requestBody = "", int attempt = 1, int maxAttempts = 5)
{
if (attempt > maxAttempts)
{
return null;
}
var client = InitialiseHttpClient(key, secret);
var request = new HttpRequestMessage
{
RequestUri = string.IsNullOrEmpty(url) ? new Uri(uri) : new Uri(url),
Method = method,
};
if (!string.IsNullOrWhiteSpace(requestBody))
{
request.Content = new StringContent(requestBody, Encoding.UTF8, "application/json");
}
SetOutputText($"Attempting to communicate with {uri}...{Environment.NewLine}");
using (var response = await client.SendAsync(request).ConfigureAwait(false))
{
using (var content = response.Content)
{
try
{
response.EnsureSuccessStatusCode();
}
catch (HttpRequestException ex)
{
if (attempt > maxAttempts)
{
SetOutputText(errorMessage);
}
return await SendRequest(url, uri, key, secret, method, requestBody, attempt + 1).ConfigureAwait(false);
}
var responseBody = await content.ReadAsStringAsync().ConfigureAwait(false);
var isSuccessResponseButEmptyBody = response.IsSuccessStatusCode &&
(string.IsNullOrEmpty(responseBody) ||
string.IsNullOrWhiteSpace(responseBody));
if (!isSuccessResponseButEmptyBody)
{
return responseBody;
}
if (attempt > maxAttempts)
{
SetOutputText(errorMessage);
}
return await SendRequest(url, uri, key, secret, method, requestBody, attempt + 1).ConfigureAwait(false);
}
}
}
when i execute the .Result on the task the call never proceeds and it stuck forever.
Yes. This is a common deadlock situation. When code running on the UI thread blocks on asynchronous code, a deadlock usually occurs.
The same .Result call from a separate project (windows service) in the same solution works.
It works because the Win32 service code does not run on a UI thread.
The proper solution is to remove the blocking code; use await instead. This in turn will cause the calling methods to become async (e.g., StringResourceApiTask.Start), and they should also be awaited, etc. The usage of async and await should "grow" through your code; this is natural.
Alternatively, you can block in a thread pool thread, e.g., Task.Run(() => a = controller.CallStringResourceApi(stringResourceApi)).GetAwaiter().GetResult();. This is a bit of a hack (consuming an unnecessary thread), but it's a quick way to remove the deadlock. Note that this hack is not appropriate for ASP.NET apps; it's acceptable here since this is a UI app.
In my Web API app, I'm using HttpServer to contain my controller in unit tests, and I'm using HttpClient to call it directly, eg:
[Fact]
public void TestMyController()
{
var config = new HttpConfiguration();
config.Routes.MapHttpRoute("default", "api/{controller}/{action}/{id}", new { id = RouteParameter.Optional });
var server = new HttpServer(config);
var client = new HttpClient(server);
var response = client.GetAsync("http://localhost/api/test/values").Result;
}
I've noticed (by stepping through the debugger, and confirmed on other SO posts), that the JsonFormatter is not really running - it's initialized, but not exercised. Since this test isn't opening a socket, and the HttpClient is directly invoking the HttpServer through the HttpMessageHandler API, it does make sense that formatting/serialization isn't run because it's not needed.
In my case, I have some custom formatting/serialization/deserialization code that isn't being hit during these tests, but it's hit when I run in a real web server. I'd like to exercise that code in these tests; and it also just seems risky to exclude the serialization/deserialization code path when testing. Any advice on this?
Following is a quick example of what you could do to force formatters to go through serialization/deserialization. Here we are converting ObjectContent to StreamContent. In the below code, the call to CopyToAsync triggers a path where formatters are forced to serialize. In case of deserilization, in order to make sure we go through formatters we want the content to be of type other than ObjectContent as ReadAsAsync has internal logic which special cases ObjectContnent and we want to circumvent it.
HttpClient client = new HttpClient(new InMemoryHttpContentSerializationHandler(new HttpServer(config)));
public class InMemoryHttpContentSerializationHandler : DelegatingHandler
{
public InMemoryHttpContentSerializationHandler(HttpMessageHandler innerHandler)
: base(innerHandler)
{
}
protected override async Task<HttpResponseMessage> SendAsync(HttpRequestMessage request, CancellationToken cancellationToken)
{
request.Content = await ConvertToStreamContentAsync(request.Content);
HttpResponseMessage response = await base.SendAsync(request, cancellationToken);
response.Content = await ConvertToStreamContentAsync(response.Content);
return response;
}
private async Task<StreamContent> ConvertToStreamContentAsync(HttpContent originalContent)
{
if (originalContent == null)
{
return null;
}
StreamContent streamContent = originalContent as StreamContent;
if (streamContent != null)
{
return streamContent;
}
MemoryStream ms = new MemoryStream();
await originalContent.CopyToAsync(ms);
// Reset the stream position back to 0 as in the previous CopyToAsync() call,
// a formatter for example, could have made the position to be at the end
ms.Position = 0;
streamContent = new StreamContent(ms);
// copy headers from the original content
foreach (KeyValuePair<string, IEnumerable<string>> header in originalContent.Headers)
{
streamContent.Headers.TryAddWithoutValidation(header.Key, header.Value);
}
return streamContent;
}
}
We were developing a sample MWC application with the logic of Business ,DataAccess,Data Layers.
In core Project we used the following code for consuming data from json parsing. This code works fine for Xamarin.Android and Xamarin.iOS, but for windows phone it shows error as 'System.Net.WebRequest does not contain a definition for GetResponse and no extension method for GetResponse...'
We tried to use Async methods for consuming WCF Rest service json data, but it returned as null before the completed method called.
Is it possible to wait and get data from completed method to return the json collection? if no please suggest how to achieve the same.
public String login<T>(T item) where T : BusinessLayer.Contracts.IBusinessEntity, new()
{
var request = HttpWebRequest.Create(url);
request.ContentType = "application/json";
request.Method = "get";
using (HttpWebResponse response = request.GetResponse() as HttpWebResponse)
{
using (StreamReader reader = new StreamReader(response.GetResponseStream()))
{
var content = reader.ReadToEnd();
string nss = content.ToString();
check = nss;
return nss;
}
}
return Check;
}
Edit: I have included the sample code of Async function.
Before Executing the DownloadStringCompleted event it returns null value. We need that DownloadStringCompleted output string for further process.
Note: We were following the logic of Tasky in Xamarin
async Task<string> AccessTheWebAsync(string url)
{
var webClient = new WebClient();
webClient.DownloadStringCompleted += (sender, e) =>
{
string data = (string)e.Result;
check = data;
};
webClient.DownloadStringAsync(new Uri(url));
return check;
}
public async Task<string> login<T>(T item) where T : BusinessLayer.Contracts.IBusinessEntity, new()
{
return check = await AccessTheWebAsync(item.url);
}
Your asynchronous code is not using Task correctly. It should use TaskCompletionSource to get the job done:
Task<string> AccessTheWebAsync(string url)
{
var source = new TaskCompletionSource<string>();
var webClient = new WebClient();
webClient.DownloadStringCompleted += (sender, e) =>
{
source.TrySetResult((string)e.Result);
};
webClient.DownloadStringAsync(new Uri(url));
return source.Task;
}
Before, your function was returning before the event fired. Using the task source wraps it in a task properly and fixes this problem.
You will also need to hook up the error event and call TrySetException to finish the implementation.
I'm trying to validate a user in my WP7 app by validating username/pass on a server.
In my "validate user" method I create a HTTPWebRequest with the stuff I need validated.
Then as far as I can see, the only option to execute is to use request.BeginGetResponse, with an async callback.
But I want to return the response from the request in the same method that I created the request in, how can I accomplish this?
I'm not sure why you would like to make it return in the calling method. I would just use the standard WebClient behavior like below:
public void ValidateUser()
{
WebClient webClient = new WebClient();
Uri uri = new Uri(url);
webClient.DownloadStringCompleted += new DownloadStringCompletedEventHandler(webClient_DownloadStringCompleted);
webClient.DownloadStringAsync(new Uri(url));
}
void webClient_DownloadStringCompleted(object sender, DownloadStringCompletedEventArgs e)
{
//Validate result from response by using e.Result
}
BeginGetResponse returns an IAsyncResult which has a wait handle that is signalled when the request completes. All you have to do is wait on this handle: [ doesn't work on WP7 ]
UPDATE: use a ManualResetEvent
var mre = new ManualResetEvent( false );
var iar = myHttpWebRequest.BeginGetResponse( state => mre.Set(), null );
mre.WaitOne();
var response = myHttpWebRequest.EndGetResponse( iar );
The only real way to achieve what you need would be to create an event in your worker class, and fire that event when the async call completes, so (pseudocode)...
void StartDownload(object[] parameters)
{
var req = HttpWebRequest.Create("http://google.com");
req.BeginGetRequestStream(Completed_handler, req);
}
void Completed_handler(object sender, DownloadStringCompletedEventArgs e)
{
var request = (HttpWebRequest)result.AsyncState;
var response = request.EndGetResponse(result);
using (var stream = response.GetResponseStream())
using (var reader = new StreamReader(stream))
{
var contents = reader.ReadToEnd();
NotifyCallerOfContent(contents)
}
}
public event EventHandler<MyCustomEventArgsClass> DownloadHasFinished;
public NotifyCallerOfDownload(string content)
{
if(null != DownloadHasFinished)
{
DownloadHasFinished(this, new MyCustomEventArgsClass(content));
}
}
and then subscribe to the DownloadHasFinished event in your calling class.
As for calling out specifically by a HttpWebRequest, take a look at this question, where you will find some working samples.
I found an example about HTTP POST in msdn, but I am wondering how can I make use of reactive extensions here.
using System;
using System.Net;
using System.IO;
using System.Text; using System.Threading;
class HttpWebRequestBeginGetRequest
{
private static ManualResetEvent allDone = new ManualResetEvent(false);
public static void Main(string[] args)
{
// Create a new HttpWebRequest object.
HttpWebRequest request = (HttpWebRequest)WebRequest.Create("http://www.contoso.com/example.aspx");
request.ContentType = "application/x-www-form-urlencoded";
// Set the Method property to 'POST' to post data to the URI.
request.Method = "POST";
// start the asynchronous operation
request.BeginGetRequestStream(new AsyncCallback(GetRequestStreamCallback), request);
// Keep the main thread from continuing while the asynchronous
// operation completes. A real world application
// could do something useful such as updating its user interface.
allDone.WaitOne();
}
private static void GetRequestStreamCallback(IAsyncResult asynchronousResult)
{
HttpWebRequest request = (HttpWebRequest)asynchronousResult.AsyncState;
// End the operation
Stream postStream = request.EndGetRequestStream(asynchronousResult);
Console.WriteLine("Please enter the input data to be posted:");
string postData = Console.ReadLine();
// Convert the string into a byte array.
byte[] byteArray = Encoding.UTF8.GetBytes(postData);
// Write to the request stream.
postStream.Write(byteArray, 0, postData.Length);
postStream.Close();
// Start the asynchronous operation to get the response
request.BeginGetResponse(new AsyncCallback(GetResponseCallback), request);
}
private static void GetResponseCallback(IAsyncResult asynchronousResult)
{
HttpWebRequest request = (HttpWebRequest)asynchronousResult.AsyncState;
// End the operation
HttpWebResponse response = (HttpWebResponse)request.EndGetResponse(asynchronousResult);
Stream streamResponse = response.GetResponseStream();
StreamReader streamRead = new StreamReader(streamResponse);
string responseString = streamRead.ReadToEnd();
Console.WriteLine(responseString);
// Close the stream object
streamResponse.Close();
streamRead.Close();
// Release the HttpWebResponse
response.Close();
allDone.Set();
}
}
I am trying to use the following code, but it does not work. Can anyone help me out on this?
Thanks in advance -Peng
return (from request in
Observable.Return((HttpWebRequest)WebRequest.Create(new Uri(postUrl))).Catch(Observable.Empty<HttpWebRequest>())
.Do(req =>
{
// Set up the request properties
req.Method = "POST";
req.ContentType = contentType;
req.UserAgent = userAgent;
req.CookieContainer = new CookieContainer();
Observable.FromAsyncPattern<Stream>(req.BeginGetRequestStream, req.EndGetRequestStream)()
.ObserveOnDispatcher()
.Subscribe(stream =>
{
stream.Write(formData, 0,
formData.Length);
stream.Close();
})
;
})
from response in
Observable.FromAsyncPattern<WebResponse>(request.BeginGetResponse, request.EndGetResponse)().Catch(Observable.Empty<WebResponse>())
from item in GetPostResponse(response.GetResponseStream()).ToObservable().Catch(Observable.Empty<string>())
select item).ObserveOnDispatcher();
Edit: To make it clear, I want to use the rx to implement the same logic in MSDN example.
in the MSDN example, it seems it first makes async call to write RequestStream, and then in the GetRequestStreamCallback, fires another async call to get the response.
Using Rx, I am able to create 2 observables
1. Observable.FromAsyncPattern(request.BeginGetRequestStream, request.EndGetRequestStream)()
2. Observable.FromAsyncPattern(request.BeginGetResponse, request.EndGetResponse)()
The problem is the second observable depends on the first one's result, so how can I do this in Rx?
In the first observable's subcribe method to create the seond observable? is it the good way?
This is how I am doing it. I configure the two Async patters up front, then use SelectMany to chain them together.
I have cut out the error handling etc from this code to keep it simple and show only the bare minimum to get it working. You should append a .Catch() similar to your own code, and if you want to get more than just a string out (say the response code) then you'll need to create a class/struct to hold all the bits of data you need and return that instead.
public IObservable<string> BeginPost(Uri uri, string postData) {
var request = HttpWebRequest.CreateHttp(uri);
request.Method = "POST";
request.ContentType = "application/x-www-form-urlencoded";
var fetchRequestStream = Observable.FromAsyncPattern<Stream>(request.BeginGetRequestStream, request.EndGetRequestStream);
var fetchResponse = Observable.FromAsyncPattern<WebResponse>(request.BeginGetResponse, request.EndGetResponse);
return fetchRequestStream().SelectMany(stream => {
using (var writer = new StreamWriter(stream)) writer.Write(postData);
return fetchResponse();
}).Select(result => {
var response = (HttpWebResponse)result;
string s = "";
if (response.StatusCode == HttpStatusCode.OK) {
using (var reader = new StreamReader(response.GetResponseStream())) s = reader.ReadToEnd();
}
return s;
});
}
Your problem is your use of Do() here, you need to move the GetRequestStream into your SelectMany (into your "from bla in, from bla in"...), since it only makes sense to get the response stream after you've written the full request. Right now, you're trying to do both concurrently.