Null response and cross session issue in chat bot - botframework

I am using the bot framework of V4. using an API call, I'm trying to get data and display it to the user and I defined a custom method to capture data from API and preprocess it, before sending it to the user through waterfall dialog and I made this method async and also using await where it is being called.
There are 2 scenarios where I'm facing the issue -
When two users send questions at an instance one of the responses captured as null instead of value acquired from API.
We are using the suggestive card to display result and when the user clicks on button cross-session was observed sporadically.
Help in this area is much appreciated.
The custom method defined to make API call and get data:
public static async Task<string>
GetEPcallsDoneAsync(ConversationData conversationData)
{
LogWriter.LogWrite("Info: Acessing end-points");
string responseMessage = null;
try
{
conversationData.fulFillmentMap = await
AnchorUtil.GetFulfillmentAsync(xxxxx);//response from API call to get data
if (conversationData.fulFillmentMap == null || (conversationData.fulFillmentMap.ContainsKey("status") && conversationData.fulFillmentMap["status"].ToString() != "200"))
{
responseMessage = "Sorry, something went wrong. Please try again later!";
}
else
{
conversationData.NLGresultMap = await
AnchorUtil.GetNLGAsync(conversationData.fulFillmentMap ,xxxx);//API call to get response to be displayed
if (conversationData.errorCaptureDict.ContainsKey("fulfillmentError") || conversationData.NLGresultMap.ContainsKey("NLGError"))
{
responseMessage = "Sorry, something went wrong:( Please try again later!!!";
}
else
{
responseMessage = FormatDataResponse(conversationData.NLGresultMap["REPLY"].ToString()); //response message
}
}
return responseMessage;
}
catch (HttpRequestException e)
{
LogWriter.LogWrite("Error: " + e.Message);
System.Console.WriteLine("Error: " + e.Message);
return null;
}
}
And the waterfall step of dialog class where the above function is being called:
private async Task<DialogTurnResult> DoProcessInvocationStep(WaterfallStepContext stepContext, CancellationToken cancellationToken)
{
conversationData.index = 0; //var for some other purpose
conversationData.result = await
AnchorUtil.GetEPcallsDoneAsync(conversationData);
await _conversationStateAccessor.SetAsync(stepContext.Context, conversationData, cancellationToken);
return await stepContext.NextAsync(cancellationToken);
}
ConversationData contains variables that are required to process data through waterfall dialogs, values to the object has been set and accessed through accessor as below in each step:
In a dialog class,
public class TopLevelDialog : ComponentDialog
{
private readonly IStatePropertyAccessor<ConversationData> _conversationStateAccessor;
ConversationData conversationData;
public TopLevelDialog(ConversationState conversationState)
: base(nameof(TopLevelDialog))
{
_conversationStateAccessor = conversationState.CreateProperty<ConversationData>(nameof(ConversationData));
AddDialog(new TextPrompt(nameof(TextPrompt)));
AddDialog(new ChoicePrompt(nameof(ChoicePrompt)));
AddDialog(new ReviewSelectionDialog(conversationState));
AddDialog(new ESSelectionDialog());
AddDialog(new WaterfallDialog(nameof(WaterfallDialog), new WaterfallStep[]
{
StartSelectionStepAsync,
GetESResultStep,
DoProcessInvocationStep,
ResultStepAsync,
IterationStepAsync
}));
InitialDialogId = nameof(WaterfallDialog);
}
private async Task<DialogTurnResult> StartSelectionStepAsync (WaterfallStepContext stepContext, CancellationToken cancellationToken)
{
conversationData = await _conversationStateAccessor.GetAsync(stepContext.Context, () => new ConversationData());
//code for functionality
await _conversationStateAccessor.SetAsync(stepContext.Context, conversationData, cancellationToken);
return await stepContext.NextAsync(null, cancellationToken);
}
//other dialog steps
}

Both of your issues likely stem from the same thing. You can't declare conversationData as a class-level property. You'll run into concurrency issues like this as every user will overwrite the conversationData for every other user. You must re-declare conversationData in each step function.
For example,
User A starts the waterfall dialog and gets half of the way through. conversationData is correct at this point and represents exactly what it should.
Now User B starts a dialog. At the StartSelectionStepAsync, they just reset conversationData for everybody because of conversationData = await _conversationStateAccessor.GetAsync(stepContext.Context, () => new ConversationData()); and all users share the same conversationData because of ConversationData conversationData;.
So now, when User A continues their conversation, conversationData will be null/empty.
How State Should Be Saved
// Copyright (c) Microsoft Corporation. All rights reserved.
// Licensed under the MIT License.
const { Channels, MessageFactory } = require('botbuilder');
const {
AttachmentPrompt,
ChoiceFactory,
ChoicePrompt,
ComponentDialog,
ConfirmPrompt,
DialogSet,
DialogTurnStatus,
NumberPrompt,
TextPrompt,
WaterfallDialog
} = require('botbuilder-dialogs');
const { UserProfile } = require('../userProfile');
const ATTACHMENT_PROMPT = 'ATTACHMENT_PROMPT';
const CHOICE_PROMPT = 'CHOICE_PROMPT';
const CONFIRM_PROMPT = 'CONFIRM_PROMPT';
const NAME_PROMPT = 'NAME_PROMPT';
const NUMBER_PROMPT = 'NUMBER_PROMPT';
const USER_PROFILE = 'USER_PROFILE';
const WATERFALL_DIALOG = 'WATERFALL_DIALOG';
/**
* This is a "normal" dialog, where userState is stored properly using the accessor, this.userProfile.
* In this dialog example, we create the userProfile using the accessor in the first step, transportStep.
* We then pass prompt results through the remaining steps using step.values.
* In the final step, summaryStep, we save the userProfile using the accessor.
*/
class UserProfileDialogNormal extends ComponentDialog {
constructor(userState) {
super('userProfileDialogNormal');
this.userProfileAccessor = userState.createProperty(USER_PROFILE);
this.addDialog(new TextPrompt(NAME_PROMPT));
this.addDialog(new ChoicePrompt(CHOICE_PROMPT));
this.addDialog(new ConfirmPrompt(CONFIRM_PROMPT));
this.addDialog(new NumberPrompt(NUMBER_PROMPT, this.agePromptValidator));
this.addDialog(new AttachmentPrompt(ATTACHMENT_PROMPT, this.picturePromptValidator));
this.addDialog(new WaterfallDialog(WATERFALL_DIALOG, [
this.transportStep.bind(this),
this.nameStep.bind(this),
this.nameConfirmStep.bind(this),
this.ageStep.bind(this),
this.pictureStep.bind(this),
this.confirmStep.bind(this),
this.saveStep.bind(this)
]));
this.initialDialogId = WATERFALL_DIALOG;
}
/**
* The run method handles the incoming activity (in the form of a TurnContext) and passes it through the dialog system.
* If no dialog is active, it will start the default dialog.
* #param {*} turnContext
* #param {*} accessor
*/
async run(turnContext, accessor) {
const dialogSet = new DialogSet(accessor);
dialogSet.add(this);
const dialogContext = await dialogSet.createContext(turnContext);
const results = await dialogContext.continueDialog();
if (results.status === DialogTurnStatus.empty) {
await dialogContext.beginDialog(this.id);
}
}
async transportStep(step) {
// Get the userProfile if it exists, or create a new one if it doesn't.
const userProfile = await this.userProfileAccessor.get(step.context, new UserProfile());
// Pass the userProfile through step.values.
// This makes it so we don't have to call this.userProfileAccessor.get() in every step.
step.values.userProfile = userProfile;
// Skip this step if we already have the user's transport.
if (userProfile.transport) {
// ChoicePrompt results will show in the next step with step.result.value.
// Since we don't need to prompt, we can pass the ChoicePrompt result manually.
return await step.next({ value: userProfile.transport });
}
// WaterfallStep always finishes with the end of the Waterfall or with another dialog; here it is a Prompt Dialog.
// Running a prompt here means the next WaterfallStep will be run when the user's response is received.
return await step.prompt(CHOICE_PROMPT, {
prompt: 'Please enter your mode of transport.',
choices: ChoiceFactory.toChoices(['Car', 'Bus', 'Bicycle'])
});
}
async nameStep(step) {
// Retrieve the userProfile from step.values.
const userProfile = step.values.userProfile;
// Set the transport property of the userProfile.
userProfile.transport = step.result.value;
// Pass the userProfile through step.values.
// This makes it so we don't have to call this.userProfileAccessor.get() in every step.
step.values.userProfile = userProfile;
// Skip the prompt if we already have the user's name.
if (userProfile.name) {
// We pass in a skipped bool so we know whether or not to send messages in the next step.
return await step.next({ value: userProfile.name, skipped: true });
}
return await step.prompt(NAME_PROMPT, 'Please enter your name.');
}
async nameConfirmStep(step) {
// Retrieve the userProfile from step.values and set the name property
const userProfile = step.values.userProfile;
// If userState is working correctly, we'll have userProfile.transport from the previous step.
if (!userProfile || !userProfile.transport) {
throw new Error(`transport property does not exist in userProfile.\nuserProfile:\n ${ JSON.stringify(userProfile) }`);
}
// Text prompt results normally end up in step.result, but if we skipped the prompt, it will be in step.result.value.
userProfile.name = step.result.value || step.result;
// step.values.userProfile.name is already set by reference, so there's no need to set it again to pass it to the next step.
// We can send messages to the user at any point in the WaterfallStep. Only do this if we didn't skip the prompt.
if (!step.result.skipped) {
await step.context.sendActivity(`Thanks ${ step.result }.`);
}
// WaterfallStep always finishes with the end of the Waterfall or with another dialog; here it is a Prompt Dialog.
// Skip the prompt if we already have the user's age.
if (userProfile.age) {
return await step.next('yes');
}
return await step.prompt(CONFIRM_PROMPT, 'Do you want to give your age?', ['yes', 'no']);
}
async ageStep(step) {
// Retrieve the userProfile from step.values
const userProfile = step.values.userProfile;
// If userState is working correctly, we'll have userProfile.name from the previous step.
if (!userProfile || !userProfile.name) {
throw new Error(`name property does not exist in userProfile.\nuserProfile:\n ${ JSON.stringify(userProfile) }`);
}
// Skip the prompt if we already have the user's age.
if (userProfile.age) {
// We pass in a skipped bool so we know whether or not to send messages in the next step.
return await step.next({ value: userProfile.age, skipped: true });
}
if (step.result) {
// User said "yes" so we will be prompting for the age.
// WaterfallStep always finishes with the end of the Waterfall or with another dialog; here it is a Prompt Dialog.
const promptOptions = { prompt: 'Please enter your age.', retryPrompt: 'The value entered must be greater than 0 and less than 150.' };
return await step.prompt(NUMBER_PROMPT, promptOptions);
} else {
// User said "no" so we will skip the next step. Give -1 as the age.
return await step.next(-1);
}
}
async pictureStep(step) {
// Retrieve the userProfile from step.values and set the age property
const userProfile = step.values.userProfile;
// We didn't set any additional properties on userProfile in the previous step, so no need to check for them here.
// Confirm prompt results normally end up in step.result, but if we skipped the prompt, it will be in step.result.value.
userProfile.age = step.result.value || step.result;
// step.values.userProfile.age is already set by reference, so there's no need to set it again to pass it to the next step.
if (!step.result.skipped) {
const msg = userProfile.age === -1 ? 'No age given.' : `I have your age as ${ userProfile.age }.`;
// We can send messages to the user at any point in the WaterfallStep. Only send it if we didn't skip the prompt.
await step.context.sendActivity(msg);
}
// Skip the prompt if we already have the user's picture.
if (userProfile.picture) {
return await step.next(userProfile.picture);
}
if (step.context.activity.channelId === Channels.msteams) {
// This attachment prompt example is not designed to work for Teams attachments, so skip it in this case
await step.context.sendActivity('Skipping attachment prompt in Teams channel...');
return await step.next(undefined);
} else {
// WaterfallStep always finishes with the end of the Waterfall or with another dialog; here it is a Prompt Dialog.
var promptOptions = {
prompt: 'Please attach a profile picture (or type any message to skip).',
retryPrompt: 'The attachment must be a jpeg/png image file.'
};
return await step.prompt(ATTACHMENT_PROMPT, promptOptions);
}
}
async confirmStep(step) {
// Retrieve the userProfile from step.values and set the picture property
const userProfile = step.values.userProfile;
// If userState is working correctly, we'll have userProfile.age from the previous step.
if (!userProfile || !userProfile.age) {
throw new Error(`age property does not exist in userProfile.\nuserProfile:\n ${ JSON.stringify(userProfile) }`);
}
userProfile.picture = (step.result && typeof step.result === 'object' && step.result[0]) || 'no picture provided';
// step.values.userProfile.picture is already set by reference, so there's no need to set it again to pass it to the next step.
let msg = `I have your mode of transport as ${ userProfile.transport } and your name as ${ userProfile.name }`;
if (userProfile.age !== -1) {
msg += ` and your age as ${ userProfile.age }`;
}
msg += '.';
await step.context.sendActivity(msg);
if (userProfile.picture && userProfile.picture !== 'no picture provided') {
try {
await step.context.sendActivity(MessageFactory.attachment(userProfile.picture, 'This is your profile picture.'));
} catch (err) {
await step.context.sendActivity('A profile picture was saved but could not be displayed here.');
}
}
// WaterfallStep always finishes with the end of the Waterfall or with another dialog; here it is a Prompt Dialog.
return await step.prompt(CONFIRM_PROMPT, { prompt: 'Would you like me to save this information?' });
}
async saveStep(step) {
if (step.result) {
// Get the current profile object from user state.
const userProfile = step.values.userProfile;
// Save the userProfile to userState.
await this.userProfileAccessor.set(step.context, userProfile);
await step.context.sendActivity('User Profile Saved.');
} else {
// Ensure the userProfile is cleared
await this.userProfileAccessor.set(step.context, {});
await step.context.sendActivity('Thanks. Your profile will not be kept.');
}
// WaterfallStep always finishes with the end of the Waterfall or with another dialog; here it is the end.
return await step.endDialog();
}
async agePromptValidator(promptContext) {
// This condition is our validation rule. You can also change the value at this point.
return promptContext.recognized.succeeded && promptContext.recognized.value > 0 && promptContext.recognized.value < 150;
}
async picturePromptValidator(promptContext) {
if (promptContext.recognized.succeeded) {
var attachments = promptContext.recognized.value;
var validImages = [];
attachments.forEach(attachment => {
if (attachment.contentType === 'image/jpeg' || attachment.contentType === 'image/png') {
validImages.push(attachment);
}
});
promptContext.recognized.value = validImages;
// If none of the attachments are valid images, the retry prompt should be sent.
return !!validImages.length;
} else {
await promptContext.context.sendActivity('No attachments received. Proceeding without a profile picture...');
// We can return true from a validator function even if Recognized.Succeeded is false.
return true;
}
}
}
module.exports.UserProfileDialogNormal = UserProfileDialogNormal;

Related

Xamarin Forms picker SelectedIndexChange null exception

I have created an application that asks for a username and returns the stats for this user. When a counter is more than 1 it focuses on a picker and asks to choose one of the options. After, the SelectedIndexChanged (Platformselect_SelectedIndexChanged) method is being called and gets the stats. The problem is that if the user enters a username, the first time it works. When the user goes back and re-enters the username it doesn't focus (it doesn't show the picker) and automatically calls the SelectedIndexChanged method with platformselect.SelectedIndex being -1 causing my program to crash. Could you please help me?
I have tried to set focus on the main thread, I have tested it in both iOS and Android and the same problem appears.
public partial class MainPage : ContentPage
{
public MainPage()
{
InitializeComponent();
platformselect.HeightRequest = 0;
this.search.Clicked += Button_Clicked;
}
public async void searchStats(string usernameText)
{
search.IsEnabled = false;
activityIndicator.IsRunning = true;
var id = await GetDataLogic.GetID(usernameText);
publicid = id;
activityIndicator.IsRunning = false;
if (id.uid == "error")
{
await DisplayAlert("Error", "Could not find username: " + usernameText, "OK");
search.IsEnabled = true;
return;
}
else if (id.uid == "errornointernet")
{
await DisplayAlert("Error", "Could not connect to the Internet. Please check your connection and try again.", "OK");
search.IsEnabled = true;
return;
}
else
{
if(id.platforms.Count > 1)
{
platformselect.ItemsSource = publicid.platforms.ToList();
//we have to show alert to choose platform and then call get user stats.
**Device.BeginInvokeOnMainThread(() =>
{
if (platformselect.IsFocused)
platformselect.Unfocus();
platformselect.Focus();
});**
}
else
{
//we call it without asking because there is only one platform.
var userstats = await GetDataLogic.GetStats(id, 0);
await Navigation.PushAsync(new Stats(userstats, favorites));
}
}
}
private async void Platformselect_SelectedIndexChanged(object sender, EventArgs e)
{
//the second time it doesnt show the picker and it automatically enters -1... it returns null exception. Has to be investigated.
var userstats = await GetDataLogic.GetStats(publicid, platformselect.SelectedIndex);
await Navigation.PushAsync(new Stats(userstats, favorites));
}
}
I would like it to always focus on the picker and wait for the response for the user, not automatically call SelectedIndexChanged with -1 as the selectedindex
I suggest change your event handler this way
private async void Platformselect_SelectedIndexChanged(object sender, EventArgs e)
{
//Check for unselection event
if(platformselect.SelectedIndex == -1) return;
var userstats = await GetDataLogic.GetStats(publicid, platformselect.SelectedIndex);
await Navigation.PushAsync(new Stats(userstats, favorites));
}

How to flag that an exception has been handled in bot code?

I have a root dialog that creates a child dialog like so...
private async Task MessageReceivedAsync(IDialogContext context, IAwaitable<IMessageActivity> result)
{
var message = await result;
string userName = context.Activity?.From.Name;
var customerForm = new FormDialog<CarValuationDialog>(
new CarValuationDialog(userName),
() => CarValuationDialog.BuildForm(),
FormOptions.PromptInStart);
context.Call(customerForm, FormSubmitted);
}
The FormSubmitted method looks like....
public async Task FormSubmitted(IDialogContext context, IAwaitable<CarValuationDialog> result)
{
try
{
var form = await result;
}
catch (FormCanceledException<CarValuationDialog> e)
{
string reply;
if (e.InnerException == null)
{
reply = e.Message;
}
else
{
reply = e.InnerException.Message;
}
context.Reset();
await context.PostAsync(reply);
}
}
When an exception occurs in the child dialog, the method FormSubmitted is executed and goes into the catch block. However, when that method finishes, I still see the "Sorry my bot had an issue" type message appear to the user.
How can I tell the bot code not to fire the unhandled exception code, I believe is in PostUnhandledExceptionToUser? Is there a flag type property I need to set to true or something?
It looks like you are making the dialog stack empty when you have an exception: you should not have this context.Reset() below:
public async Task FormSubmitted(IDialogContext context, IAwaitable<CarValuationDialog> result)
{
try
{
var form = await result;
}
catch (FormCanceledException<CarValuationDialog> e)
{
string reply;
if (e.InnerException == null)
{
reply = e.Message;
}
else
{
reply = e.InnerException.Message;
}
context.Reset();
await context.PostAsync(reply);
}
}
Remove this line and the next message will be handled by your root dialog
in your message controller in POST method, use defaultifexception
await Conversation.SendAsync(activity, () => new Dialogs.DialogLUIS().DefaultIfException());

Await Async : how to get one method completed before the second one starts

How to I make the call so that the GetRecords is completed before _reportViewerService.ShowReport starts. Using SignalR the setUi updates a txt field which displays names from part of the result calculated in getRecords, and the rest should be printed in report there after.
(whats happening now is both running the same time, then the report being showed before I see the live update status)
Thanks in advance
public async Task ViewReport()
{
var reportData = await _apiCallExecutor.ExecuteAsync(new GetRecords(queryModel, setUiHooks));
try
{
if (reportData.Count > 0)
{
var settings = new ReportSettings();
settings.ReportPath = "Utilities/SetDeliveryIdByBatchReport";
settings.ReportTitle = "Set Delivery ID By Batch - Exception Listing";
settings.DataSources.Add("DeliveryIdExceptionRecords", reportData);
ReportStatus = "Printing Exception Report...";
await _reportViewerService.ShowReport(settings);
}
}
finally
{
ViewModelState = ViewModelStates.Edit;
}
ReportStatus = "Done...";
}
You want to use some kind of "signal" from GetRecords, e.g., an IObservable or Task that completes when the data has arrived.
class GetRecords
{
...
public Task Done { get; }
// or: public IObservable<Unit> Done { get; }
}
then:
var getRecords = new GetRecords(queryModel, setUiHooks);
var reportData = await _apiCallExecutor.ExecuteAsync(getRecords);
await getRecords.Done;
...

How to allow the user only if he/she type the “ivr” or “IVR” using Form Flow concept

I am working on bot framework technology, in one of my current project I want to allow the user only if he/she type the ‘’ivr” or “IVR” otherwise it shows some feedback to the user.
For that I have wrote below lines of code, but this code shows some wrong output to the user. Even if the user enter ivr or IVR it shows feedback to the user for the first time, but from second time onwards its working correctly.
[Serializable]
class Customer
{
//Create Account Template
[Prompt("Please send any of these commands like **IVR** (or) **ivr**.")]
public string StartingWord;
public static IForm<Customer> BuildForm()
{
OnCompletionAsyncDelegate<Customer> accountStatus = async (context, state) =>
{
await Task.Delay(TimeSpan.FromSeconds(5));
await context.PostAsync("We are currently processing your account details. We will message you the status.");
};
var builder = new FormBuilder<Customer>();
return builder
//.Message("Welcome to the BankIVR bot! To start an conversation with this bot send **ivr** or **IVR** command.\r \n if you need help, send the **Help** command")
.Field(nameof(Customer.StartingWord), validate: async (state, response) =>
{
var result = new ValidateResult { IsValid = true, Value = response };
string str = (response as string);
if (str.ToLower() != "ivr")
{
result.Feedback = "I'm sorry. I didn't understand you.";
result.IsValid = false;
return result;
}
else if (str.ToLower() == "ivr")
{
result.IsValid = true;
return result;
}
else
{
return result;
}
})
.OnCompletion(accountStatus)
.Build();
}
};
Please tell me how to resolve this issue using Form Flow concept.
-Pradeep
Your code looks correct to me - I can only suggest you debug your code with a step-through debugger and see where the logic tests are failing.
That said, if it's not working for people in Turkey, it's because you shouldn't use .ToLower() for normalizing text, for example the .ToLower() method does not work for text that contains the Turkish dotless 'I' character: http://archives.miloush.net/michkap/archive/2004/12/02/273619.html
Also, your else case will never be hit because your two prior checks (!= and ==) cover every possible case (the C# compiler is currently not sophisticated enough to flag the else case as unreachable code).
The correct way to do a case-insensitive comparison is with String.Equals:
if( "ivr".Equals( str, StringComparison.InvariantCultureIgnoreCase ) ) {
result.IsValid = true;
return result;
}
else {
result.Feedback = "I'm sorry. I didn't understand you.";
result.IsValid = false;
}
Finally, I got the result with out any issue.
here is my updated code for to allow only the user enter "ivr or IVR" word, to start a form flow conversation with bot.
.Field(nameof(Customer.StartingWord), validate: async (state, response) =>
{
var result = new ValidateResult { IsValid = true, Value = response };
string str = (response as string);
if ("ivr".Equals(str, StringComparison.InvariantCultureIgnoreCase))
{
//result.IsValid = true;
//return result;
}
else
{
result.Feedback = "I'm sorry. I didn't understand you.";
result.IsValid = false;
//return result;
}
return result;
})
-Pradeep

Azure Notification Hub and WP8 Intermitant notifications

This is a fairly long piece of code but I am getting nowhere with this and cannot see any issues, although I am new to using notification hubs. I am trying to register for targeted notifications (the logged on user) using the notification hub in Azure. After the registration, a test notification is sent.
The issue I am having is that sometimes the notification is sent to the device, and sometimes it is not. It mostly isn't but occasionally when I step through the code on the server, i will get the notification on the emulator come through. Once when I deployed the app to my phone the notification came though on the emulator! I cannot discover a pattern.
My Controller class looks like this;
private NotificationHelper hub;
public RegisterController()
{
hub = NotificationHelper.Instance;
}
public async Task<RegistrationDescription> Post([FromBody]JObject registrationCall)
{
var obj = await hub.Post(registrationCall);
return obj;
}
And the helper class (which is used elsewhere so is not directly in the controller) looks like this;
public static NotificationHelper Instance = new NotificationHelper();
public NotificationHubClient Hub { get; set; }
// Create the client in the constructor.
public NotificationHelper()
{
var cn = "<my-cn>";
Hub = NotificationHubClient.CreateClientFromConnectionString(cn, "<my-hub>");
}
public async Task<RegistrationDescription> Post([FromBody] JObject registrationCall)
{
// Get the registration info that we need from the request.
var platform = registrationCall["platform"].ToString();
var installationId = registrationCall["instId"].ToString();
var channelUri = registrationCall["channelUri"] != null
? registrationCall["channelUri"].ToString()
: null;
var deviceToken = registrationCall["deviceToken"] != null
? registrationCall["deviceToken"].ToString()
: null;
var userName = HttpContext.Current.User.Identity.Name;
// Get registrations for the current installation ID.
var regsForInstId = await Hub.GetRegistrationsByTagAsync(installationId, 100);
var updated = false;
var firstRegistration = true;
RegistrationDescription registration = null;
// Check for existing registrations.
foreach (var registrationDescription in regsForInstId)
{
if (firstRegistration)
{
// Update the tags.
registrationDescription.Tags = new HashSet<string>() {installationId, userName};
// We need to handle each platform separately.
switch (platform)
{
case "windows":
var winReg = registrationDescription as MpnsRegistrationDescription;
winReg.ChannelUri = new Uri(channelUri);
registration = await Hub.UpdateRegistrationAsync(winReg);
break;
case "ios":
var iosReg = registrationDescription as AppleRegistrationDescription;
iosReg.DeviceToken = deviceToken;
registration = await Hub.UpdateRegistrationAsync(iosReg);
break;
}
updated = true;
firstRegistration = false;
}
else
{
// We shouldn't have any extra registrations; delete if we do.
await Hub.DeleteRegistrationAsync(registrationDescription);
}
}
// Create a new registration.
if (!updated)
{
switch (platform)
{
case "windows":
registration = await Hub.CreateMpnsNativeRegistrationAsync(channelUri,
new string[] {installationId, userName});
break;
case "ios":
registration = await Hub.CreateAppleNativeRegistrationAsync(deviceToken,
new string[] {installationId, userName});
break;
}
}
// Send out a test notification.
await SendNotification(string.Format("Test notification for {0}", userName), userName);
return registration;
And finally, my SendNotification method is here;
internal async Task SendNotification(string notificationText, string tag)
{
try
{
var toast = PrepareToastPayload("<my-hub>", notificationText);
// Send a notification to the logged-in user on both platforms.
await NotificationHelper.Instance.Hub.SendMpnsNativeNotificationAsync(toast, tag);
//await hubClient.SendAppleNativeNotificationAsync(alert, tag);
}
catch (ArgumentException ex)
{
// This is expected when an APNS registration doesn't exist.
Console.WriteLine(ex.Message);
}
}
I suspect the issue is in my phone client code, which is here and SubscribeToService is called immediately after WebAPI login;
public void SubscribeToService()
{
_channel = HttpNotificationChannel.Find("mychannel");
if (_channel == null)
{
_channel = new HttpNotificationChannel("mychannel");
_channel.Open();
_channel.BindToShellToast();
}
_channel.ChannelUriUpdated += async (o, args) =>
{
var hub = new NotificationHub("<my-hub>", "<my-cn>");
await hub.RegisterNativeAsync(args.ChannelUri.ToString());
await RegisterForMessageNotificationsAsync();
};
}
public async Task RegisterForMessageNotificationsAsync()
{
using (var client = GetNewHttpClient(true))
{
// Get the info that we need to request registration.
var installationId = LocalStorageManager.GetInstallationId(); // a new Guid
var registration = new Dictionary<string, string>()
{
{"platform", "windows"},
{"instId", installationId},
{"channelUri", _channel.ChannelUri.ToString()}
};
var request = new HttpRequestMessage(HttpMethod.Post, new Uri(ApiUrl + "api/Register/RegisterForNotifications"));
request.Content = new StringContent(JsonConvert.SerializeObject(registration), Encoding.UTF8, "application/json");
string message;
try
{
HttpResponseMessage response = await client.SendAsync(request);
message = await response.Content.ReadAsStringAsync();
}
catch (Exception ex)
{
message = ex.Message;
}
_registrationId = message;
}
}
Any help would be greatly appriciated as I have been stuck on this now for days! I know this is a lot of code to paste up here but it is all relevant.
Thanks,
EDIT: The SubscribeToService() method is called when the user logs in and authenticates with the WebAPI. The method is here;
public async Task<User> SendSubmitLogonAsync(LogonObject lo)
{
_logonObject = lo;
using (var client = GetNewHttpClient(false))
{
var logonString = String.Format("grant_type=password&username={0}&password={1}", lo.username, lo.password);
var sc = new StringContent(logonString, Encoding.UTF8);
var response = await client.PostAsync("Token", sc);
if (response.IsSuccessStatusCode)
{
_logonResponse = await response.Content.ReadAsAsync<TokenResponseModel>();
var userInfo = await GetUserInfoAsync();
if (_channel == null)
SubscribeToService();
else
await RegisterForMessageNotificationsAsync();
return userInfo;
}
// ...
}
}
I have solved the issue. There are tons of fairly poorly organised howto's for azure notification hubs and only one of them has this note toward the bottom;
NOTE:
You will not receive the notification when you are still in the app.
To receive a toast notification while the app is active, you must
handle the ShellToastNotificationReceived event.
This is why I was experiencing intermittent results, as i assumed you would still get a notification if you were in the app. And this little note is pretty well hidden.
Have you used proper tag / tag expressions while register/send the message. Also, Where are you storing the id back from the notification hub. It should be used when you update the channel uri (it will expire).
I would suggest to start from scratch.
Ref: http://msdn.microsoft.com/en-us/library/dn530749.aspx

Resources