WPF – j2i.net

Connecting to AWS IoT’s MQTT Server with .Net

A client I was working with wanted communication between systems for a solution to use AWS IoT for broadcasting messages among the computers making up the solution and controlling various computers. I worked on the solution for the system, but had minimal access to their AWS resources, which is consisten with their security policies. Usually, if I have access to the AWS subscription I could use an MQTT viewer that is part of the service for performing some diagnostic tasks. With this client, I didn’t have that access and had to make my own viewer when performing diagnostics.

Making a viewer is pretty easy once you have the resources that you need. I chose WPF because of the speed at which a functional UI could be build along with M2Mqtt as an MQTT client. Before writing any code, there was some work that needed to be done with the certificates. When accessing an AWS MQTT Instance, the information that you will need includes the domain name for the MQTT instance, an Amazon root certificate, a certificate and private key file. You’ll need this information packaged in a pfx file to easily use it with M2Mqtt. Packaging these certificates that way is just a matter of running a command. For the files I had, let’s use these names.

000-certificate.pem.crt – The certificate file for the AWS MQTT Instance
000-private.pem.key – The private key for the AWS MQTT instance
AmazonRootCA1.cer – The AWS root certificate

Using those file names, the command that I used was as follows.

openssl pkcs12 -export -in .\000-certificate.pem.crt -inkey .\000-private.pem.key -out certificate.pfx -certfile .\AmazonRootCA1.cer

After this command runs, I have a file named certificate.pfx. I’ll be using this in my .Net viewer.

In the interest of keeping the program reusable, I’ve placed information on the paths to the certificate files in the applications settings. If I needed to change these files post-compilation, they are in a JSON file. These settings include the following.

BrokerDomainName – The domain name of the MQTT Instance that the application connects to
BrokerCertificatePath – The relative file path to the
BrokerRootCertificatePath – The relative path to the AWS root certificate
BrokerPort – the port that the MQTT Instance is using. Amazon uses port 883
ClientPrefix – Prefix for the name that will be used for identifying this client.
DefaultTopic – The topic that the client will automatically subscribe to upon starting

Concerning the client prefix, every client is identified by a unique string. To ensure the string is unique, I am generating a GUID when the application starts. It’s possible that someone starts more than one instance of the program. For this reason I don’t persist the GUID. A second instance of the program will have a different GUID. But I still want the string to be recognizable as having come from this program. The ClientPrefix string puts a recognizable string before the GUID to satisfy this need.

The paths in the settings are relative to the application. To load the certificates using these paths and to generate a new client name, I use the following code.

Settings ds = Settings.Default;
var rootCertificatePath = Path.Combine(AppDomain.CurrentDomain.BaseDirectory, ds.BrokerRootCertificatePath);
var deviceCertificatePath = Path.Combine(AppDomain.CurrentDomain.BaseDirectory, ds.BrokerCertificatePath);
var rootCertificate = X509Certificate.CreateFromCertFile(rootCertificatePath);
var deviceCertificate = X509Certificate.CreateFromCertFile(deviceCertificatePath);
clientName = $"{ds.ClientPrefix}-${Guid.NewGuid().ToString()}";

With that, we have all the information that we need for connecting to the MQTT broker. We can instantiate a MqttClient, subscribe to it’s events, and start showing the message topics. The call to establish a connection is a blocking call. When it returns, a connection will have been established.

client = new MqttClient(ds.BrokerDomainName, ds.BrokerPort, true, rootCertificate, deviceCertificate, MqttSslProtocols.TLSv1_2);
client.MqttMsgSubscribed += Client_MqttMsgSubscribed;            
client.MqttMsgPublishReceived += Client_MqttMsgPublishReceived;
try
{
    client.Connect(clientName);

    var defaultTopic = ds.DefaultTopic;
    if (!String.IsNullOrEmpty(defaultTopic))
    {
        client.Subscribe(new string[] { defaultTopic }, new byte[] { MqttMsgBase.QOS_LEVEL_AT_LEAST_ONCE });
    }
} catch(Exception ex)
{
    MessageBox.Show(ex.Message, "Could not connect");
}

The primary information of concern in a message is the topic. The message may also have a payload. I capture both of those items. It is possible that the payload contains data that isn’t a string. I’ve decided not to show it, but captured it anyway should I change my mind. To hold the information I use the following class.

public class ReceivedMessage: ViewModelBase
{

    private string _topic;
    public String Topic { 
        get { return _topic; }
        set
        {
            SetValueIfChanged(() => Topic, () => _topic, value);
        }
    }

    private byte[] _payload;
    public byte[] Payload
    {
        get { return _payload; }
        set
        {
            SetValueIfChanged(() => Payload, () => _payload, value);
        }
    }

    DateTimeOffset _timestamp = DateTimeOffset.Now;
    public DateTimeOffset Timestamp
    {
        get { return _timestamp; }
        set
        {
            SetValueIfChanged(()=>Timestamp, () => _timestamp, value);
        }
    }
}

The base class from which this derives, ViewModelBase, is something I’ve talked about before. See this post if you need more information on how this base class allows this class to be bound to the UI. As new messages come in, I add their content to instances of ReceivedMessage and add them to an ObservableCollection. The collection is bound to a ListView on the UI. The entirety of the main UI window declaration follows. This UI uses only the default code for its code-behind.

<Window x:Class="MessageWatcher.MainWindow"
        xmlns="http://schemas.microsoft.com/winfx/2006/xaml/presentation"
        xmlns:x="http://schemas.microsoft.com/winfx/2006/xaml"
        xmlns:d="http://schemas.microsoft.com/expression/blend/2008"
        xmlns:mc="http://schemas.openxmlformats.org/markup-compatibility/2006"
        xmlns:vm="clr-namespace:MessageWatcher.ViewModels"
        xmlns:local="clr-namespace:MessageWatcher"
        mc:Ignorable="d"
        Title="MainWindow" Height="450" Width="800">

    <Grid>
        <Grid.DataContext>
            <vm:MainViewModel />
        </Grid.DataContext>

        <ListView ItemsSource="{Binding MessageList}">
            <ListView.ItemTemplate>
                <DataTemplate>
                    <Grid>
                        <TextBlock Text="{Binding Topic}" />
                    </Grid>
                </DataTemplate>
            </ListView.ItemTemplate>
        </ListView>
    
    </Grid>
</Window>

With that, I had a working viewer for monitoring the messages as they went by.

Posts may contain products with affiliate links. When you make purchases using these links, we receive a small commission at no extra cost to you. Thank you for your support.

Mastodon: @j2inet@masto.ai
Instagram: @j2inet
Facebook: @j2inet
YouTube: @j2inet
Telegram: j2inet
Twitter: @j2inet

Updated ViewModelBase for my WPF Projects

When I’m working on projects that use WPF, there’s a set of base classes that I usually use for classes that are bindable to the UI. Back in 2016 I shared the code for these classes. In 2018, I shared an updated version of these classes for UWP. I have another updated version of these classes that I’m sharing in part because it is used in some code for a future post.

For those unfamiliar, the WPF derived Windows UI technologies support binding UI elements to code classes so that the UI is automatically updated with a value in the class changes. Classes that are bound to the UI must implement the INotifyPropertyChanged interface. This interface defines a single event of type PropertyChangedEventHandler. This event is raised to indicate that a property on the class has changed. The event data contains the name of the class member that has changed. UI elements bound to that member get updated in response.

The class that I’ve defined, ViewModelBase, eases management of that event. The PropertyChanged event expects the name of the property to be passed as a string. If the name of the property is passed directly as a string in code, there’s opportunity for bugs from mistyping the property name. Or, if a property name is changed, there is the possibility that a string value is not appropriately updated. To reduce the possibility of such bugs occurring the ViewModelBase that I’m using uses Expressions instead of strings. With a bit of reflection, I extract the name of the field/property referenced, which gaurantees that only valid values are used. If something invalid is passed it would result in failure at compile time. Using an expression allows Intellisense to be used which reduces the chance of typos. If a field or property name is changed through the VisualStudio IDE, these expressions are recognized too. While I did this in previous versions of this classes, it was necessary to make some updates to how it works based on changes in .Net.

Additionally, this version of the class can be used in code with multiple threads. It uses a Dispatcher to ensure that the PropertyChanged events are routed to the UI thread before being raised.

public abstract class ViewModelBase : INotifyPropertyChanged
{
    public ViewModelBase()
    {
        this.Dispatcher = Dispatcher.CurrentDispatcher;
    }
    [IgnoreDataMember]
    [JsonIgnore]
    public Dispatcher Dispatcher { get; set; }

    protected void DoOnMain(Action a)
    {
        Dispatcher.Invoke(a);
    }

    public event PropertyChangedEventHandler PropertyChanged;
    protected void OnPropertyChanged(String propertyName)
    {
        if (PropertyChanged != null)
        {
            Action a = () => { PropertyChanged(this, new PropertyChangedEventArgs(propertyName)); };
            if (Dispatcher != null)
            {
                Dispatcher.Invoke(a);
            }
            else
            {
                a();
            }
        }
    }

    protected void OnPropertyChanged(Expression expression)
    {
        OnPropertyChanged(((MemberExpression)expression).Member.Name);
    }

    protected bool SetValueIfChanged<T>(Expression<Func<T>> propertyExpression, Expression<Func<T>> fieldExpression, object value)
    {
        var property = (PropertyInfo)((MemberExpression)propertyExpression.Body).Member;
        var field = (FieldInfo)((MemberExpression)fieldExpression.Body).Member;
        return SetValueIfChanged(property, field, value);
    }

    protected bool SetValueIfChanged(PropertyInfo pi, FieldInfo fi, object value)
    {
        var currentValue = pi.GetValue(this, new object[] { });
        if ((currentValue == null && value == null) || (currentValue != null && currentValue.Equals(value)))
            return false;
        fi.SetValue(this, value);
        OnPropertyChanged(pi.Name);
        return true;
    }
}

Commands are classes that implement the ICommand interface. ICommand objects are bound to UI elements such as buttons and used for invoking code. This interfaces defines three members. There is an event named CanExecuteChange and two methods, CanExecuteChange() and Execute(). The Execute() method contains the code that is meant to be invoked when the user clicks on the button. CanExecute() returns true or false to indicate whether the command should be presently invokable. You may want a command to not be invokable until the user has satisfied certain conditions. For example, on a UI that uploads a file might want the upload button disabled until the user selected a file, and that file is a valid file. For that scenario, your implementation of CanExecute() would check if a file had been selected and if the file is valid and only return true if both conditions are satisfied. Or, if the command started a long running task for which only one instance is allowed, the CanExecute() function may only return true if there are now instances of the task running.

In this code, I’ve got a DelegateCommand class defined. This class has been around in its current form for longer than I remember. I believe it was originally in something that Microsoft provided, and it hasn’t changed much. To facilitate creating objects that contain the code, the DelegateCommand class acceptions an action (an action is a function that can be easily passed around as an object). The constructor for the class accepts either just an action to execute, or an action and a function used by CanExecute() (if no function is passed for CanExecute() it is assumed that the command can always be execute.

There are two versions of the DelegateCommand class. One is a Generic implementation used when the execute command must also accept a typed parameter. The non-generic implementation receives an Object parameter.

    public class DelegateCommand : ICommand
    {
        public DelegateCommand(Action execute)
            : this(execute, null)
        {
        }

        public DelegateCommand(Action execute, Func<object, bool> canExecute)
        {
            _execute = execute;
            _canExecute = canExecute;
        }

        public bool CanExecute(object parameter)
        {
            if (_canExecute != null)
                return _canExecute(parameter);
            return true;
        }

        public void Execute(object parameter)
        {
            _execute();
        }

        public void RaiseCanExecuteChanged()
        {
            if (CanExecuteChanged != null)
                CanExecuteChanged(this, EventArgs.Empty);
        }

        public event EventHandler CanExecuteChanged;

        private Action _execute;
        private Func<object, bool> _canExecute;
    }


    public class DelegateCommand<T> : ICommand
    {
        public DelegateCommand(Action<T> execute)
            : this(execute, null)
        {
        }

        public DelegateCommand(Action<T> execute, Func<T, bool> canExecute)
        {
            _execute = execute;
            _canExecute = canExecute;
        }

        public bool CanExecute(object parameter)
        {
            if (_canExecute != null)
            {

                return _canExecute((T)parameter);
            }

            return true;
        }

        public void Execute(object parameter)
        {
            _execute((T)parameter);
        }

        public void RaiseCanExecuteChanged()
        {
            if (CanExecuteChanged != null)
                CanExecuteChanged(this, EventArgs.Empty);
        }

        public event EventHandler CanExecuteChanged;

        private Action<T> _execute;
        private Func<T, bool> _canExecute;
    }

By itself, this code might not be meaningful. But I’ll be referencing it in future content, starting with a post scheduled for a couple of weeks after this one.

Posts may contain products with affiliate links. When you make purchases using these links, we receive a small commission at no extra cost to you. Thank you for your support.

Mastodon: @j2inet@masto.ai
Instagram: @j2inet
Facebook: @j2inet
YouTube: @j2inet
Telegram: j2inet
Twitter: @j2inet