如何为C＃Azure功能编写单元测试？

Question

I just started writing some C# function code these days and I have to send track event using Application Insight(AI). Here is the sample code that I wrote.

namespace BlobTrigger {
    public static class Main {
        private static string sKey = TelemetryConfiguration.Active.InstrumentationKey = System.Environment.GetEnvironmentVariable("APPINSIGHTS_INSTRUMENTATIONKEY", EnvironmentVariableTarget.Process);
        private static TelemetryClient sTelemetry;
       [FunctionName("BlobTrigger")]
       public static void Run(
            [BlobTrigger("upload/{name}.wav")] Stream myBlob,
            string name,
            Microsoft.Azure.WebJobs.ExecutionContext context,
            TraceWriter log) {

            sTelemetry  = new TelemetryClient() { InstrumentationKey = sKey };
            sTelemetry.Context.Operation.Id = context.InvocationId.ToString();
            sTelemetry.Context.Operation.Name = name;
            sTelemetry.TrackEvent("File is uploaded");
            .....
    }
}

This function works fine. But my problem is writing some unit test for this. I create some mock class for four arguments of the Run method and overrode its method already. This was easy. but I don't know how to mock TelemetryClient#TrackEvent because I NEW that instance in the Run method.

I saw the page below using DI for this but I couldn't understand that how to write unit test properly.

Using Application Insights with Unit Tests?

So can you show me the example unit test code for this?

Answer 1

First of all, Azure Functions supports Application Insights out-of-the-box.

Azure Functions now has direct integration with Application Insights

Therefore, I wouldn't recommend you to directly implement TelemetryClient in your code. Instead, replace your TraceWriter parameter to ILogger to get benefits from Application Insights.

However, if you really want to use the TelemetryClient within your code, I would recommend creating a wrapper interface like ITelemetryClientWrapper , implementing it and injecting it through the dependency injection approach.

I've written a blog post about dependency injection around Azure Functions:

Azure Functions with IoC Container