问题
.NET Core configuration allows so many options to add values (environment variables, json files, command line args).
I just can't figure out and find an answer how to populate it via code.
I am writing unit tests for extension methods to configurations and I thought populating it in the unit tests via code would be easier than loading dedicated json files for each test.
My current code:
[Fact]
public void Test_IsConfigured_Positive()
{
// test against this configuration
IConfiguration config = new ConfigurationBuilder()
// how to populate it via code
.Build();
// the extension method to test
Assert.True(config.IsConfigured());
}
Update:
One special case is the "empty section" which would look like this in json.
{
"MySection": {
// the existence of the section activates something triggering IsConfigured to be true but does not overwrite any default value
}
}
Update 2:
As Matthew pointed out in the comments having an empty section in the json gives the same result as not having the section at all. I distilled an example and yes, that is the case. I was wrong expecting a different behaviour.
So what do I do and what did I expect:
I am writing unit tests for 2 extension methods for IConfiguration (actually because the binding of values in Get...Settings method does not work for some reason (but thats a different topic). They look like this:
public static bool IsService1Configured(this IConfiguration configuration)
{
return configuration.GetSection("Service1").Exists();
}
public static MyService1Settings GetService1Settings(this IConfiguration configuration)
{
if (!configuration.IsService1Configured()) return null;
MyService1Settings settings = new MyService1Settings();
configuration.Bind("Service1", settings);
return settings;
}
My missunderstanding was that if I place an empty section in the appsettings the IsService1Configured() method would return true (which is obviously wrong now). The difference I expected is with having an empty section now the GetService1Settings() method returns null and not as I expected the MyService1Settings with all default values.
Fortunately this still works for me since I won't have empty sections (or now know that I have to avoid those cases). It was just one theoretical case I came across while writing the unit tests.
Further down the road (for those interested in).
For what do I use it? Configuration based service activation/deactivation.
I have an application which has a service / some services compiled into it. Depending on the deployment I need to activate/deactivate the services completly. This is because some (local or testings setups) do not have full access to a complete infrastructure (helper services like caching, metrics...). And I do that via the appsettings. If the service is configured (the config section exists) it will be added. If the config section is not present it will not be used.
The full code for the distilled example is below.
- in Visual Studio create a new API named WebApplication1 from the templates (without HTTPS and Authentication)
- delete the Startup class and appsettings.Development.json
- replace the code in Program.cs with the code below
- now in appsettings.json you can activate/deactivate the services by adding/removing
Service1andService2section
using Microsoft.AspNetCore;
using Microsoft.AspNetCore.Builder;
using Microsoft.AspNetCore.Hosting;
using Microsoft.AspNetCore.Mvc;
using Microsoft.Extensions.Configuration;
using Microsoft.Extensions.DependencyInjection;
using Microsoft.Extensions.Logging;
using Newtonsoft.Json;
using System;
namespace WebApplication1
{
public class MyService1Settings
{
public int? Value1 { get; set; }
public int Value2 { get; set; }
public int Value3 { get; set; } = -1;
}
public static class Service1Extensions
{
public static bool IsService1Configured(this IConfiguration configuration)
{
return configuration.GetSection("Service1").Exists();
}
public static MyService1Settings GetService1Settings(this IConfiguration configuration)
{
if (!configuration.IsService1Configured()) return null;
MyService1Settings settings = new MyService1Settings();
configuration.Bind("Service1", settings);
return settings;
}
public static IServiceCollection AddService1(this IServiceCollection services, IConfiguration configuration, ILogger logger)
{
MyService1Settings settings = configuration.GetService1Settings();
if (settings == null) throw new Exception("loaded MyService1Settings are null (did you forget to check IsConfigured in Startup.ConfigureServices?) ");
logger.LogAsJson(settings, "MyServiceSettings1: ");
// do what ever needs to be done
return services;
}
public static IApplicationBuilder UseService1(this IApplicationBuilder app, IConfiguration configuration, ILogger logger)
{
// do what ever needs to be done
return app;
}
}
public class Program
{
public static void Main(string[] args)
{
CreateWebHostBuilder(args).Build().Run();
}
public static IWebHostBuilder CreateWebHostBuilder(string[] args) =>
WebHost.CreateDefaultBuilder(args)
.ConfigureLogging
(
builder =>
{
builder.AddDebug();
builder.AddConsole();
}
)
.UseStartup<Startup>();
}
public class Startup
{
public IConfiguration Configuration { get; }
public ILogger<Startup> Logger { get; }
public Startup(IConfiguration configuration, ILoggerFactory loggerFactory)
{
Configuration = configuration;
Logger = loggerFactory.CreateLogger<Startup>();
}
// This method gets called by the runtime. Use this method to add services to the container.
public void ConfigureServices(IServiceCollection services)
{
// flavour 1: needs check(s) in Startup method(s) or will raise an exception
if (Configuration.IsService1Configured()) {
Logger.LogInformation("service 1 is activated and added");
services.AddService1(Configuration, Logger);
} else
Logger.LogInformation("service 1 is deactivated and not added");
// flavour 2: checks are done in the extension methods and no Startup cluttering
services.AddOptionalService2(Configuration, Logger);
services.AddMvc().SetCompatibilityVersion(CompatibilityVersion.Version_2_2);
}
// This method gets called by the runtime. Use this method to configure the HTTP request pipeline.
public void Configure(IApplicationBuilder app, IHostingEnvironment env)
{
if (env.IsDevelopment()) app.UseDeveloperExceptionPage();
// flavour 1: needs check(s) in Startup method(s) or will raise an exception
if (Configuration.IsService1Configured()) {
Logger.LogInformation("service 1 is activated and used");
app.UseService1(Configuration, Logger); }
else
Logger.LogInformation("service 1 is deactivated and not used");
// flavour 2: checks are done in the extension methods and no Startup cluttering
app.UseOptionalService2(Configuration, Logger);
app.UseMvc();
}
}
public class MyService2Settings
{
public int? Value1 { get; set; }
public int Value2 { get; set; }
public int Value3 { get; set; } = -1;
}
public static class Service2Extensions
{
public static bool IsService2Configured(this IConfiguration configuration)
{
return configuration.GetSection("Service2").Exists();
}
public static MyService2Settings GetService2Settings(this IConfiguration configuration)
{
if (!configuration.IsService2Configured()) return null;
MyService2Settings settings = new MyService2Settings();
configuration.Bind("Service2", settings);
return settings;
}
public static IServiceCollection AddOptionalService2(this IServiceCollection services, IConfiguration configuration, ILogger logger)
{
if (!configuration.IsService2Configured())
{
logger.LogInformation("service 2 is deactivated and not added");
return services;
}
logger.LogInformation("service 2 is activated and added");
MyService2Settings settings = configuration.GetService2Settings();
if (settings == null) throw new Exception("some settings loading bug occured");
logger.LogAsJson(settings, "MyService2Settings: ");
// do what ever needs to be done
return services;
}
public static IApplicationBuilder UseOptionalService2(this IApplicationBuilder app, IConfiguration configuration, ILogger logger)
{
if (!configuration.IsService2Configured())
{
logger.LogInformation("service 2 is deactivated and not used");
return app;
}
logger.LogInformation("service 2 is activated and used");
// do what ever needs to be done
return app;
}
}
public static class LoggerExtensions
{
public static void LogAsJson(this ILogger logger, object obj, string prefix = null)
{
logger.LogInformation(prefix ?? string.Empty) + ((obj == null) ? "null" : JsonConvert.SerializeObject(obj, Formatting.Indented)));
}
}
}
回答1:
You can use MemoryConfigurationBuilderExtensions to provide it via a dictionary.
var myConfiguration = new Dictionary<string, string>
{
{"Key1", "Value1"},
{"Nested:Key1", "NestedValue1"},
{"Nested:Key2", "NestedValue2"}
}
var configuration = new ConfigurationBuilder()
.AddInMemoryCollection(myConfiguration)
.Build();
回答2:
Would AddInMemoryCollection extension method help?
You can pass a key-value collection into it:
IEnumerable<KeyValuePair<String,String>> with the data you might need for a test.
var builder = new ConfigurationBuilder();
builder.AddInMemoryCollection(new Dictionary<string, string>
{
{ "key", "value" }
});
回答3:
The solution I went for (which answers the question title at least!) is to use a settings file in the solution testsettings.json and set it to "Copy Always".
private IConfiguration _config;
public UnitTestManager()
{
IServiceCollection services = new ServiceCollection();
services.AddSingleton<IConfiguration>(Configuration);
}
public IConfiguration Configuration
{
get
{
if (_config == null)
{
var builder = new ConfigurationBuilder().AddJsonFile($"testsettings.json", optional: false);
_config = builder.Build();
}
return _config;
}
}
回答4:
I prefer not to have my application classes dependent on IConfiguration. Instead I create a configuration class to hold the config, with a constructor that can initialise it from IConfiguration like this:
public class WidgetProcessorConfig
{
public int QueueLength { get; set; }
public WidgetProcessorConfig(IConfiguration configuration)
{
configuration.Bind("WidgetProcessor", this);
}
public WidgetProcessorConfig() { }
}
then in your ConfigureServices, you just have to do:
services.AddSingleton<WidgetProcessorConfig>();
services.AddSingleton<WidgetProcessor>();
and for testing:
var config = new WidgetProcessorConfig
{
QueueLength = 18
};
var widgetProcessor = new WidgetProcessor(config);
回答5:
You can use the following technique to mock IConfiguration.GetValue<T>(key) extension method.
var configuration = new Mock<IConfiguration>();
var configSection = new Mock<IConfigurationSection>();
configSection.Setup(x => x.Value).Returns("fake value");
configuration.Setup(x => x.GetSection("MySection")).Returns(configSection.Object);
//OR
configuration.Setup(x => x.GetSection("MySection:Value")).Returns(configSection.Object);
来源：https://stackoverflow.com/questions/55497800/populate-iconfiguration-for-unit-tests