👁 Image
BoxOfYellow.ConsoleMarkdownRenderer.Fakes 0.11.1

ConsoleMarkdownRenderer.Fakes

Test fakes for ConsoleMarkdownRenderer, enabling dependency injection and unit testing of code that displays markdown.

Installation

dotnet add package BoxOfYellow.ConsoleMarkdownRenderer.Fakes

Usage

FakeMarkdownDisplayer implements IMarkdownDisplayer and records all calls for assertion:

using BoxOfYellow.ConsoleMarkdownRenderer;
using BoxOfYellow.ConsoleMarkdownRenderer.Fakes;

var fake = new FakeMarkdownDisplayer();

// Pass the fake wherever IMarkdownDisplayer is expected
await fake.DisplayMarkdownAsync(new Uri("https://example.com/readme.md"));

// Assert calls were made as expected
Assert.AreEqual(1, fake.Calls.Count);
Assert.AreEqual("https://example.com/readme.md", fake.Calls[0].Uri?.ToString());

Both overloads (URI-based and text-based) are recorded:

await fake.DisplayMarkdownAsync("# Hello", new Uri("file:///base/"));
Assert.AreEqual("# Hello", fake.Calls[0].Text);

Asserting no warnings with ValidatingFakeMarkdownDisplayer

ValidatingFakeMarkdownDisplayer implements IMarkdownDisplayer like FakeMarkdownDisplayer, but under the covers each call is delegated to a real MarkdownDisplayer wired up against an isolated Spectre.Console.Testing.TestConsole. That means:

• The validation logic stays in lock-step with the real renderer/displayer — there is no parallel "what counts as a warning" implementation to drift.
• Nothing is written to the surrounding test console; the previous AnsiConsole.Console is restored on exit (even on exception).
• The displayer is forced into the non-interactive code path so it never tries to prompt during validation.

For each call, the fake captures structured warning data from the renderer:

• Unhandled markdown object types — markdown elements no ConsoleObjectRenderer knows how to render (the IncludeDebug "Unhandled <Name>" warning).
• Unusable links — followable links present together with allowFollowingLinks: true, which would trigger the "Non-interactive terminal detected. The following links are available but cannot be followed interactively" warning in CI/non-interactive terminals.
• Unknown emphasis delimiters — emphasis inlines that fall into the catch-all branch in ConsoleEmphasisInlineRenderer (rendered as e.g. (!1)).

using BoxOfYellow.ConsoleMarkdownRenderer;
using BoxOfYellow.ConsoleMarkdownRenderer.Fakes;

var fake = new ValidatingFakeMarkdownDisplayer();

await fake.DisplayMarkdownAsync("# Title\n\nSome **bold** text.", allowFollowingLinks: true);

// Throws MarkdownValidationException if any warning condition is detected.
fake.AssertNoWarnings();

// Or assert each category individually:
fake.AssertNoUnhandledTypes();
fake.AssertNoUnknownEmphasisDelimiters();
fake.AssertNoUnusableLinkWarnings();

You can also inspect the per-call validation results — useful for learning exactly what tripped the unknown-emphasis catch-all:

foreach (var call in fake.Calls)
{
 foreach (var d in call.Validation.UnknownEmphasisDelimiters)
 Console.WriteLine($"Unknown emphasis delimiter: {d.DelimiterChar} x{d.DelimiterCount}");
 foreach (var t in call.Validation.UnhandledTypes)
 Console.WriteLine($"Unhandled markdown type: {t.Name}");
 foreach (var link in call.Validation.FollowableLinks)
 Console.WriteLine($"Followable link: {link.Url}");
}

Recursive validation

Set recursive: true in the constructor and the fake will follow every markdown link discovered in each rendered document, validate it the same way, and record a child call. Visited absolute URIs are tracked per top-level call to avoid cycles. Recursion is bounded by maxDepth (default 10) and maxFiles (default 100) guardrails — AssertNoWarnings() will also fail if either was hit, and you can inspect MaxDepthReached/FilesProcessed/ExceededMaxDepth/ExceededMaxFiles directly.

var fake = new ValidatingFakeMarkdownDisplayer(httpClientFactory, recursive: true);
await fake.DisplayMarkdownAsync(new Uri("https://example.com/index.md"), allowFollowingLinks: true);

fake.AssertNoWarnings(); // covers the root document AND every linked .md
foreach (var call in fake.Calls.Where(c => c.IsRecursive))
 Console.WriteLine($"Recursively validated (depth {call.Depth}): {call.Uri}");

Supplying an IHttpClientFactory

Because each call is delegated to a real MarkdownDisplayer, the URI overload (and recursive link-following) will perform real HTTP requests against any web URI. Hand the fake an IHttpClientFactory to stub those requests out:

var fake = new ValidatingFakeMarkdownDisplayer(myFactory);

A few easy ways to build a factory that maps URIs to canned responses:

• RichardSzalay.MockHttp — mockHttp.When("https://example.com/*.md").Respond("text/plain", "# stub");, then wrap the resulting HttpClient in a one-line IHttpClientFactory.

• A small hand-rolled DelegatingHandler when you don't want a new dependency:

  sealed class StubHttpClientFactory : IHttpClientFactory
  {
   readonly Func<Uri, HttpResponseMessage> _responder;
   public StubHttpClientFactory(Func<Uri, HttpResponseMessage> responder) => _responder = responder;
   public HttpClient CreateClient(string name) => new(new Handler(_responder));
  
   sealed class Handler(Func<Uri, HttpResponseMessage> responder) : HttpMessageHandler
   {
   protected override Task<HttpResponseMessage> SendAsync(HttpRequestMessage req, CancellationToken ct)
   => Task.FromResult(responder(req.RequestUri!));
   }
  }
  

• Moq + a custom HttpMessageHandler if you already use Moq in your test suite.

Thread safety: ValidatingFakeMarkdownDisplayer temporarily swaps AnsiConsole.Console on every call. Like other things that touch AnsiConsole.Console, do not run multiple validations concurrently in the same process.

See Also

• ConsoleMarkdownRenderer README — full library documentation