Subset equivalency assertion

Description

I have a container (superset) with 3 strings in it. I need to make sure that 2 strings are present in that container. Usually this comparison requires doing either a substring search or a case insensitive comparison. In either case, this involves specifying a lambda predicate for the test in the assertion.

BeSubsetOf() exists, but this requires the equals method be implemented on the objects. It would be ideal to have a similar method for testing for equivalency, especially when the subset check involves complex objects. Something like BeEquivalentSubsetOf()?

Complete minimal example reproducing the issue

Below is an example of how I execute my test today.

var ctx = TestCorrelator.GetLogEventsFromCurrentContext().ToList();
foreach (var log in expectedLogs)
{
    ctx.Should().Contain(evt => evt.MessageTemplate.Text.Contains(log));
}

Expected behavior:

What would be ideal is to do something like this:

var myStrings = new List<string> { "Hello World", "Foo Bar", "No Match" };
var expectedSubstrings = new List<string> { "Foo", "World" };
expectedSubstrings.Should().BeEquivalentSubsetOf(myStrings, (a, b) => b.Contains(a));

Using my earlier example:

expectedLogs.Should().BeEquivalentSubsetOf(TestCorrelator.GetLogEventsFromCurrentContext(),
    (log, evt) => evt.MessageTemplate.Text.Contains(log));

Versions

Using version 5.* with .NET 5

[rcdailey]

For example, I want to verify that a particular method I call logs a warning. There’s other logs in there I don’t care about, like debug and informational logs. But I need to make sure one or two warnings appear in that list.

For those, you can use the overload of collection.Should().Contain() that takes a lambda expression, right?

Sorry I’ve unintentionally over-simplified my example. But what I need is to be able to specify the expected container and a lambda. Right now that option doesn’t seem to exist:

The current version that takes a lambda will stop after the first match. I need it to continue evaluating the lambda until the number of true returns count matches the size of the expected container.

I remember looking at HasCount() to do this, which does take a lambda, but it passes in a count not each element. So it wouldn’t work either.

Furthermore, the dataset returned is a list of some complex ref type. The test should only pass if there exists 2 objects in the list that match a specific subset, and that subset may only look at specific properties on each object (not the full object graph).

I guess you could use two calls of collection.Should().ContainEquivalentOf for that one.

Yeah that’s essentially what I’m doing now, except with a foreach instead of unrolling the loop. It matters more with larger data sets. My examples here are simplified for discussion purposes but I think you get the idea. It’s all about that extra sugar coating. I’ve started to view explicit foreach in a test as “bad” since it isn’t fluent. Maybe that’s self-imposed and wrong of me to think that way.

I’m hoping you can tell me I’m doing something wrong here.

You’re not doing anything wrong here. There are so many different ways of testing. I’m already happy that you are writing unit tests in the first place. And to be fair, I regularly run into scenarios where I wonder if there could not be a simpler way to assert some complicated scenario. But often, either I can come up with a syntax that would be incredibly hard to explain to others, or I simply can’t find anything that the compiler would be able to handle.

Thanks, I needed this reassurance. It’s definitely worth mentioning (and I should have made this clear at the start if I forgot to) that there’s no real issue here. Fluent Assertions is an amazing library that already makes things flow in a more human readable way. At the scale of problems it is solving, this is really just a nitpick.

Part of this was me feeling like I wasn’t fitting into the fluent model very well. But you’re right there’s a lot of variations. A single foreach, at the end of the day, isn’t a big deal.

I very much value and appreciate your feedback. Thank you very much.

[dennisdoomen]

@meeple142 maybe this is what you need? It’s using a custom extension method that tries to deserialize the JSON into an anonymous type defined by an existing anonymous type and then uses FA to compare that anonymous type with itself.