Hi! I'm struggling with UI tests... I'm trying to get all nodes with a certain content description, and I want to check all these nodes are displayed. I don't know in advance how many nodes there can be, so I can't assert on the count.

composeTestRule
    .onAllNodesWithContentDescription("Foobar")
    .assertAreDisplayed() // <-- I want this kind of function

Any idea?

1. strange test if you don’t know the expected number of nodes. 2. What is hindering you on using a foreach loop in the nodes?

1. server-driven UI, anything is possible 🤷 2. what do you suggest? AFAIK, there are 2 options to get nodes a. semantics nodes by doing something like that

composeTestRule
  .onAllNodesWithContentDescription("Foobar")
  .fetchSemanticsNodes()

But I get a

List<SemanticsNode>

, I don't know what to do with that to call

assertIsDisplayed()

b.

SemanticsNodeInteractionCollection

by doing this

composeTestRule
  .onAllNodesWithContentDescription("Foobar")

But there is nothing interesting I can do with

SemanticsNodeInteractionCollection

if I don't expect any particular size.

Ah, okay. But on the collection there is an

assertAll

that you can pass a matcher. Which in your case will be the

isDisplayed

isDisplayed()

doesn't seem to exist

Yeah, it was just a „placeholder“ for a matcher that should exist.

That's the whole problem 😄

There’s also a

fetchSemanticNodes

which return the list of nodes that you can definitely assert with `isDisplayed`()

Oh, I just had an idea

private fun SemanticsNodeInteractionCollection.assertAreDisplayed(): SemanticsNodeInteractionCollection {
        val semNodes = fetchSemanticsNodes()
        semNodes.forEachIndexed { index, _ ->
            val semanticsNodeInteraction = get(index)
            semanticsNodeInteraction.assertIsDisplayed()
        }
        return this
    }

It should work, I guess

👍🏻

Thanks for your help 👍

Looks like you solved your issue, but I am wondering about this:

server-driven UI, anything is possible 🤷

We use our own flavor of server-driven UI, but tests are still deterministic. The server responses are hardcoded and thus we know the number of elements that a particular test should have. Why is that not the case for you? Are you using a real server for your UI tests? Does your test involve testing for a random number of elements?

What I meant by "anything is possible" is that the server can tell the frontend to display 3 times the same image in different places. In my tests I'm not able to determine this count, but when I build the UI I will give the same content description to each image, as they are the same.