Replace usages of AssertLegacyTrait::assertNoText, which is deprecated

This problem is being caused by case sensitivity and not [3175718]. \Drupal\FunctionalTests\AssertLegacyTrait::assertNoText() is case sensitive - it uses \PHPUnit\Framework\Assert::assertStringNotContainsString(). The replacement code eventually uses \Behat\Mink\WebAssert::elementTextNotContains which does $regex = '/'.preg_quote($text, '/').'/ui'; and then $this->assertElementText(!preg_match($regex, $actual), $message, $element);

It's true that [3175718] does help because it removes words you would not expect to be there and I also think that doing a case insensitive check is a good idea because asserting for missing text is fraught enough without having to worry about that.

+++ b/core/tests/Drupal/Tests/WebAssert.php
@@ -894,7 +894,7 @@ public function pageTextNotContains($text) {
-    return parent::pageTextNotContains($text);
+    return $this->elementTextNotContains('xpath', '//body', $text);

You need to revert this change.

This all looks OK to me. I read through the entire patch, the conversions are mostly 1:1 and the ones that are not appear to make sense in that they are now looking for more specific elements to avoid false positives.

Reviewed locally with git diff --color-words. Overall, the changes make sense. For the few cases that were not exactly a 1:1 replacement, my main concern is that using negative assertions tied to specific xpaths of page markup is pretty fragile -- we could easily get false passes if the page markup or theming is changed. It would be better to find a different way to fix those assertions, and I'd suggest splitting those tests into their own blocking issue that this could be postponed on. In particular:

1. FieldUIDeleteTest and UserAdminTest pair negative and positive xpath assertions related to page markup, either as a before/after or including some things but not other things with a given markup pattern, so those might be okay.
2. However, UpdateContribTest only has a negative assertion based on specific page markup. This is a concern because the test could have a false pass if the markup or theme changes. Can we do this a different way?
3. Similarly, ExposedFormTest only has a negative assertion, so I don't think an xpath assertion is the right choice here for the same reasons as above.

As for the other things that are not the 1:1 replacement:

• CommentRestExportTest asserts that the response does not contain the text, which makes sense given that it's a REST test.
• The updated inline docs in LocaleTranslationUiTest make sense.

Finally, there are a few tests that still have assertNoText() following this patch, so they probably need to be addressed as well:

core/modules/comment/tests/src/Kernel/Views/CommentAdminViewTest.php
core/modules/user/tests/src/Kernel/WhosOnlineBlockTest.php
core/modules/field/tests/src/Kernel/DisplayApiTest.php
core/modules/system/tests/src/Kernel/Render/ClassyTest.php
core/modules/views/tests/src/Kernel/Handler/AreaResultTest.php
core/modules/views/tests/src/Kernel/Handler/AreaEmptyTest.php
core/modules/views/tests/src/Kernel/Handler/FieldFieldAccessTestBase.php

I think this is really close -- just would like a smaller first step to potentially solve the negative xpath assertions a different way, and address the remaining usages.

Thanks!

Re: #23 -- Aha, thanks for clarifying. Are there separate plans to deal with the kernel tests?

I addressed #22 2,3

NW for recent feedback.

All feedback was addressed, changes look correct to me.

Needs a re-roll.

Committed 068f00c and pushed to 9.3.x. Thanks!

We should backport this to 9.2.x to keep other patches in sync.

On it.

Please review.

Needs another reroll.

The MR looks perfect to me - thanks!

Committed/pushed to 9.2.x, thanks!