[nazem-droubi]

Description

Fixes #24337.

XML.status should be -9 ("a start-tag was not matched with an end-tag") in two cases where Ruffle reported something else:

• An end-tag closes the element directly above the open one, as in <a><b>foo</a>. Ruffle gave -10.
• Elements are still open when the input ends, as in <a>foo. Ruffle gave 0.

I ran a set of inputs in the Flash Player 32 debug projector first. They show that -9 applies only when the end-tag names the element one level up. If it names an element further up, as in <a><b><c>foo</a>, or an element that isn't open, Flash still returns -10. So mapping every quick-xml MismatchedEndTag to -9 would be wrong. This PR returns -9 only when the end-tag matches the parent of the open element, and sets -9 at EOF when elements are still open. The parsed tree doesn't change: in every case the test covers, Ruffle and Flash already produce the same toString().

Testing

• New test avm1/xml_status_mismatched_tags, compiled with Rascal. It covers 21 inputs: the -9 cases (including crossed tags like <a><b>foo</a></b>), the -10 cases next to them, unmatched end-tags, input ending with elements still open, and well-formed input. It traces each case in a multi-line case / returned / toString format. output.txt was recorded with the Flash Player 32 Linux debug projector (cargo testutils execute -s). The test fails without the fix.

Checklist

• I, a human, have self-reviewed this PR and fully understand the changes within.
• I have made or updated tests where possible.
• All of my commits are properly scoped, compile successfully, and pass all tests.
• This PR does not make sense to split up into smaller PRs.
• An LLM was involved in the authoring of this code.

AI disclosure: an AI coding agent working on my behalf wrote the code, the test, and this description, and ran the test in Flash Player. I'll handle any review discussion personally.

[kjarosh]

I'm opening this as a draft so I can review it myself before marking it ready.

Please don't do that. It's fine to open as a draft so that others can review a WIP, but you should review the code before opening the PR. We already have too many PRs open.

Not to mention reviewers won't get notified when you are ready and it's wasting our CI resources.

[kjarosh]

The full tests suite passes locally: 4295 passed, 0 failed. cargo fmt --check and cargo clippy -p ruffle_core -- -D warnings are clean.

As mentioned in the template, you don't need to say that, CI takes care of it. This sentence means nothing because every PR has to pass it.

[nazem-droubi]

Thanks — removed that sentence from the description. CI covers it.

[evilpie]

Only the second commit should have tests. Also make the commit message that actually describes what is being test.

[evilpie]

Please remove the following from the commit message. You can't be your own co-author.

Co-authored-by: Nazem nazem-droubi@users.noreply.github.com

Otherwise looks good.

[nazem-droubi]

Removed the Co-authored-by trailer from both commits. Thanks for the review.