Unopened HTML tag causes exception in budoux 0.6 about budoux HOT 4 CLOSED

Thanks for reporting this issue!

I think the parser should gracefully handle unpaired close tags. Let me look into this, similar to how browsers handle such case.

For self-closing tags such as <img>, I think it's supported, but we don't have tests. I'll look into it too.

from budoux.

Thanks for reporting this. The change is due to the non-breaking markup support we introduced at #251, where we need to track the elements in a queue.

Could you elaborate your specific use case that you want to include a close tag with no corresponding open tag? I acknowledge that we need to improve the error message, but BudouX is intended to work with a valid document fragment.

from budoux.

Thanks. It was actually a content bug to have a missing open tag, but I thought worth reporting since it's a change in behaviour.

I don't need this to be a supported case, though maybe a nicer exception would be useful - otherwise happy for you to close as expected behaviour.

from budoux.

Thanks for your input. Raising a better exepction sounds like a plan.
That said, the element queue's emptiness is not enough to detect if the given HTML string is valid or not. The string like below will work with budoux==0.6.0 CLI even thought the string is not a valid HTML.

budoux --html "あの空を見る限り<img>今日はとてもいい天気ですね。</p>"

We may need to check void elements as well to manage the element queue and the document's validity better.
@kojiishi any thoughts on this?

from budoux.