Bug: p tags in anchor tags are not handled correct

[SuPuL]

If you paste something like

<a class="mdn" href="https://developer.mozilla.org/en-US/docs/Web/API/DataTransfer"><p>event.dataTransfer</p></a>

within a richtext. The editor gets confused, and the cursor will be stuck in the LinkNode. I'm not sure what the best solution is for such a case.

Lexical version: latest

Steps To Reproduce

1. Goto any website. Create the example text. Copy the element.
2. Goto https://playground.lexical.dev/
3. Paste the html content.
4. Start typing. You'll see that you are in a invalid state.

The current behavior

The editor renders the Link, but is stuck afterwards.

The expected behavior

The editor works as expected, like if I paste an anchor with plain text.

Impact of fix

Copy html from any website might lead to an invalid state. User will likely loose data.

[semimikoh]

@etrepum

Hi! I'd like to work on this issue.

I've reproduced the problem where the cursor gets stuck in the LinkNode when pasting HTML with block elements inside anchor tags.

Since Lexical doesn't allow block elements (like <p>) inside inline elements (like <a>), I'm considering a few approaches to handle this invalid HTML structure:

1. Restructure the DOM: Move the anchor inside the paragraph: <a><p>text</p></a> → <p><a>text</a></p>
2. Remove block wrapper: Strip the <p> tag and keep only the text content inside the link
3. Convert to inline: Replace <p> with <span> to maintain some semantic structure
4. Split into separate nodes: Create separate LinkNode and ParagraphNode
5. Preserve line breaks: If the <p> was meant for line breaks, convert to <br> instead

Which approach would align best with Lexical's architecture and philosophy? I'm leaning towards option 1 (restructuring) as it produces valid HTML while preserving the original intent, but I'd appreciate guidance on the preferred solution.

[etrepum]

Most of the approaches here are really coming at it from the wrong direction. What matters is the lexical tree (ParagraphNode, TextNode, LinkNode, LineBreakNode) not really the DOM. 2 & 3 are indistinguishable from the perspective of a lexical document for example. 4 doesn’t really make sense, you would have to more precisely define what you mean by separate for anyone else to understand.

The first thing that should be done is to come up with use cases where anchor tags are on the outside of block tags, then you need to decide what the best strategy is to accommodate all of those use cases with valid lexical documents, and then finally you worry about the implementation details of exactly where and how to do that import or transform. Possibly it is some combination of all of the strategies depending on the structure of the input.

[etrepum]

Here is a related PR that might be of some inspiration, it's unclear if they intend to address the feedback or not. #7982