[lexical][auto-link] Fix auto link crash editor

[MaxPlav]

Fixed bug introduced in #6146. Improving the way a potential top-level domain is parsed for email autolinks. Adding more e2e tests, apparently the email auto-link use case wasn't covered by tests.

Closes #6430

Before

Screen.Recording.2024-07-19.at.18.32.12.mov

After

Screen.Recording.2024-07-19.at.18.32.44.mov

[vercel]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

Name	Status	Preview	Comments	Updated (UTC)
lexical	✅ Ready (Inspect)	Visit Preview	💬 Add feedback	Jul 19, 2024 11:35pm
lexical-playground	✅ Ready (Inspect)	Visit Preview	💬 Add feedback	Jul 19, 2024 11:35pm

[github-actions]

size-limit report 📦

Path	Size
lexical - cjs	28.76 KB (0%)
lexical - esm	28.61 KB (0%)
@lexical/rich-text - cjs	37.2 KB (0%)
@lexical/rich-text - esm	28.21 KB (0%)
@lexical/plain-text - cjs	35.85 KB (0%)
@lexical/plain-text - esm	25.5 KB (0%)
@lexical/react - cjs	39.12 KB (0%)
@lexical/react - esm	29.49 KB (0%)

[etrepum]

It would probably be a good idea to add a unit test suite for this sort of thing, e2e is a bit overkill to test a regex

[ivailop7]

I agree with Bob here, these should be perfectly fine to be unit tests, otherwise LGTM.

[MaxPlav]

hi @etrepum @ivailop7 thanks for the review, the e2e tests for the Autolink node have existed for quite a while (added probably here #1156) but they didn’t cover all use cases specifically for the Autolink node created for a recognized email pattern. The way the text is transformed to an Autolink node is spread in multiple functions, it’s a bit complicated and not easy to troubleshoot. In this diff (besides the bug fix) I’m just covering all possible positive and negative use cases for the whole autolink creation/update flow transformed from a text email pattern. So I would suggest keeping the e2e tests as it might be not sufficient to cover the functions with unit tests. :) wdyt?