Bug: Additional break line when initializing or copy/pasting in the editor with break lines

[mariofpalb]

When I initialize the Lexical editor from an DOM like <p><br></p>, passing through generateNodesFromDOM and inserting the generated nodes to the editor, it imports it with an additional break line like <p><br><br></p>. This makes the editor have more break lines than expected. It is also curious that both
count as one character if I delete them and count the characters.

@zeitstein mentioned in a discord chat that when trying to pass a single <p></p> without break lines, it also imports it with an additional break line like <p><br></p>. They also mention that that this PR can be related -> #3581 (thanks for rising this up in the discord chat, here is the full thread: https://discord.com/channels/953974421008293909/1060331843578249376)

I also managed to reproduce it when copying some text with break lines from the Notes Application from my Mac.

Lexical version: 0.7.6

Steps To Reproduce

1. Write something like this (with some break lines) in a Note in the Mac Notes Application:

First line



Fifth line

2. Copy the five lines (or the block of text you have written) and paste it in the lexical playground editor
3. You can find that there are more than 5 lines in this particular case.

---

Here is an example of the InitialState plugin to initialize the editor from a preview HTML string (not sure why is duplicating the text in the codesandbox, in my code only runs once): https://codesandbox.io/s/break-line-example-w1ic2x?file=/src/plugins/InitialState.js

The current behavior

When copying or initializing some text with break lines, it duplicates them, making the text different from the expected one.

The expected behavior

When copying or initializing some text with break lines, the number of break lines is exactly the same as the one inserted in the editor.

[chalecki]

I'm facing the same issue. I've noticed $generateNodesFromDOM is adding an extra <br /> tag for each "empty" paragraph. If you put any character, eg. period or even empty space - the problem does not exist

[rabishah]

Facing the same issue, exactly as @chalecki .

[willwang888]

Facing the same issue!

[chalecki]

I think it was introduced in #590 in commit f38d4d5 .

Commenting on this line https://github.com/facebook/lexical/blob/main/packages/lexical/src/LexicalReconciler.ts#L324 does the trick I think but I lack knowledge on the overall flow so would be best to hear from @trueadm more on this behavior

[almazkaliyev]

maybe this can helps u

const removeLineBreaks = (inputHtml: string): string => {
  let copy = inputHtml
  copy = copy.replaceAll('<br>', '')
  return copy
}

const onChange = (_: EditorState, editorInstance: LexicalEditor) => {
  editorInstance.update(() => {
    const generatedHtml = $generateHtmlFromNodes(editorInstance, null)
    let resultHtmlString = wrapTables(generatedHtml)
    resultHtmlString = removeLineBreaks(generatedHtml)
    setHtml(resultHtmlString)
  })
}

[mariofpalb]

@almazkaliyev the point is not to remove every <br> from the generated HTML, I want to keep the exact number of break lines from a copy/paste and when initialising the editor with some previous HTML saved from the database.

[sagarchoudhary96]

@thegreatercurve do we have any updates for the fix for this issue ?

[obinmatt]

@thegreatercurve any updates? 😄

[oleksandr-danylchenko]

Having the same issue. An empty text field has the <p><br></p by default

Rich text field	$generateHtmlFromNodes result

[ferugi]

I've found that a single empty paragraph converted to HTML with $generateHtmlFromNodes will result in <p><br></p>.
However when converting back from HTML to Lexical nodes with $generateNodesFromDOM and then updating the contents of the editor's root node, I end up with the following HTML in the ContentEditable element: <p><br><br></p>.

[keithhackbarth]

@oleksandr-danylchenko and @ferugi - I am experiencing both of your observations reproducibly as well.
@thegreatercurve - Let me know if anything we can do to help (ie test cases, repo steps, etc.)

[rjullian]

I'm facing a similar issue and would really appreciate any updates on the progress made toward resolving it. Thanks.

[alexz-spotnana]

This is my function to initialize the editor state from HTML. Since the lexical's renderer adds additional
tags, this function does the opposite – decreases the number of
tags by 1 in the parsed DOM and then parses DOM to nodes and inserts them into the editor.

const parser = new DOMParser();
const dom = parser.parseFromString(initialHtml, 'text/html');

const elementsWithBr = dom.querySelectorAll('*:not(br) > br');
for (let i = 0; i < elementsWithBr.length; i += 1) {
  const brElement = elementsWithBr[i];
  const parentElement = brElement.parentNode;
  const lastBr = parentElement?.querySelectorAll('br:last-child')[0];
  if (lastBr === brElement) {
    parentElement?.removeChild(brElement);
  }
}

const nodes = $generateNodesFromDOM(editor, dom);

$getRoot().select();
$insertNodes(nodes);

Does work for me, but checked on some basic HTML fragments that fit my needs.

[gajanan999]

HI ,

I am facing this issue still in version 0.12.0

[kazimirchuk]

Facing the same issue version 0.12.4

[rtrevisanlectra]

Hello, same issue in version 0.13.1

[mustafamoe]

Same issue in version 0.13.1

[dpr10]

Hello, I'm having the same issue but found a workaround, if you clear the editor before the insertion, the line break is removed.

$getRoot().clear();
$getRoot().select();
$insertNodes(nodes);

[Michael7141]

Hello, I'm having the same issue but found a workaround, if you clear the editor before the insertion, the line break is removed.

$getRoot().clear();
$getRoot().select();
$insertNodes(nodes);

bless your soul;
cuz u set mine on fire