fix(rome_js_formatter): restrict export line break for single specifier

[unvalley]

Summary

• a part of fix this comment for prettier compatibility
  • noted as export line break
  • playground
• restrict line breaks when specifier of export is long
• format output is below

// given code
export { loooooooooooooooooooooooooooooooooooooooooooooooooong } from "loooooooooooooooooooooooooooooooooooooooooooooong"

// formated (Before merging this PR)
export { 
  loooooooooooooooooooooooooooooooooooooooooooooooooong,
} from "loooooooooooooooooooooooooooooooooooooooooooooong"


// formated (After merging this PR ideally)
export { loooooooooooooooooooooooooooooooooooooooooooooooooong } from "loooooooooooooooooooooooooooooooooooooooooooooong"

Test Plan

• updated js/module/export/named_from_clause.js snapshot
• updated prettier_tests prettier/js/export-extension/export.js snapshot

[netlify]

✅ Deploy Preview for docs-rometools canceled.

Built without sensitive environment variables

Name	Link
🔨 Latest commit	c2042dc
🔍 Latest deploy log	https://app.netlify.com/sites/docs-rometools/deploys/6373ad87cbb2700008096180

[unvalley]

I've referenced import_named_clause.

tools/crates/rome_js_formatter/src/js/module/import_named_clause.rs

Line 58 in 58fbb93

// SAFETY: we know that the `specifiers.specifiers().len() == 1`, so unwrap `iter().next()` is safe.

[MichaReiser]

We should fall back to the default implementation if specifier or separator isn't ok, and we have to call format_removed if the separator is not none.

We should also check that the specifier has no comments

It can then be written like this (you have to make the node and trailing_separator fields pub on AstSeparatedElement

match specifiers.elements().next() {
    Some(AstSeparatedElement {
        node: Ok(node),
        trailing_separator: Ok(separator),
    }) if specifiers.len() == 1 && !f.comments().has_comments(node.syntax()) => {
        write!(f, [space(), node.format()])?;
        
        if let Some(separator) = separator {
            write!(f, [format_removed(&separator)])?;
        }
        
        write!(f, [space()])?;
    }
    _ => {
        if node_has_leading_newline(specifiers.syntax()) {
            write!(f, [block_indent(&specifiers.format()),])?;
        } else {
            write!(
                f,
                [group(&soft_space_or_block_indent(&specifiers.format())),]
            )?;
        };
    }
}

Thanks for pointing to the import specifier code. I noticed that I missed a check when refactoring how Rome handles comments.

Would you mind changing

  if specifier.syntax().has_comments_direct()
      || separator
          .map(|sep| {
              sep.has_leading_comments() || sep.has_trailing_comments()
          })
          .unwrap_or(false)
  {
      write!(f, [named_import.format()])
  } else {

to

 if f.comments().has_comments(specifier.syntax()) {
      write!(f, [named_import.format()])
  } else {

[unvalley]

@MichaReiser
Thanks for reviewing and clear explanation! I've updated the code.
I also edited import specifier code and its test.

[MichaReiser]

Thank you!