Fixes #7673: document symbol responses by language server

[andreamocci]

As described in #7673, this PR fixes the document symbol responses in the language server.

• Fixed range of symbols (using symbol position and not symbol name position), thus enabling breadcrumbs
• Removed top level container object from the returned symbols
• Improved symbol kinds (added enums, enum value/cases, traits as interfaces, and type params/fields)
• Added local members (as in metals)
• Fixed container names (stripping the module class suffix)

Screenshot of breadcrumbs working:

As a side modification, we found that in Typer.scala there is a likely missing Synthetic flag for the anon class.

[dottybot]

Hello, and thank you for opening this PR! 🎉

All contributors have signed the CLA, thank you! ❤️

Commit Messages

We want to keep history, but for that to actually be useful we have
some rules on how to format our commit messages (relevant xkcd).

Please stick to these guidelines for commit messages:

1. Separate subject from body with a blank line
2. When fixing an issue, start your commit message with Fix #<ISSUE-NBR>:
3. Limit the subject line to 72 characters
4. Capitalize the subject line
5. Do not end the subject line with a period
6. Use the imperative mood in the subject line ("Add" instead of "Added")
7. Wrap the body at 80 characters
8. Use the body to explain what and why vs. how

adapted from https://chris.beams.io/posts/git-commit

Have an awesome day! ☀️

[smarter]

There's a few test failures:

[error] Failed tests:
[error] 	dotty.tools.languageserver.WorksheetTest
[error] 	dotty.tools.languageserver.DocumentSymbolTest

You can run one of them from sbt like this:

> dotty-language-server/testOnly *WorksheetTest

Or to run a single test method:

> dotty-language-server/testOnly *WorksheetTest -- *worksheetDocumentSymbol

Let me know if you need help updating the tests.

[andreamocci]

@smarter Thank you, and apologies for these failures. Indeed, these are expected failures since I didn't change any test (I've run just a plain sbt test). I've realized that too late. I've looked at the tests now and I am pretty confident I understood how they work (the interpolators for code markers are pretty amazing, by the way).
I have done the fixes but I'm going to check them again tomorrow morning before committing, and definitely add a few new ones that exercise and elicit the things I've changed.

[smarter]

I think this makes too many assumptions about which trees to include or not, there might be non-synthetic definitions inside synthetic trees in other places than enums. It might not be enough but as a first step maybe we can just exclude type and term parameters of synthetic definitions:

Suggested change

	n.symbol.is(Case) || {
	!n.symbol.is(Param) || {

[andreamocci]

Makes sense. I'm going to commit this soon with the change below.

[smarter]

I don't think we ever want to return these ERROR trees, so we could add this check to the ones we do by default in https://github.com/lampepfl/dotty/blob/3a7dfd2c14e1bc146f79faef8fcad42efadcdaeb/compiler/src/dotty/tools/dotc/interactive/Interactive.scala#L187

[andreamocci]

I'm going to move the condition. I wasn't sure this was the best way to check for an erroneous tree. Can you confirm this? Does this change impact tests other than the language server / worksheets?

[smarter]

I wasn't sure this was the best way to check for an erroneous tree. Can you confirm this?

It's the best way to deal with the error trees produced by Parser currently, typed error trees should already be handled.

Does this change impact tests other than the language server / worksheets?

I guess we'll have to run the tests and see :).

[smarter]

LGTM, thanks!