CC/ @nkcsgexi
I have a couple of questions.

1) Assuming this is CURSOR_REFACTORING, which cursor position should be handled?
For instance, when user want to refactor this:

foo
  .bar(x: { _ in .. })
  .baz(y: { _ in .. })

to:

foo
  .bar { _ in .. }
  .baz(y: { _ in .. })

Where the cursor should be on? Anywhere in range of ".bar(x: {" and "})" ?

2) Is there convenient way to retrieve context info from SemaToken?
When the cursor is on "(", it's Kind is ExprStart and TrailingExpr is just a TupleExpr. Since current SemaToken struct doesn't have any context info, I couldn't detect this TupleExpr is an argument for CallExpr.
Should I extend SemaLocResolver and SemaToken so that it preserves surrounding expressions?

Or should I make it RANGE_REFACTORING? then handle:

• Info.ContainedNodes[0] of RangeKind::SingleExpression; or

• Info.CommonExprParent of RangeKind::PartOfExpression

Where the cursor should be on? Anywhere in range of ".bar(x: {" and "})" ?

I think this make sense, essentially:

• in the range of the call

• excluding call arguments, but including the braces of the last closure argument

• excluding the inner range of the closure

Is there convenient way to retrieve context info from SemaToken? [..] Should I extend SemaLocResolver and SemaToken so that it preserves surrounding expressions?

I think that's a good idea. But instead of keeping track more expressions (which can be arbitrary how many parent expressions you want to look at) I recommend that SemaToken keeps track of the "containing" declaration (e.g. the function that contains the expression). Or perhaps the top-level containing statement. Then have a getParentMap() method that delegates to Expr::getParentMap() to get a map useful for getting the parents in the AST tree.
Also consider that we may need a more generic map that also can provide the parent Stmt of an Expr

Thank you! I'll try that.

Comment by KacperHarasim (JIRA)

Hi @rintaro, are you working on this or is it available to implement? Thanks for letting me know 🙂

Hi kacperh (JIRA User), feel free to take this task 🙂
I've partially implemented this, but I'm so busy these days that I can't finish them up until October.

Comment by KacperHarasim (JIRA)

@rintaro Thanks for the answer! Since you have implemented it partially I will find another refactoring task to take. 🙂

Implemented in #12268