Handle EmptyRelation during SQL unparsing

[goldmedal]

Which issue does this PR close?

Closes #10799 .

Rationale for this change

What changes are included in this PR?

Are these changes tested?

Yes

Are there any user-facing changes?

No

[devinjdangelo]

Thanks @goldmedal this looks good to me 🚀. I have one minor suggestion to avoid a panic.

[devinjdangelo]

I think this change will result in a panic on a UninitializedFieldError due to the unwrap. E.g. if self.relation is None rather than Some(None). I think we could avoid this by using Result::transpose rather than .ok().unwrap() here and then continuing to collect::<Result<T>, _>.

[goldmedal]

Thanks, @devinjdangelo. Nice idea. I've learned more about Rust.

[alamb]

Looks great to me -- thank you @goldmedal and @devinjdangelo

[alamb]

❤️

[alamb]

I think you could also do this (which is more concise)

Suggested change

	let new = self;
	new.relation = Some(TableFactorBuilder::Empty);
	new
	self.relation = Some(TableFactorBuilder::Empty);
	new

Though I see it follows the same pattern as the code above so if we are going to change any of the we should change them all

[alamb]

This is totally fine, but I figured I would point out another way to express this logic that you might prefer in future PRs

Instead of

   let ast_join = ast::Join {
                    relation: match right_relation.build()? {
                        Some(relation) => relation,
                        None => return internal_err!("Failed to build right relation"),
                    },
                    join_operator: self
                        .join_operator_to_sql(join.join_type, join_constraint),
                };

You can do something like this with a let .. else to match or return

   let Some(relation) = match right_relation.build()? else {
    return internal_err!("Failed to build right relation"),
   };

   let ast_join = ast::Join {
                    relation,
                    join_operator: self
                        .join_operator_to_sql(join.join_type, join_constraint),
                };

[goldmedal]

Thanks @alamb @devinjdangelo
I think I will address @alamb's comments in the follow-up PR.