libsyntax: cleanup and refactor pat.rs

[Centril]

A smaller refactoring & cleanup of pat.rs (best read commit by commit).

r? @petrochenkov

[Centril]

(I started writing up this PR before the note in #63469 (comment) but I'm happy to wait for moving things around.)

[petrochenkov]

I personally don't extract code into functions unless they are called at least twice or take a whole page of text, so I'd rather back out a half of the "extract" commits.
In general, I find it harder to read code if it's broken into small fragments in different places.

[Centril]

so I'd rather back out a half of the "extract" commits.

Which ones would you like to keep?

In general, I find it harder to read code if it's broken into small fragments in different places.

Heh; I personally think (and imo this is conventional SE thinking) that large methods forces you to hold more state in your head, that they obscure the overall structure re. the "big picture" (parse_pat_with_range_pat) and "details" (the definitions of the alternatives in the production), and that they violate the single responsibility principle. I think it's a good idea to refactor eagerly (both into more functions and files) and plan for growth; otherwise you tend you grow huge messes like parser.rs was before. Moreover, by extracting into functions, you get the benefits of the internal rustdoc as well as being able to search for functions in there and jump to where you want directly. Debuginfo also gets improved by smaller functions -- instead of being thrown into a large one the backtrace points out a smaller context (up to inlining).

[petrochenkov]

I don't want to argue about this too much, it still looks ok, in particular because every match arm in parse_pat is moved into a separate function (something I didn't notice in commit by commit reading), even if the resulting functions sometimes look like do_this_thing_specifically_while_also_doing_that_thing.

@bors r+

[bors]

📌 Commit c8fc4c1 has been approved by petrochenkov

[dlrobertson]

🎉