fix: reject leading ., ) without prefix as item marker

[xxchan]

fix #5835

[xxchan]

cc @anforowicz to also take a look (who authorized #5423)

[xxchan]

@ytmimi Could you take a look? Since the bug and fix are both very straightforward, and it's newly introduced. Thanks.

[ytmimi]

@xxchan I'm traveling over the next few days, but I should have some time to review this early next week.

[ytmimi]

Oh I see. If the prefix.len() == 0 the condition would still pass. Good catch. I'm wondering if there's a more obvious way to write this since it took me a few seconds to figure out what (1..=2).contains(..) was doing.

Maybe assigning the condition to a meaningful variable name would help.

let has_leading_digits = (1..=2).contains(&prefix.len()) && prefix.chars().all(|c| char::is_ascii_digit(&c));

[ytmimi]

@xxchan thanks for making the edit!

To further prevent this regression it would be great to add the example from #5835 (comment) as a test case in tests/target/.

Happy to merge this after we add that test case and rebase on the latest master 😁

[calebcartwright]

I haven't had a chance to look at this, but before merging, can we confirm this won't introduce a formatting diff (i.e. that this won't try to modify the output/result of a prior rustfmt run)?

[xxchan]

can we confirm this won't introduce a formatting diff (i.e. that this won't try to modify the output/result of a prior rustfmt run)?

Yes, because it makes a condition triggering the fmt change more restrictive.

[ytmimi]

@calebcartwright I ran the diff check job before the rebase and it failed ❌ for the same reason I explained in #5853 (comment).

I'm running the job again now that the branch has been rebased. I'm assuming we should be good now 🤞🏼, but I'll follow up once the job completes.

Edit:
diff check job ran successfully after the rebase ✅

[ytmimi]

@xxchan the issue mentions that wrap_comments=true needs to be set in order to trigger the bug. Can you update your test cases with // rustfmt-wrap_comments: true, and could you also add a test case using a doc comment.

[ytmimi]

@xxchan thanks again for your help on this one!