avoid pre-reading files

[OneSadCookie]

Per #5947, files are being read unnecessarily when they have been previously linted and appear in the cache.

To avoid this,

• remove the test for file emptiness from Linter.init
• extract the redundant check for this in Rule.lint to a helper, which simplifies Rule.lint and allows it to return non-optional
• check the helper before calling either Rule.lint or Rule.correct.

[SwiftLintBot]

	1 Warning
⚠️	If this is a user-facing change, please include a CHANGELOG entry to credit yourself! You can find it at CHANGELOG.md.

	18 Messages
📖	Building this branch resulted in a binary size of 29488.51 KiB vs 29488.71 KiB when built on main (-1% smaller).
📖	Linting Aerial with this PR took 1.77 s vs 0.95 s on main (86% slower).
📖	Linting Alamofire with this PR took 1.29 s vs 1.29 s on main (0% slower).
📖	Linting Brave with this PR took 8.68 s vs 8.72 s on main (0% faster).
📖	Linting DuckDuckGo with this PR took 5.61 s vs 5.66 s on main (0% faster).
📖	Linting Firefox with this PR took 11.67 s vs 11.77 s on main (0% faster).
📖	Linting Kickstarter with this PR took 10.49 s vs 10.63 s on main (1% faster).
📖	Linting Moya with this PR took 0.54 s vs 0.54 s on main (0% slower).
📖	Linting NetNewsWire with this PR took 2.94 s vs 2.97 s on main (1% faster).
📖	Linting Nimble with this PR took 0.8 s vs 0.8 s on main (0% slower).
📖	Linting PocketCasts with this PR took 8.77 s vs 8.86 s on main (1% faster).
📖	Linting Quick with this PR took 0.45 s vs 0.46 s on main (2% faster).
📖	Linting Realm with this PR took 4.67 s vs 4.56 s on main (2% slower).
📖	Linting Sourcery with this PR took 2.4 s vs 2.37 s on main (1% slower).
📖	Linting Swift with this PR took 4.65 s vs 4.67 s on main (0% faster).
📖	Linting VLC with this PR took 1.25 s vs 1.27 s on main (1% faster).
📖	Linting Wire with this PR took 18.56 s vs 18.89 s on main (1% faster).
📖	Linting WordPress with this PR took 11.57 s vs 11.64 s on main (0% faster).

Here's an example of your CHANGELOG entry:

* avoid pre-reading files.  
  [OneSadCookie](https://github.com/OneSadCookie)
  [#issue_number](https://github.com/realm/SwiftLint/issues/issue_number)

note: There are two invisible spaces after the entry's text.

Generated by 🚫 Danger

[SimplyDanny]

Basically looks good to me!

Would be great to have a test for linting and correcting empty files with (mock) rules that do/don't support this.

[SimplyDanny]

Stylistically, I would prefer a separate if statement in the for block to check for shouldRun. Or you can make use of a where clause.

We could even check for CorrectableRule in the for block to avoid creating a new list and iterating twice.

[OneSadCookie]

I've tried to write those tests, but pretty low confidence that that's "the right way"; pointers appreciated.

I tried to see if there was a way to write an integration test that the files' contents aren't read when the cache is populated, but short of modifying SourceKitten to report contents reads, or magical unix shenanigans (I think you can somehow get a pipe at a path so it gets opened like a regular file, maybe you could tell if data was consumed on the pipe?), I don't see how. Definitely open to ideas here, as regressions will be easy.