legacy_objc_type rule triggers on NSNumber in the context of NumberFormatter which requires it

[tylerbarnesTrexis]

New Issue Checklist

• Updated SwiftLint to the latest version
• I searched for existing GitHub issues

Describe the bug

legacy_objc_type rule triggers on NSNumber in the context of NumberFormatter

Complete output when running SwiftLint, including the stack trace and command used

% swiftlint
Linting Swift files in current working directory
Linting 
...
Linting
warning: Legacy Objective-C Reference Type Violation: Prefer Swift value types to bridged Objective-C reference types (legacy_objc_type)
Done linting! Found 1 violation, 0 serious in files.

Environment

• SwiftLint version: 0.44.0
• Installation method used: Homebrew and SPM via Custom Package
• Paste your configuration file:

only_rules:
  - legacy_objc_type

// This triggers a violation:
NumberFormatter.string(from: Double as NSNumber)

Originally posted by @PH9 in #3555 (comment)

[CraigSiemens]

We recently ran into this as well with the following APIs.

• DateComponentsFormatter.allowedUnits
• NumberFormatter.string(from:)
• UNMutableNotificationContent.badge

[SimplyDanny]

In my opinion, there is nothing we can really do about that. When working with certain frameworks it is simply unavoidable to reference these types in some way. The rule is opt-in and can be disabled by comments in the code in case of (rare) exceptions.

[userException]

I got an error for NSNumber violation so changed:
NSNumber.init(value: someValue)
to
NSNumber.(value: someValue)

Worked !

[rgoldberg]

@SimplyDanny Could an ignore or exclude config key accept an array of type names to ignore?

If you want to ignore this rule for NSNumbers because your Swift code must use many NSNumbers, just configure:

legacy_objc_type:
  ignore: [NSNumber]

The rule will still trigger for all other bridged Objective-C types.

[SimplyDanny]

@SimplyDanny Could an ignore or exclude config key accept an array of type names to ignore?

That's still not a fix but at least a way to reduce the number of false positives. I'd be supportive if someone wants to add such an option.

[kapitoshka438]

@SimplyDanny Could an ignore or exclude config key accept an array of type names to ignore?

That's still not a fix but at least a way to reduce the number of false positives. I'd be supportive if someone wants to add such an option.

@SimplyDanny please review