Add rule to migrate away from ktlint-disable directives

[paul-dingemans]

Description

Add rule to migrate ktlint-disable directives from comments to @Suppress or @SuppressWarnings annotations

Note that the SuppressionLocatorBuilder never ignore this rule as otherwise the suppression hint to ignore all rules would also ignore the rule which has to migrate this suppression hint.

REVIEW suggestion:

• The functional change for Drop support for ktlint-enable / ktlint-disable directive #1947 is contained in this commit
• Other commits are supportive to the change

Closes #1947

Checklist

• PR description added
• tests are added
• KtLint has been applied on source code itself and violations are fixed
• documentation is updated
• CHANGELOG.md is updated

In case of adding a new rule:

• Rule is added to rules documentation

[paul-dingemans]

@atulgpt I you also want to contribute by reviewing changes, you're welcome to review this PR. You can review it in detail or just briefly. Normally, I would just merge the change as there are almost no regular contributors on the project that would like to do this. I you don't want to review, just let me know.

REVIEW suggestion:

• The functional change for is contained in this commit
• Other commits are supportive to the change

[atulgpt]

@atulgpt I you also want to contribute by reviewing changes, you're welcome to review this PR. You can review it in detail or just briefly. Normally, I would just merge the change as there are almost no regular contributors on the project that would like to do this. I you don't want to review, just let me know.

REVIEW suggestion:

* The functional change for is contained in [this commit](https://github.com/pinterest/ktlint/pull/2068/commits/428d52e637d931754b3756f8912f64cea884f8d5)

* Other commits are supportive to the change

Hi, @paul-dingemans I have added a few more TC suggestions. Will take a look in the logic as well

[paul-dingemans]

[paul-dingemans]

@atulgpt I you also want to contribute by reviewing changes, you're welcome to review this PR. You can review it in detail or just briefly. Normally, I would just merge the change as there are almost no regular contributors on the project that would like to do this. I you don't want to review, just let me know.
REVIEW suggestion:

* The functional change for is contained in [this commit](https://github.com/pinterest/ktlint/pull/2068/commits/428d52e637d931754b3756f8912f64cea884f8d5)

* Other commits are supportive to the change

Hi, @paul-dingemans I have added a few more TC suggestions. Will take a look in the logic as well

Wow, I am really amazed at the good quality of the review comments. Tnx! I have left some review comments unanswered. I have to think about them.

[paul-dingemans]

Can you please "resolve" the remarks on which I have responded and on which you agree? In that way, it is more clear for both of us what still needs to be fixed or discussed. Tnx. You're doing really great as reviewer!

[atulgpt]

Can you please "resolve" the remarks on which I have responded and on which you agree? In that way, it is more clear for both of us what still needs to be fixed or discussed. Tnx. You're doing really great as a reviewer!

Somehow it doesn't give me that option

[paul-dingemans]

Hmm, I can not change that. If you leave a thumbs up or 'ok' message then I will resolve them.

[paul-dingemans]

Hi @atulgpt ,
I hope you can find some time for review of logic.

[atulgpt]

Hi @paul-dingemans somehow the below TC is failing(it is adding multiple annotations)

val code =
    """
    class Foo() {
        val foo: Int
            get() = 1
            set(value: Int) { // ktlint-disable standard:foo
                field = value // ktlint-disable standard:foo
            }
    }
    """.trimIndent()
val formattedCode =
    """
    class Foo() {
        val foo: Int
            get() = 1
            @Suppress("ktlint:standard:foo")
            set(value: Int) {
                field = value
            }
    }
    """.trimIndent()

[atulgpt]

class Foo() {
    var foo: Int = 0
        get() = 1
        @Suppress(names = ["abc"])
        set(value: Int) {
            field = value // ktlint-disable standard:foo
        }
}

this is also failing(seems to be an issue with named parameter)

[atulgpt]

I think currently we are also not handling type aliases like below

typealias A = Int // ktlint-disable standard:foo

[atulgpt]

And below code behaves strangely(maybe I am not setting the TC correctly)

val code =
    """
    private val a by lazy(LazyThreadSafetyMode.PUBLICATION) { // ktlint-disable standard:foo
        val t = 0 // ktlint-disable standard:foo
        2
    }
    """.trimIndent()
val formattedCode =
    """
    private val a by lazy(LazyThreadSafetyMode.PUBLICATION) { // ktlint-disable standard:foo
        val t = 0 // ktlint-disable standard:foo
        2
    }
    """.trimIndent()

above code doesn't show any violations but if I change the code like below

val code =
    """
    private val a by lazy(LazyThreadSafetyMode.PUBLICATION) { // ktlint-disable standard:foo
        val t = 0 // ktlint-disable standard:foo
        2
    }
    fun t() { // ktlint-disable standard:foo

    }
    """.trimIndent()

it gets itself corrected as below code

private val a by lazy(LazyThreadSafetyMode.PUBLICATION) @Suppress("ktlint:standard:foo")
  {
      @Suppress("ktlint:standard:foo")
      val t = 0
      2
  }
  @Suppress("ktlint:standard:foo")
  fun t() {
  
  }

[paul-dingemans]

Hi @paul-dingemans somehow the below TC is failing(it is adding multiple annotations)

val code =
    """
    class Foo() {
        val foo: Int
            get() = 1
            set(value: Int) { // ktlint-disable standard:foo
                field = value // ktlint-disable standard:foo
            }
    }
    """.trimIndent()
val formattedCode =
    """
    class Foo() {
        val foo: Int
            get() = 1
            @Suppress("ktlint:standard:foo")
            set(value: Int) {
                field = value
            }
    }
    """.trimIndent()

Fixed

[paul-dingemans]

class Foo() {
var foo: Int = 0
get() = 1
@Suppress(names = ["abc"])
set(value: Int) {
field = value // ktlint-disable standard:foo
}
}

I have simplified your example to:

            @Suppress(names = ["unused"])
            val foo = "foo" // ktlint-disable standard:foo

Fixed. When merging the annotations, I have chosen to drop the usage of the name argument for the sake of simplicity. So the result of above is:

            @Suppress("ktlint:standard:foo", "unused")
            val foo = "foo"

[paul-dingemans]

I think currently we are also not handling type aliases like below

typealias A = Int // ktlint-disable standard:foo

Type aliases are handled. No fix required.

[paul-dingemans]

And below code behaves strangely(maybe I am not setting the TC correctly)

val code =
    """
    private val a by lazy(LazyThreadSafetyMode.PUBLICATION) { // ktlint-disable standard:foo
        val t = 0 // ktlint-disable standard:foo
        2
    }
    """.trimIndent()
val formattedCode =
    """
    private val a by lazy(LazyThreadSafetyMode.PUBLICATION) { // ktlint-disable standard:foo
        val t = 0 // ktlint-disable standard:foo
        2
    }
    """.trimIndent()

above code doesn't show any violations but if I change the code like below

val code =
    """
    private val a by lazy(LazyThreadSafetyMode.PUBLICATION) { // ktlint-disable standard:foo
        val t = 0 // ktlint-disable standard:foo
        2
    }
    fun t() { // ktlint-disable standard:foo

    }
    """.trimIndent()

it gets itself corrected as below code

private val a by lazy(LazyThreadSafetyMode.PUBLICATION) @Suppress("ktlint:standard:foo")
  {
      @Suppress("ktlint:standard:foo")
      val t = 0
      2
  }
  @Suppress("ktlint:standard:foo")
  fun t() {
  
  }

Nice TC. I have use a simplified example to reproduce:

val foo by lazy(LazyThreadSafetyMode.PUBLICATION) { // ktlint-disable standard:indent
    // do something
}

which is now formatted as:

val foo by @Suppress("ktlint:standard:indent")
    lazy(LazyThreadSafetyMode.PUBLICATION) {
        // do something
    }

This example also learned that in case of nested expression, the annotation was not created at the outer most expression which resulted in code which could not be compiled. For example when using FUNCTION_LITERAL or LAMBDA_EXPRESSION:

val foo =
    setOf("a")
        .map {
            bar(it) // ktlint-disable standard:indent
        }

val foo = foo() {
    bar(it) // ktlint-disable standard:indent
}

This is also fixed.

[atulgpt]

@paul-dingemans under which rule this got formatted?

[paul-dingemans]

I am not sure whether I understand.

But the IndentationRule would format:

                return mapOf(
                       1 to "   1 ms",
                      10 to "  10 ms",
                     999 to " 999 ms",
                    1000 to "   1 sec",
                )

to

                return mapOf(
                    1 to "   1 ms",
                    10 to "  10 ms",
                    999 to " 999 ms",
                    1000 to "   1 sec",
                )

if the rule would not be suppressed.

Does that answer the question?

[atulgpt]

Actually, I am wondering why can't ruleProviders = setOf( be on the same line? Is this coz ktlint official recommends to always(in case of multiline chain) to newline after =

[paul-dingemans]

Yes, see https://pinterest.github.io/ktlint/0.49.1/rules/experimental/#multiline-expression-wrapping. Rationale is that when using long identifiers (or small screens) that the first part of the expression gets disconnected from the other lines. So it is still allowed to write rulesproviders = setOf(....), if its fits on a single line.