Add 'Generate enum associated value accessors' code action by ayush-that · Pull Request #2680 · swiftlang/sourcekit-lsp

ayush-that · 2026-06-01T08:03:56Z

Implements #2522.

Adds a syntactic code action on an enum that generates, per case, an is<Case> Boolean property and (for cases with associated values) an as<Case> property returning the value(s) as an optional.

For example, on:

enum Value {
    case text(String)
    case number(Int)
}

it generates asText / isText / asNumber / isNumber.

Question (from #2544): should is<Case> be generated for every case(current behavior), or only for cases with associated values?

Testing

Adds testGenerateEnumAssociatedValueAccessors.

Copilot

Pull request overview

Note

Copilot was unable to run its full agentic suite in this review.

Adds a new SwiftSyntax-based refactoring code action to generate as<Case>/is<Case> accessors for enum cases, and wires it into the code action registry with a corresponding test.

Changes:

Register the new GenerateEnumAssociatedValueAccessors provider in the global syntax code action list.
Implement the enum accessor generation refactoring provider.
Add a code action test asserting the expected workspace edit.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 2 comments.

File	Description
Tests/SourceKitLSPTests/CodeActionTests.swift	Adds a test validating the new “Generate enum associated value accessors” code action and its edit.
Sources/SwiftSyntaxCodeActions/SyntaxCodeActions.swift	Registers the new code action provider in the aggregated provider list.
Sources/SwiftSyntaxCodeActions/GenerateEnumAssociatedValueAccessors.swift	Implements the refactoring provider that inserts `asX`/`isX` computed properties for enum cases.
Sources/SwiftSyntaxCodeActions/CMakeLists.txt	Includes the new Swift source file in the CMake build.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

+      markers: ["1️⃣"],
+      exhaustive: false
+    ) { uri, positions in


+    var accessors: [String] = []
+    for element in caseElements {
+      let caseName = element.name.identifier?.name ?? element.name.text
+      let capitalized = caseName.prefix(1).uppercased() + caseName.dropFirst()
+      let casePattern = element.name.text
+
+      if let parameters = element.parameterClause?.parameters, !parameters.isEmpty {
+        let asName = "as\(capitalized)"
+        if usedNames.insert(asName).inserted {
+          accessors.append(
+            makeAsAccessor(
+              name: asName,
+              casePattern: casePattern,
+              parameters: Array(parameters),
+              memberIndentation: memberIndentation,
+              bodyIndentation: bodyIndentation
+            )
+          )
+        }
+      }
+
+      let isName = "is\(capitalized)"
+      if usedNames.insert(isName).inserted {
+        accessors.append(
+          makeIsAccessor(
+            name: isName,
+            casePattern: casePattern,
+            memberIndentation: memberIndentation,
+            bodyIndentation: bodyIndentation
+          )
+        )
+      }
+    }


ahoppen · 2026-06-01T23:38:43Z

+///
+/// After:
+///
+/// ```swift


I think we should have two separate refactorings to generate the as and is names.

okie. split into two.

ahoppen · 2026-06-01T23:39:59Z

+    // Attached macros, `#if` blocks, and freestanding macro expansions can
+    // introduce cases or members that are not visible to a purely syntactic
+    // walk, which could produce incomplete accessors or collide with members
+    // added by macro expansion. Bail conservatively in those cases.
+    if !syntax.attributes.isEmpty {
+      throw RefactoringNotApplicableError("Enum has attributes that may be attached macros")
+    }


I don’t think we need this bail-out. If we fail to generate one is accessor, that’s not a correctness issue and if it collides with a member generated by the macro, the user will get a compilation issue, which they can also decide how to fix, we shouldn’t try to make assumptions on their behalf.

Similarly, we can generate accessors if the enum contains #if since we also won’t produce a correctness issue.

ahoppen · 2026-06-01T23:42:05Z

+    // Infer indentation from the source rather than hardcoding it: members are
+    // indented one step deeper than the enum, and property bodies one step
+    // deeper than the members. Deriving the step from the existing members
+    // honors whatever width (or tabs) the file already uses.


I don’t think this comment provides much value

ahoppen · 2026-06-01T23:42:55Z

+      syntax.memberBlock.members.first?.firstToken(viewMode: .sourceAccurate)?.indentationOfLine.description
+      ?? (enumIndentation + "    ")


We shouldn’t assume 4 space indentation. Use BasicFormat.inferIndentation to determine the file’s indentation.

ahoppen · 2026-06-01T23:43:53Z

+
+      if let parameters = element.parameterClause?.parameters, !parameters.isEmpty {
+        let asName = "as\(capitalized)"
+        if usedNames.insert(asName).inserted {


We shouldn’t generate the same name twice, so I think it’s sufficient to check usedNames.contains(asName).

ahoppen · 2026-06-11T21:19:01Z

+      guard !usedNames.contains(name) else { continue }
+      usedNames.insert(name)


You should be able to combine this in one call as follows.

Suggested change

guard !usedNames.contains(name) else { continue }

usedNames.insert(name)

guard !usedNames.insert(name).inserted else { continue }

this code is gone now, along with the rest of the name tracking.

ahoppen · 2026-06-11T21:19:33Z

+        continue
+      }
+      let name = "as\(capitalizedCaseName(of: element))"
+      guard !usedNames.contains(name) else { continue }


Why do we need to check for duplicate names here? Do you have an example?

no. i removed the whole name tracking, including the skip for existing members. duplicate names now end up as compiler errors.

ahoppen · 2026-06-11T21:20:51Z

+      ]
+    }
+  }
+


Could you add a test case for the following source code?

enum Foo { case value(int: Int) case value(string: String) }

done. i've added testGenerateEnumCaseAsAccessorsForOverloadedCaseNames.

ahoppen · 2026-06-11T21:21:34Z

+    }
+    let types = parameters.map { $0.type.trimmedDescription }
+    let returnType: String
+    if parameters.count == 1 {


If you write it as follows, you don’t need the [0] access below.

Suggested change

if parameters.count == 1 {

if let parameter = parameters.only {

ahoppen · 2026-06-11T21:24:26Z

+    name: String,
+    casePattern: String,
+    parameters: [EnumCaseParameterSyntax],
+    indentation: (member: String, body: String)


Instead of passing the member and body indentation here, it would be cleaner to pass a base indentation and the indentation step so that we can compute the effective indentation in here. Otherwise you need to understand how indentation.member and indentation.body are used to correctly set it.

ahoppen · 2026-06-11T21:25:53Z

+  }
+}
+
+extension GenerateEnumCaseAsAccessors: SyntaxRefactoringCodeActionProvider {


Instead of having extensions of the same type in the same file, I think it’s clearer if we just throw all members into on struct declaration.

ahoppen · 2026-06-11T21:30:44Z

+                  newText: """
+
+
+                        var asPoint: (Int, Int)? {


I think this should also contain the tuple labels, ie. this should be

Suggested change

var asPoint: (Int, Int)? {

var asPoint: (x: Int, y: Int)? {

ahoppen · 2026-06-11T21:31:53Z

+                  newText: """
+
+
+                        var isFoo: Bool {


I would rather generate duplicate cases here and have a compiler error that the user needs to fix instead of implicitly assuming that isFoo should refer to the underscore version of foo.

i agree. i've removed the name tracking, both isFoo accessors get generated and the foo/Foo test checks the duplicate output.

ayush-that · 2026-06-13T06:32:04Z

the name tracking is gone completely
is accessors are generated for every case, as accessors for cases with associated values
duplicate names end up as compiler errors
tuple return types keep their labels
added overloaded test case

ahoppen

Thanks for the update. Now, if an accessor already exists, we will create a second copy of it, right? I don’t think we should do that. Also: If all accessors already exist, we shouldn’t suggest the refactoring action.

ayush-that · 2026-06-20T07:44:15Z

You're right 😓. It duplicated, so I've added back the skip for accessor names that already exist, and when all of them already exist the action isn't offered anymore.

Copilot AI review requested due to automatic review settings June 1, 2026 08:03

ayush-that requested review from ahoppen, hamishknight and rintaro as code owners June 1, 2026 08:03

Copilot AI reviewed Jun 1, 2026

View reviewed changes

ahoppen reviewed Jun 1, 2026

View reviewed changes

ayush-that force-pushed the feature/generate-enum-associated-value-accessors branch from f77ad8a to ff61ae1 Compare June 5, 2026 10:21

ayush-that requested a review from ahoppen June 5, 2026 10:27

ahoppen reviewed Jun 11, 2026

View reviewed changes

ayush-that force-pushed the feature/generate-enum-associated-value-accessors branch from ff61ae1 to f431821 Compare June 12, 2026 18:16

ayush-that requested a review from ahoppen June 13, 2026 06:38

ahoppen reviewed Jun 19, 2026

View reviewed changes

ahoppen mentioned this pull request Jun 19, 2026

Add swap binary operands code action #2694

Open

Add 'as'/'is' enum case accessor code actions

4e6aeb2

ayush-that force-pushed the feature/generate-enum-associated-value-accessors branch from f431821 to 4e6aeb2 Compare June 20, 2026 07:38

ayush-that requested a review from ahoppen June 20, 2026 07:44

		syntax.memberBlock.members.first?.firstToken(viewMode: .sourceAccurate)?.indentationOfLine.description
		?? (enumIndentation + " ")

		guard !usedNames.contains(name) else { continue }
		usedNames.insert(name)

	guard !usedNames.contains(name) else { continue }
	usedNames.insert(name)
	guard !usedNames.insert(name).inserted else { continue }

	if parameters.count == 1 {
	if let parameter = parameters.only {

Conversation

ayush-that commented Jun 1, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Testing

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

ayush-that commented Jun 13, 2026

Uh oh!

ahoppen left a comment

Choose a reason for hiding this comment

Uh oh!

ayush-that commented Jun 20, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

ayush-that commented Jun 1, 2026 •

edited

Loading