Generic Reference type and match result subscript

Sources/VariadicsGenerator/VariadicsGenerator.swift

@@ -90,10 +90,7 @@ var standardError = StandardErrorStream()
 
 typealias Counter = Int64
 let regexProtocolName = "RegexProtocol"
-let concatenationStructTypeBaseName = "Concatenate"
-let capturingGroupTypeBaseName = "CapturingGroup"
 let matchAssociatedTypeName = "Match"
-let captureAssociatedTypeName = "Capture"
 let patternBuilderTypeName = "RegexBuilder"
 let patternProtocolRequirementName = "regex"
 let regexTypeName = "Regex"
@@ -535,108 +532,169 @@ struct VariadicsGenerator: ParsableCommand {
         : "(\(baseMatchTypeName), \(newCaptureType), " + (0..<arity).map { "C\($0)" }.joined(separator: ", ") + ")"
     }
     let whereClause = "where R.\(matchAssociatedTypeName) == \(matchType)"
+    let rawNewMatchType = newMatchType(newCaptureType: "W")
+    let transformedNewMatchType = newMatchType(newCaptureType: "NewCapture")
     output("""
       // MARK: - Non-builder capture arity \(arity)
 
       public func capture<\(genericParams)>(
-        _ component: R, as reference: Reference? = nil
-      ) -> \(regexTypeName)<\(newMatchType(newCaptureType: "W"))> \(whereClause) {
-        .init(node: .group(.capture, component.regex.root, reference?.id))
+        _ component: R
+      ) -> \(regexTypeName)<\(rawNewMatchType)> \(whereClause) {
+        .init(node: .group(.capture, component.regex.root))
+      }
+
+      public func capture<\(genericParams)>(
+        _ component: R, as reference: Reference<W>
+      ) -> \(regexTypeName)<\(rawNewMatchType)> \(whereClause) {
+        .init(node: .group(.capture, component.regex.root, reference.id))
+      }
+
+      public func capture<\(genericParams), NewCapture>(
+        _ component: R,
+        transform: @escaping (Substring) -> NewCapture
+      ) -> \(regexTypeName)<\(transformedNewMatchType)> \(whereClause) {
+        .init(node: .groupTransform(
+          .capture,
+          component.regex.root,
+          CaptureTransform(resultType: NewCapture.self) {
+            transform($0) as Any
+          }))
       }
 
       public func capture<\(genericParams), NewCapture>(
         _ component: R,
-        as reference: Reference? = nil,
+        as reference: Reference<NewCapture>,
         transform: @escaping (Substring) -> NewCapture
-      ) -> \(regexTypeName)<\(newMatchType(newCaptureType: "NewCapture"))> \(whereClause) {
+      ) -> \(regexTypeName)<\(transformedNewMatchType)> \(whereClause) {
         .init(node: .groupTransform(
           .capture,
           component.regex.root,
           CaptureTransform(resultType: NewCapture.self) {
             transform($0) as Any
           },
-          reference?.id))
+          reference.id))
       }
 
       public func tryCapture<\(genericParams), NewCapture>(
         _ component: R,
-        as reference: Reference? = nil,
         transform: @escaping (Substring) throws -> NewCapture
-      ) -> \(regexTypeName)<\(newMatchType(newCaptureType: "NewCapture"))> \(whereClause) {
+      ) -> \(regexTypeName)<\(transformedNewMatchType)> \(whereClause) {
+        .init(node: .groupTransform(
+          .capture,
+          component.regex.root,
+          CaptureTransform(resultType: NewCapture.self) {
+            try transform($0) as Any
+          }))
+      }
+
+      public func tryCapture<\(genericParams), NewCapture>(
+        _ component: R,
+        as reference: Reference<NewCapture>,
+        transform: @escaping (Substring) throws -> NewCapture
+      ) -> \(regexTypeName)<\(transformedNewMatchType)> \(whereClause) {
         .init(node: .groupTransform(
           .capture,
           component.regex.root,
           CaptureTransform(resultType: NewCapture.self) {
             try transform($0) as Any
           },
-          reference?.id))
+          reference.id))
       }
 
       public func tryCapture<\(genericParams), NewCapture>(
         _ component: R,
-        as reference: Reference? = nil,
         transform: @escaping (Substring) -> NewCapture?
-      ) -> \(regexTypeName)<\(newMatchType(newCaptureType: "NewCapture"))> \(whereClause) {
+      ) -> \(regexTypeName)<\(transformedNewMatchType)> \(whereClause) {
+        .init(node: .groupTransform(
+          .capture,
+          component.regex.root,
+          CaptureTransform(resultType: NewCapture.self) {
+            transform($0) as Any?
+          }))
+      }
+
+      public func tryCapture<\(genericParams), NewCapture>(
+        _ component: R,
+        as reference: Reference<NewCapture>,
+        transform: @escaping (Substring) -> NewCapture?
+      ) -> \(regexTypeName)<\(transformedNewMatchType)> \(whereClause) {
         .init(node: .groupTransform(
           .capture,
           component.regex.root,
           CaptureTransform(resultType: NewCapture.self) {
             transform($0) as Any?
           },
-          reference?.id))
+          reference.id))
       }
 
       // MARK: - Builder capture arity \(arity)
 
       public func capture<\(genericParams)>(
-        as reference: Reference? = nil,
         @RegexBuilder _ component: () -> R
-      ) -> \(regexTypeName)<\(newMatchType(newCaptureType: "W"))> \(whereClause) {
-        .init(node: .group(.capture, component().regex.root, reference?.id))
+      ) -> \(regexTypeName)<\(rawNewMatchType)> \(whereClause) {
+        .init(node: .group(.capture, component().regex.root))
+      }
+
+      public func capture<\(genericParams)>(
+        as reference: Reference<W>,
+        @RegexBuilder _ component: () -> R
+      ) -> \(regexTypeName)<\(rawNewMatchType)> \(whereClause) {
+        .init(node: .group(.capture, component().regex.root, reference.id))
       }
 
       public func capture<\(genericParams), NewCapture>(
-        as reference: Reference? = nil,
         @RegexBuilder _ component: () -> R,
         transform: @escaping (Substring) -> NewCapture
-      ) -> \(regexTypeName)<\(newMatchType(newCaptureType: "NewCapture"))> \(whereClause) {
+      ) -> \(regexTypeName)<\(transformedNewMatchType)> \(whereClause) {
         .init(node: .groupTransform(
           .capture,
           component().regex.root,
           CaptureTransform(resultType: NewCapture.self) {
             transform($0) as Any
-          },
-          reference?.id))
+          }))
       }
 
       public func tryCapture<\(genericParams), NewCapture>(
-        as reference: Reference? = nil,
+        as reference: Reference<NewCapture>,
         @RegexBuilder _ component: () -> R,
         transform: @escaping (Substring) throws -> NewCapture
-      ) -> \(regexTypeName)<\(newMatchType(newCaptureType: "NewCapture"))> \(whereClause) {
+      ) -> \(regexTypeName)<\(transformedNewMatchType)> \(whereClause) {
         .init(node: .groupTransform(
           .capture,
           component().regex.root,
           CaptureTransform(resultType: NewCapture.self) {
             try transform($0) as Any
           },
-          reference?.id))
+          reference.id))
+      }
+
+      public func tryCapture<\(genericParams), NewCapture>(
+        @RegexBuilder _ component: () -> R,
+        transform: @escaping (Substring) -> NewCapture?
+      ) -> \(regexTypeName)<\(transformedNewMatchType)> \(whereClause) {
+        .init(node: .groupTransform(
+          .capture,
+          component().regex.root,
+          CaptureTransform(resultType: NewCapture.self) {
+            transform($0) as Any?
+          }))
       }
 
       public func tryCapture<\(genericParams), NewCapture>(
-        as reference: Reference? = nil,
+        as reference: Reference<NewCapture>,
         @RegexBuilder _ component: () -> R,
         transform: @escaping (Substring) -> NewCapture?
-      ) -> \(regexTypeName)<\(newMatchType(newCaptureType: "NewCapture"))> \(whereClause) {
+      ) -> \(regexTypeName)<\(transformedNewMatchType)> \(whereClause) {
         .init(node: .groupTransform(
           .capture,
           component().regex.root,
           CaptureTransform(resultType: NewCapture.self) {
             transform($0) as Any?
           },
-          reference?.id))
+          reference.id))
       }
 
+
       """)
   }
 }

Sources/_StringProcessing/ByteCodeGen.swift

@@ -298,7 +298,7 @@ extension Compiler.ByteCodeGen {
   mutating func emitGroup(
     _ kind: AST.Group.Kind,
     _ child: DSLTree.Node,
-    _ referenceID: Reference.ID?
+    _ referenceID: ReferenceID?
   ) throws -> CaptureRegister? {
     guard kind.isCapturing || referenceID == nil else {
       throw Unreachable("Reference ID shouldn't exist for non-capturing groups")

Sources/_StringProcessing/Capture.swift

@@ -52,9 +52,11 @@ func constructExistentialMatchComponent(
     underlying = Optional<Any>(nil) as Any
     someCount = optionalCount - 1
   }
-
   for _ in 0..<someCount {
-    underlying = Optional(underlying) as Any
+    func wrap<T>(_ x: T) {
+      underlying = Optional(x) as Any
+    }
+    _openExistential(underlying, do: wrap)
   }
   return underlying
 }

Sources/_StringProcessing/Engine/Consume.swift

@@ -63,7 +63,8 @@ extension Engine where Input == String {
     guard let result = result else { return nil }
 
     let capList = cpu.storedCaptures
-    return (result, CaptureList(caps: capList))
+    return (result, CaptureList(
+      values: capList, referencedCaptureOffsets: program.referencedCaptureOffsets))
   }
 }
 

Sources/_StringProcessing/Engine/MEBuilder.swift

@@ -43,8 +43,8 @@ extension MEProgram where Input.Element: Hashable {
     var captureStructure: CaptureStructure = .empty
 
     // Symbolic reference resolution
-    var unresolvedReferences: [Reference.ID: [InstructionAddress]] = [:]
-    var captureOffsets: [Reference.ID: Int] = [:]
+    var unresolvedReferences: [ReferenceID: [InstructionAddress]] = [:]
+    var referencedCaptureOffsets: [ReferenceID: Int] = [:]
     var captureCount: Int {
       // We currently deduce the capture count from the capture register number.
       nextCaptureRegister.rawValue
@@ -274,7 +274,7 @@ extension MEProgram.Builder {
       .init(.backreference, .init(capture: cap)))
   }
 
-  public mutating func buildUnresolvedReference(id: Reference.ID) {
+  public mutating func buildUnresolvedReference(id: ReferenceID) {
     buildBackreference(.init(0))
     unresolvedReferences[id, default: []].append(lastInstructionAddress)
   }
@@ -352,7 +352,8 @@ extension MEProgram.Builder {
       staticTransformFunctions: transformFunctions,
       staticMatcherFunctions: matcherFunctions,
       registerInfo: regInfo,
-      captureStructure: captureStructure)
+      captureStructure: captureStructure,
+      referencedCaptureOffsets: referencedCaptureOffsets)
   }
 
   public mutating func reset() { self = Self() }
@@ -424,7 +425,7 @@ extension MEProgram.Builder {
 fileprivate extension MEProgram.Builder {
   mutating func resolveReferences() throws {
     for (id, uses) in unresolvedReferences {
-      guard let offset = captureOffsets[id] else {
+      guard let offset = referencedCaptureOffsets[id] else {
         throw RegexCompilationError.uncapturedReference
       }
       for use in uses {
@@ -437,11 +438,11 @@ fileprivate extension MEProgram.Builder {
 
 // Register helpers
 extension MEProgram.Builder {
-  public mutating func makeCapture(id: Reference.ID?) -> CaptureRegister {
+  public mutating func makeCapture(id: ReferenceID?) -> CaptureRegister {
     defer { nextCaptureRegister.rawValue += 1 }
     // Register the capture for later lookup via symbolic references.
     if let id = id {
-      let preexistingValue = captureOffsets.updateValue(
+      let preexistingValue = referencedCaptureOffsets.updateValue(
         captureCount, forKey: id)
       assert(preexistingValue == nil)
     }

Sources/_StringProcessing/Engine/MECapture.swift

@@ -143,14 +143,15 @@ extension Processor._StoredCapture: CustomStringConvertible {
 }
 
 public struct CaptureList {
-  var caps: Array<Processor<String>._StoredCapture>
+  var values: Array<Processor<String>._StoredCapture>
+  var referencedCaptureOffsets: [ReferenceID: Int]
 
 //  func extract(from s: String) -> Array<Array<Substring>> {
 //    caps.map { $0.map { s[$0] }  }
 //  }
 //
   func latestUntyped(from s: String) -> Array<Substring?> {
-    caps.map {
+    values.map {
       guard let last = $0.latest else {
         return nil
       }

Sources/_StringProcessing/Engine/MEProgram.swift

@@ -35,6 +35,7 @@ public struct MEProgram<Input: Collection> where Input.Element: Equatable {
   var enableTracing: Bool = false
 
   let captureStructure: CaptureStructure
+  let referencedCaptureOffsets: [ReferenceID: Int]
 }
 
 extension MEProgram: CustomStringConvertible {

Sources/_StringProcessing/Engine/Processor.swift

@@ -14,6 +14,8 @@ public enum MatchMode {
   case partialFromFront
 }
 
+typealias Program = MEProgram<String>
+
 /// A concrete CU. Somehow will run the concrete logic and
 /// feed stuff back to generic code
 struct Controller {

Sources/_StringProcessing/Engine/Structuralize.swift

@@ -62,17 +62,17 @@ extension CaptureStructure {
     }
 
     switch self {
-    case let .tuple(caps):
-      assert(list.caps.count == caps.count)
+    case let .tuple(values):
+      assert(list.values.count == values.count)
       var result = Array<StructuredCapture>()
-      for (cap, storedCap) in zip(caps, list.caps) {
+      for (cap, storedCap) in zip(values, list.values) {
         result.append(mapCap(cap, storedCap))
       }
       return result
 
     default:
-      assert(list.caps.count == 1)
-      return [mapCap(self, list.caps.first!)]
+      assert(list.values.count == 1)
+      return [mapCap(self, list.values.first!)]
     }
   }
 }