Merge pull request #442 from hamishknight/main-merge

Author: hamishknight

Package.swift

@@ -7,10 +7,18 @@ let availabilityDefinition = PackageDescription.SwiftSetting.unsafeFlags([
     "-Xfrontend",
     "-define-availability",
     "-Xfrontend",
-    #"SwiftStdlib 5.7:macOS 9999, iOS 9999, watchOS 9999, tvOS 9999"#,
+    "SwiftStdlib 5.7:macOS 9999, iOS 9999, watchOS 9999, tvOS 9999",
 ])
 
-let stdlibSettings: [PackageDescription.SwiftSetting] = [
+/// Swift settings for building a private stdlib-like module that is to be used
+/// by other stdlib-like modules only.
+let privateStdlibSettings: [PackageDescription.SwiftSetting] = [
+    .unsafeFlags(["-Xfrontend", "-disable-implicit-concurrency-module-import"]),
+    .unsafeFlags(["-Xfrontend", "-disable-implicit-string-processing-module-import"]),
+]
+
+/// Swift settings for building a user-facing stdlib-like module.
+let publicStdlibSettings: [PackageDescription.SwiftSetting] = [
     .unsafeFlags(["-enable-library-evolution"]),
     .unsafeFlags(["-Xfrontend", "-disable-implicit-concurrency-module-import"]),
     .unsafeFlags(["-Xfrontend", "-disable-implicit-string-processing-module-import"]),
@@ -43,7 +51,7 @@ let package = Package(
         .target(
             name: "_RegexParser",
             dependencies: [],
-            swiftSettings: stdlibSettings),
+            swiftSettings: privateStdlibSettings),
         .testTarget(
             name: "MatchingEngineTests",
             dependencies: [
@@ -55,11 +63,11 @@ let package = Package(
         .target(
             name: "_StringProcessing",
             dependencies: ["_RegexParser", "_CUnicode"],
-            swiftSettings: stdlibSettings),
+            swiftSettings: publicStdlibSettings),
         .target(
             name: "RegexBuilder",
             dependencies: ["_StringProcessing", "_RegexParser"],
-            swiftSettings: stdlibSettings),
+            swiftSettings: publicStdlibSettings),
         .testTarget(
             name: "RegexTests",
             dependencies: ["_StringProcessing"],

Sources/PatternConverter/PatternConverter.swift

@@ -70,7 +70,8 @@ struct PatternConverter: ParsableCommand {
 
     print()
     if !skipDSL {
-      let render = ast.renderAsBuilderDSL(
+      let render = renderAsBuilderDSL(
+        ast: ast,
         maxTopDownLevels: topDownConversionLimit,
         minBottomUpLevels: bottomUpConversionLimit
       )

Sources/RegexBuilder/Algorithms.swift

@@ -9,7 +9,7 @@
 //
 //===----------------------------------------------------------------------===//
 
-import _StringProcessing
+@_spi(RegexBuilder) import _StringProcessing
 
 // FIXME(rdar://92459215): We should be using 'some RegexComponent' instead of
 // <R: RegexComponent> for the methods below that don't impose any additional
@@ -313,3 +313,31 @@ where Self: BidirectionalCollection, SubSequence == Substring {
     try replace(content(), maxReplacements: maxReplacements, with: replacement)
   }
 }
+
+// String split overload breakers
+
+extension StringProtocol where SubSequence == Substring {
+  @available(SwiftStdlib 5.7, *)
+  public func split(
+    separator: String,
+    maxSplits: Int = .max,
+    omittingEmptySubsequences: Bool = true
+  ) -> [Substring] {
+    return _split(
+      separator: separator,
+      maxSplits: maxSplits,
+      omittingEmptySubsequences: omittingEmptySubsequences)
+  }
+  
+  @available(SwiftStdlib 5.7, *)
+  public func split(
+    separator: Substring,
+    maxSplits: Int = .max,
+    omittingEmptySubsequences: Bool = true
+  ) -> [Substring] {
+    return _split(
+      separator: separator,
+      maxSplits: maxSplits,
+      omittingEmptySubsequences: omittingEmptySubsequences)
+  }
+}

Sources/RegexBuilder/DSL.swift

@@ -127,6 +127,19 @@ extension DSLTree.Node {
   }
 }
 
+/// A regex component that matches exactly one occurrence of its underlying
+/// component.
+@available(SwiftStdlib 5.7, *)
+public struct One<Output>: RegexComponent {
+  public var regex: Regex<Output>
+
+  public init<Component: RegexComponent>(
+    _ component: Component
+  ) where Component.RegexOutput == Output {
+    self.regex = component.regex
+  }
+}
+
 @available(SwiftStdlib 5.7, *)
 public struct OneOrMore<Output>: _BuiltinRegexComponent {
   public var regex: Regex<Output>

Sources/_RegexParser/Regex/AST/AST.swift

@@ -29,7 +29,6 @@ extension AST {
 
 extension AST {
   /// A node in the regex AST.
-  @frozen
   public indirect enum Node:
     Hashable, _TreeNode //, _ASTPrintable ASTValue, ASTAction
   {
@@ -53,6 +52,9 @@ extension AST {
     /// Comments, non-semantic whitespace, etc
     case trivia(Trivia)
 
+    /// Intepolation `<{...}>`, currently reserved for future use.
+    case interpolation(Interpolation)
+
     case atom(Atom)
 
     case customCharacterClass(CustomCharacterClass)
@@ -78,6 +80,7 @@ extension AST.Node {
     case let .quantification(v):        return v
     case let .quote(v):                 return v
     case let .trivia(v):                return v
+    case let .interpolation(v):         return v
     case let .atom(v):                  return v
     case let .customCharacterClass(v):  return v
     case let .empty(v):                 return v
@@ -130,7 +133,7 @@ extension AST.Node {
     case .conditional, .customCharacterClass, .absentFunction:
       return true
     case .alternation, .concatenation, .quantification, .quote, .trivia,
-        .empty:
+        .empty, .interpolation:
       return false
     }
   }
@@ -194,6 +197,16 @@ extension AST {
     }
   }
 
+  public struct Interpolation: Hashable, _ASTNode {
+    public let contents: String
+    public let location: SourceLocation
+
+    public init(_ contents: String, _ location: SourceLocation) {
+      self.contents = contents
+      self.location = location
+    }
+  }
+
   public struct Empty: Hashable, _ASTNode {
     public let location: SourceLocation
 
@@ -249,7 +262,6 @@ extension AST {
   }
 
   public struct Reference: Hashable {
-    @frozen
     public enum Kind: Hashable {
       // \n \gn \g{n} \g<n> \g'n' (?n) (?(n)...
       // Oniguruma: \k<n>, \k'n'

Sources/_RegexParser/Regex/AST/Atom.swift

@@ -19,7 +19,6 @@ extension AST {
       self.location = loc
     }
 
-    @frozen
     public enum Kind: Hashable {
       /// Just a character
       ///
@@ -146,7 +145,6 @@ extension AST.Atom {
 
   // Characters, character types, literals, etc., derived from
   // an escape sequence.
-  @frozen
   public enum EscapedBuiltin: Hashable {
     // TODO: better doc comments
 
@@ -399,7 +397,6 @@ extension AST.Atom {
 }
 
 extension AST.Atom.CharacterProperty {
-  @frozen
   public enum Kind: Hashable {
     /// Matches any character, equivalent to Oniguruma's '\O'.
     case any
@@ -430,27 +427,73 @@ extension AST.Atom.CharacterProperty {
     /// Character name in the form `\p{name=...}`
     case named(String)
 
+    /// Numeric type.
+    case numericType(Unicode.NumericType)
+    
+    /// Numeric value.
+    case numericValue(Double)
+    
+    /// Case mapping.
+    case mapping(MapKind, String)
+    
+    /// Canonical Combining Class.
+    case ccc(Unicode.CanonicalCombiningClass)
+    
+    /// Character age, as per UnicodeScalar.Properties.age.
+    case age(major: Int, minor: Int)
+
+    /// A block property.
+    case block(Unicode.Block)
+
     case posix(Unicode.POSIXProperty)
 
     /// Some special properties implemented by PCRE and Oniguruma.
     case pcreSpecial(PCRESpecialCategory)
-    case onigurumaSpecial(OnigurumaSpecialProperty)
+
+    /// Some special properties implemented by Java.
+    case javaSpecial(JavaSpecial)
+
+    public enum MapKind: Hashable {
+      case lowercase
+      case uppercase
+      case titlecase
+    }
   }
 
-  // TODO: erm, separate out or fold into something? splat it in?
-  @frozen
   public enum PCRESpecialCategory: String, Hashable {
     case alphanumeric     = "Xan"
     case posixSpace       = "Xps"
     case perlSpace        = "Xsp"
     case universallyNamed = "Xuc"
     case perlWord         = "Xwd"
   }
+
+  /// Special Java properties that correspond to methods on
+  /// `java.lang.Character`, with the `java` prefix replaced by `is`.
+  public enum JavaSpecial: String, Hashable, CaseIterable {
+    case alphabetic             = "javaAlphabetic"
+    case defined                = "javaDefined"
+    case digit                  = "javaDigit"
+    case identifierIgnorable    = "javaIdentifierIgnorable"
+    case ideographic            = "javaIdeographic"
+    case isoControl             = "javaISOControl"
+    case javaIdentifierPart     = "javaJavaIdentifierPart" // not a typo, that's actually the name
+    case javaIdentifierStart    = "javaJavaIdentifierStart" // not a typo, that's actually the name
+    case javaLetter             = "javaLetter"
+    case javaLetterOrDigit      = "javaLetterOrDigit"
+    case lowerCase              = "javaLowerCase"
+    case mirrored               = "javaMirrored"
+    case spaceChar              = "javaSpaceChar"
+    case titleCase              = "javaTitleCase"
+    case unicodeIdentifierPart  = "javaUnicodeIdentifierPart"
+    case unicodeIdentifierStart = "javaUnicodeIdentifierStart"
+    case upperCase              = "javaUpperCase"
+    case whitespace             = "javaWhitespace"
+  }
 }
 
 extension AST.Atom {
   /// Anchors and other built-in zero-width assertions.
-  @frozen
   public enum AssertionKind: String {
     /// \A
     case startOfSubject = #"\A"#
@@ -824,7 +867,7 @@ extension AST.Node {
     case .alternation, .concatenation, .group,
         .conditional, .quantification, .quote,
         .trivia, .customCharacterClass, .empty,
-        .absentFunction:
+        .absentFunction, .interpolation:
       return nil
     }
   }

Sources/_RegexParser/Regex/AST/CustomCharClass.swift

@@ -27,7 +27,6 @@ extension AST {
       self.location = sr
     }
 
-    @frozen
     public enum Member: Hashable {
       /// A nested custom character class `[[ab][cd]]`
       case custom(CustomCharacterClass)
@@ -52,20 +51,23 @@ extension AST {
       public var lhs: Atom
       public var dashLoc: SourceLocation
       public var rhs: Atom
+      public var trivia: [AST.Trivia]
 
-      public init(_ lhs: Atom, _ dashLoc: SourceLocation, _ rhs: Atom) {
+      public init(
+        _ lhs: Atom, _ dashLoc: SourceLocation, _ rhs: Atom,
+        trivia: [AST.Trivia]
+      ) {
         self.lhs = lhs
         self.dashLoc = dashLoc
         self.rhs = rhs
+        self.trivia = trivia
       }
     }
-    @frozen
     public enum SetOp: String, Hashable {
       case subtraction = "--"
       case intersection = "&&"
       case symmetricDifference = "~~"
     }
-    @frozen
     public enum Start: String {
       case normal = "["
       case inverted = "[^"
@@ -98,6 +100,11 @@ extension CustomCC.Member {
     return false
   }
 
+  public var asTrivia: AST.Trivia? {
+    guard case .trivia(let t) = self else { return nil }
+    return t
+  }
+
   public var isSemantic: Bool {
     !isTrivia
   }

Sources/_RegexParser/Regex/AST/Quantification.swift

@@ -36,7 +36,6 @@ extension AST {
       self.trivia = trivia
     }
 
-    @frozen
     public enum Amount: Hashable {
       case zeroOrMore              // *
       case oneOrMore               // +
@@ -47,7 +46,6 @@ extension AST {
       case range(Located<Int>, Located<Int>) // {n,m}
     }
 
-    @frozen
     public enum Kind: String, Hashable {
       case eager      = ""
       case reluctant  = "?"

Sources/_RegexParser/Regex/Parse/CaptureList.swift

@@ -42,6 +42,21 @@ extension CaptureList {
   }
 }
 
+extension CaptureList {
+  /// Retrieve the capture index of a given named capture, or `nil` if there is
+  /// no such capture.
+  public func indexOfCapture(named name: String) -> Int? {
+    // Named references are guaranteed to be unique for literal ASTs by Sema.
+    // The DSL tree does not use named references.
+    captures.indices.first(where: { captures[$0].name == name })
+  }
+
+  /// Whether the capture list has a given named capture.
+  public func hasCapture(named name: String) -> Bool {
+    indexOfCapture(named: name) != nil
+  }
+}
+
 // MARK: Generating from AST
 
 extension AST.Node {
@@ -103,7 +118,7 @@ extension AST.Node {
         break
       }
 
-    case .quote, .trivia, .atom, .customCharacterClass, .empty:
+    case .quote, .trivia, .atom, .customCharacterClass, .empty, .interpolation:
       break
     }
   }