Adds RegexBuilder.CharacterClass.anyUnicodeScalar (swiftlang#315)

This provides a RegexBuilder API that represents the same as `\O`
in regex syntax.

Author: natecook1000

Sources/RegexBuilder/CharacterClass.swift

@@ -60,6 +60,10 @@ extension RegexComponent where Self == CharacterClass {
   public static var anyGrapheme: CharacterClass {
     .init(unconverted: .anyGrapheme)
   }
+  
+  public static var anyUnicodeScalar: CharacterClass {
+    .init(unconverted: .anyUnicodeScalar)
+  }
 
   public static var whitespace: CharacterClass {
     .init(unconverted: .whitespace)

Sources/_StringProcessing/_CharacterClassModel.swift

@@ -33,6 +33,8 @@ public struct _CharacterClassModel: Hashable {
     case any
     /// Any grapheme cluster
     case anyGrapheme
+    /// Any Unicode scalar
+    case anyScalar
     /// Character.isDigit
     case digit
     /// Character.isHexDigit
@@ -155,8 +157,12 @@ public struct _CharacterClassModel: Hashable {
     case .graphemeCluster:
       let c = str[i]
       var matched: Bool
+      var next = str.index(after: i)
       switch cc {
       case .any, .anyGrapheme: matched = true
+      case .anyScalar:
+        matched = true
+        next = str.unicodeScalars.index(after: i)
       case .digit:
         matched = c.isNumber && (c.isASCII || !options.usesASCIIDigits)
       case .hexDigit:
@@ -174,12 +180,13 @@ public struct _CharacterClassModel: Hashable {
       if isInverted {
         matched.toggle()
       }
-      return matched ? str.index(after: i) : nil
+      return matched ? next : nil
     case .unicodeScalar:
       let c = str.unicodeScalars[i]
       var matched: Bool
       switch cc {
       case .any: matched = true
+      case .anyScalar: matched = true
       case .anyGrapheme: fatalError("Not matched in this mode")
       case .digit:
         matched = c.properties.numericType != nil && (c.isASCII || !options.usesASCIIDigits)
@@ -224,6 +231,10 @@ extension _CharacterClassModel {
     .init(cc: .anyGrapheme, matchLevel: .graphemeCluster)
   }
 
+  public static var anyUnicodeScalar: _CharacterClassModel {
+    .init(cc: .any, matchLevel: .unicodeScalar)
+  }
+
   public static var whitespace: _CharacterClassModel {
     .init(cc: .whitespace, matchLevel: .graphemeCluster)
   }
@@ -275,6 +286,7 @@ extension _CharacterClassModel.Representation: CustomStringConvertible {
     switch self {
     case .any: return "<any>"
     case .anyGrapheme: return "<any grapheme>"
+    case .anyScalar: return "<any scalar>"
     case .digit: return "<digit>"
     case .hexDigit: return "<hex digit>"
     case .horizontalWhitespace: return "<horizontal whitespace>"
@@ -431,6 +443,7 @@ extension AST.Atom.EscapedBuiltin {
     case .notWordCharacter: return .word.inverted
 
     case .graphemeCluster: return .anyGrapheme
+    case .trueAnychar: return .anyUnicodeScalar
 
     default:
       return nil

Tests/RegexTests/MatchTests.swift

@@ -1483,13 +1483,11 @@ extension RegexTests {
       (eDecomposed, false))
 
     // FIXME: \O is unsupported
-    firstMatchTest(#"\O\u{301}"#, input: eDecomposed, match: eDecomposed,
-              xfail: true)
-    firstMatchTest(#"e\O"#, input: eDecomposed, match: eDecomposed,
-              xfail: true)
-    firstMatchTest(#"\O\u{301}"#, input: eComposed, match: nil,
-              xfail: true)
-    firstMatchTest(#"e\O"#, input: eComposed, match: nil,
+    firstMatchTest(#"(?u)\O\u{301}"#, input: eDecomposed, match: eDecomposed)
+    firstMatchTest(#"(?u)e\O"#, input: eDecomposed, match: eDecomposed,
+      xfail: true)
+    firstMatchTest(#"\O"#, input: eComposed, match: eComposed)
+    firstMatchTest(#"\O"#, input: eDecomposed, match: nil,
               xfail: true)
 
     matchTest(