Merge pull request #540 from hamishknight/scalar-support

Update to match Unicode proposal

Author: hamishknight

Sources/RegexBuilder/CharacterClass.swift

@@ -48,14 +48,10 @@ extension RegexComponent where Self == CharacterClass {
     .init(DSLTree.CustomCharacterClass(members: [.atom(.any)]))
   }
 
-  public static var anyGrapheme: CharacterClass {
+  public static var anyGraphemeCluster: CharacterClass {
     .init(unconverted: .anyGrapheme)
   }
 
-  public static var anyUnicodeScalar: CharacterClass {
-    .init(unconverted: .anyUnicodeScalar)
-  }
-
   public static var whitespace: CharacterClass {
     .init(unconverted: .whitespace)
   }

Sources/_RegexParser/Regex/Parse/Sema.swift

@@ -220,14 +220,14 @@ extension RegexValidator {
     _ esc: AST.Atom.EscapedBuiltin, at loc: SourceLocation
   ) {
     switch esc {
-    case .resetStartOfMatch, .singleDataUnit,
+    case .resetStartOfMatch, .singleDataUnit, .trueAnychar,
         // '\N' needs to be emitted using 'emitAny'.
         .notNewline:
       error(.unsupported("'\\\(esc.character)'"), at: loc)
 
     // Character classes.
     case .decimalDigit, .notDecimalDigit, .whitespace, .notWhitespace,
-        .wordCharacter, .notWordCharacter, .graphemeCluster, .trueAnychar,
+        .wordCharacter, .notWordCharacter, .graphemeCluster,
         .horizontalWhitespace, .notHorizontalWhitespace,
         .verticalTab, .notVerticalTab:
       break

Tests/RegexBuilderTests/AlgorithmsTests.swift

@@ -259,7 +259,7 @@ class AlgorithmsResultBuilderTests: XCTestCase {
   func testStartsAndContains() throws {
     let fam = "👨‍👩‍👧‍👦👨‍👨‍👧‍👧  we Ⓡ family"
     let startsWithGrapheme = fam.starts {
-      OneOrMore(.anyGrapheme)
+      OneOrMore(.anyGraphemeCluster)
       OneOrMore(.whitespace)
     }
     XCTAssertEqual(startsWithGrapheme, true)
@@ -271,7 +271,7 @@ class AlgorithmsResultBuilderTests: XCTestCase {
 
     let content = {
       Regex {
-        OneOrMore(.anyGrapheme)
+        OneOrMore(.anyGraphemeCluster)
         OneOrMore(.whitespace)
       }
     }
@@ -320,7 +320,7 @@ class AlgorithmsResultBuilderTests: XCTestCase {
 
     var mutable = "👨‍👩‍👧‍👦  we Ⓡ family"
     mutable.trimPrefix {
-      .anyGrapheme
+      .anyGraphemeCluster
       ZeroOrMore(.whitespace)
     }
     XCTAssertEqual(mutable, "we Ⓡ family")

Tests/RegexTests/MatchTests.swift

@@ -1815,7 +1815,7 @@ extension RegexTests {
       match: eDecomposed,
       xfail: true
     )
-    firstMatchTest(#"\O"#, input: eComposed, match: eComposed)
+    firstMatchTest(#"\O"#, input: eComposed, match: eComposed, xfail: true)
     firstMatchTest(#"\O"#, input: eDecomposed, match: nil,
               xfail: true)
 

Tests/RegexTests/ParseTests.swift

@@ -806,6 +806,8 @@ extension RegexTests {
     parseTest(#"\M-\C--"#, atom(.keyboardMetaControl("-")), unsupported: true)
     parseTest(#"\M-a"#, atom(.keyboardMeta("a")), unsupported: true)
 
+    parseTest(#"\O"#, escaped(.trueAnychar), unsupported: true)
+
     // MARK: Comments
 
     parseTest(