Merge pull request #440 from hamishknight/chunk-loader

Additional character property parsing

Author: hamishknight

Sources/_RegexParser/Regex/AST/Atom.swift

@@ -441,28 +441,55 @@ extension AST.Atom.CharacterProperty {
 
     /// 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?
   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 {

Sources/_RegexParser/Regex/Parse/CharacterPropertyClassification.swift

@@ -62,6 +62,7 @@ enum ParseError: Error, Hashable {
   case unknownProperty(key: String?, value: String)
   case unrecognizedScript(String)
   case unrecognizedCategory(String)
+  case unrecognizedBlock(String)
   case invalidAge(String)
   case invalidNumericValue(String)
   case unrecognizedNumericType(String)
@@ -195,6 +196,8 @@ extension ParseError: CustomStringConvertible {
       return "unrecognized script '\(value)'"
     case .unrecognizedCategory(let value):
       return "unrecognized category '\(value)'"
+    case .unrecognizedBlock(let value):
+      return "unrecognized block '\(value)'"
     case .unrecognizedNumericType(let value):
       return "unrecognized numeric type '\(value)'"
     case .invalidAge(let value):

Sources/_RegexParser/Regex/Parse/Diagnostics.swift

@@ -173,8 +173,10 @@ extension RegexValidator {
       break
     case .pcreSpecial:
       throw error(.unsupported("PCRE property"), at: loc)
-    case .onigurumaSpecial:
+    case .block:
       throw error(.unsupported("Unicode block property"), at: loc)
+    case .javaSpecial:
+      throw error(.unsupported("Java property"), at: loc)
     }
   }