Fixing issue #2009

src/org/rascalmpl/library/lang/rascal/tests/concrete/Character.rsc

@@ -49,6 +49,41 @@ test bool charClassOrderedRanges() = (#[a-z A-Z]).symbol == \char-class([range(6
 test bool charClassMergedRanges() = (#[A-Z F-G]).symbol == \char-class([range(65,90)]);
 test bool charClassExtendedRanges() = (#[A-M N-Z]).symbol == \char-class([range(65,90)]);
 
+test bool asciiEscape() = \char-class([range(0,127)]) == #[\a00-\a7F].symbol;
+test bool utf16Escape() = \char-class([range(0,65535)]) == #[\u0000-\uFFFF].symbol;
+test bool utf24Escape() = \char-class([range(0,1114111)]) == #[\U000000-\U10FFFF].symbol;
+test bool highLowSurrogateRange1() = \char-class([range(9312,12991)]) == #[①-㊿].symbol;
+test bool highLowSurrogateRange2() = \char-class([range(127829,127829)]) == #[🍕].symbol;
+test bool differentEscapesSameResult1() = #[\a00-\a7F] == #[\u0000-\u007F];
+test bool differentEscapesSameResult2() = #[\a00-\a7F] == #[\U000000-\U00007F];
+
+/* to avoid a known ambiguity */
+alias NotAZ = ![A-Z];
+
+test bool unicodeCharacterClassSubtype1() {
+  Tree t = char(charAt("⑭", 0));
+
+  if ([①-㊿] circled := t) {
+    assert [⑭] _ := circled;
+    assert NotAZ _ := circled;
+    return true;
+  }
+
+  return false;
+}
+
+test bool unicodeCharacterClassSubtype2() {
+  Tree t = char(charAt("🍕", 0));
+
+  if ([🍕] pizza := t) {
+    assert [\a00-🍕] _ := pizza;
+    assert NotAZ _ := pizza;
+    return true;
+  }
+
+  return false;
+}
+
 // ambiguity in this syntax must be resolved first
 //test bool differenceCC() = (#[a-zA-Z] - [A-Z]).symbol == (#[a-z]).symbol;
 //test bool unionCC()      = (#[a-z] || [A-Z]).symbol == (#[A-Za-z]).symbol;

src/org/rascalmpl/values/parsetrees/SymbolFactory.java

@@ -337,31 +337,30 @@
 
 	private static IValue char2int(Char character) {
 		String s = ((Char.Lexical) character).getString();
-		if (s.startsWith("\\")) {
-			if (s.length() > 1 && java.lang.Character.isDigit(s.charAt(1))) { // octal escape
-				// TODO
-				throw new NotYetImplemented("octal escape sequence in character class types");
-			}
-			if (s.length() > 1 && s.charAt(1) == 'u') { // octal escape
-				// TODO
-				throw new NotYetImplemented("unicode escape sequence in character class types");
-			}
-			char cha = s.charAt(1);
+		if (s.matches("\\\\[auU][0-9A-F]+")) {
+			// ascii escape (a), utf16 escape (u) or utf24 escape (U)
+			return factory.integer(Integer.parseInt(s.substring(2), 16));
+		}
+		else if (s.startsWith("\\")) {
+			// builtin escape
+			int cha = s.codePointAt(1);
 			switch (cha) {
-			case 't': return factory.integer('\t');
-			case 'n': return factory.integer('\n');
-			case 'r': return factory.integer('\r');
-			case '\"' : return factory.integer('\"');
-			case '\'' : return factory.integer('\'');
-			case '-' : return factory.integer('-');
-			case '<' : return factory.integer('<');
-			case '>' : return factory.integer('>');
-			case '\\' : return factory.integer('\\');
+				case 't': return factory.integer('\t');
+				case 'n': return factory.integer('\n');
+				case 'r': return factory.integer('\r');
+				case '\"' : return factory.integer('\"');
+				case '\'' : return factory.integer('\'');
+				case '-' : return factory.integer('-');
+				case '<' : return factory.integer('<');
+				case '>' : return factory.integer('>');
+				case '\\' : return factory.integer('\\');
+				default: return factory.integer(cha);
 			}
-			s = s.substring(1);
 		}
-		char cha = s.charAt(0);
-		return factory.integer(cha);
+		else {
+			// just a single character (but possibly two char's)
+			return factory.integer(s.codePointAt(0));
+		}
 	}
 
 	public static IConstructor charClass(int ch) {