Fix some more unicode edge cases with character entities.

renggli · Mar 27, 2022 · f4c84cb · f4c84cb
1 parent 043f989
commit f4c84cb
Show file tree

Hide file tree

Showing 2 changed files with 22 additions and 4 deletions.
diff --git a/lib/src/xml/entities/default_mapping.dart b/lib/src/xml/entities/default_mapping.dart
@@ -27,19 +27,23 @@ class XmlDefaultEntityMapping extends XmlEntityMapping {
     if (input.length > 1 && input[0] == '#') {
       if (input.length > 2 && (input[1] == 'x' || input[1] == 'X')) {
         // Hexadecimal character reference.
-        final value = int.tryParse(input.substring(2), radix: 16);
-        return value != null ? String.fromCharCode(value) : null;
+        return _decodeNumericEntity(input.substring(2), 16);
       } else {
         // Decimal character reference.
-        final value = int.tryParse(input.substring(1));
-        return value != null ? String.fromCharCode(value) : null;
+        return _decodeNumericEntity(input.substring(1), 10);
       }
     } else {
       // Named character reference.
       return entities[input];
     }
   }
 
+  String? _decodeNumericEntity(String input, int radix) {
+    final value = int.tryParse(input, radix: radix);
+    if (value == null || value < 0 || 0x10FFFF < value) return null;
+    return String.fromCharCode(value);
+  }
+
   @override
   String encodeText(String input) =>
       input.replaceAllMapped(_textPattern, _textReplace);

diff --git a/test/entity_test.dart b/test/entity_test.dart
@@ -27,6 +27,20 @@ void testDefaultMapping(XmlEntityMapping entityMapping) {
       expect(entityMapping.decode('&#xInvalid;'), '&#xInvalid;');
       expect(entityMapping.decode('&#XInvalid;'), '&#XInvalid;');
     });
+    test('unicode', () {
+      // https://www.compart.com/en/unicode/U+0000
+      expect(entityMapping.decode('&#0;'), '\u0000');
+      expect(entityMapping.decode('&#x0000;'), '\u0000');
+      // https://www.compart.com/en/unicode/U+10FFFF
+      expect(entityMapping.decode('&#1114111;'), '\uDBFF\uDFFF');
+      expect(entityMapping.decode('&#x10FFFF;'), '\uDBFF\uDFFF');
+    });
+    test('unicode invalid', () {
+      expect(entityMapping.decode('&#-1;'), '&#-1;');
+      expect(entityMapping.decode('&#x-1;'), '&#x-1;');
+      expect(entityMapping.decode('&#1114112;'), '&#1114112;');
+      expect(entityMapping.decode('&#x110000;'), '&#x110000;');
+    });
     test('incomplete', () {
       expect(entityMapping.decode('&'), '&');
       expect(entityMapping.decode('&amp'), '&amp');