Fix #140: Entity decoding broken if it crosses a chunk

renggli · Mar 26, 2022 · 8afc3a4 · 8afc3a4
1 parent ca19b86
commit 8afc3a4
Show file tree

Hide file tree

Showing 3 changed files with 31 additions and 2 deletions.
diff --git a/lib/src/xml_events/events/text.dart b/lib/src/xml_events/events/text.dart
@@ -1,3 +1,4 @@
+import '../../xml/entities/entity_mapping.dart';
 import '../../xml/utils/node_type.dart';
 import '../event.dart';
 import '../visitor.dart';
@@ -20,3 +21,27 @@ class XmlTextEvent extends XmlEvent {
   @override
   bool operator ==(Object other) => other is XmlTextEvent && other.text == text;
 }
+
+/// Internal event of an XML text node that is lazily decoded.
+class XmlRawTextEvent extends XmlEvent implements XmlTextEvent {
+  XmlRawTextEvent(this.raw, this.entityMapping);
+
+  final String raw;
+
+  final XmlEntityMapping entityMapping;
+
+  @override
+  late final String text = entityMapping.decode(raw);
+
+  @override
+  XmlNodeType get nodeType => XmlNodeType.TEXT;
+
+  @override
+  void accept(XmlEventVisitor visitor) => visitor.visitTextEvent(this);
+
+  @override
+  int get hashCode => Object.hash(nodeType, text);
+
+  @override
+  bool operator ==(Object other) => other is XmlTextEvent && other.text == text;
+}
diff --git a/lib/src/xml_events/parser.dart b/lib/src/xml_events/parser.dart
@@ -38,7 +38,7 @@ class XmlEventParser {
 
   Parser<XmlTextEvent> characterData() =>
       XmlCharacterDataParser(XmlToken.openElement, 1)
-          .map((each) => XmlTextEvent(entityMapping.decode(each)));
+          .map((each) => XmlRawTextEvent(each, entityMapping));
 
   Parser<XmlStartElementEvent> startElement() => [
         XmlToken.openElement.toParser(),

diff --git a/lib/src/xml_events/streams/normalizer.dart b/lib/src/xml_events/streams/normalizer.dart
@@ -37,10 +37,14 @@ class _XmlNormalizeEventsSink extends ChunkedConversionSink<List<XmlEvent>> {
     for (var i = 0; i < buffer.length - 1;) {
       final event1 = buffer[i], event2 = buffer[i + 1];
       if (event1 is XmlTextEvent && event2 is XmlTextEvent) {
-        final event = XmlTextEvent(event1.text + event2.text);
+        // Combine text nodes, decode the combined input.
+        final event = event1 is XmlRawTextEvent && event2 is XmlRawTextEvent
+            ? XmlRawTextEvent(event1.raw + event2.raw, event1.entityMapping)
+            : XmlTextEvent(event1.text + event2.text);
         // Propagate annotations.
         event.attachBuffer(event1.buffer);
         event.attachLocation(event1.start, event2.stop);
+        assert(event1.parent == event2.parent);
         event.attachParent(event1.parent);
         // Update the buffer.
         buffer[i] = event;