Fix the handling by KtXmlReader of text with entities in it. It will now

never be recorded as ignorable whitespace, even when parsed as separate parts. This should fix #241.
pdvrieze · Nov 23, 2024 · 1438e45 · 1438e45
1 parent 51c7c0f
commit 1438e45
Show file tree

Hide file tree

Showing 3 changed files with 26 additions and 2 deletions.
diff --git a/core/base/src/commonMain/kotlin/nl/adaptivity/xmlutil/core/KtXmlReader.kt b/core/base/src/commonMain/kotlin/nl/adaptivity/xmlutil/core/KtXmlReader.kt
@@ -426,7 +426,7 @@ public class KtXmlReader internal constructor(
             _eventType = COMMENT
             return
         }
-
+        val lastEvent = _eventType
         val eventType = peekType()
         _eventType = eventType
         when (eventType) {
@@ -445,7 +445,10 @@ public class KtXmlReader internal constructor(
                 if (depth == 1) state = State.POST
             }
 
-            TEXT -> {
+            TEXT -> if (lastEvent == ENTITY_REF) { // Entity refs are part of text, so don't
+                // consider the following text whitespace at all
+                pushRegularText('<', false)
+            } else {
                 pushText('<')
                 if (isWhitespace) _eventType = IGNORABLE_WHITESPACE
             }

diff --git a/testutil/api/testutil.api b/testutil/api/testutil.api
@@ -38,5 +38,6 @@ public abstract class nl/adaptivity/xmlutil/test/TestCommonReader {
 	protected final fun testReadUnknownEntity (Lkotlin/jvm/functions/Function1;)V
 	public fun testReaderWithBOM ()V
 	protected final fun testReaderWithBOM (Lkotlin/jvm/functions/Function1;)V
+	public fun testWhiteSpaceWithEntity ()V
 }
 
diff --git a/testutil/src/commonMain/kotlin/nl/adaptivity/xmlutil/test/TestCommonReader.kt b/testutil/src/commonMain/kotlin/nl/adaptivity/xmlutil/test/TestCommonReader.kt
@@ -328,4 +328,24 @@ abstract class TestCommonReader {
     open fun testProcessingInstructionDom() {
         testProcessingInstruction(::createReader) { DomWriter() }
     }
+
+    @Test
+    open fun testWhiteSpaceWithEntity() {
+        val data = "<x>   &amp;   </x>"
+        val r = createReader(data)
+        r.nextTag()
+        r.require(EventType.START_ELEMENT, "", "x")
+        assertEquals(EventType.TEXT, r.next())
+        r.require(EventType.TEXT, null)
+        if (r.text.isBlank()) { // either parse as 3 parts or as a single text
+            assertEquals("   ", r.text)
+            assertEquals(EventType.ENTITY_REF, r.next())
+            assertEquals("&", r.text)
+            assertEquals(EventType.TEXT, r.next())
+            assertEquals("   ", r.text)
+        } else {
+            assertEquals("   &   ", r.text)
+        }
+
+    }
 }