refactor reopening of dangling inline elements, make it lazy to be able to supress the repopen in case of empty elements

rbri · rbri · commit 7c31107452ef · 2025-11-02T12:27:19.000+01:00
diff --git a/src/main/java/org/htmlunit/cyberneko/HTMLTagBalancer.java b/src/main/java/org/htmlunit/cyberneko/HTMLTagBalancer.java
@@ -613,6 +613,8 @@ public void startElement(final QName elem, XMLAttributes attrs, final Augmentati
         final HTMLElements.Element element = getElement(elem);
         final short elementCode = element.code;
 
+        reopenInlineElements(null);
+
         if (elementCode == HTMLElements.TEMPLATE) {
             fTemplateFragment = true;
         }
@@ -980,6 +982,7 @@ public void startCDATA(final Augmentations augs) throws XNIException {
         fSeenAnything = true;
 
         consumeEarlyTextIfNeeded();
+        reopenInlineElements(null);
 
         // check for end of document
         if (fSeenRootElementEnd) {
@@ -1023,6 +1026,8 @@ public void characters(final XMLString text, final Augmentations augs) throws XN
             return;
         }
 
+        reopenInlineElements(null);
+
         if (!fDocumentFragment) {
             // handle bare characters
             if (!fSeenRootElement) {
@@ -1091,6 +1096,10 @@ public void endElement(final QName element, final Augmentations augs) throws XNI
         final HTMLElements.Element elem = getElement(element);
         final short elementCode = elem.code;
 
+        if (reopenInlineElements(elem)) {
+            return;
+        }
+
         if (!fTemplateFragment && fOpenedSelect) {
             if (elementCode == HTMLElements.SELECT) {
                 fOpenedSelect = false;
@@ -1199,21 +1208,42 @@ else if (!elem.isEmpty()) {
                 callEndElement(info.qname, i < depth - 1 ? synthesizedAugs() : augs);
             }
         }
+    }
 
-        // re-open inline elements
-        if (depth > 1) {
-            final int size = fInlineStack.top;
-            for (int i = 0; i < size; i++) {
-                final Info info = fInlineStack.pop();
-                final XMLAttributes attributes = info.attributes;
-                if (fReportErrors) {
-                    final String iname = info.qname.getRawname();
-                    fErrorReporter.reportWarning("HTML2008", new Object[]{iname});
-                }
-                forceStartElement(info.qname, attributes, synthesizedAugs());
-            }
+    // re-open inline elements
+    protected boolean reopenInlineElements(final HTMLElements.Element element) {
+        final int size = fInlineStack.top;
+
+        if (size == 0) {
+            return false;
         }
 
+        int i = 0;
+        Info info = fInlineStack.pop();
+        XMLAttributes attributes = info.attributes;
+        if (fReportErrors) {
+            final String iname = info.qname.getRawname();
+            fErrorReporter.reportWarning("HTML2008", new Object[]{iname});
+        }
+
+        if (element != null && info.element.code == element.code) {
+            return true;
+        }
+
+        forceStartElement(info.qname, attributes, synthesizedAugs());
+        i++;
+
+        for ( ; i < size; i++) {
+            info = fInlineStack.pop();
+            attributes = info.attributes;
+            if (fReportErrors) {
+                final String iname = info.qname.getRawname();
+                fErrorReporter.reportWarning("HTML2008", new Object[]{iname});
+            }
+
+            forceStartElement(info.qname, attributes, synthesizedAugs());
+        }
+        return false;
     }
 
     // Returns an HTML element.
diff --git a/src/test/resources/org/htmlunit/cyberneko/testfiles/malformed/test-overlapping-tags.html b/src/test/resources/org/htmlunit/cyberneko/testfiles/malformed/test-overlapping-tags.html
@@ -1,6 +1,7 @@
 <html>
 <body>
 <b><i>Bold and italic</b></i>
-<strong><em>Strong and emphasis</strong></em>
+<strong><em>Strong and emphasis</strong>  </em>
+<strong><em>text between</strong>some</em>
 </body>
 </html>
diff --git a/src/test/resources/org/htmlunit/cyberneko/testfiles/malformed/test-overlapping-tags.html.canonical b/src/test/resources/org/htmlunit/cyberneko/testfiles/malformed/test-overlapping-tags.html.canonical
@@ -15,6 +15,16 @@
 )em
 )strong
 (em
+"  
+)em
+"\n
+(strong
+(em
+"text between
+)em
+)strong
+(em
+"some
 )em
 "\n\n
 )body

-Original file line number
+Diff line change
 )em
 )strong
 (em
 +"
 +)em
 +"\n
 +(strong
 +(em
 +"text between
 +)em
 +)strong
 +(em
 +"some
 )em
 "\n\n
 )body