Make the RTE handle inline elements gracefully for JSON (#17983)

umbraco · Jan 15, 2025 · 8f42a3a · 8f42a3a
1 parent 4dd8fbe
commit 8f42a3a
Show file tree

Hide file tree

Showing 2 changed files with 44 additions and 1 deletion.
diff --git a/src/Umbraco.Infrastructure/DeliveryApi/ApiRichTextElementParser.cs b/src/Umbraco.Infrastructure/DeliveryApi/ApiRichTextElementParser.cs
@@ -101,8 +101,9 @@ private T ParseElement<T>(HtmlNode element, IPublishedSnapshot publishedSnapshot
         // - non-#comment nodes
         // - non-#text nodes
         // - non-empty #text nodes
+        // - empty #text between inline elements (see #17037)
         HtmlNode[] childNodes = element.ChildNodes
-            .Where(c => c.Name != CommentNodeName && (c.Name != TextNodeName || string.IsNullOrWhiteSpace(c.InnerText) is false))
+            .Where(c => c.Name != CommentNodeName && (c.Name != TextNodeName || c.NextSibling is not null || string.IsNullOrWhiteSpace(c.InnerText) is false))
             .ToArray();
 
         var tag = TagName(element);

diff --git a/tests/Umbraco.Tests.UnitTests/Umbraco.Core/DeliveryApi/RichTextParserTests.cs b/tests/Umbraco.Tests.UnitTests/Umbraco.Core/DeliveryApi/RichTextParserTests.cs
@@ -357,6 +357,48 @@ public void ParseElement_CanHandleMixedInlineAndBlockLevelBlocks()
         Assert.IsEmpty(blockLevelBlock.Elements);
     }
 
+    [Test]
+    public void ParseElement_CanHandleWhitespaceAroundInlineElemements()
+    {
+        var parser = CreateRichTextElementParser();
+
+        var element = parser.Parse("<p>What follows from <strong>here</strong> <em>is</em> <a href=\"#\">just</a> a bunch of text.</p>") as RichTextRootElement;
+        Assert.IsNotNull(element);
+        var paragraphElement = element.Elements.Single() as RichTextGenericElement;
+        Assert.IsNotNull(paragraphElement);
+
+        var childElements = paragraphElement.Elements.ToArray();
+        Assert.AreEqual(7, childElements.Length);
+
+        var childElementCounter = 0;
+
+        void AssertNextChildElementIsText(string expectedText)
+        {
+            var textElement = childElements[childElementCounter++] as RichTextTextElement;
+            Assert.IsNotNull(textElement);
+            Assert.AreEqual(expectedText, textElement.Text);
+        }
+
+        void AssertNextChildElementIsGeneric(string expectedTag, string expectedInnerText)
+        {
+            var genericElement = childElements[childElementCounter++] as RichTextGenericElement;
+            Assert.IsNotNull(genericElement);
+            Assert.AreEqual(expectedTag, genericElement.Tag);
+            Assert.AreEqual(1, genericElement.Elements.Count());
+            var textElement = genericElement.Elements.First() as RichTextTextElement;
+            Assert.IsNotNull(textElement);
+            Assert.AreEqual(expectedInnerText, textElement.Text);
+        }
+
+        AssertNextChildElementIsText("What follows from ");
+        AssertNextChildElementIsGeneric("strong", "here");
+        AssertNextChildElementIsText(" ");
+        AssertNextChildElementIsGeneric("em", "is");
+        AssertNextChildElementIsText(" ");
+        AssertNextChildElementIsGeneric("a", "just");
+        AssertNextChildElementIsText(" a bunch of text.");
+    }
+
     [Test]
     public void ParseMarkup_CanParseContentLink()
     {