#24 Write \n when formatting <br /> or <hr /> into plaintext.

Author: CXuesong

MwParserFromScratch/MwParserFromScratch.csproj

@@ -2,6 +2,7 @@
 
   <PropertyGroup>
     <TargetFrameworks>netstandard2.0;net8.0</TargetFrameworks>
+    <LangVersion>latest</LangVersion>
     <GenerateDocumentationFile>true</GenerateDocumentationFile>
 
     <AssemblyName>MwParserFromScratch</AssemblyName>

MwParserFromScratch/Nodes/Inline.cs

@@ -889,7 +889,7 @@ public override TagStyle TagStyle
         {
             set
             {
-                if (value == TagStyle.SelfClosing && value == TagStyle.CompactSelfClosing)
+                if (value is TagStyle.SelfClosing or TagStyle.CompactSelfClosing)
                 {
                     if (Content != null && Content.Lines.Count > 0)
                         throw new InvalidOperationException("Cannot self-close a tag with non-empty content.");
@@ -904,8 +904,30 @@ public override TagStyle TagStyle
         /// <inheritdoc />
         internal override void ToPlainTextCore(StringBuilder builder, NodePlainTextFormatter formatter)
         {
-            if (Content != null)
-                formatter(Content, builder);
+            if (string.Equals(Name, "br", StringComparison.OrdinalIgnoreCase)
+                || string.Equals(Name, "hr", StringComparison.OrdinalIgnoreCase))
+            {
+                // for <br /> or <hr />, by default, we want to actually render a line break.
+                // Abnormal cases in MW:
+                // <br>abc</br> will be rendered as <br />abc<br />.
+                // <br></br> will be rendered as <br /><br />.
+                // In practice, however, <br>abc</br> won't be parsed correctly, as <br> itself as self-closing.
+                // See WikitextParserOptions.DefaultSelfClosingOnlyTags
+                builder.Append('\n');
+                if (Content != null)
+                {
+                    formatter(Content, builder);
+                    builder.Append('\n');
+                }
+                return;
+            }
+
+            // We are not wrapping other block-style elements, e.g., <div>, with \n,
+            // as these elements could be overridden into inline-style. Inline elements, e.g., <span>
+            // could be otherwise overridden into block-style.
+            // We leave such triaging responsibility to the library consumer (`formatter` arg).
+
+            if (Content != null) formatter(Content, builder);
         }
 
     }

MwParserFromScratch/Nodes/Node.cs

@@ -364,6 +364,8 @@ public Node Clone()
 
         private static void DefaultNodePlainTextFormatter(Node node, StringBuilder builder)
         {
+            Debug.Assert(node != null);
+            Debug.Assert(builder != null);
             node.ToPlainTextCore(builder, DefaultNodePlainTextFormatter);
         }
 

UnitTestProject1/RegressionTests.cs

@@ -0,0 +1,32 @@
+using UnitTestProject1.Primitive;
+using Xunit;
+using Xunit.Abstractions;
+
+namespace UnitTestProject1;
+
+public class RegressionTests : ParserTestBase
+{
+
+    /// <inheritdoc />
+    public RegressionTests(ITestOutputHelper output) : base(output)
+    {
+    }
+
+    /// <summary>
+    /// ToPlainText containing <br /> does not translate such tag in a \n new line
+    /// </summary>
+    [Fact]
+    public void Issue24()
+    {
+        var root = this.ParseWikitext(
+            "<span>Member of the [[Pennsylvania Senate]]<br/> from the [[Pennsylvania Senate, District 48|48th]] district</span>"
+        );
+        Assert.Equal("Member of the Pennsylvania Senate\n from the 48th district", root.ToPlainText());
+
+        // Abnormal case 1 -- actually the second </br> won't be treated as closing tag, as <br> itself is self-closing.
+        // See WikitextParserOptions.DefaultSelfClosingOnlyTags
+        root = this.ParseWikitext("test<br>foo</br>bar");
+        Assert.Equal("test\nfoo</br>bar", root.ToPlainText());
+    }
+
+}