From 8872e22c3aff7da7765c83ae3e4b252eb51e532e Mon Sep 17 00:00:00 2001
From: Immortal <marek.stoj@gmail.com>
Date: Wed, 29 May 2013 13:34:18 +0200
Subject: [PATCH] Updated SgmlDomBuilder so that it respects significant
 whitespaces.

---
 .gitignore                                         |  1 +
 .../NReadability.Tests/SgmlDomBuilderTests.cs      | 14 +++++++++++++-
 Src/NReadability/NReadability/SgmlDomBuilder.cs    |  2 +-
 3 files changed, 15 insertions(+), 2 deletions(-)

diff --git a/.gitignore b/.gitignore
index a4ce609..4b25b62 100644
--- a/.gitignore
+++ b/.gitignore
@@ -36,3 +36,4 @@
 /Src/NReadability/NReadability.6.1.ReSharper.user
 /Src/NReadability/NReadability.sln.DotSettings.user
 *.suo
+Src/NReadability/NReadability.sln.DotSettings
diff --git a/Src/NReadability/NReadability.Tests/SgmlDomBuilderTests.cs b/Src/NReadability/NReadability.Tests/SgmlDomBuilderTests.cs
index d78e5fa..0543f0f 100644
--- a/Src/NReadability/NReadability.Tests/SgmlDomBuilderTests.cs
+++ b/Src/NReadability/NReadability.Tests/SgmlDomBuilderTests.cs
@@ -1,4 +1,5 @@
-using System.Xml.Linq;
+using System;
+using System.Xml.Linq;
 using NUnit.Framework;
 
 namespace NReadability.Tests
@@ -54,6 +55,17 @@ public void Builder_handles_invalid_entity_references()
       Assert.DoesNotThrow(() => _sgmlDomSerializer.SerializeDocument(document));
     }
 
+    [Test]
+    public void Builder_respects_significant_whitespaces()
+    {
+      XDocument document =
+        _sgmlDomBuilder.BuildDocument("<a>Link1</a> <a>Link2</a>");
+
+      string html = _sgmlDomSerializer.SerializeDocument(document);
+
+      Assert.IsTrue(html.Contains("<a>Link1</a> <a>Link2</a>"));
+    }
+
     #endregion
   }
 }
diff --git a/Src/NReadability/NReadability/SgmlDomBuilder.cs b/Src/NReadability/NReadability/SgmlDomBuilder.cs
index 44d918d..68cecc4 100644
--- a/Src/NReadability/NReadability/SgmlDomBuilder.cs
+++ b/Src/NReadability/NReadability/SgmlDomBuilder.cs
@@ -96,7 +96,7 @@ private static XDocument LoadDocument(string htmlContent)
       {
         sgmlReader.CaseFolding = CaseFolding.ToLower;
         sgmlReader.DocType = "HTML";
-        sgmlReader.WhitespaceHandling = WhitespaceHandling.None;
+        sgmlReader.WhitespaceHandling = WhitespaceHandling.Significant;
 
         using (var sr = new StreamReader(new MemoryStream(Encoding.UTF8.GetBytes(htmlContent))))
         {