♻️ refactoring and adding a few more tests

Author: Rohland

src/HtmlDiff.Tests/UtilsTests.cs

@@ -0,0 +1,35 @@
+using NExpect;
+using NUnit.Framework;
+using static NExpect.Expectations;
+
+namespace HtmlDiff.Tests
+{
+    [TestFixture]
+    public class UtilsTests
+    {
+        [TestFixture]
+        public class GetTagName
+        {
+            [TestCase("", "")]
+            [TestCase(null, "")]
+            [TestCase("test", "")]
+            [TestCase("<test", "")]
+            [TestCase("< div >", "")]
+            [TestCase("</ div>", "")]
+            [TestCase("<div>", "div")]
+            [TestCase(" \t<div> \t", "div")]
+            [TestCase("<DIV>", "div")]
+            [TestCase("</div>", "div")]
+            [TestCase("<div attr='test'>", "div")]
+            [TestCase("<div attr=test>", "div")]
+            public void WithTestCase_ShouldMapToTagName(string input, string expected)
+            {
+                // arrange and act
+                var result = Utils.GetTagName(input);
+                
+                // assert
+                Expect(result).To.Equal(expected);
+            }
+        }
+    }
+}

src/HtmlDiff/HtmlDiff.cs

@@ -12,6 +12,8 @@ public class HtmlDiff
         /// This value defines balance between speed and memory utilization. The higher it is the faster it works and more memory consumes.
         /// </summary>
         private const int MatchGranularityMaximum = 4;
+        private const string DelTag = "del";
+        private const string InsTag = "ins";
 
         private readonly StringBuilder _content;
         private string _newText;
@@ -188,13 +190,13 @@ private void ProcessReplaceOperation(Operation operation)
         private void ProcessInsertOperation(Operation operation, string cssClass)
         {
             var text = _newWords.Where((s, pos) => pos >= operation.StartInNew && pos < operation.EndInNew).ToList();
-            InsertTag("ins", cssClass, text);
+            InsertTag(InsTag, cssClass, text);
         }
 
         private void ProcessDeleteOperation(Operation operation, string cssClass)
         {
             var text = _oldWords.Where((s, pos) => pos >= operation.StartInOld && pos < operation.EndInOld).ToList();
-            InsertTag("del", cssClass, text);
+            InsertTag(DelTag, cssClass, text);
         }
 
         private void ProcessEqualOperation(Operation operation)
@@ -247,12 +249,11 @@ private void InsertTag(string tag, string cssClass, List<string> words)
                 }
                 else
                 {
-                    // Check if the tag is a special case
                     if (SpecialCaseOpeningTagRegex.IsMatch(words[0]))
                     {
                         _specialTagDiffStack.Push(words[0]);
                         specialCaseTagInjection = "<ins class='mod'>";
-                        if (tag == "del")
+                        if (tag == DelTag)
                         {
                             words.RemoveAt(0);
 
@@ -263,23 +264,18 @@ private void InsertTag(string tag, string cssClass, List<string> words)
                             }
                         }
                     }
-
                     else if (_specialCaseClosingTags.ContainsKey(words[0]))
                     {
                         var openingTag = _specialTagDiffStack.Count == 0 ? null : _specialTagDiffStack.Pop();
-
-                        // If we didn't have an opening tag, and we don't have a match with the previous tag used 
-                        if (openingTag == null || Utils.GetTagName(openingTag) != Utils.GetTagName(words.Last()))
-                        {
-                            // do nothing
-                        }
-                        else
+                        var hasOpeningTag = !(openingTag is null);
+                        var openingAndClosingTagsMatch = Utils.GetTagName(openingTag) == Utils.GetTagName(words.Last());
+                        if (hasOpeningTag && openingAndClosingTagsMatch)
                         {
                             specialCaseTagInjection = "</ins>";
                             specialCaseTagInjectionIsBefore = true;
                         }
 
-                        if (tag == "del")
+                        if (tag == DelTag)
                         {
                             words.RemoveAt(0);
 

src/HtmlDiff/Utils.cs

@@ -11,7 +11,7 @@ public static class Utils
         private static Regex tagWordRegex = new Regex(@"<[^\s>]+", RegexOptions.Compiled);
         private static Regex whitespaceRegex = new Regex("^(\\s|&nbsp;)+$", RegexOptions.Compiled);
         private static Regex wordRegex = new Regex(@"[\w\#@]+", RegexOptions.Compiled | RegexOptions.ECMAScript);
-        private static Regex tagRegex = new Regex(@"\s*</?(?<name>[^\s>]+)[^>]*>\s*", RegexOptions.Compiled);
+        private static Regex tagRegex = new Regex(@"</?(?<name>[^\s/>]+)[^>]*>", RegexOptions.Compiled);
 
         private static readonly string[] SpecialCaseWordTags = { "<img" };
 
@@ -79,6 +79,7 @@ public static string StripAnyAttributes(string word)
             {
                 return StripTagAttributes(word);
             }
+
             return word;
         }
 
@@ -89,13 +90,15 @@ public static bool IsWord(char text)
 
         public static string GetTagName(string word)
         {
-            var match = tagRegex.Match(word);
-
-            // returns empty string if it's not a tag
-            if (!match.Success)
-                return string.Empty;
-
-            return match.Groups["name"].Value.ToLower();
+            var noResult = string.Empty;
+            if (word is null)
+            {
+                return noResult;
+            }
+            var tagMatch = tagRegex.Match(word);
+            return tagMatch.Success
+                ? tagMatch.Groups["name"].Value.ToLower()
+                : noResult;
         }
     }
 }