Refactoring and handling to cleanup unnecessary spaces using CleanupU…

…nnecessarySpaces
mysticmind · Jun 19, 2024 · dfa306a · dfa306a
1 parent 2e3d52f
commit dfa306a
Show file tree

Hide file tree

Showing 8 changed files with 38 additions and 32 deletions.
diff --git a/README.md b/README.md
@@ -19,15 +19,15 @@ If you have used and benefitted from this library. Please feel free to buy me a
 Install the package from NuGet using `Install-Package ReverseMarkdown` or clone the repository and built it yourself.
 
 <!-- snippet: Usage -->
-<a id='snippet-usage'></a>
+<a id='snippet-Usage'></a>
 ```cs
 var converter = new ReverseMarkdown.Converter();
 
 string html = "This a sample <strong>paragraph</strong> from <a href=\"http://test.com\">my site</a>";
 
 string result = converter.Convert(html);
 ```
-<sup><a href='/src/ReverseMarkdown.Test/Snippets.cs#L11-L19' title='Snippet source file'>snippet source</a> | <a href='#snippet-usage' title='Start of snippet'>anchor</a></sup>
+<sup><a href='/src/ReverseMarkdown.Test/Snippets.cs#L11-L19' title='Snippet source file'>snippet source</a> | <a href='#snippet-Usage' title='Start of snippet'>anchor</a></sup>
 <!-- endSnippet -->
 
 Will result in:
@@ -43,7 +43,7 @@ This a sample **paragraph** from [my site](http://test.com)
 The conversion can be customized:
 
 <!-- snippet: UsageWithConfig -->
-<a id='snippet-usagewithconfig'></a>
+<a id='snippet-UsageWithConfig'></a>
 ```cs
 var config = new ReverseMarkdown.Config
 {
@@ -59,7 +59,7 @@ var config = new ReverseMarkdown.Config
 
 var converter = new ReverseMarkdown.Converter(config);
 ```
-<sup><a href='/src/ReverseMarkdown.Test/Snippets.cs#L27-L43' title='Snippet source file'>snippet source</a> | <a href='#snippet-usagewithconfig' title='Start of snippet'>anchor</a></sup>
+<sup><a href='/src/ReverseMarkdown.Test/Snippets.cs#L27-L43' title='Snippet source file'>snippet source</a> | <a href='#snippet-UsageWithConfig' title='Start of snippet'>anchor</a></sup>
 <!-- endSnippet -->
 
 ## Configuration options

diff --git a/README.source.md b/README.source.md
@@ -12,15 +12,15 @@ If you have used and benefitted from this library. Please feel free to buy me a
 Install the package from NuGet using `Install-Package ReverseMarkdown` or clone the repository and build it yourself.
 
 <!-- snippet: Usage -->
-<a id='snippet-usage'></a>
+<a id='snippet-Usage'></a>
 ```cs
 var converter = new ReverseMarkdown.Converter();
 
 string html = "This a sample <strong>paragraph</strong> from <a href=\"http://test.com\">my site</a>";
 
 string result = converter.Convert(html);
 ```
-<sup><a href='/src/ReverseMarkdown.Test/Snippets.cs#L11-L19' title='Snippet source file'>snippet source</a> | <a href='#snippet-usage' title='Start of snippet'>anchor</a></sup>
+<sup><a href='/src/ReverseMarkdown.Test/Snippets.cs#L11-L19' title='Snippet source file'>snippet source</a> | <a href='#snippet-Usage' title='Start of snippet'>anchor</a></sup>
 <!-- endSnippet -->
 
 Will result in:
@@ -36,7 +36,7 @@ This a sample **paragraph** from [my site](http://test.com)
 The conversion can also be customized:
 
 <!-- snippet: UsageWithConfig -->
-<a id='snippet-usagewithconfig'></a>
+<a id='snippet-UsageWithConfig'></a>
 ```cs
 var config = new ReverseMarkdown.Config
 {
@@ -52,7 +52,7 @@ var config = new ReverseMarkdown.Config
 
 var converter = new ReverseMarkdown.Converter(config);
 ```
-<sup><a href='/src/ReverseMarkdown.Test/Snippets.cs#L27-L43' title='Snippet source file'>snippet source</a> | <a href='#snippet-usagewithconfig' title='Start of snippet'>anchor</a></sup>
+<sup><a href='/src/ReverseMarkdown.Test/Snippets.cs#L27-L43' title='Snippet source file'>snippet source</a> | <a href='#snippet-UsageWithConfig' title='Start of snippet'>anchor</a></sup>
 <!-- endSnippet -->
 
 ## Configuration options

diff --git a/src/ReverseMarkdown.Test/ChildConverterTests.cs b/src/ReverseMarkdown.Test/ChildConverterTests.cs
@@ -20,7 +20,7 @@ public void WhenConverter_A_IsReplacedByConverter_IgnoreAWhenHasClass()
             var converter = new ReverseMarkdown.Converter(new Config(), typeof(IgnoreAWhenHasClass).Assembly);
 
             var type = converter.GetType();
-            var prop = type.GetField("_converters", BindingFlags.NonPublic | BindingFlags.Instance);
+            var prop = type.GetField("Converters", BindingFlags.NonPublic | BindingFlags.Instance);
 
             Assert.NotNull(prop);
 

diff --git a/src/ReverseMarkdown/Config.cs b/src/ReverseMarkdown/Config.cs
@@ -83,6 +83,8 @@ public enum TableWithoutHeaderRowHandlingOption
         /// </summary>
         public bool TableHeaderColumnSpanHandling { get; set; } = true;
 
+        public bool CleanupUnnecessarySpaces { get; set; } = true;
+
 
         /// <summary>
         /// Determines whether url is allowed: WhitelistUriSchemes contains no elements or contains passed url.

diff --git a/src/ReverseMarkdown/Converter.cs b/src/ReverseMarkdown/Converter.cs
@@ -10,10 +10,10 @@ namespace ReverseMarkdown
 {
     public class Converter
     {
-        protected readonly IDictionary<string, IConverter> _converters = new Dictionary<string, IConverter>();
-        protected readonly IConverter _passThroughTagsConverter;
-        protected readonly IConverter _dropTagsConverter;
-        protected readonly IConverter _byPassTagsConverter;
+        protected readonly IDictionary<string, IConverter> Converters = new Dictionary<string, IConverter>();
+        protected readonly IConverter PassThroughTagsConverter;
+        protected readonly IConverter DropTagsConverter;
+        protected readonly IConverter ByPassTagsConverter;
 
         public Converter() : this(new Config()) {}
 
@@ -23,19 +23,19 @@ public Converter(Config config, params Assembly[] additionalAssemblies)
         {
             Config = config;
 
-            List<Assembly> assemblies = new List<Assembly>()
+            var assemblies = new List<Assembly>()
             {
                 typeof(IConverter).GetTypeInfo().Assembly
             };
 
             if (!(additionalAssemblies is null))
                 assemblies.AddRange(additionalAssemblies);
 
-            List<Type> types = new List<Type>();
+            var types = new List<Type>();
             // instantiate all converters excluding the unknown tags converters
             foreach (var assembly in assemblies)
             {
-                foreach (var ctype in assembly.GetTypes()
+                foreach (var converterType in assembly.GetTypes()
                     .Where(t => t.GetTypeInfo().GetInterfaces().Contains(typeof(IConverter)) &&
                     !t.GetTypeInfo().IsAbstract
                     && t != typeof(PassThrough)
@@ -44,32 +44,32 @@ public Converter(Config config, params Assembly[] additionalAssemblies)
                 {
                     // Check to see if any existing types are children/equal to
                     // the type to add.
-                    if (types.Any(e => ctype.IsAssignableFrom(e)))
+                    if (types.Any(e => converterType.IsAssignableFrom(e)))
                         // If they are, ignore the type.
                         continue;
 
                     // See if there is a type that is a parent of the
                     // current type.
-                    Type toRemove = types.FirstOrDefault(e => e.IsAssignableFrom(ctype));
+                    var toRemove = types.FirstOrDefault(e => e.IsAssignableFrom(converterType));
                     // if there is ...
                     if (!(toRemove is null))
                         // ... remove the parent.
                         types.Remove(toRemove);
 
                     // finally, add the type.
-                    types.Add(ctype);
+                    types.Add(converterType);
                 }
             }
 
             // For each type to register ...
-            foreach (var ctype in types)
+            foreach (var converterType in types)
                 // ... activate them
-                Activator.CreateInstance(ctype, this);
+                Activator.CreateInstance(converterType, this);
 
             // register the unknown tags converters
-            _passThroughTagsConverter = new PassThrough(this);
-            _dropTagsConverter = new Drop(this);
-            _byPassTagsConverter = new ByPass(this);
+            PassThroughTagsConverter = new PassThrough(this);
+            DropTagsConverter = new Drop(this);
+            ByPassTagsConverter = new ByPass(this);
         }
 
         public Config Config { get; protected set; }
@@ -94,35 +94,35 @@ public virtual string Convert(string html)
             // cleanup multiple new lines
             result = Regex.Replace( result, @"(^\p{Zs}*(\r\n|\n)){2,}", Environment.NewLine, RegexOptions.Multiline);
 
-            return result.Trim().FixMultipleNewlines();
+            return Config.CleanupUnnecessarySpaces ? result.Trim().FixMultipleNewlines() : result;
         }
 
         public virtual void Register(string tagName, IConverter converter)
         {
-            _converters[tagName] = converter;
+            Converters[tagName] = converter;
         }
 
         public virtual IConverter Lookup(string tagName)
         {
             // if a tag is in the pass through list then use the pass through tags converter
             if (Config.PassThroughTags.Contains(tagName))
             {
-                return _passThroughTagsConverter;
+                return PassThroughTagsConverter;
             }
 
-            return _converters.ContainsKey(tagName) ? _converters[tagName] : GetDefaultConverter(tagName);
+            return Converters.TryGetValue(tagName, out var converter) ? converter : GetDefaultConverter(tagName);
         }
 
         private IConverter GetDefaultConverter(string tagName)
         {
             switch (Config.UnknownTags)
             {
                 case Config.UnknownTagsOption.PassThrough:
-                    return _passThroughTagsConverter;
+                    return PassThroughTagsConverter;
                 case Config.UnknownTagsOption.Drop:
-                    return _dropTagsConverter;
+                    return DropTagsConverter;
                 case Config.UnknownTagsOption.Bypass:
-                    return _byPassTagsConverter;
+                    return ByPassTagsConverter;
                 default:
                     throw new UnknownTagException(tagName);
             }

diff --git a/src/ReverseMarkdown/Converters/ConverterBase.cs b/src/ReverseMarkdown/Converters/ConverterBase.cs
@@ -23,7 +23,7 @@ protected string TreatChildren(HtmlNode node)
         }
 
         private string Treat(HtmlNode node) {
-            //TrimNewLine(node);
+            // TrimNewLine(node);
             var converter = Converter.Lookup(node.Name);
             return converter.Convert(node);
         }

diff --git a/src/ReverseMarkdown/Converters/Div.cs b/src/ReverseMarkdown/Converters/Div.cs
@@ -36,6 +36,8 @@ public override string Convert(HtmlNode node)
                 "table"
             };
 
+            content = Converter.Config.CleanupUnnecessarySpaces ? content.Trim() : content;
+
             // if there is a block child then ignore adding the newlines for div
             if ((node.ChildNodes.Count == 1 && blockTags.Contains(node.FirstChild.Name)))
             {

diff --git a/src/ReverseMarkdown/Converters/P.cs b/src/ReverseMarkdown/Converters/P.cs
@@ -16,6 +16,8 @@ public override string Convert(HtmlNode node)
             var indentation = IndentationFor(node);
             var newlineAfter = NewlineAfter(node);
 
+            var content = Converter.Config.CleanupUnnecessarySpaces ? TreatChildren(node).Trim() : TreatChildren(node); 
+
             return $"{indentation}{TreatChildren(node)}{newlineAfter}";
         }