Merge pull request #363 from baynezy/feature/issue-362-refactor

Create concrete implementations of replacers

src/Html2Markdown/Converter.cs

@@ -17,17 +17,17 @@ public partial class Converter
 	/// </summary>
 	public Converter() 
 	{
-			_replacers = new Markdown().Replacers();
-		}
+		_replacers = new Markdown().Replacers();
+	}
 
 	/// <summary>
 	/// Create a converter with a custom conversion scheme
 	/// </summary>
 	/// <param name="scheme">Conversion scheme to control conversion</param>
 	public Converter(IScheme scheme)
 	{
-			_replacers = scheme.Replacers();
-		}
+		_replacers = scheme.Replacers();
+	}
 
 	/// <summary>
 	/// Converts Html contained in a file to a Markdown string

src/Html2Markdown/Replacement/AnchorTagReplacer.cs

@@ -0,0 +1,11 @@
+namespace Html2Markdown.Replacement;
+/// <summary>
+/// Replaces an anchor tag with the link text and the link URL in Markdown format.
+/// </summary>
+public class AnchorTagReplacer : CustomReplacer
+{
+    public AnchorTagReplacer()
+    {
+        CustomAction = HtmlParser.ReplaceAnchor;
+    }
+}

src/Html2Markdown/Replacement/BlockquoteTagReplacer.cs

@@ -0,0 +1,11 @@
+namespace Html2Markdown.Replacement;
+/// <summary>
+/// Replaces an anchor tag with the link text and the link URL in Markdown format.
+/// </summary>
+public class BlockquoteTagReplacer : CustomReplacer
+{
+    public BlockquoteTagReplacer()
+    {
+        CustomAction = HtmlParser.ReplaceBlockquote;
+    }
+}

src/Html2Markdown/Replacement/BodyTagReplacer.cs

@@ -0,0 +1,12 @@
+namespace Html2Markdown.Replacement;
+/// <summary>
+/// Removes the body tag.
+/// </summary>
+public class BodyTagReplacer : PatternReplacer
+{
+    public BodyTagReplacer()
+    {
+        Pattern = "</?body[^>]*>";
+        Replacement = "";
+    }
+}

src/Html2Markdown/Replacement/BreakTagReplacer.cs

@@ -0,0 +1,12 @@
+namespace Html2Markdown.Replacement;
+/// <summary>
+/// Replaces the HTML break tag with its Markdown equivalent.
+/// </summary>
+public class BreakTagReplacer : PatternReplacer
+{
+    public BreakTagReplacer()
+    {
+        Pattern = "<br[^>]*>";
+        Replacement = "  " + Environment.NewLine;
+    }
+}

src/Html2Markdown/Replacement/CodeTagReplacer.cs

@@ -0,0 +1,16 @@
+namespace Html2Markdown.Replacement;
+/// <summary>
+/// Replaces the HTML code tag with its Markdown equivalent.
+/// </summary>
+public class CodeTagReplacer : CustomReplacer
+{
+    public CodeTagReplacer()
+    {
+        CustomAction = HtmlParser.ReplaceCode;
+    }
+
+    public CodeTagReplacer(bool supportSyntaxHighlighting)
+    {
+        CustomAction = html => HtmlParser.ReplaceCode(html, supportSyntaxHighlighting);
+    }
+}

src/Html2Markdown/Replacement/CommonMark/CommonMarkLayoutReplacementGroup.cs

@@ -7,41 +7,16 @@ namespace Html2Markdown.Replacement.CommonMark;
 public class CommonMarkLayoutReplacementGroup : IReplacementGroup
 {
 	private readonly IList<IReplacer> _replacements = new List<IReplacer> {
-		new PatternReplacer
-		{
-			Pattern = "<hr[^>]*>",
-			Replacement = Environment.NewLine + Environment.NewLine + "* * *" + Environment.NewLine
-		},
-		new CustomReplacer
-		{
-			CustomAction = ReplaceCode
-		},
-		new CustomReplacer
-		{
-			CustomAction = HtmlParser.ReplacePre
-		},
-		new CustomReplacer
-		{
-			CustomAction = HtmlParser.ReplaceParagraph
-		},
-		new PatternReplacer
-		{
-			Pattern = "<br[^>]*>",
-			Replacement = "  " + Environment.NewLine
-		},
-		new CustomReplacer
-		{
-			CustomAction = HtmlParser.ReplaceBlockquote
-		}
+		new HorizontalRuleTagReplacer(),
+		new CodeTagReplacer(true),
+		new PreTagReplacer(),
+		new ParagraphTagReplacer(),
+		new BreakTagReplacer(),
+		new BlockquoteTagReplacer()
 	};
 
-	private static string ReplaceCode(string html)
-	{
-		return HtmlParser.ReplaceCode(html, true);
-	}
-
 	public IEnumerable<IReplacer> Replacers()
 	{
-			return _replacements;
-		}
+		return _replacements;
+	}
 }

src/Html2Markdown/Replacement/CompositeReplacer.cs

@@ -0,0 +1,18 @@
+namespace Html2Markdown.Replacement;
+/// <summary>
+/// Allows for multiple replacements to be applied to the HTML.
+/// </summary>
+public abstract class CompositeReplacer : IReplacer
+{
+    private readonly IList<IReplacer> _replacements = new List<IReplacer>();
+
+    protected void AddReplacer(IReplacer replacer)
+    {
+        _replacements.Add(replacer);
+    }
+
+    public string Replace(string html)
+    {
+        return _replacements.Aggregate(html, (current, replacer) => replacer.Replace(current));
+    }
+}

src/Html2Markdown/Replacement/CustomReplacer.cs

@@ -1,11 +1,13 @@
 namespace Html2Markdown.Replacement;
-
-internal class CustomReplacer : IReplacer
+/// <summary>
+/// Allows custom replacement of HTML tags utilising external functions.
+/// </summary>
+public class CustomReplacer : IReplacer
 {
 	public string Replace(string html)
 	{
 		return CustomAction.Invoke(html);
 	}
 
-	public Func<string, string> CustomAction { get; init; }
+	protected Func<string, string> CustomAction { get; init; }
 }

src/Html2Markdown/Replacement/DocTypeReplacer.cs

@@ -0,0 +1,12 @@
+namespace Html2Markdown.Replacement;
+/// <summary>
+/// Removes the doctype tag.
+/// </summary>
+public class DocTypeReplacer : PatternReplacer
+{
+    public DocTypeReplacer()
+    {
+        Pattern = "<!DOCTYPE[^>]*>";
+        Replacement = "";
+    }
+}

src/Html2Markdown/Replacement/EmphasisTagReplacer.cs

@@ -0,0 +1,33 @@
+namespace Html2Markdown.Replacement;
+/// <summary>
+/// Replaces the HTML emphasis tag with its Markdown equivalent.
+/// </summary>
+public class EmphasisTagReplacer : CompositeReplacer
+{
+    public EmphasisTagReplacer()
+    {
+        AddReplacer(new PatternReplacer
+        {
+            Pattern = @"<(?:em|i)>(\s+)",
+            Replacement = " *"
+        });
+
+        AddReplacer(new PatternReplacer
+        {
+            Pattern = "<(?:em|i)>",
+            Replacement = "*"
+        });
+
+        AddReplacer(new PatternReplacer
+        {
+            Pattern = @"(\s+)</(em|i)>",
+            Replacement = "* "
+        });
+
+        AddReplacer(new PatternReplacer
+        {
+            Pattern = "</(em|i)>",
+            Replacement = "*"
+        });
+    }
+}

src/Html2Markdown/Replacement/EntitiesReplacementGroup.cs

@@ -6,14 +6,11 @@ namespace Html2Markdown.Replacement;
 public class EntitiesReplacementGroup : IReplacementGroup
 {
 	private readonly IList<IReplacer> _replacements = new List<IReplacer> {
-		new CustomReplacer
-		{
-			CustomAction = HtmlParser.ReplaceEntities
-		}
+		new HtmlEntitiesReplacer()
 	};
 
 	public IEnumerable<IReplacer> Replacers()
 	{
-			return _replacements;
-		}
+		return _replacements;
+	}
 }

src/Html2Markdown/Replacement/HeadTagReplacer.cs

@@ -0,0 +1,12 @@
+namespace Html2Markdown.Replacement;
+/// <summary>
+/// Removes the doctype tag.
+/// </summary>
+public class HeadTagReplacer : PatternReplacer
+{
+    public HeadTagReplacer()
+    {
+        Pattern = "</?head[^>]*>";
+        Replacement = "";
+    }
+}

src/Html2Markdown/Replacement/Heading.cs

@@ -0,0 +1,13 @@
+namespace Html2Markdown.Replacement;
+/// <summary>
+/// HTML Heading Tag Enum
+/// </summary>
+public enum Heading
+{
+    H1 = 1,
+    H2 = 2,
+    H3 = 3,
+    H4 = 4,
+    H5 = 5,
+    H6 = 6
+}

src/Html2Markdown/Replacement/HeadingReplacementGroup.cs

@@ -6,45 +6,16 @@ namespace Html2Markdown.Replacement;
 public class HeadingReplacementGroup : IReplacementGroup
 {
 	private readonly IList<IReplacer> _replacements = new List<IReplacer> {
-		new PatternReplacer
-		{
-			Pattern = "</h[1-6]>",
-			Replacement = Environment.NewLine + Environment.NewLine
-		},
-		new PatternReplacer
-		{
-			Pattern = "<h1[^>]*>",
-			Replacement = Environment.NewLine + Environment.NewLine + "# "
-		},
-		new PatternReplacer
-		{
-			Pattern = "<h2[^>]*>",
-			Replacement = Environment.NewLine + Environment.NewLine + "## "
-		},
-		new PatternReplacer
-		{
-			Pattern = "<h3[^>]*>",
-			Replacement = Environment.NewLine + Environment.NewLine + "### "
-		},
-		new PatternReplacer
-		{
-			Pattern = "<h4[^>]*>",
-			Replacement = Environment.NewLine + Environment.NewLine + "#### "
-		},
-		new PatternReplacer
-		{
-			Pattern = "<h5[^>]*>",
-			Replacement = Environment.NewLine + Environment.NewLine + "##### "
-		},
-		new PatternReplacer
-		{
-			Pattern = "<h6[^>]*>",
-			Replacement = Environment.NewLine + Environment.NewLine + "###### "
-		}
+		new HeadingTagReplacer(Heading.H1),
+		new HeadingTagReplacer(Heading.H2),
+		new HeadingTagReplacer(Heading.H3),
+		new HeadingTagReplacer(Heading.H4),
+		new HeadingTagReplacer(Heading.H5),
+		new HeadingTagReplacer(Heading.H6)
 	};
 
 	public IEnumerable<IReplacer> Replacers()
 	{
-			return _replacements;
-		}
+		return _replacements;
+	}
 }

src/Html2Markdown/Replacement/HeadingTagReplacer.cs

@@ -0,0 +1,22 @@
+namespace Html2Markdown.Replacement;
+/// <summary>
+/// Replaces the HTML heading tag with its Markdown equivalent.
+/// </summary>
+public class HeadingTagReplacer : CompositeReplacer
+{
+    public HeadingTagReplacer(Heading heading)
+    {
+        var headingNumber = (int) heading;
+        AddReplacer(new PatternReplacer
+        {
+            Pattern = $"</h{headingNumber}>",
+            Replacement = Environment.NewLine + Environment.NewLine
+        });
+
+        AddReplacer(new PatternReplacer
+        {
+            Pattern = $"<h{headingNumber}[^>]*>",
+            Replacement = Environment.NewLine + Environment.NewLine + new string('#', headingNumber) + " "
+        });
+    }
+}

src/Html2Markdown/Replacement/HorizontalRuleTagReplacer.cs

@@ -0,0 +1,12 @@
+namespace Html2Markdown.Replacement;
+/// <summary>
+/// Replaces the HTML horizontal rule tag with its Markdown equivalent.
+/// </summary>
+public class HorizontalRuleTagReplacer : PatternReplacer
+{
+    public HorizontalRuleTagReplacer()
+    {
+        Pattern = "<hr[^>]*>";
+        Replacement = Environment.NewLine + Environment.NewLine + "* * *" + Environment.NewLine;
+    }
+}

src/Html2Markdown/Replacement/HtmlCommentReplacer.cs

@@ -0,0 +1,12 @@
+namespace Html2Markdown.Replacement;
+/// <summary>
+/// Removes the HTML comment tag.
+/// </summary>
+public class HtmlCommentReplacer : PatternReplacer
+{
+    public HtmlCommentReplacer()
+    {
+        Pattern = "<!--[^-]+-->";
+        Replacement = "";
+    }
+}

src/Html2Markdown/Replacement/HtmlEntitiesReplacer.cs

@@ -0,0 +1,11 @@
+namespace Html2Markdown.Replacement;
+/// <summary>
+/// Replaces HTML entities with their Markdown equivalent.
+/// </summary>
+public class HtmlEntitiesReplacer : CustomReplacer
+{
+    public HtmlEntitiesReplacer()
+    {
+        CustomAction = HtmlParser.ReplaceEntities;
+    }
+}