Parser.jay

%{
using System;
using System.IO;
using System.Linq;
using System.Collections.Generic;

namespace MarkdownParser
{
	class Counter
	{
		public Counter (int count = 0)
		{
			Count = count;
		}
	
		public int Count {
			get;
			set;
		}
	}
	
	public class Parser
	{
		int yacc_verbose_flag = 0;

        public IEnumerable<IBlock> Parse (string input)
        {
			return Parse (new StringReader (input));
        }
        
        public IEnumerable<IBlock> Parse (TextReader reader)
        {
        	var lexer = new Tokenizer (reader);
			return (IEnumerable<IBlock>)this.yyparse (lexer);
        }
%}

%token ERROR
%token TEXT /* A string of character until the next newline/span */
%token EMPTY_LINE
%token NEWLINE

%token TRIPLE_QUOTE

%token HASH

%token HEADING_LINE
%token HEADING_LINE_DASH

%token EMPH_MARKER
%token STRONG_MARKER
%token BULLET_MARKER
%token INLINE_CODE_MARKER
%token URL_MARKER

%start document

%%

document: opt_block_list { $$ = $1; ((List<IBlock>)$$).Reverse (); }

opt_block_list
	: /* empty */ { $$ = new List<IBlock> (); }
	| block opt_block_list { ((List<IBlock>)$2).Add ((IBlock)$1); $$ = $2; }

block
	: block_inner EMPTY_LINE { $$ = $1; }

block_inner
	: paragraph { $$ = $1; }
	| heading { $$ = $1; }
	| code { $$ = $1; }
	| bullet_list { $$ = $1; }

paragraph: span_list NEWLINE { $$ = new Paragraph ((IList<ISpan>)$1); }

heading
	: atx_heading { $$ = $1; }
/*	| setex_heading { $$ = $1; }*/

code
	: TRIPLE_QUOTE TEXT NEWLINE opt_text_list TRIPLE_QUOTE NEWLINE { $$ = new Code (language: ((string)$2).Trim (), content: string.Join (Environment.NewLine, (IEnumerable<string>)$4)); }

opt_text_list: opt_text_list_inner { $$ = $1; ((List<string>)$1).Reverse (); }

opt_text_list_inner
	: /* empty */ { $$ = new List<string> (); }
	| text opt_text_list_inner { ((List<string>)$2).Add ((string)$1); $$ = $2; }

bullet_list
	: BULLET_MARKER 

span_list: span_list_inner { $$ = $1; ((List<ISpan>)$$).Reverse (); }

span_list_inner
	: span { $$ = new List<ISpan> { (ISpan)$1 }; }
	| span span_list_inner { ((List<ISpan>)$2).Add ((ISpan)$1); $$ = $2; }

span
	: TEXT { $$ = new FormatedSpan (SpanFormating.None, (string)$1); }
	| url { $$ = $1; }
	| emphasis { $$ = $1; }
	| strong { $$ = $1; }
	| inline_code { $$ = $1; }

url: URL_MARKER TEXT URL_MARKER URL_MARKER TEXT URL_MARKER { $$ = new UrlSpan (href: (string)$5, content: (string)$2); }

inline_code: INLINE_CODE_MARKER TEXT INLINE_CODE_MARKER { $$ = new FormatedSpan (SpanFormating.Code, (string)$1); }

emphasis: EMPH_MARKER TEXT EMPH_MARKER { $$ = new FormatedSpan (SpanFormating.Emphasis, (string)$2); }

strong: STRONG_MARKER TEXT STRONG_MARKER { $$ = new FormatedSpan (SpanFormating.Strong, (string)$2); }

text: TEXT NEWLINE { $$ = $1; }

atx_heading
	: hash_list text { $$ = new Heading (HeadingStyle.Atx, depth: ((Counter)$1).Count, content: ((string)$2).Trim ()); }

hash_list
	: HASH { $$ = new Counter (1); }
	| HASH hash_list { ((Counter)$2).Count++; $$ = $2; }

/*setex_heading: TEXT NEWLINE setex_heading_line NEWLINE { ((Heading)$3).Content = (string)$1; $$ = $3; }

setex_heading_line
	: HEADING_LINE { $$ = new Heading (HeadingStyle.Setex, depth: 0); }
	| HEADING_LINE_DASH { $$ = new Heading (HeadingStyle.Setex, depth: 1); }*/

%%

}

ParserTest.cs

using System;
using System.IO;
using System.Linq;
using System.Collections.Generic;

using NUnit.Framework;

using MarkdownParser;

namespace MarkdownParserTests
{
	[TestFixture]
	public class Test
	{
		[Test]
		public void TestMarkdownAstTest ()
		{
			var ast = new IBlock[] {
				new Heading (HeadingStyle.Atx, 1, "My Test Markdown file"),
				new Heading (HeadingStyle.Atx, 2, "Introduction"),
				new Paragraph (new ISpan[] { new FormatedSpan ("This is a test Markdown file") }),
				new Paragraph (new ISpan[] {
					new FormatedSpan ("It has some "),
					new FormatedSpan (SpanFormating.Emphasis, "emphasis"),
					new FormatedSpan (" and some "),
					new FormatedSpan (SpanFormating.Strong, "strong"),
					new FormatedSpan (" stuff to talk about"),
				}),
				new Code (language: "csharp", content: @"SomeCodeIsEvenDisplayed ()
{
	public void IntoSomething ()
	{
		Console.WriteLine (""That looks like code"");
	}
}"),
				new Paragraph (new ISpan[] {
					new FormatedSpan ("You can even embed cool url like "),
					new UrlSpan (href: "http://neteril.org", content: "my website"),
					new FormatedSpan (".")
				})
			};

			AssertAst (ast, "MarkdownParserTests.MarkdownSnippets.test_markdown.md");
		}

		void AssertAst (IEnumerable<IBlock> expectedAst, string markdownRes)
		{
			var reader = new StreamReader (GetType ().Assembly.GetManifestResourceStream (markdownRes));
			var parser = new Parser ();
			IEnumerable<IBlock> ast = null;
			try {
				ast = parser.Parse (reader);
			} catch (Exception e) {
				Assert.Fail ("Parser failed: {0}", e.ToString ());
			}
			CollectionAssert.AreEqual (expectedAst, ast);
		}
	}
}

test_markdown.md

My Test Markdown file

Introduction

This is a test Markdown file

It has some emphasis and some strong stuff to talk about

SomeCodeIsEvenDisplayed ()
{
	public void IntoSomething ()
	{
		Console.WriteLine ("That looks like code");
	}
}

You can even embed cool url like my website.