markdown table reformat.cs

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

namespace MarkDownTableReformatter
{
	class Program
	{
		static void Main(string[] args)
		{
			var path = args[0];
			var result = FormatBlock(File.ReadLines(path).ToArray());
			var text = MakeString(result);
			File.WriteAllText(path + "2", text, Encoding.UTF8);
		}

		private static string MakeString(List<List<string>> lines)
		{
			StringBuilder sb = new StringBuilder();
			foreach (var line in lines)
			{
				foreach (var cell in line)
					sb.Append(cell);
				sb.AppendLine();
			}

			return sb.ToString();
		}

		private static List<List<string>> FormatBlock(string[] textLines)
		{
			var cellLines = textLines
				.Select((x,i) => new {line=x.Trim(), lineno=i})
				.Where(x=>!string.IsNullOrWhiteSpace(x.line))
				.Select(x =>  SplitLine(x.line, x.lineno))
				.ToList();

			var maxColumnCount = cellLines.Select(line => line.Count).Max();

			NormalizeAllLinesToSameColumnCount(cellLines, maxColumnCount);

			PadColumnsToSameWidth(maxColumnCount, cellLines);

			return cellLines;
		}

		private static void PadColumnsToSameWidth(int maxColumnCount, List<List<string>> cellLines)
		{
			for (int col = 0; col < maxColumnCount; col++)
			{
				var maxColumnWidth = cellLines.Select(x => x[col].Length).Max();

				foreach (var cellLine in cellLines)
					cellLine[col] = PadColumnToWidth(cellLine, col, maxColumnWidth);
			}
		}

		private static string PadColumnToWidth(List<string> cellLine, int col, int maxColumnWidth)
		{
			var cell = cellLine[col];
			if (cell.Length < maxColumnWidth)
				return cell.Substring(0, cell.Length - 1).PadRight(maxColumnWidth-1) + cell.Last();
			return cell;
		}

		private static void NormalizeAllLinesToSameColumnCount(List<List<string>> cellLines, int maxColumns)
		{
			foreach (var cellLine in cellLines)
			{
				if (cellLine.Count < maxColumns)
				{
					for (int i = 0; i < maxColumns - cellLine.Count; i++)
						cellLine.Add("|");
				}
			}
		}

		private static List<string> SplitLine(string line, int lineno)
		{
			if (!line.StartsWith("^") && !line.StartsWith("|"))
				throw new Exception(string.Format("Cannot parse line. Needs to start with characters ^ or | at line {0}. Content: '{1}'", lineno, line));

			char separator = line[0];

			List<string> elements = new List<string>();
			elements.Add(separator.ToString());

			int pos = 1;
			while (pos < line.Length)
			{
				var cell = GetCell(line, pos, separator, lineno);
				pos += cell.Length;
				elements.Add(" "+cell.Trim());
			}

			return elements;
		}

		private static string GetCell(string line, int pos, char separator, int lineno)
		{
			bool isInsideLinkScope = false;

			for (int i = pos; i < line.Length; i++)
			{
				if (isInsideLinkScope)
				{
					if (line[i] == ']' && line.Length > i + 1 && line[i + 1] == ']')
					{
						isInsideLinkScope = false;
						i++;
					}

					continue;
				}

				if (line[i] == '[' && line.Length > i + 1 && line[i + 1] == '[')
				{
					isInsideLinkScope = true;
					i++;
					continue;
				}

				if (line[i] == separator)
					return line.Substring(pos, i - pos + 1);
			}

			if (isInsideLinkScope)
				throw new Exception(string.Format("Did not find a closing link scope ']]' at line {0}", lineno));

			throw new Exception(string.Format("Did not find a separator '{0}' for cell at line {1}", separator, lineno));
		}
	}
}