CSVManipulation

CSVManipulation

public class Manipulation
{
	private static void ValidateSource(string source)
	{
		// Validate path...
		bool sourceexists = File.Exists(source);

		if (!sourceexists)
		{
			Console.WriteLine("Unable to find " + source + "!");
			// For exit codes, see: https://msdn.microsoft.com/en-us/library/ms681382(v=vs.85)
			Environment.Exit(3);
		}
	}

	private static string[] GetHeaders(string source)
	{
		ValidateSource(source);
		string csvsource = @source;

		StreamReader sourcereader = new StreamReader(csvsource);
		var csvreadsource = new CsvReader(sourcereader);
		var headers = csvreadsource.FieldHeaders;
		
		sourcereader.Close();
		sourcereader.Dispose();
		
		return headers;
	}

	private static IEnumerable<CSVRowObj> GetRecords(string source)
	{
		ValidateSource(source);
		string csvsource = @source;

		// NOTE: See difference between System.IO.StreamReader and System.IO.TextReader here (either one is fine for this program):
		// https://stackoverflow.com/questions/6669885/difference-between-streamwriter-reader-and-stringwriter-readerll?rq=1
		StreamReader sourcereader = new StreamReader(csvsource);
		var csvreadsource = new CsvReader(sourcereader);
		var records = csvreadsource.GetRecords<CSVRowObj>();

		sourcereader.Close();
		sourcereader.Dispose();
		return records;
	}

	public static void WriteUnique(string oldsource, string newsource, string dest)
	{
		// Validate paths...
		ValidateSource(oldsource);
		ValidateSource(newsource);
		string destparentdir = Path.GetDirectoryName(dest);
		bool destparentdirexists = Directory.Exists(destparentdir);
		
		if (!destparentdirexists)
		{
			Console.WriteLine("Unable to find" + destparentdir + "!");
			Environment.Exit(3);
		}
		
		string oldcsvsource = @oldsource;
		string newcsvsource = @newsource;
		string outputdiff = @dest;

		// NOTE: See difference between System.IO.StreamReader and System.IO.TextReader here (either one is fine for this program):
		// https://stackoverflow.com/questions/6669885/difference-between-streamwriter-reader-and-stringwriter-readerll?rq=1
		StreamReader oldreader = new StreamReader(oldcsvsource);
		StreamReader newreader = new StreamReader(newcsvsource);
		StreamWriter outputwriter = new StreamWriter(outputdiff);
		
		var oldcsvreadsource = new CsvReader(oldreader);
		var newcsvreadsource = new CsvReader(newreader);
		var csvwritedest = new CsvWriter(outputwriter);
		
		// WARNING: The below .ToList() puts everything in memory!
		var oldrecords = oldcsvreadsource.GetRecords<CSVRowObj>().ToList();
		var newrecords = newcsvreadsource.GetRecords<CSVRowObj>().ToList();
		var allrecords = oldrecords.Union(newrecords);
		var distinctrecords = allrecords.Distinct(new CSVRowObj.Comparer());
		csvwritedest.WriteRecords(distinctrecords);
		
		oldreader.Close();
		oldreader.Dispose();
		newreader.Close();
		newreader.Dispose();
		outputwriter.Close();
		outputwriter.Dispose();
	}

	public static void JoinOnTypeandReference(string sourcea, string sourceb, string dest)
	{	
		// Validate Output Destination
		string outputdest = @dest;
		string destparentdir = Path.GetDirectoryName(outputdest);
		bool destparentdirexists = Directory.Exists(destparentdir);
		if (!destparentdirexists)
		{
			Console.WriteLine("Unable to find" + destparentdir + "!");
			Environment.Exit(3);
		}

		// Get Records from both csv files
		// NOTE: The IEnumerables are NOT entirely loaded in memory until you actually do something with them
		var leftrecords = GetRecords(sourcea);
		var rightrecords = GetRecords(sourceb);
		
		// Prepare to write to destination
		StreamWriter outputwriter = new StreamWriter(outputdest);
		var csvwritedest = new CsvWriter(outputwriter);

		// Join right records to left records if Type and Reference values are the same
		var matches = from left in leftrecords
					  join right in rightrecords
					  on
					  new { left.Type, left.Reference} equals
					  new { right.Type, right.Reference }
					  select new CSVRowObj
					  {
						  Date = left.Date,
						  Amount = left.Amount,
						  Description = left.Description,
						  Description2 = left.Description2 + right.Description2,
						  Reference = left.Reference,
						  Type = left.Type,
					  };
					  
		csvwritedest.WriteRecords(matches);
	}
}

class CSVRowObj
{
	public string Date { get; set; }
	public string Amount { get; set; }
	public string Description { get; set; }
	public string Description2 { get; set; }
	public string Reference { get; set; }
	public string Type { get; set; }

	// From: https://stackoverflow.com/questions/16197290/checking-for-duplicates-in-a-list-of-objects-c-sharp
	public class Comparer : IEqualityComparer<CSVRowObj>
	{
		public bool Equals(CSVRowObj x, CSVRowObj y)
		{
			if (x == null || y == null)
				return false;

			return x.Date == y.Date && x.Amount == y.Amount && x.Description == y.Description &&
				x.Description2 == y.Description2 && x.Reference == y.Reference && x.Type == y.Type;
		}

		public int GetHashCode(CSVRowObj obj)
		{
			if (obj == null)
				return 0;

			// NOTE: The below return statement returns true if and only if ONE of the operands returns true
			// This behavior is due to the carrot '^' between the operands
			return (obj.Date == null ? 0 : obj.Date.GetHashCode()) ^
				(obj.Amount == null ? 0 : obj.Amount.GetHashCode()) ^
				(obj.Description == null ? 0 : obj.Description.GetHashCode()) ^
				(obj.Description2 == null ? 0 : obj.Description2.GetHashCode()) ^
				(obj.Reference == null ? 0 : obj.Reference.GetHashCode()) ^
				(obj.Type == null ? 0 : obj.Type.GetHashCode());
		}
	}
}

public void Main()
{
	string csvorig = @"C:\Users\pddomain\Downloads\csv\April2016.csv";
	string csvnew = @"C:\Users\pddomain\Downloads\csv\April2016_New.csv";
	string outputfile = @"C:\Users\pddomain\Output.csv";
	//Manipulation.WriteUnique(csvorig, csvnew, outputfile);
	Manipulation.JoinOnTypeandReference(csvorig, csvnew, outputfile);
}