Parsing CSV files in C#, with header

Question

Is there a default/official/recommended way to parse CSV files in C#? I don't want to roll my own parser.

Also, I've seen instances of people using ODBC/OLE DB to read CSV via the Text driver, and a lot of people discourage this due to its "drawbacks." What are these drawbacks?

Ideally, I'm looking for a way through which I can read the CSV by column name, using the first record as the header / field names. Some of the answers given are correct but work to basically deserialize the file into classes.

Answer 1

Single source file solution for straightforward parsing needs, useful. Deals with all the nasty edge cases. Such as new line normalization and handling new lines in quoted string literals. Your welcome!

If you CSV file has a header you just read out the column names (and compute column indexes) from the first row. Simple as that.

Note that Dump is a LINQPad method, you might want to remove that if you are not using LINQPad.

void Main()
{
    var file1 = "a,b,c\r\nx,y,z";
    CSV.ParseText(file1).Dump();

    var file2 = "a,\"b\",c\r\nx,\"y,z\"";
    CSV.ParseText(file2).Dump();

    var file3 = "a,\"b\",c\r\nx,\"y\r\nz\"";
    CSV.ParseText(file3).Dump();

    var file4 = "\"\"\"\"";
    CSV.ParseText(file4).Dump();
}

static class CSV
{
    public struct Record
    {
        public readonly string[] Row;

        public string this[int index] => Row[index];

        public Record(string[] row)
        {
            Row = row;
        }
    }

    public static List<Record> ParseText(string text)
    {
        return Parse(new StringReader(text));
    }

    public static List<Record> ParseFile(string fn)
    {
        using (var reader = File.OpenText(fn))
        {
            return Parse(reader);
        }
    }

    public static List<Record> Parse(TextReader reader)
    {
        var data = new List<Record>();

        var col = new StringBuilder();
        var row = new List<string>();
        for (; ; )
        {
            var ln = reader.ReadLine();
            if (ln == null) break;
            if (Tokenize(ln, col, row))
            {
                data.Add(new Record(row.ToArray()));
                row.Clear();
            }
        }

        return data;
    }

    public static bool Tokenize(string s, StringBuilder col, List<string> row)
    {
        int i = 0;

        if (col.Length > 0)
        {
            col.AppendLine(); // continuation

            if (!TokenizeQuote(s, ref i, col, row))
            {
                return false;
            }
        }

        while (i < s.Length)
        {
            var ch = s[i];
            if (ch == ',')
            {
                row.Add(col.ToString().Trim());
                col.Length = 0;
                i++;
            }
            else if (ch == '"')
            {
                i++;
                if (!TokenizeQuote(s, ref i, col, row))
                {
                    return false;
                }
            }
            else
            {
                col.Append(ch);
                i++;
            }
        }

        if (col.Length > 0)
        {
            row.Add(col.ToString().Trim());
            col.Length = 0;
        }

        return true;
    }

    public static bool TokenizeQuote(string s, ref int i, StringBuilder col, List<string> row)
    {
        while (i < s.Length)
        {
            var ch = s[i];
            if (ch == '"')
            {
                // escape sequence
                if (i + 1 < s.Length && s[i + 1] == '"')
                {
                    col.Append('"');
                    i++;
                    i++;
                    continue;
                }
                i++;
                return true;
            }
            else
            {
                col.Append(ch);
                i++;
            }
        }
        return false;
    }
}

Answer 2

A CSV parser is now a part of .NET Framework.

Add a reference to Microsoft.VisualBasic.dll (works fine in C#, don't mind the name)

using (TextFieldParser parser = new TextFieldParser(@"c:\temp\test.csv"))
{
    parser.TextFieldType = FieldType.Delimited;
    parser.SetDelimiters(",");
    while (!parser.EndOfData)
    {
        //Process row
        string[] fields = parser.ReadFields();
        foreach (string field in fields)
        {
            //TODO: Process field
        }
    }
}

The docs are here - TextFieldParser Class

Answer 3

Some time ago I had wrote simple class for CSV read/write based on Microsoft.VisualBasic library. Using this simple class you will be able to work with CSV like with 2 dimensions array. You can find my class by the following link: https://github.com/ukushu/DataExporter

Simple example of usage:

Csv csv = new Csv("\t");//delimiter symbol

csv.FileOpen("c:\\file1.csv");

var row1Cell6Value = csv.Rows[0][5];

csv.AddRow("asdf","asdffffff","5")

csv.FileSave("c:\\file2.csv");

For reading header only you need is to read csv.Rows[0] cells :)

Answer 4

Here is a helper class I use often, in case any one ever comes back to this thread (I wanted to share it).

I use this for the simplicity of porting it into projects ready to use:

public class CSVHelper : List<string[]>
{
  protected string csv = string.Empty;
  protected string separator = ",";

  public CSVHelper(string csv, string separator = "\",\"")
  {
    this.csv = csv;
    this.separator = separator;

    foreach (string line in Regex.Split(csv, System.Environment.NewLine).ToList().Where(s => !string.IsNullOrEmpty(s)))
    {
      string[] values = Regex.Split(line, separator);

      for (int i = 0; i < values.Length; i++)
      {
        //Trim values
        values[i] = values[i].Trim('\"');
      }

      this.Add(values);
    }
  }
}

And use it like:

public List<Person> GetPeople(string csvContent)
{
  List<Person> people = new List<Person>();
  CSVHelper csv = new CSVHelper(csvContent);
  foreach(string[] line in csv)
  {
    Person person = new Person();
    person.Name = line[0];
    person.TelephoneNo = line[1];
    people.Add(person);
  }
  return people;
}

[Updated csv helper: bug fixed where the last new line character created a new line]