C# - Light and Fast CSV Parser

Introduction

Parsing CSV files may sound like an easy task, but in reality it is not that trivial. Below is a CsvParser class implementation that I use in my own projects. It supports the following features that I find critical:

• Custom Delimiter and Qualifier characters
• Supports quoting notation (allows delimiter character to be part of a value)
• Supports quote escaping (allows quote character to be part of a value)
• Supports both '\n' and '\r\n' line endings
• Designed to return IEnumerable via yield return (no memory buffers)
• Designed to return Header and the rest of lines separately (using Tuple)

Source Code

public static class CsvParser
{
    private static Tuple<T, IEnumerable<T>> HeadAndTail<T>(this IEnumerable<T> source)
    {
        if (source == null)
            throw new ArgumentNullException("source");
        var en = source.GetEnumerator();
        en.MoveNext();
        return Tuple.Create(en.Current, EnumerateTail(en));
    }

    private static IEnumerable<T> EnumerateTail<T>(IEnumerator<T> en)
    {
        while (en.MoveNext()) yield return en.Current;
    }

    public static IEnumerable<IList<string>> 
           Parse(string content, char delimiter, char qualifier)
    {
        using (var reader = new StringReader(content))
            return Parse(reader, delimiter, qualifier);
    }

    public static Tuple<IList<string>, IEnumerable<IList<string>>> 
           ParseHeadAndTail(TextReader reader, char delimiter, char qualifier)
    {
        return HeadAndTail(Parse(reader, delimiter, qualifier));
    }

    public static IEnumerable<IList<string>> 
           Parse(TextReader reader, char delimiter, char qualifier)
    {
        var inQuote = false;
        var record = new List<string>();
        var sb = new StringBuilder();

        while (reader.Peek() != -1)
        {
            var readChar = (char) reader.Read();

            if (readChar == '\n' || (readChar == '\r' && (char) reader.Peek() == '\n'))
            {
                // If it's a \r\n combo consume the \n part and throw it away.
                if (readChar == '\r')
                    reader.Read();

                if (inQuote)
                {
                    if (readChar == '\r')
                        sb.Append('\r');
                    sb.Append('\n');
                }
                else
                {
                    if (record.Count > 0 || sb.Length > 0)
                    {
                        record.Add(sb.ToString());
                        sb.Clear();
                    }

                    if (record.Count > 0)
                        yield return record;

                    record = new List<string>(record.Count);
                }
            }
            else if (sb.Length == 0 && !inQuote)
            {
                if (readChar == qualifier)
                    inQuote = true;
                else if (readChar == delimiter)
                {
                    record.Add(sb.ToString());
                    sb.Clear();
                }
                else if (char.IsWhiteSpace(readChar))
                {
                    // Ignore leading whitespace
                }
                else
                    sb.Append(readChar);
            }
            else if (readChar == delimiter)
            {
                if (inQuote)
                    sb.Append(delimiter);
                else
                {
                    record.Add(sb.ToString());
                    sb.Clear();
                }
            }
            else if (readChar == qualifier)
            {
                if (inQuote)
                {
                    if ((char) reader.Peek() == qualifier)
                    {
                        reader.Read();
                        sb.Append(qualifier);
                    }
                    else
                        inQuote = false;
                }
                else
                    sb.Append(readChar);
            }
            else
                sb.Append(readChar);
        }

        if (record.Count > 0 || sb.Length > 0)
            record.Add(sb.ToString());

        if (record.Count > 0)
            yield return record;
    }
}

Using the Code

Here is an example of reading CSV file. The following code snippet parses out the first 5 records and prints them out to the Console in form of key/value pairs:

const string fileName = @"C:\Temp\file.csv";
using (var stream = File.OpenRead(fileName))
using (var reader = new StreamReader(stream))
{
    var data = CsvParser.ParseHeadAndTail(reader, ',', '"');

    var header = data.Item1;
    var lines = data.Item2;

    foreach (var line in lines.Take(5))
    {
        for (var i = 0; i < header.Count; i++)
            if (!string.IsNullOrEmpty(line[i]))
                Console.WriteLine("{0}={1}", header[i], line[i]);
        Console.WriteLine();
    }
}
Console.ReadLine();

History

• 27^th September, 2014: Initial version