Looping through tabular CSV files doesn't return intended fields

Question

I have a 20,000 line CSV file with this data:

Location,	Light,	Proximity,	Ax,	Ay,	Az,	Gx,	Gy,	Gz
SidePocket	2.0	0.0	-1.1259307861328125	-10.622817993164063	0.8393707275390625	0.7456817626953125	-2.3446502685546875	-0.6551361083984375
HandBag	2.0	0.0	-1.1259307861328125	-10.622817993164063	0.8393707275390625	0.8383636474609375	-3.1872711181640625	-0.064971923828125
SidePocket	2.0	0.0	0.5566253662109375	-9.675201416015625	1.7905426025390625	0.8383636474609375	-3.1872711181640625	-0.064971923828125
SidePocket	2.0	0.0	0.5566253662109375	-9.675201416015625	1.7905426025390625	0.170440673828125	-2.976348876953125	0.05218505859375
BackPocket	2.0	0.0	-0.3665771484375	-9.739242553710938	2.12567138671875	0.170440673828125	-2.976348876953125	0.05218505859375
SidePocket	2.0	0.0	-0.3665771484375	-9.739242553710938	2.12567138671875	-0.1981201171875	-1.846099853515625	0.290802001953125
Ear	2.0	0.0	-0.490264892578125	-9.91455078125	1.34954833984375	-0.1981201171875	-1.846099853515625	0.290802001953125

I want something to loop through all rows of a column based on the column header and do some computation. I tried using the CsvHelper library:

private void btnBrowse_Click(object sender, RoutedEventArgs e)
{
    OpenFileDialog openFileDialog = new OpenFileDialog();
    if (openFileDialog.ShowDialog() == true)
        textReader = File.ReadAllText(openFileDialog.FileName);
    stringParse = new StringReader(textReader);
    txtOutput.Text = "";
}

private void btnParse_Click(object sender, RoutedEventArgs e)
{
    var csv = new CsvReader(stringParse);
    while( csv.Read() )
    {               
        var stringField = csv.GetField<string>( "Location" );
        txtOutput.Text += DoSomething(stringField.ToString()) + "\n";
    }
}

The problem I face are:

• Firstly, this is very, very slow. Halts my small WPF app.
• Secondly, it doesn't return the values for the intended field. It dumps the whole CSV file in the output.

Where have I made a mistake?

Here's how the actual CSV file looks:

Location, Light, Proximity, Ax, Ay, Az, Gx, Gy, Gz
"SidePocket" 2.0 0.0 -1.1259307861328125 -10.622817993164063 0.8393707275390625 0.7456817626953125 -2.3446502685546875 -0.6551361083984375
"HandBag" 2.0 0.0 -1.1259307861328125 -10.622817993164063 0.8393707275390625 0.8383636474609375 -3.1872711181640625 -0.064971923828125
"SidePocket" 2.0 0.0 0.5566253662109375 -9.675201416015625 1.7905426025390625 0.8383636474609375 -3.1872711181640625 -0.064971923828125

Answer 1

I have no problem reading a csv file of 4 times the size of your 20,000 lines in under 2 seconds. I don't know what is happening in DoSomething so that might be your next call of investigation.

I have made the following changes to address the two issues you describe.

First: use a StringBuilder to prevent the creation and discarding of lots of strings in your loop.

Second: Set the Delimiter in the Configuration object of the CsvHelper to a Space but do include the comma of your header field name in that case:

var sw = new Stopwatch();
sw.Start();
using(var csv = new CsvReader(new StreamReader(@"csv-test.txt")))
{
    csv.Configuration.Delimiter=" "; // space

    var sb = new StringBuilder();

    while (csv.Read())
    {
        var stringField = csv.GetField<string>("Location,"); // the comma is relevant
        // or use sb.AppendFormat("{0}\n", DoSomething(stringField));
        sb.AppendLine(stringField);
    }
    txtOutput.Text = sb.ToString();
}
sw.Stop();
Console.WriteLine(sw.ElapsedMilliseconds);

For 102,000 lines above code runs on my box in 1,7 seconds.