Just Some Code

I'm Cesar Aguirre. Software engineer

© 2026. All rights reserved.

TIL: How to Read Invalid Fields as Null with CsvHelper

10 Apr 2026 #csharp #todayilearned

In another episode of the Excel paradox of Coding…

TL;DR: To read an invalid field from a CSV file as null with CsvHelper, create a map class and use Default() with useOnConversionFailure set to true on the appropriate field.

The cleanest approach

To honor the 20-minute rule and preserve my keystrokes, let’s say we need to read a CSV file with CsvHelper. By default, it throws an exception when it fails to parse a column, for example when it finds text on a numeric column.

Here’s how to read that invalid field as null instead of throwing an exception,

using CsvHelper;
using CsvHelper.Configuration;
using System.Globalization;
using System.Text;

namespace TestProject1;

[TestClass]
public class CsvHelperTests
{
    private class MovieRating
    {
        public string Name { get; set; }
        public float? Rating { get; set; }
    }

    private sealed class MovieRatingMap : ClassMap<MovieRating>
    {
        public MovieRatingMap()
        {
            Map(m => m.Name);
            Map(m => m.Rating)
                .Default(defaultValue: default(float?),
                         useOnConversionFailure: true);
            //           ^^^^^
            // If the field is invalid, use the default value
        }
    }

    [TestMethod]
    public void NullWhenInvalidFloats()
    {
        var csv = new StringBuilder()
            .AppendLine("Name,Rating")
            .AppendLine("Inception,9.0")
            .AppendLine("Titanic,ThisIsNotAValidRating")  /* <-- */
            .ToString();

        using var reader = new StringReader(csv);
        using var csvReader = new CsvReader(reader, CultureInfo.InvariantCulture);
        csvReader.Context.RegisterClassMap<MovieRatingMap>();
        //                ^^^^^

        var records = csvReader.GetRecords<MovieRating>().ToList();

        Assert.AreEqual(2, records.Count);

        var last = records.Last();
        Assert.AreEqual("Titanic", last.Name);
        Assert.IsNull(last.Rating);
        //            ^^^^^
        // Look, ma! It's null
    }
}

The magic happens in MovieRatingMap. The Rating map uses Default() with useOnConversionFailure set to true.

The manual approach

As an alternative, let’s read the CSV file manually while processing the records as needed,

[TestMethod]
public void ByHand()
{
	var csv = new StringBuilder()
		.AppendLine("Name,Rating")
		.AppendLine("Inception,9.0")
		.AppendLine("Titanic,ThisIsNotAValidRating")
		.ToString();

	using var reader = new StringReader(csv);
	using var csvReader = new CsvReader(reader, CultureInfo.InvariantCulture);
	csvReader.Context.RegisterClassMap<MovieRatingMap>();

	csvReader.Read();
	csvReader.ReadHeader();
	//        ^^^^^
	// Open the reader and read the header

	var records = new List<MovieRating>();
	while (csvReader.Read())
	//               ^^^^^
	// Read it by hand...
	{
		var name = csvReader.GetField<string>("Name");
		var rating = csvReader.GetField("Rating");
		//                     ^^^^^
		// Read the field as string...

		var parsedRating = !float.TryParse(rating, NumberStyles.Float, CultureInfo.InvariantCulture, out var parsed)
								? default(float?)
								: parsed;
		// Parse it as needed...

		records.Add(new MovieRating
		{
			Name = name,
			Rating = parsedRating
		});
	}

	Assert.AreEqual(2, records.Count);

	var last = records.Last();
	Assert.AreEqual("Titanic", last.Name);
	Assert.IsNull(last.Rating);
	//            ^^^^^
	// Look, ma! It's still null
}

A lot more work! We have full control, but it’s less clean.

In case you’re wondering, I asked Copilot about this but, with a straight face, it hallucinated suggesting a flag on CsvConfiguration, which didn’t even exist.

For scenarios like this, you still need strong debugging and code reading skills. I cover those skills and more in Street-Smart Coding—A roadmap with 30 lessons to help you code like a pro.

Related Posts