dotnet/machinelearning · error · ArgumentNullException

Value cannot be null. (Parameter 'input')

Error message

Value cannot be null. (Parameter 'input')

What it means

Utils.Normalize converts a string into a valid C# identifier (sanitizing, prefixing, capitalizing). Passing null reaches the switch on input and throws ArgumentNullException('input'). The library requires a non-null string because it must produce a valid identifier name for generated code.

Source

Thrown at src/Microsoft.ML.CodeGenerator/Utils.cs:142

            {
                var f = val as bool?;
                return f.GetValueOrDefault() ? "true" : "false";
            }

            return val.ToString();
        }

        internal static string Normalize(string input)
        {
            //check if first character is int
            if (!string.IsNullOrEmpty(input) && int.TryParse(input.Substring(0, 1), out int val))
            {
                input = "_" + input;
                return Normalize(input);
            }
            switch (input)
            {
                case null: throw new ArgumentNullException(nameof(input));
                case "": throw new ArgumentException($"{nameof(input)} cannot be empty", nameof(input));
                default:
                    var sanitizedInput = Sanitize(input);
                    return sanitizedInput.First().ToString().ToUpper() + sanitizedInput.Substring(1);
            }
        }

        internal static Type GetCSharpType(DataKind labelType)
        {
            switch (labelType)
            {
                case Microsoft.ML.Data.DataKind.String:
                    return typeof(string);
                case Microsoft.ML.Data.DataKind.Boolean:
                    return typeof(bool);
                case Microsoft.ML.Data.DataKind.Single:
                    return typeof(float);
                case Microsoft.ML.Data.DataKind.Double:

View on GitHub (pinned to 7b76e69cf9)

Solutions

1. Ensure the name source is non-null before calling Normalize; supply a fallback name when the column/metadata name is missing.
2. Check your input data headers — replace empty/null headers with valid names.
3. If calling Normalize directly, guard with string.IsNullOrWhiteSpace before the call.
4. Catch ArgumentNullException and substitute a generated placeholder name.

Example fix

// before
var identifier = Utils.Normalize(columnName);
// after
var identifier = Utils.Normalize(columnName ?? "Column_" + index);

if (name is null)
    throw new InvalidOperationException("Column name was null; check data source headers");
var identifier = Utils.Normalize(name);

static bool IsNormalizable(string s) => !string.IsNullOrEmpty(s);

try
{
    identifier = Utils.Normalize(rawName);
}
catch (ArgumentNullException)
{
    identifier = "UnnamedColumn";
}

When it happens

Trigger: Calling Utils.Normalize(null) directly, or indirectly when a column/metadata name used to derive an identifier is null (e.g. a schema column with null name passed through code generation helpers that call Normalize).

Common situations: Data files with missing/blank header names producing null column names; programmatic PipelineNode/slot-name construction where the name field was never set; codegen invoked on a schema whose text metadata is absent.

Related errors

• input cannot be empty (Parameter 'input')
• columns
• rawType
• throw new ArgumentNullException(nameof(ids));
• throw new ArgumentNullException(nameof(tokenIds));

AI-assisted analysis of dotnet/machinelearning@7b76e69cf9 (2026-09-11). Data as JSON: /api/errors/66f0ee82b3aec063. Report an issue: GitHub.