Fix IndexOutOfRangeException in LoadFromText when text file has more columns than specified DataTypes

[lievendf]

Description

This pull request resolves a System.IndexOutOfRangeException that occurs during ExcelRangeBase.LoadFromText execution when importing text or CSV files.

Root Cause

In src/EPPlus/LoadFunctions/LoadFromTextBase.cs, the ConvertData method checks if the current column index col is out of bounds for the dataType array:

if (isText && (dataType == null || dataType.Length < col))
{
    return string.IsNullOrEmpty(v) ? null : v;
}
else
{
    if(dataType == null || dataType.Length < col)
        return ConvertData(Format, eDataTypes.Unknown, v, col, isText);
    return ConvertData(Format, dataType[col], v, col, isText);
}

The bounds checks use dataType.Length < col. Because col is a 0-based column index, when col is equal to dataType.Length, it is already out of bounds. The off-by-one check allowed execution to bypass the guard and attempt to access dataType[col], resulting in a System.IndexOutOfRangeException.

Solution

Simplified and corrected the bounds checking logic in src/EPPlus/LoadFunctions/LoadFromTextBase.cs's ConvertData method. Introduced a clean boolean flag isOutOfBounds checking col >= dataType.Length and cleanly structured the branches:

protected object ConvertData(T Format, eDataTypes[] dataType, string v, int col, bool isText)
{
    bool isOutOfBounds = dataType == null || col >= dataType.Length;

    if (isOutOfBounds)
    {
        if (isText)
        {
            return string.IsNullOrEmpty(v) ? null : v;
        }
        return ConvertData(Format, eDataTypes.Unknown, v, col, isText);
    }

    return ConvertData(Format, dataType[col], v, col, isText);
}

This ensures that any trailing columns beyond the length of the specified DataTypes array are handled gracefully and default to eDataTypes.Unknown (General) rather than causing a crash.

Unit Tests

Added a new unit test ShouldLoadCsvFormatWithTrailingColumns to src/EPPlusTest/LoadFunctions/LoadFromTextTests.cs to verify the fix and explain the historical issue with clear comments:

[TestMethod]
public void ShouldLoadCsvFormatWithTrailingColumns()
{
    // This test verifies that importing a text/CSV file with more columns than specified in DataTypes does not crash.
    // Previously, an off-by-one bounds check (dataType.Length < col) allowed the loop to attempt accessing dataType[col]
    // when 'col' was equal to 'dataType.Length', resulting in a System.IndexOutOfRangeException.
    // With the fix (col >= dataType.Length), trailing columns beyond the DataTypes array are gracefully imported as Unknown (General).
    AddLine("a;2;extra");
    _format.Delimiter = ';';
    _format.DataTypes = new eDataTypes[] { eDataTypes.String, eDataTypes.Number };
    _worksheet.Cells["A1"].LoadFromText(_lines.ToString(), _format);
    Assert.AreEqual("a", _worksheet.Cells["A1"].Value);
    Assert.AreEqual(2d, _worksheet.Cells["B1"].Value);
    Assert.AreEqual("extra", _worksheet.Cells["C1"].Value);
}

This test runs and passes successfully on both .NET 8.0 and .NET Framework 4.6.2 targets.

[swmal]

Hello @lievendf - Many thanks for this PR, see my comment on your previous PR about signing our CAA.