Gfrktyl

I'm using C# .NET Core 2.1 and Excel lib DotnetCore.NPOI

I'm converting xls file to xlsx by loading xls and copying sheet by sheet, cell by cell to xlsx - How can I improve performance / ram usage here?

private string ConvertToXSLX(string path) 

{

    using (var fs = File.OpenRead(path))

    {

        var result = XLS_to_XLSX_Converter.Convert(fs);



        // (file.xls) + x = file.xlsx

        path = $"{path}x";



        using (var fs2 = new FileStream(path, FileMode.OpenOrCreate, FileAccess.Write))

        {

            fs2.Write(result, 0, result.Length);

        }

    }

    return path;

}

public static byte Convert(Stream sourceStream)

{

    var source = new HSSFWorkbook(sourceStream);

    var destination = new XSSFWorkbook();



    for (int i = 0; i < source.NumberOfSheets; i++)

    {

        var hssfSheet = (HSSFSheet)source.GetSheetAt(i);



        var xssfSheet = (XSSFSheet)destination.CreateSheet(source.GetSheetAt(i).SheetName);



        CopyRows(hssfSheet, xssfSheet);

    }



    using (var ms = new MemoryStream())

    {

        destination.Write(ms);

        return ms.ToArray();

    }

}



private static void CopyRows(HSSFSheet hssfSheet, XSSFSheet destination)

{

    for (int i = 0; i < hssfSheet.PhysicalNumberOfRows; i++)

    {

        destination.CreateRow(i);



        var cells = hssfSheet.GetRow(i)?.Cells ?? new List<ICell>();



        for (int j = 0; j < cells.Count; j++)

        {

            var row = destination.GetRow(i);

            row.CreateCell(j);



            CopyCell((HSSFCell)cells[j], (XSSFCell)row.Cells[j]);

        }   

    }

}



private static void CopyCell(HSSFCell oldCell, XSSFCell newCell)

{

    CopyCellValue(oldCell, newCell);

}



private static void CopyCellValue(HSSFCell oldCell, XSSFCell newCell)

{

    switch (oldCell.CellType)

    {

        case CellType.String:

            newCell.SetCellValue(oldCell.StringCellValue);

            break;

        case CellType.Numeric:

            newCell.SetCellValue(oldCell.NumericCellValue);

            break;

        case CellType.Blank:

            newCell.SetCellType(CellType.Blank);

            break;

        case CellType.Boolean:

            newCell.SetCellValue(oldCell.BooleanCellValue);

            break;

        case CellType.Error:

            newCell.SetCellErrorValue(oldCell.ErrorCellValue);

            break;

        default:

            break;

    }

} 

You're writing to a MemoryStream only to get a byte to write to another stream, you can definitely avoid this intermediate conversion:

private string ConvertToXSLX(string inputPath) 

{

    string outputPath = Path.ChangeExtension(inputPath, ".xlsx");



    using (var input = File.OpenRead(inputPath))

    using (var output = new FileStream(outputPath, FileMode.OpenOrCreate, FileAccess.Write))

    {

        XLS_to_XLSX_Converter.Convert(input, output);

    }



    return outputPath;

}

The change in Convert() should be trivial.

Also note that I'm using Path.ChangeExtension() instead of manually adding a character, code is clear without any comment and it handles special cases for you (for example a trailing space).

In CopyRows() you create an empty List<ICell>, you do not need to and you can avoid the initial allocation simply using Enumerable.Empty<ICell>() (and changing your code to work with an enumeration) or - at least - reusing the same empty object (move it outside the loop and create it with an initial capacity of 0).

In CopyCellValue() you have an empty default case for your switch. It's a good thing to always have default but it has a purpose: detect unknown cases. If you put a break then you're silently ignoring them. Is it on purpose? Write a comment to explain you ignore unknown cells. It's an error? Throw an exception.

I never used DotnetCore.NPOI then I can't comment about the way you use it, be sure to dispose all the intermediate IDisposable objects.