monorepo/LCU.API.Internal.FileOperations/LCU.Api.Internal.FileOperations/Services/ExcelServices/ExcelService.cs

using System;
using System.IO;
using System.Linq;
using Serilog;
using DocumentFormat.OpenXml.Packaging;
using DocumentFormat.OpenXml.Spreadsheet;
using LCU.Library.Internal.Exceptions;
using LCU.Library.Internal.FileOperationsModels.Models.ExcelModels.Responses;
using LCU.Library.Internal.FileOperationsModels.Models.ExcelModels.Requests;

namespace LCU.Api.Internal.FileOperations.Services.ExcelServices;

public class ExcelService : IExcelService
{
    public ExcelDataResponse ExtractExcelWorkBookData(ExcelDataRequest excelDataRequest)
    {
        var excelDataResponse = new ExcelDataResponse();
        try
        {
            var fileBytes = Convert.FromBase64String(excelDataRequest.FileContent);
            using var memoryStream = new MemoryStream(fileBytes);
            
            using SpreadsheetDocument spreadsheetDocument = SpreadsheetDocument.Open(memoryStream, false);
            WorkbookPart workbookPart = spreadsheetDocument.WorkbookPart;
            Workbook workbook = workbookPart.Workbook;
            foreach (Sheet sheet in workbook.GetFirstChild<Sheets>().Elements<Sheet>())
            {
                SheetRequest sheetRequest = null;
                if (excelDataRequest.SkipSheetForExtraction(sheet.Name, out sheetRequest))
                {
                    continue;
                }

                var sheetResponse = new SheetResponse(sheet.Name);
                var worksheetPart = (WorksheetPart)workbookPart.GetPartById(sheet.Id);
                var sheetData = worksheetPart.Worksheet.Elements<SheetData>().FirstOrDefault();
                if (sheetData == null)
                {
                    Log.Information($"Sheet named {sheet.Name} does not have any sheet data.");
                    continue;
                }

                sheetResponse = ExtractRowData(sheetData, sheetRequest, sheetResponse);
                excelDataResponse.WorkBook.SheetResponses.Add(sheetResponse);
            }

            return excelDataResponse;
        }
        catch (Exception exception)
        {
            var message = $"Failure during ExtractExcelWorkBookData. Exception Message: {exception.Message}";
            Log.Error(message, exception);
            throw new ServiceException(message);
        }
    }

    private SheetResponse ExtractRowData(SheetData sheetData, SheetRequest sheetRequest, SheetResponse sheetResponse)
    {
        foreach (var row in sheetData.Elements<Row>())
        {
            if (!int.TryParse(row.RowIndex, out int rowIndex))
            {
                continue;
            }

            if (sheetRequest.SkipRowForExtraction(rowIndex))
            {
                continue;
            }

            sheetResponse = ExtractCellData(row, rowIndex, sheetRequest, sheetResponse);
        }

        return sheetResponse;
    }

    private SheetResponse ExtractCellData(Row row, int rowIndex, SheetRequest sheetRequest, SheetResponse sheetResponse)
    {
        foreach (var cell in row.Elements<Cell>())
        {
            var columnName = GetColumnName(cell.CellReference);
            if (sheetRequest.SkipColumnForExtraction(columnName))
            {
                continue;
            }

            if(cell is null || cell.CellValue is null)
            {
                continue;
            }

            var cellResponse = new CellResponse()
            {
                Row = rowIndex,
                ColumnName = columnName,
                Value = cell.CellValue.InnerText,
                Formula = cell.CellFormula is null ? string.Empty : cell.CellFormula.InnerText
            };
            sheetResponse.CellResponses.Add(cellResponse);
        }

        return sheetResponse;
    }

    private static string GetColumnName(string cellReference)
    {
        string columnName = "";
        foreach (char character in cellReference)
        {
            if (!char.IsLetter(character))
            {
                break;
            }

            columnName += character;
        }
        return columnName;
    }
}