BDF3/XLimport.cs

using System;
using System.Collections.Generic;
using System.Data;
using System.IO;
using System.Linq;
using System.Text.RegularExpressions;

namespace bdf
{
    class CXLFile
    {
        //public static List<XLWorkbook> ServiceOrders = new List<XLWorkbook> { };

        public static List<Spire.Xls.Workbook> S_ServiceOrders = new List<Spire.Xls.Workbook> { };


        public static void S_getXLS(string task, string dir)
        {
            //string folderPath = dir;
            string filePrefix1 = "ROGERS_";
            string filePrefix2 = task + "_SO";
            var excelFiles = getLatestVersionFiles(dir, filePrefix1, filePrefix2);

            if ( (excelFiles.Length == 0) || (excelFiles == null) )
            {
                Console.WriteLine("No Excel files found with prefix: " + filePrefix1 + "*" + filePrefix2);
                //throw new System.ArgumentNullException("filePaths");
                return;
            }

            foreach (string path in excelFiles)
            {
                if (string.IsNullOrWhiteSpace(path))
                    continue;

                if (!System.IO.File.Exists(path))
                    throw new System.IO.FileNotFoundException(
                        "Excel file not found.", path);

                Logger.Log(1, "Scanning: {0}", Path.GetFileName(path));

                // Create a new workbook and load the file
                Spire.Xls.Workbook wb = new Spire.Xls.Workbook();

                //straight up load from file
                //wb.LoadFromFile(path);  // FreeSpire autodetects .xls vs .xlsx

                //more network savvy way that handles sharepoint's idiosyncrasies
                FileStream stream = null;

                // OneDrive sync engine frequently locks files briefly:
                const int maxRetries = 12;
                const int retryDelayMs = 150;

                for (int i = 0; i < maxRetries; i++)
                {
                    try
                    {
                        stream = new FileStream(
                            path,
                            FileMode.Open,
                            FileAccess.Read,
                            FileShare.ReadWrite | FileShare.Delete
                        );
                        break; // opened successfully
                    }
                    catch (IOException)
                    {
                        if (i == maxRetries - 1)
                            throw;

                        System.Threading.Thread.Sleep(retryDelayMs);
                    }
                }
                wb.LoadFromStream(stream);

                Logger.Log(5, "  -loaded SO={0}", wb.Worksheets[1][2, 1].Value.ToString() );

                S_ServiceOrders.Add(wb);
            }
            Logger.Log("All SO Excel files processed successfully.");
            return; // S_ServiceOrders;
        }


        // This routine is wholly dependent on the filename structure being ROGERS*LDF*_SO*VER*number
        // in future should just open each workbook and read exact version from the first worksheet.
        public static string[] getLatestVersionFiles(string folder, string prefixA, string prefixB)
        {
            var files = Directory.GetFiles(folder);

            // Build wildcard pattern:
            // file must contain prefixA ... anything ... prefixB
            Func<string, bool> matchesPrefix = name =>
                name.IndexOf(prefixA, StringComparison.OrdinalIgnoreCase) >= 0 &&
                name.IndexOf(prefixB, StringComparison.OrdinalIgnoreCase) >
                    name.IndexOf(prefixA, StringComparison.OrdinalIgnoreCase);

            var filtered = files
                .Where(f => matchesPrefix(Path.GetFileName(f)))
                .ToList();

            // Serial = 6 digits
            // After "VER" allow ANY non-digit characters, then the version number
            var regex = new Regex(
                @"(?<serial>\d{6}).*?VER\D*(?<ver>\d+)",
                RegexOptions.IgnoreCase
            );

            Dictionary<string, (int version, string file)> best =
                new Dictionary<string, (int, string)>();

            foreach (var file in files)
            {
                string name = Path.GetFileName(file);
                var match = regex.Match(name);

                if (!match.Success)
                {
                    Logger.Log(1, " Did not find SO and Version for file {0}", name);
                    continue;
                }

                string serial = match.Groups["serial"].Value;

                if (!int.TryParse(match.Groups["ver"].Value, out int ver))
                {
                    Logger.Log(1, " Did not extract Version for file {0}", name);
                    continue;
                }

                // Keep highest version per serial
                if (!best.ContainsKey(serial)) // || ver > best[serial].version)
                {
                    best[serial] = (ver, file);
                    Logger.Log(5, " Adding: {0} ver {1}", serial, ver);
                }
                else if (ver > best[serial].version)
                {
                    Logger.Log(2, " Replacing: {0} ver {1} with ver {2}", serial, best[serial].version, ver);
                    best[serial] = (ver, file);
                }
                else
                {
                    Logger.Log(4, " Ignoring: {0} v.{1}", serial, ver);
                }
            }

            return best.Values.Select(v => v.file).ToArray();
        }

        /*
        public static void getXLS(string task, string dir)
        {
            //string folderPath = @"C:\Data\ExcelFiles";   // Folder to scan
            string folderPath = dir;
            string filePrefix1 = "ROGERS_";
            string filePrefix2 = task + "_SO";

            //List<XLWorkbook> ServiceOrders = new List<XLWorkbook> { };

            try
            {
                // Required for ExcelDataReader in .NET 4
                System.Text.Encoding.RegisterProvider(
                    System.Text.CodePagesEncodingProvider.Instance);

                // Get all Excel files that match the prefix
                //var excelFiles = Directory.GetFiles(folderPath, filePrefix1 + "*" + filePrefix2 + "*.xls*", SearchOption.TopDirectoryOnly);

                var excelFiles = getLatestVersionFiles(dir, filePrefix1, filePrefix2);

                if (excelFiles.Length == 0)
                {
                    Console.WriteLine("No Excel files found with prefix: " + filePrefix1 + "*" + filePrefix2);
                    return;
                }

                DataSet _dataSet = null;

                foreach (var filePath in excelFiles)
                {
                    Logger.Log(1, "Scanning: {0}", Path.GetFileName(filePath));
                    // DataSet dataSet = ReadExcelFile(filePath);

                    // STEP 1: Read file into DataSet using ExcelDataReader
                    try
                    {
                        _dataSet = ReadExcelFile(filePath);
                        //_dataSet = LoadOneDriveDataSet(filePath);
                    }
                    catch
                    {
                        try
                        {
                            _dataSet = LoadOneDriveDataSet(filePath);
                        }
                        catch (Exception e)
                        {
                            Logger.Log(0, "Error trying to access file {0}.  ({1})", filePath, e.Message);
                        }
                    }


                    // STEP 2: Convert DataSet into a ClosedXML workbook for manipulation
                    var workbook = new XLWorkbook();
                    //using (var workbook = new XLWorkbook())

                    {
                        foreach (DataTable table in _dataSet.Tables)
                        {
                            var worksheet = workbook.Worksheets.Add(table.TableName);

                            // Insert headers
                            for (int col = 0; col < table.Columns.Count; col++)
                            {
                                worksheet.Cell(1, col + 1).Value = table.Columns[col].ColumnName;
                                worksheet.Cell(1, col + 1).Style.Font.Bold = true;
                            }

                            // Insert rows
                            for (int row = 0; row < table.Rows.Count; row++)
                            {
                                for (int col = 0; col < table.Columns.Count; col++)
                                {
                                    //Logger.Log(5, "type is {0}", table.Rows[row][col].GetType().Name);

                                    string valuetype = table.Rows[row][col].GetType().Name;

                                    if (valuetype == "Int32")
                                    {
                                        worksheet.Cell(row + 2, col + 1).Value = (Int32)table.Rows[row][col];
                                    }
                                    else if (valuetype == "String")
                                    {
                                        worksheet.Cell(row + 2, col + 1).Value = (String)table.Rows[row][col];
                                    }
                                    else if (valuetype == "Decimal")
                                    {
                                        worksheet.Cell(row + 2, col + 1).Value = (Decimal)table.Rows[row][col];
                                    }
                                    else if (valuetype == "Double")
                                    {
                                        worksheet.Cell(row + 2, col + 1).Value = (Double)table.Rows[row][col];
                                    }
                                    else if (valuetype == "DBNull")
                                    {
                                        worksheet.Cell(row + 2, col + 1).Value = "";
                                    }
                                    else
                                    {
                                        worksheet.Cell(row + 2, col + 1).Value = "fix me";
                                    }

                                    //worksheet.Cell(row + 2, col + 1).Value = table.Rows[row][col];
                                }
                            }

                            worksheet.Columns().AdjustToContents();
                            //Logger.Log(0, "Sheet{0} Value={1}", workbook.Worksheets.Count, workbook.Worksheet(workbook.Worksheets.Count).Cell("C2").Value);
                        }

                        // Example: Save modified workbook (optional)
                        string outputFile = Path.Combine(folderPath, "" + Path.GetFileName(filePath) + "x");

                        //workbook.SaveAs(outputFile);
                        Logger.Log(7, "Processed file: {0}", outputFile);
                        //Logger.Log(0, "Book has {0} sheets", workbook.Worksheets.Count );

                        CXLFile.ServiceOrders.Add(workbook);
                        //Logger.Log("test: " + workbook.Worksheet(1).Cell("C2").Value);
                    }

                    //Logger.Log("Finished: " + Path.GetFileName(filePath));
                    //Logger.Log(new string('-', 50));
                }

                Logger.Log("All SO Excel files processed successfully.");
            }
            catch (Exception ex)
            {
                Logger.Log(0, "SO processing error: {0}" + ex.Message);
                //Logger.Log("Press any key to exit...");
                //Console.ReadKey();
            }

            //
            //return ServiceOrders;

        }

        /// <summary>
        /// Reads an Excel file into a DataSet using ExcelDataReader.
        /// Each worksheet becomes a DataTable.
        /// </summary>
        static DataSet ReadExcelFile(string filePath)
        {
            // Required for ExcelDataReader to process XLS files
            System.Text.Encoding.RegisterProvider(
                System.Text.CodePagesEncodingProvider.Instance);

            using (var stream = System.IO.File.Open(filePath, FileMode.Open, FileAccess.Read))
            {
                IExcelDataReader reader;

                // Auto-detect format (XLS or XLSX)
                if (Path.GetExtension(filePath).Equals(".xls", StringComparison.OrdinalIgnoreCase))
                    reader = ExcelReaderFactory.CreateBinaryReader(stream);
                else
                    reader = ExcelReaderFactory.CreateOpenXmlReader(stream);

                // Convert to DataSet (with first row as header)
                var result = reader.AsDataSet(new ExcelDataSetConfiguration()
                {
                    ConfigureDataTable = _ => new ExcelDataTableConfiguration()
                    {
                        UseHeaderRow = true
                    }
                });

                reader.Close();
                return result;
            }
        }

        public static DataSet LoadOneDriveDataSet(string path)
        {
            if (!System.IO.File.Exists(path))
                throw new FileNotFoundException("OneDrive file not found.", path);

            // Required for ExcelDataReader to process XLS files
            System.Text.Encoding.RegisterProvider(
                System.Text.CodePagesEncodingProvider.Instance);

            FileStream stream = null;

            // OneDrive sync engine frequently locks files briefly:
            const int maxRetries = 12;
            const int retryDelayMs = 150;

            for (int i = 0; i < maxRetries; i++)
            {
                try
                {
                    stream = new FileStream(
                        path,
                        FileMode.Open,
                        FileAccess.Read,
                        FileShare.ReadWrite | FileShare.Delete
                    );
                    break; // opened successfully
                }
                catch (IOException)
                {
                    if (i == maxRetries - 1)
                        throw;

                    System.Threading.Thread.Sleep(retryDelayMs);
                }
            }

            // STEP 1: Read the workbook into a DataSet (ExcelDataReader)
            DataSet dataSet;

            using (var reader = ExcelReaderFactory.CreateReader(stream))
            {
                dataSet = reader.AsDataSet(new ExcelDataSetConfiguration
                {
                    ConfigureDataTable = _ => new ExcelDataTableConfiguration
                    {
                        UseHeaderRow = true
                    }
                });
            }

            return dataSet;
        }
        */
    }
}