c# 读取excel数据的两种方法（转）

转载自：http://developer.51cto.com/art/201302/380622.htm，

方法一：OleDb： 用这种方法读取Excel速度还是非常的快的，但这种方式读取数据的时候不太灵活，不过可以在 DataTable 中对数据进行一些删减修改。

优点：读取方式简单、读取速度快

缺点：除了读取过程不太灵活之外，这种读取方式还有个弊端就是，当Excel数据量很大时。会非常占用内存，当内存不够时会抛出内存溢出的异常。

不过一般情况下还是非常不错的。

DataTable GetDataFromExcelByConn(bool hasTitle = false)
{
    OpenFileDialog openFile = new OpenFileDialog();
    openFile.Filter = "Excel(*.xlsx)|*.xlsx|Excel(*.xls)|*.xls";
    openFile.InitialDirectory = Environment.GetFolderPath(Environment.SpecialFolder.Desktop);
    openFile.Multiselect = false;
    if (openFile.ShowDialog() == DialogResult.Cancel) return null;
    var filePath = openFile.FileName;
    string fileType = System.IO.Path.GetExtension(filePath);
    if (string.IsNullOrEmpty(fileType)) return null;

    using (DataSet ds = new DataSet())
    {
        string strCon = string.Format("Provider=Microsoft.Jet.OLEDB.{0}.0;" +
                        "Extended Properties=\"Excel {1}.0;HDR={2};IMEX=1;\";" +
                        "data source={3};",
                        (fileType == ".xls" ?  : ), (fileType == ".xls" ?  : ), (hasTitle ? "Yes" : "NO"), filePath);
        string strCom = " SELECT * FROM [Sheet1$]";
        using (OleDbConnection myConn = new OleDbConnection(strCon))
        using (OleDbDataAdapter myCommand = new OleDbDataAdapter(strCom, myConn))
        {
            myConn.Open();
            myCommand.Fill(ds);
        }
        if (ds == null || ds.Tables.Count <= ) return null;
        return ds.Tables[];
    }
}

方法二：Com组件的方式读取Excel

这种方式需要先引用 Microsoft.Office.Interop.Excel 。首选说下这种方式的优缺点

优点：可以非常灵活的读取Excel中的数据

缺点：如果是Web站点部署在IIS上时，还需要服务器机子已安装了Excel，有时候还需要为配置IIS权限。最重要的一点因为是基于单元格方式读取的，所以数据很慢（曾做过试验，直接读取千行、200多列的文件，直接读取耗时15分钟。即使采用多线程分段读取来提高CPU的利用率也需要8分钟。PS：CPU I3）

需要读取大文件的的童鞋们慎重。。。

DataTable GetDataFromExcelByCom(bool hasTitle = false)
{
    OpenFileDialog openFile = new OpenFileDialog();
    openFile.Filter = "Excel(*.xlsx)|*.xlsx|Excel(*.xls)|*.xls";
    openFile.InitialDirectory = Environment.GetFolderPath(Environment.SpecialFolder.Desktop);
    openFile.Multiselect = false;
    if (openFile.ShowDialog() == DialogResult.Cancel) return null;
    var excelFilePath = openFile.FileName;

    Excel.Application app = new Excel.Application();
    Excel.Sheets sheets;
    object oMissiong = System.Reflection.Missing.Value;
    Excel.Workbook workbook = null;
    DataTable dt = new DataTable();

    try
    {
        if (app == null) return null;
        workbook = app.Workbooks.Open(excelFilePath, oMissiong, oMissiong, oMissiong, oMissiong, oMissiong,
            oMissiong, oMissiong, oMissiong, oMissiong, oMissiong, oMissiong, oMissiong, oMissiong, oMissiong);
        sheets = workbook.Worksheets;

        //将数据读入到DataTable中
        Excel.Worksheet worksheet = (Excel.Worksheet)sheets.get_Item();//读取第一张表
        if (worksheet == null) return null;

        int iRowCount = worksheet.UsedRange.Rows.Count;
        int iColCount = worksheet.UsedRange.Columns.Count;
        //生成列头
        for (int i = ; i < iColCount; i++)
        {
            var name = "column" + i;
            if (hasTitle)
            {
                var txt = ((Excel.Range)worksheet.Cells[, i + ]).Text.ToString();
                if (!string.IsNullOrWhiteSpace(txt)) name = txt;
            }
            while (dt.Columns.Contains(name)) name = name + "_1";//重复行名称会报错。
            dt.Columns.Add(new DataColumn(name, typeof(string)));
        }
        //生成行数据
        Excel.Range range;
        int rowIdx = hasTitle ?  : ;
        for (int iRow = rowIdx; iRow <= iRowCount; iRow++)
        {
            DataRow dr = dt.NewRow();
            for (int iCol = ; iCol <= iColCount; iCol++)
            {
                range = (Excel.Range)worksheet.Cells[iRow, iCol];
                dr[iCol - ] = (range.Value2 == null) ? "" : range.Text.ToString();
            }
            dt.Rows.Add(dr);
        }
        return dt;
    }
    catch { return null; }
    finally
    {
        workbook.Close(false, oMissiong, oMissiong);
        System.Runtime.InteropServices.Marshal.ReleaseComObject(workbook);
        workbook = null;
        app.Workbooks.Close();
        app.Quit();
        System.Runtime.InteropServices.Marshal.ReleaseComObject(app);
        app = null;
    }
}

原文的方法二还提供了多线程处理数据的代码，一并复制到此(此处出现了一个SheetOptions的类型，无法考证其来源，如果知晓，请留言，谢谢。)：

/// <summary>
/// 使用COM，多线程读取Excel（1 主线程、4 副线程）
/// </summary>
/// <param name="excelFilePath">路径</param>
/// <returns>DataTabel</returns>
public System.Data.DataTable ThreadReadExcel(string excelFilePath)
{
    Excel.Application app = new Excel.Application();
    Excel.Sheets sheets = null;
    Excel.Workbook workbook = null;
    object oMissiong = System.Reflection.Missing.Value;
    System.Data.DataTable dt = new System.Data.DataTable();

    try
    {
        if (app == null)
        {
            return null;
        }

        workbook = app.Workbooks.Open(excelFilePath, oMissiong, oMissiong, oMissiong, oMissiong, oMissiong, oMissiong,
            oMissiong, oMissiong, oMissiong, oMissiong, oMissiong, oMissiong, oMissiong, oMissiong);

        //将数据读入到DataTable中——Start
        sheets = workbook.Worksheets;
        Excel.Worksheet worksheet = (Excel.Worksheet)sheets.get_Item();//读取第一张表
        if (worksheet == null)
            return null;

        string cellContent;
        int iRowCount = worksheet.UsedRange.Rows.Count;
        int iColCount = worksheet.UsedRange.Columns.Count;
        Excel.Range range;

        //负责列头Start
        DataColumn dc;
        int ColumnID = ;
        range = (Excel.Range)worksheet.Cells[, ];
        //while (range.Text.ToString().Trim() != "")
        while (iColCount >= ColumnID)
        {
            dc = new DataColumn();
            dc.DataType = System.Type.GetType("System.String");

            string strNewColumnName = range.Text.ToString().Trim();
            if (strNewColumnName.Length == ) strNewColumnName = "_1";
            //判断列名是否重复
            for (int i = ; i < ColumnID; i++)
            {
                if (dt.Columns[i - ].ColumnName == strNewColumnName)
                    strNewColumnName = strNewColumnName + "_1";
            }

            dc.ColumnName = strNewColumnName;
            dt.Columns.Add(dc);

            range = (Excel.Range)worksheet.Cells[, ++ColumnID];
        }
        //End  

        //数据大于500条，使用多进程进行读取数据
        if (iRowCount -  > )
        {
            //开始多线程读取数据
            //新建线程
            int b2 = (iRowCount - ) / ;
            DataTable dt1 = new DataTable("dt1");
            dt1 = dt.Clone();
            SheetOptions sheet1thread = new SheetOptions(worksheet, iColCount, , b2 + , dt1);
            Thread othread1 = new Thread(new ThreadStart(sheet1thread.SheetToDataTable));
            othread1.Start();

            //阻塞 1 毫秒，保证第一个读取 dt1
            Thread.Sleep();

            DataTable dt2 = new DataTable("dt2");
            dt2 = dt.Clone();
            SheetOptions sheet2thread = new SheetOptions(worksheet, iColCount, b2 + , b2 *  + , dt2);
            Thread othread2 = new Thread(new ThreadStart(sheet2thread.SheetToDataTable));
            othread2.Start();

            DataTable dt3 = new DataTable("dt3");
            dt3 = dt.Clone();
            SheetOptions sheet3thread = new SheetOptions(worksheet, iColCount, b2 *  + , b2 *  + , dt3);
            Thread othread3 = new Thread(new ThreadStart(sheet3thread.SheetToDataTable));
            othread3.Start();

            DataTable dt4 = new DataTable("dt4");
            dt4 = dt.Clone();
            SheetOptions sheet4thread = new SheetOptions(worksheet, iColCount, b2 *  + , b2 *  + , dt4);
            Thread othread4 = new Thread(new ThreadStart(sheet4thread.SheetToDataTable));
            othread4.Start();

            //主线程读取剩余数据
            for (int iRow = b2 *  + ; iRow <= iRowCount; iRow++)
            {
                DataRow dr = dt.NewRow();
                for (int iCol = ; iCol <= iColCount; iCol++)
                {
                    range = (Excel.Range)worksheet.Cells[iRow, iCol];
                    cellContent = (range.Value2 == null) ? "" : range.Text.ToString();
                    dr[iCol - ] = cellContent;
                }
                dt.Rows.Add(dr);
            }

            othread1.Join();
            othread2.Join();
            othread3.Join();
            othread4.Join();

            //将多个线程读取出来的数据追加至 dt1 后面
            foreach (DataRow dr in dt.Rows)
                dt1.Rows.Add(dr.ItemArray);
            dt.Clear();
            dt.Dispose();

            foreach (DataRow dr in dt2.Rows)
                dt1.Rows.Add(dr.ItemArray);
            dt2.Clear();
            dt2.Dispose();

            foreach (DataRow dr in dt3.Rows)
                dt1.Rows.Add(dr.ItemArray);
            dt3.Clear();
            dt3.Dispose();

            foreach (DataRow dr in dt4.Rows)
                dt1.Rows.Add(dr.ItemArray);
            dt4.Clear();
            dt4.Dispose();

            return dt1;
        }
        else
        {
            for (int iRow = ; iRow <= iRowCount; iRow++)
            {
                DataRow dr = dt.NewRow();
                for (int iCol = ; iCol <= iColCount; iCol++)
                {
                    range = (Excel.Range)worksheet.Cells[iRow, iCol];
                    cellContent = (range.Value2 == null) ? "" : range.Text.ToString();
                    dr[iCol - ] = cellContent;
                }
                dt.Rows.Add(dr);
            }
        }
        //将数据读入到DataTable中——End
        return dt;
    }
    catch
    {
        return null;
    }
    finally
    {
        workbook.Close(false, oMissiong, oMissiong);
        System.Runtime.InteropServices.Marshal.ReleaseComObject(workbook);
        System.Runtime.InteropServices.Marshal.ReleaseComObject(sheets);
        workbook = null;
        app.Workbooks.Close();
        app.Quit();
        System.Runtime.InteropServices.Marshal.ReleaseComObject(app);
        app = null;
        GC.Collect();
        GC.WaitForPendingFinalizers();
    }
}

原文还提供了第三种方法，感兴趣的可以关心一下：

方法三：NPOI方式读取Excel，NPOI是一组开源的组件，类似Java的 POI。包括：NPOI、NPOI.HPSF、NPOI.HSSF、NPOI.HSSF.UserModel、NPOI.POIFS、NPOI.Util，下载的时候别只下一个噢

优点：读取Excel速度较快，读取方式操作灵活性

缺点：只支持03的Excel，xlsx的无法读取。由于这点，使用这种方式的人不多啊，没理由要求客户使用03版Excel吧，再说03版Excel对于行数还有限制，只支持65536行。

（听他们的开发人员说会在2012年底推出新版，支持xlsx的读取。但一直很忙没时间去关注这个事情，有兴趣的同学可以瞧瞧去）

using System;
using System.Data;
using System.IO;
using System.Web;
using NPOI;
using NPOI.HPSF;
using NPOI.HSSF;
using NPOI.HSSF.UserModel;
using NPOI.POIFS;
using NPOI.Util;
using System.Text;
using System.Configuration;

public class NPOIHelper
{
    private static int ExcelMaxRow = Convert.ToInt32(ConfigurationManager.AppSettings["ExcelMaxRow"]);
    /// <summary>
    /// 由DataSet导出Excel
    /// </summary>
    /// <param name="sourceTable">要导出数据的DataTable</param>
    /// <param name="sheetName">工作表名称</param>
    /// <returns>Excel工作表</returns>
    private static Stream ExportDataSetToExcel(DataSet sourceDs)
    {
        HSSFWorkbook workbook = new HSSFWorkbook();
        MemoryStream ms = new MemoryStream();

        for (int i = ; i < sourceDs.Tables.Count; i++)
        {
            HSSFSheet sheet = (HSSFSheet)workbook.CreateSheet(sourceDs.Tables[i].TableName);
            HSSFRow headerRow = (HSSFRow)sheet.CreateRow();
            // handling header.
            foreach (DataColumn column in sourceDs.Tables[i].Columns)
                headerRow.CreateCell(column.Ordinal).SetCellValue(column.ColumnName);
            // handling value.
            int rowIndex = ;
            foreach (DataRow row in sourceDs.Tables[i].Rows)
            {
                HSSFRow dataRow = (HSSFRow)sheet.CreateRow(rowIndex);
                foreach (DataColumn column in sourceDs.Tables[i].Columns)
                {
                    dataRow.CreateCell(column.Ordinal).SetCellValue(row[column].ToString());
                }
                rowIndex++;
            }
        }
        workbook.Write(ms);
        ms.Flush();
        ms.Position = ;
        workbook = null;
        return ms;
    }
    /// <summary>
    /// 由DataSet导出Excel
    /// </summary>
    /// <param name="sourceTable">要导出数据的DataTable</param>
    /// <param name="fileName">指定Excel工作表名称</param>
    /// <returns>Excel工作表</returns>
    public static void ExportDataSetToExcel(DataSet sourceDs, string fileName)
    {
        //检查是否有Table数量超过65325
        for (int t = ; t < sourceDs.Tables.Count; t++)
        {
            if (sourceDs.Tables[t].Rows.Count > ExcelMaxRow)
            {
                DataSet ds = GetdtGroup(sourceDs.Tables[t].Copy());
                sourceDs.Tables.RemoveAt(t);
                //将得到的ds插入 sourceDs中
                for (int g = ; g < ds.Tables.Count; g++)
                {
                    DataTable dt = ds.Tables[g].Copy();
                    sourceDs.Tables.Add(dt);
                }
                t--;
            }
        }

        MemoryStream ms = ExportDataSetToExcel(sourceDs) as MemoryStream;
        HttpContext.Current.Response.AppendHeader("Content-Disposition", "attachment;filename=" + fileName);
        HttpContext.Current.Response.BinaryWrite(ms.ToArray());
        HttpContext.Current.ApplicationInstance.CompleteRequest();
        //HttpContext.Current.Response.End();
        ms.Close();
        ms = null;
    }
    /// <summary>
    /// 由DataTable导出Excel
    /// </summary>
    /// <param name="sourceTable">要导出数据的DataTable</param>
    /// <returns>Excel工作表</returns>
    private static Stream ExportDataTableToExcel(DataTable sourceTable)
    {
        HSSFWorkbook workbook = new HSSFWorkbook();
        MemoryStream ms = new MemoryStream();
        HSSFSheet sheet = (HSSFSheet)workbook.CreateSheet(sourceTable.TableName);
        HSSFRow headerRow = (HSSFRow)sheet.CreateRow();
        // handling header.
        foreach (DataColumn column in sourceTable.Columns)
            headerRow.CreateCell(column.Ordinal).SetCellValue(column.ColumnName);
        // handling value.
        int rowIndex = ;
        foreach (DataRow row in sourceTable.Rows)
        {
            HSSFRow dataRow = (HSSFRow)sheet.CreateRow(rowIndex);
            foreach (DataColumn column in sourceTable.Columns)
            {
                dataRow.CreateCell(column.Ordinal).SetCellValue(row[column].ToString());
            }
            rowIndex++;
        }
        workbook.Write(ms);
        ms.Flush();
        ms.Position = ;
        sheet = null;
        headerRow = null;
        workbook = null;
        return ms;
    }
    /// <summary>
    /// 由DataTable导出Excel
    /// </summary>
    /// <param name="sourceTable">要导出数据的DataTable</param>
    /// <param name="fileName">指定Excel工作表名称</param>
    /// <returns>Excel工作表</returns>
    public static void ExportDataTableToExcel(DataTable sourceTable, string fileName)
    {
        //如数据超过65325则分成多个Table导出
        if (sourceTable.Rows.Count > ExcelMaxRow)
        {
            DataSet ds = GetdtGroup(sourceTable);
            //导出DataSet
            ExportDataSetToExcel(ds, fileName);
        }
        else
        {
            MemoryStream ms = ExportDataTableToExcel(sourceTable) as MemoryStream;
            HttpContext.Current.Response.AppendHeader("Content-Disposition", "attachment;filename=" + fileName);
            HttpContext.Current.Response.BinaryWrite(ms.ToArray());
            HttpContext.Current.ApplicationInstance.CompleteRequest();
            //HttpContext.Current.Response.End();
            ms.Close();
            ms = null;
        }
    }

    /// <summary>
    /// 传入行数超过65325的Table，返回DataSet
    /// </summary>
    /// <param name="dt"></param>
    /// <returns></returns>
    public static DataSet GetdtGroup(DataTable dt)
    {
        string tablename = dt.TableName;

        DataSet ds = new DataSet();
        ds.Tables.Add(dt);

        double n = dt.Rows.Count / Convert.ToDouble(ExcelMaxRow);

        //创建表
        for (int i = ; i < n; i++)
        {
            DataTable dtAdd = dt.Clone();
            dtAdd.TableName = tablename + "_" + i.ToString();
            ds.Tables.Add(dtAdd);
        }

        //分解数据
        for (int i = ; i < ds.Tables.Count; i++)
        {
            //新表行数达到最大 或 基表数量不足
            while (ds.Tables[i].Rows.Count != ExcelMaxRow && ds.Tables[].Rows.Count != ExcelMaxRow)
            {
                ds.Tables[i].Rows.Add(ds.Tables[].Rows[ExcelMaxRow].ItemArray);
                ds.Tables[].Rows.RemoveAt(ExcelMaxRow);

            }
        }

        return ds;
    }

    /// <summary>
    /// 由DataTable导出Excel
    /// </summary>
    /// <param name="sourceTable">要导出数据的DataTable</param>
    /// <param name="fileName">指定Excel工作表名称</param>
    /// <returns>Excel工作表</returns>
    public static void ExportDataTableToExcelModel(DataTable sourceTable, string modelpath, string modelName, string fileName, string sheetName)
    {
        int rowIndex = ;//从第二行开始，因为前两行是模板里面的内容
        int colIndex = ;
        FileStream file = new FileStream(modelpath + modelName + ".xls", FileMode.Open, FileAccess.Read);//读入excel模板
        HSSFWorkbook hssfworkbook = new HSSFWorkbook(file);
        HSSFSheet sheet1 = (HSSFSheet)hssfworkbook.GetSheet("Sheet1");
        sheet1.GetRow().GetCell().SetCellValue("excelTitle");      //设置表头
        foreach (DataRow row in sourceTable.Rows)
        {   //双循环写入sourceTable中的数据
            rowIndex++;
            colIndex = ;
            HSSFRow xlsrow = (HSSFRow)sheet1.CreateRow(rowIndex);
            foreach (DataColumn col in sourceTable.Columns)
            {
                xlsrow.CreateCell(colIndex).SetCellValue(row[col.ColumnName].ToString());
                colIndex++;
            }
        }
        sheet1.ForceFormulaRecalculation = true;
        FileStream fileS = new FileStream(modelpath + fileName + ".xls", FileMode.Create);//保存
        hssfworkbook.Write(fileS);
        fileS.Close();
        file.Close();
    }
}