- 背景:老系统asp.net 2.0项目使用客户反应,某个业务每个月导入数据操作很慢,大致需要15-30分钟才会导入完毕;
- 分析:导入慢的原因:
.数据量过大,且采用的是同步,单个excel sheet 13万+数据;
.导入前验证,每行某列 验证,频繁和数据库交互
.使用很老的微软企业库进行批量插入操作,效率低下 ,插入后,又批量进行执行sql修改操作 ;
3 解决方案:
修改excel转table 的方案由OpenXMLHelper 转换变为 NPOI;
1 public class NPOIHelper 2 { 3 /// <summary> 4 /// 将excel导入到datatable 5 /// </summary> 6 /// <param name="filePath">excel路径</param> 7 /// <param name="isColumnName">第一行是否是列名</param> 8 /// <returns>返回datatable</returns> 9 public DataTable ExcelToDataTable(string filePath, bool isColumnName) 10 { 11 DataTable dataTable = null; 12 FileStream fs = null; 13 DataColumn column = null; 14 DataRow dataRow = null; 15 IWorkbook workbook = null; 16 ISheet sheet = null; 17 IRow row = null; 18 ICell cell = null; 19 int startRow = 0; 20 try 21 { 22 using (fs = File.OpenRead(filePath)) 23 { 24 // 2007版本 25 if (filePath.IndexOf(".xlsx") > 0) 26 workbook = new XSSFWorkbook(fs); 27 // 2003版本 28 else if (filePath.IndexOf(".xls") > 0) 29 workbook = new HSSFWorkbook(fs); 30 31 if (workbook != null) 32 { 33 sheet = workbook.GetSheetAt(0);//读取第一个sheet,当然也可以循环读取每个sheet 34 dataTable = new DataTable(); 35 if (sheet != null) 36 { 37 int rowCount = sheet.LastRowNum;//总行数 38 if (rowCount > 0) 39 { 40 IRow firstRow = sheet.GetRow(0);//第一行 41 int cellCount = firstRow.LastCellNum;//列数 42 43 //构建datatable的列 44 if (isColumnName) 45 { 46 startRow = 1;//如果第一行是列名,则从第二行开始读取 47 for (int i = firstRow.FirstCellNum; i < cellCount; ++i) 48 { 49 cell = firstRow.GetCell(i); 50 if (cell != null) 51 { 52 if (cell.StringCellValue != null) 53 { 54 column = new DataColumn(cell.StringCellValue); 55 dataTable.Columns.Add(column); 56 } 57 } 58 } 59 } 60 else 61 { 62 for (int i = firstRow.FirstCellNum; i < cellCount; ++i) 63 { 64 column = new DataColumn("column" + (i + 1)); 65 dataTable.Columns.Add(column); 66 } 67 } 68 69 //填充行 70 for (int i = startRow; i <= rowCount; ++i) 71 { 72 row = sheet.GetRow(i); 73 if (row == null) continue; 74 75 dataRow = dataTable.NewRow(); 76 for (int j = row.FirstCellNum; j < cellCount; ++j) 77 { 78 cell = row.GetCell(j); 79 if (cell == null) 80 { 81 dataRow[j] = ""; 82 } 83 else 84 { 85 //CellType(Unknown = -1,Numeric = 0,String = 1,Formula = 2,Blank = 3,Boolean = 4,Error = 5,) 86 switch (cell.CellType) 87 { 88 case CellType.BLANK: 89 dataRow[j] = ""; 90 break; 91 case CellType.NUMERIC: 92 short format = cell.CellStyle.DataFormat; 93 //对时间格式(2015.12.5、2015/12/5、2015-12-5等)的处理 94 if (format == 14 || format == 31 || format == 57 || format == 58) 95 dataRow[j] = cell.DateCellValue; 96 else 97 dataRow[j] = cell.NumericCellValue; 98 break; 99 case CellType.STRING: 100 dataRow[j] = cell.StringCellValue; 101 break; 102 } 103 } 104 } 105 if (dataRow == null) Logger.Write(string.Format("转换行失败,行数为:{0}", i.ToString())); 106 dataTable.Rows.Add(dataRow); 107 } 108 } 109 else 110 { 111 Logger.Write(string.Format("转换datarow完毕,行数:{0}", rowCount.ToString())); 112 } 113 } 114 } 115 else 116 { 117 Logger.Write("转换workbook 为空"); 118 } 119 } 120 return dataTable; 121 } 122 catch (Exception ex) 123 { 124 Logger.Write(string.Format("转换失败,异常:{0}", ex.ToString())); 125 if (fs != null) 126 { 127 fs.Close(); 128 } 129 return null; 130 } 131 } 132 }