快捷導(dǎo)航

C# DataTable數(shù)據(jù)遍歷優(yōu)化詳解

更新時(shí)間：2021年01月19日 15:08:39 作者：程序員小站

這篇文章主要介紹了C# DataTable數(shù)據(jù)遍歷優(yōu)化詳解，具有很好的參考價(jià)值，希望對(duì)大家有所幫助。一起跟隨小編過來看看吧

我們?cè)谶M(jìn)行開發(fā)時(shí)，會(huì)經(jīng)常使用DataTable來存儲(chǔ)和操作數(shù)據(jù)，我發(fā)現(xiàn)在遍歷DataTable并對(duì)數(shù)據(jù)進(jìn)行刪除和添加操作時(shí)速度非常慢，查閱相關(guān)資料并測(cè)試在添加主鍵后可以使遍歷和操作速度提高很多：

測(cè)試代碼，測(cè)試的是我們向取出來數(shù)據(jù)滿足Flag！=1條件的所有數(shù)據(jù)的后面添加一條數(shù)據(jù)(因?yàn)檫@條數(shù)據(jù)的一些字段值是根據(jù)前面的幾條滿足條件[“AccID='” + accID + “' AND Y='” + year + “' AND AbsID <= ” + absID;]數(shù)據(jù)的值累加得到的)所以需要進(jìn)行整個(gè)DataTable的遍歷來計(jì)算添加：

public static void Test2()
{
 Stopwatch watch = new Stopwatch();
 using (DbConnection conn = SqlHelper.GetConnection("ConnectionString"))
 {
  using (SqlCommand cmd = new SqlCommand())
  {
   watch.Start();
   cmd.CommandText = string.Format(@"
select ROW_NUMBER() OVER (Order by S.AccID,S.CurrID,S.AbsID,S.Flag)AS RowNum,S.* from Test S 
");
   cmd.Connection = conn as SqlConnection;
   cmd.CommandTimeout = 60000;
   conn.Open();
   DataTable table = ExecuteDataTable(cmd);
   watch.Stop();
   Console.WriteLine("從數(shù)據(jù)庫(kù)取出數(shù)據(jù){0}條", table.Rows.Count);
   Stopwatch watch2 = new Stopwatch();
   watch2.Start();
   DataTable newTable = HandleAccYear(table,true);
   watch2.Stop();
   Console.WriteLine("數(shù)據(jù){0},遍歷操作時(shí)間：毫秒：{1}，秒：{2}", newTable.Rows.Count, watch2.ElapsedMilliseconds, watch2.ElapsedMilliseconds / 1000);
  }
  conn.Close();
 }
}

填充數(shù)據(jù)到DataTable的方法

public static DataTable ExecuteDataTable(SqlCommand cmd)
{
  DataTable table = new DataTable();
  SqlDataAdapter adaper = new SqlDataAdapter(cmd);
  adaper.Fill(table);
  return table;
}

private static DataTable HandleAccYear(DataTable dt, bool isCurrency)
{
  DataTable newdt = dt.Clone();
  //不使用主鍵
  //dt.PrimaryKey = new DataColumn[] {
  // dt.Columns["AccID"],
  // dt.Columns["Flag"],
  // dt.Columns["AbsID"],
  // dt.Columns["RowNum"],
  //};
  if (dt.Rows.Count > 0)
  {
   object flag = null;
   foreach (DataRow row in dt.Rows)
   {
    flag = row["Flag"];
    if (flag != null && !Helper.AreEqual(flag.ToString(), "1"))
    {
     DataRow newRow = newdt.NewRow();
     DataRow sourceRow = newdt.NewRow();
     sourceRow.ItemArray = row.ItemArray;
     newRow.ItemArray = row.ItemArray;
     string accID = row["AccID"].ToString(),
      year = row["Y"].ToString(),
      absID = row["AbsID"].ToString();
     newRow["Flag"] = "5";
     newRow["SumInfo"] = "測(cè)試數(shù)據(jù)";
     string filter = "AccID='" + accID + "' AND Y='" + year + "' AND AbsID <= " + absID;
     if (!isCurrency)
     {
      filter = "AccID='" + accID + "'AND CurrID='" + row["CurrID"] + "' AND Y='" + year + "' AND AbsID <= " + absID;
     }
     DataRow[] selectRow = dt.Select(filter);
     double debitLC = 0, debitQty = 0, creditLC = 0, creditQty = 0, debitFC = 0, creditFC = 0;
     foreach (DataRow item in selectRow)
     {
      debitLC += ToDouble(item["YearDebitLC"]);
      debitQty += ToDouble(item["YearDebitQty"]);
      creditLC +=ToDouble(item["YearCreditLC"]);
      creditQty += ToDouble(item["YearCreditQty"]);
      if (!isCurrency)
      {
       debitFC += ToDouble(item["YearDebitFC"]);
       creditFC += ToDouble(item["YearCreditFC"]);
      }
     }
     newRow["CurDebitLC"] = debitLC;
     newRow["CurDebitQty"] = debitQty;
     newRow["CurCreditLC"] = creditLC;
     newRow["CurCreditQty"] = creditQty;
     //newRow["CurDebitLC"] = dt.Compute("Sum(YearDebitLC)", filter);
     //newRow["CurDebitQty"] = dt.Compute("Sum(YearDebitQty)", filter);
     //newRow["CurCreditLC"] = dt.Compute("Sum(YearCreditLC)", filter);
     //newRow["CurCreditQty"] = dt.Compute("Sum(YearCreditQty)", filter);
     if (!isCurrency)
     {
      //newRow["CurCreditFC"] = dt.Compute("Sum(YearCreditFC)", filter);
      //newRow["CurDebitFC"] = dt.Compute("Sum(YearDebitFC)", filter);
      newRow["CurCreditFC"] = creditFC;
      newRow["CurDebitFC"] = debitFC;
     }
     newdt.Rows.Add(sourceRow);
     newdt.Rows.Add(newRow);
    }
    else
    {
     DataRow sourceRow = newdt.NewRow();
     sourceRow.ItemArray = row.ItemArray;
     newdt.Rows.Add(sourceRow);
    }
   }
  }
  return newdt;
 }

當(dāng)不使用主鍵進(jìn)行遍歷計(jì)算插入相應(yīng)的值時(shí)所用時(shí)間竟然是這么多：

當(dāng)我使用同樣的方法，同樣的數(shù)據(jù)添加主鍵(即把HandleAccYear方法中不使用主鍵下面的注釋去掉后).進(jìn)行遍歷計(jì)算等操作，得出的結(jié)果竟然有這么大的差別：

補(bǔ)充：C# DataTable數(shù)據(jù)量大，循環(huán)處理數(shù)據(jù)的時(shí)候優(yōu)化速度

相信大家用for循環(huán)datatable數(shù)據(jù)的不會(huì)太少，這個(gè)在數(shù)據(jù)量比較小的時(shí)候可以接受，但是數(shù)據(jù)量大的時(shí)候卻會(huì)造成CPU占用過高，甚至把電腦資源耗盡卡死至無限等待，

其實(shí)一些循環(huán)耗時(shí)的操作可以用線程池分塊來處理，這樣會(huì)減輕CPU很多壓力，好比食堂打飯，當(dāng)只有一個(gè)窗口的時(shí)候勢(shì)必等待的時(shí)間會(huì)非常的長(zhǎng)，但是多開幾個(gè)窗口的時(shí)候卻大大提高效率，

C#中用線程池就可以做到，本來一開始的時(shí)候我用的是為每個(gè)區(qū)塊開一個(gè)線程，但是有一個(gè)問題就是開了那么多的線程沒辦法結(jié)束他們，后來我想到了線程池，

具體代碼如下：

int sid = dt.Rows.Count % 100 == 0 ? (dt.Rows.Count / 100) : (dt.Rows.Count / 100 + 1);
    for (int a = 1; a <= sid; a++)
    {
     object aa=a.ToString() + "," + sid.ToString();
     ThreadPool.QueueUserWorkItem(todo
      , aa);
    }
 public void todo(object aa)
 {
  string sql = "";
  int startindex = Convert.ToInt32(aa.ToString().Split(',')[0]);
  int limitstep = Convert.ToInt32(aa.ToString().Split(',')[1]);
  for (int i = (startindex > 1 ? ((startindex - 1) * 100) : 0); i < (startindex == limitstep ? (dt.Rows.Count) : startindex*100); i++)
  {
   //todo數(shù)據(jù)操作
  }
  Thread.Sleep(2000);
 }

以上為個(gè)人經(jīng)驗(yàn)，希望能給大家一個(gè)參考，也希望大家多多支持腳本之家。如有錯(cuò)誤或未考慮完全的地方，望不吝賜教。

您可能感興趣的文章: