C#用DataTable实现Groupby

http://hi.baidu.com/vhion/blog/item/8b4e23d3dd6b1f143af3cf83.html

C#用DataTable实现Groupby

2011-03-07 11:33

写程序的应该都知道SQL中group by的作用，但是有时候可能会遇到无法在SQL中group by的情况，我刚好就碰到了这样的情况。于是就尝试了下在DataTable中对数据进行分组计算的方法，有两种解决办法。
      1.用两层循环计算，前提条件是数据必须是要按分组的列排好序的，如下：

            DataTable dt = new DataTable();
            dt.Columns.AddRange(new DataColumn[] { new DataColumn("name", typeof(string)),
                                                   new DataColumn("score", typeof(int)) });

            dt.Rows.Add(new object[] { "张三", 1 });
            dt.Rows.Add(new object[] { "张三", 4 });
            dt.Rows.Add(new object[] { "李四", 100 });
            dt.Rows.Add(new object[] { "李四", 90 });
            dt.Rows.Add(new object[] { "王五", 77 });

            DataTable dtLwolf = dt.Clone();

            for (int i = 0; i < dt.Rows.Count; )
            {
                DataRow dr = dtLwolf.NewRow();
                string name = dt.Rows[i]["name"].ToString();
                dr["name"] = name;

                int score = 0;
        //内层也是循环同一个表，当遇到不同的name时跳出内层循环
                for (; i < dt.Rows.Count; )
                {
                    if (name == dt.Rows[i]["name"].ToString())
                    {
                        score += Convert.ToInt32(dt.Rows[i]["score"]);
                        dr["score"] = score;
                        i++;
                    }
                    else
                    {
                        break;
                    }
                }
                dtLwolf.Rows.Add(dr);
            }

dtLwolf中的数据即是最终结果

      2.借助DataTable的Compute方法，DataTable中数据不用事先排好序。
        下面代码中的dt是跟前面的是一样的

            DataTable dtName = dt.DefaultView.ToTable(true, "name");
            for (int i = 0; i < dtName.Rows.Count; i++)
            {
                DataRow[] rows = dt.Select("name='" + dtName.Rows[i][0] + "'");
                //temp用来存储筛选出来的数据
                DataTable temp = dtLwolf.Clone();
                foreach (DataRow row in rows)
                {
                    temp.Rows.Add(row.ItemArray);
                }

                DataRow dr = dtLwolf.NewRow();
                dr[0] = dtName.Rows[i][0].ToString();
                dr[1] = temp.Compute("sum(score)", "");
                dtLwolf.Rows.Add(dr);
            }