如何获得每组的最小值和日期？

Question

我有两个相关的 MySQL 表。我的目的是获取每个组的最小值以及与找到的每个最小值相关联的日期。

                              TABLE 1
> ------------+--------------+--------------+-----------+-----------------+
> --  Code -- | - IdGroup1 - | - IdGroup2 - | - State - | - NameProtocol -|
> ------------+--------------+--------------+-----------+-----------------+
>    ZZ-100   |   11111111   |    1110000   |     1     |       OSM1      |
> ------------+--------------+--------------+-----------+-----------------+
>    ZZ-200   |   55555555   |    5550000   |     1     |       OSM1      |
> ------------+--------------+--------------+-----------+-----------------+
>    ZZ-300   |   99999999   |    9990000   |     1     |       OSM1      |

表 1 和表 3 是相关的。

                             TABLE 3
> ---------------------+-------------------+----------------+
> --  NameProtocol  -- | -- Description -- | -- Protocol -- |
> ---------------------+-------------------+----------------+
>         ATC0         |        d1         |      UDP       |
> ---------------------+-------------------+----------------+
>         OSM1         |        d2         |      TCP       |
> ---------------------+-------------------+----------------+

表 2 是记录新值的位置。

                                TABLE 2
> ---------+-----------------------+----------------+----------------+
> - Value -| ------- Date -------- | -- IdGroup1 -- | -- IdGroup2 -- |
> ---------+-----------------------+----------------+----------------+
>    10    |  2020-08-16 02:30:10  |    99999999    |     9990000    |
> ---------+-----------------------+----------------+----------------+
>    15    |  2020-08-16 02:31:10  |    99999999    |     9990000    |
> ---------+-----------------------+----------------+----------------+
>    20    |  2020-08-16 02:32:10  |    99999999    |     9990000    |
> ---------+-----------------------+----------------+----------------+
>    115   |  2020-08-16 02:31:20  |    55555555    |     5550000    |
> ---------+-----------------------+----------------+----------------+
>    120   |  2020-08-16 02:32:20  |    55555555    |     5550000    |
> ---------+-----------------------+----------------+----------------+
>    90    |  2020-08-16 02:35:20  |    11111111    |     1110000    |
> ---------+-----------------------+----------------+----------------+
>    100   |  2020-08-16 02:30:20  |    11111111    |     1110000    |

我已经做了很多测试，我仍然设法得到正确的答案，最好的近似值是通过以下查询获得的：

var query = Table2                                              //Outer Table
            .Join(Table1,                                       //Inner Table to join
                         p => new { p.IdGroup1, p.IdGroup2 },   //Condition from outer table
                         e => new { e.IdGroup1, e.IdGroup2 },   //Condition from inner table
                         (p, e) => new {                        //Result
                                          Code = e.Code,                                          
                                          Value = p.Value,
                                          Date = p.Date })         
            .GroupBy(gb => new { gb.Code })           
            .OrderBy(ob => ob.Key.Code)
            .Select(s => new {  Code = s.Key.Code, 
                                Value = (double?)s.Min(a => a.Value),
                                Date = "?" })  // TODO: The date remains to be implemented.
            .ToList();

我的查询结果：

> -------------+-------------+------------+
> --  Code  -- | -- Value -- | -- Date -- | 
> -------------+-------------+------------+
>    ZZ-100    |      90     |     ?      | 
> -------------+-------------+------------+
>    ZZ-200    |     115     |     ?      |
> -------------+-------------+------------+
>    ZZ-300    |      10     |     ?      |

我只需要为找到的每个最小值添加日期。我应该怎么做才能将它集成到我的查询中？

代码：

var query = Table1                      
        .Join(Table2.Where(w => (w.State == 1)),            
                         h => new { h.IdGroup1, h.IdGroup2 },   
                         p => new { p.IdGroup1, p.IdGroup2 },   
                         (h, p) => new { h, p })                    
        .Join(Table3.Where(w => (w.Protocol == "TCP")),
                                     pt => pt.p.NameProtocol,
                                     p => p.NameProtocol,
                                     (pt, p) => new { pt, p })          
        .GroupBy(gb => new { gb.pt.p.Code })            
        .OrderBy(ob => ob.Key.Code)
        .Select(s => new {  Code = s.Key.Code, 
                            Value = (double?)s.Min(a => a.pt.h.Value) })
        .ToList();

Answer 1

对此类请求最有效的查询是使用 Window Functions，EF 不支持，我认为这永远不会发生。 所以只需使用 SQL 并通过 Dapper 运行它，无论如何。

SELECT 
    s.Code,
    s.Value,
    s.Date
FROM
    (
        SELECT 
           t1.Code,
           t2.Value,
           t2.Date,
           ROW_NUMBER() OVER (PARTITION BY t1.Code ORDER BY t2.Value) AS RN
        FROM TABLE1 t1
        JOIN TBALE3 t3 ON t3.NameOfProtocol = t1.NameOfProtocol
        LEFT JOIN TABLE2 t2 ON t1.IdGroup1 = t2.IdGroup1 AND t1.IdGroup2 = t2.IdGroup2 AND t2.FechaCaudalHistorico <= @dateFilter
        WHERE t3.Protocol = 'TCP'
    ) s
WHERE s.RN = 1

如果您不是纯 EF Core 专家并且仍然需要 LINQ，您可以尝试linq2db.EntityFrameworkCore 扩展，它具有这种可能性，并且可以通过 LINQ 编写查询：

var dateFilter = DateTime.Parse ("2020-09-16 03:00:00");
var rnQuery =
   from t1 in Table1
   join t3 in Table3 on t1.NameOfProtocol equals t3.NameOfProtocol
   from t2 in Table2.Where(t2 => t1.IdGroup1 == t2.IdGroup1 && t1.IdGroup2 == t2.IdGroup2 && t2.FechaCaudalHistorico <= dateFilter)
      .DefaultIfEmpty()
   where t3.Protocol == "TCP"
   select new 
   {
      t1.Code,
      Value = Sql.ToNullable(t2.Value),
      Date  = Sql.ToNullable(t2.Date),
      RN    = Sql.Ext.RowNumber().Over().PartitionBy(t1.Code).OrderBy(t2.Value).ToValue()
   };

var query = from s in rnQuery
   where s.RN == 1
   select new 
   {
      s.Code,
      s.Value,
      s.Date,
   };

// switch to alternative LINQ parser
query = query.ToLinqToDB();

var result = query.ToList();

因此，您将拥有与上述相同的 SQL。

Answer 2

您可以尝试以下方法：

.Select(s => 
{
    var values = s.OrderBy(x=>x.Value);
    var firstValue = values.First();
    return new 
    {
        Code = s.Key.Code;
        Value = (double?)firstValue.Value;
        Date = firstValue.Date;
    }
})

基本上，我们根据Value 属性对每个组中的项目进行排序。具有最小值的元素将是排序之后的第一个元素。然后我们选择那个元素并读取它的值和日期，我们就完成了。

更新

一个快速的解决方案是在Join 之后调用ToList。这会将所有数据带入应用程序的内存中，您将在那里执行所有需要的处理。

.Join(Table1
      , p => new { p.IdGroup1, p.IdGroup2 }
      , e => new { e.IdGroup1, e.IdGroup2 }
      , (p, e) => new 
        {
           Code = e.Code,                                          
           Value = p.Value,
           Date = p.Date 
    })
   .ToList()  
   .GroupBy(gb => new { gb.Code })
   .OrderBy(ob => ob.Key.Code)
   .Select(s => 
   {
       var values = s.OrderBy(x=>x.Value);
       var firstValue = values.First();
       return new 
       {
           Code = s.Key.Code;
           Value = (double?)firstValue.Value;
           Date = firstValue.Date;
       }
   }).ToList();

Answer 3

你可以这样做：

var query = Table2                                              //Outer Table
    .Join(Table1,                                       //Inner Table to join
        p => new { p.IdGroup1, p.IdGroup2 },   //Condition from outer table
        e => new { e.IdGroup1, e.IdGroup2 },   //Condition from inner table
        (p, e) => new
        {                        //Result
            Code = e.Code,
            Value = p.Value,
            Date = p.Date
        })
    .GroupBy(gb => new { gb.Code })
    .OrderBy(ob => ob.Key.Code)
    .Select(s =>
    {
        var min = s.FirstOrDefault(x=> x.Value == s.Min(a => a.Value));
        return new {Code = s.Key.Code, Value = (double?)min.Value, Date = min.Date};
    })  // TODO: The date remains to be implemented.
    .ToList();