SQL 服务器：在单个列上进行部分分组

Question

当天早些时候已经问过类似的问题，解决方案将我引向另一个问题（在示例之后）：

我有这些列，其中两列是数量。
其中一个（LotQty）是一个总和，按Productid和Lot分组
另一个 (SumQuantity) 必须是仅按 Productid 分组的前者的总和。

结果应如下所示：

SumQuantity    Productid  LotQty        Lot
----------------------------------------------------
 512           40652      256.000000    2020-12-20
 512           40652      256.000000    2020-12-21
1024           40661      512.000000    2019-12-19
1024           40661      512.000000    2019-12-20
 512           40710      256.000000    2021-03-03
 512           40710      256.000000    2021-04-04

即SumQuantity = sum(LotQty) 按 productid 分组，而 sum(LotQty) 按 productid AND lot 分组。

select 
    sum(sum(s.cuquantity)) over () SumQuantity,
    s.productid, sum(s.cuquantity) LotQty, la.Value Lot 
from 
    log l 
left join 
    logstock s on s.logid = l.id
left join 
    Logstockattributes la on la.LogStockID = s.id and la.AttributeID = 10
where  
    l.receiptid = 5950195
group by 
    productid, la.value

结果是：

SumQuantity  Productid    LotQty        Lot
---------------------------------------------------
2048         40652        256.000000    2020-12-20
2048         40652        256.000000    2020-12-21
2048         40661        512.000000    2019-12-19
2048         40661        512.000000    2019-12-20
2048         40710        256.000000    2021-03-03
2048         40710        256.000000    2021-04-04

样本table

  Logid        Productid      Cuquantity    Lot
  -----------------------------------------------------
  1            40652          256.000000    2020-12-20
  2            40652          255.000000    2020-12-21
  3            40652            1.000000    2020-12-21
  4            40661          512.000000    2019-12-19
  5            40661          512.000000    2019-12-20
  6            40710          256.000000    2021-03-03
  7            40710          255.000000    2021-04-04
  8            40710            1.000000    2021-04-04

我应该如何更改我的 select 以获得我正在寻找的结果？

Answer 1

您只需添加分区并删除多余的 SUM。本质上，您需要 window 函数而不是完整的聚合方法，以便您可以 return 行级详细信息。

select distinct
    SumQuantity = sum(Cuquantity) over (partition by Productid),
    LotQty = sum(Cuquantity) over (partition by Productid, Lot),
    Productid,
    Lot
from log l 
left join logstock s on s.logid=l.id
left join Logstockattributes la on la.LogStockID=s.id and la.AttributeID=10
where l.receiptid=5950195
la.value

Answer 2

我想你只需要 partition by:

select sum(sum(s.cuquantity)) over (partition by productid) as SumQuantity,
       s.productid, sum(s.cuquantity) as LotQty, la.Value as Lot 
from log l left join
     logstock s
     on s.logid = l.id left join
     Logstockattributes la
     on la.LogStockID = s.id and la.AttributeID = 10
where l.receiptid = 5950195
group by productid,la.value

SQL 服务器：在单个列上进行部分分组

SQL Server : partial group by on single column

sql

sql-server

group-by

sum

partial