Question

我有一个复杂的（好吧，对我来说）问题。这是简短的版本。也许如果我能从这一小部分得到一些帮助，我将不需要更大部分的帮助。

无论如何，假设我有一张包含几个样本行的表格：

CustNum - CustName - Salesman - Address - SoldTotal - SaleDate
a500 - bubby's - 03 - 123 Street - $254 - 03-13-2013
a500 - bubby's - 22 - 123 Street - $996 - 05-12-2013
a500 - bubby's - 13 - 123 Street - $1001 - 09-20-2013
b100 - squirrely's - 14 - 456 Street - $460 - 03-14-2013

老板想要的是一个列表，其中包含每位买家的一行，最近的推销员和总销售额。上面的理想输出是（日期可以用于比较目的，但他不想在报告中使用它）：

CustNum - CustName - Salesman - Address - SoldTotal 
a500 - bubby's - 13 - 123 Street - $2251 
b100 - squirrely's - 14 - 456 Street - $460

如何最好地完成？我尝试自我加入尝试比较日期，但我得到了算术溢出。我必须弄错了。有任何想法吗？

Answer 1

;WITH x AS 
(
  SELECT CustNum, CustName, SalesMan, Address, 
    s = SUM(SoldTotal) OVER (PARTITION BY CustNum),
    r = ROW_NUMBER() OVER (PARTITION BY CustNum ORDER BY SaleDate DESC)
  FROM dbo.YourTable
)
SELECT CustNum, CustName, SalesMan, Address, SoldTotal = s
  FROM x
  WHERE r = 1;

SQL Fiddle demo

Answer 2

以下是使用行号和求和窗口函数的示例：

with SampleData (CustNum, CustName, Salesman, Address,  SoldTotal, SaleDate ) as (
    select 'a500', 'bubbys', '03', '123 Street', 254, '03-13-2013' union all
    select 'a500', 'bubbys', '03', '123 Street', 996, '05-12-2013' union all
    select 'b100', 'squirrelys', '14', '456 Street', 460, '03-14-2013'
)
select CustNum, CustName, MostRecentSalesman, SoldTotal
from (
    select
        CustNum
    ,   CustName
    ,   MostRecentSalesman = Salesman
    ,   rn = row_number() over (partition by CustNum order by SaleDate desc)
    ,   SoldTotal = sum(soldTotal) over (partition by CustNum)
    from SampleData
) x
where x.rn = 1

最初的“with SampleData ...”部分只是要用一些示例数据来编写查询。

SQL根据最新日期选择行，包括较旧的总和？

2 个答案: