Question

我有一个类似的查询：

select
    CONVERT(VARCHAR(7),[startdatetime],111) AS [year-month],
    nm.nameLine1,
    sum(datediff(hour, startdatetime, enddatetime)) as total 
from  srl
inner join  sr on srl= sr.ServiceRequestId
inner join Name nm on(sr.clientCustomerId = nm.customerId and nm.nameTypeId = 'OFICE')
where (startdatetime >= '08-01-2011 00:00:00.000' and enddatetime <= '10-31-2011 00:00:00.000')
group by nm.nameLine1, [year-month]
order by nm.nameLine1, [year-month]

以上查询的输出是::

year-month nameLine1       total       
---------- ---------       ----------- 
2011/08    B               4 
2011/09    B               7 
2011/10    B               0 
2011/08    E               167 
2011/09    E               212 
2011/10    E               131 
2011/08    L               14 
2011/09    L               23 
2011/10    L               3 
2011/08    O               18 
2011/09    O               8 
2011/10    O               7 
2011/08    S               43 
2011/09    S               60 
2011/10    S               60

现在我的问题是，我应该如何在查询中为单个nameLine1total在名为nameLine1的其他列中获取总和。输出应该是这样的：

year-month nameLine1       total            nameLine1total
---------- ---------       -----------      ---------------
2011/08    B               4         
2011/09    B               7                 
2011/10    B               0                    11 
2011/08    E               167 
2011/09    E               212 
2011/10    E               131                  510 
2011/08    L               14 
2011/09    L               23 
2011/10    L               3                    40 
2011/08    O               18 
2011/09    O               8 
2011/10    O               7                    33
2011/08    S               43 
2011/09    S               60 
2011/10    S               60                   163

Answer 1

你使用的是什么样的sql - sql server，mysql等等。另外，它对你的答案无关紧要但你真的想要＆lt; = 10-31-2011 00:00:00，这样数据从10月31日起不包含在您的查询中？此外，您将如何使用该查询 - 如果是报告工具，报告工具很可能比sql更容易计算它。如果在每一行都包含名称总数是可以接受的，那将更容易 - 例如如果“S”的所有三行在最后一栏中都显示“163”，那还可以吗？

Answer 2

由于convert(.., .., 111)我认为这是一个SQL Server，如果它是2005+我将使用CTE：

with cte (ym, n, total, row) as (
    select *, row_number() over(partition by nameLine1 order by nameLine1, [year-month])  from (
        select
            convert(varchar(7),[startdatetime],111) as [year-month],
            nm.nameLine1,
            sum(datediff(hour, startdatetime, enddatetime)) as total 
        from  srl
        inner join sr on srl= sr.ServiceRequestId
        inner join Name nm on 
            sr.clientCustomerId = nm.customerId and 
            nm.nameTypeId = 'OFICE'
        where startdatetime >= '08-01-2011 00:00:00.000' and enddatetime <= '10-31-2011 00:00:00.000'
        group by nm.nameLine1, [year-month]
        order by nm.nameLine1, [year-month]
    ) t
) 
select c.ym as [year-month], c.n as [nameLine1], c.total as [total], g.total as [nameLine1total]
from cte c
left join (
    select max(row) as row, sum(total) as total, n 
    from cte group by n
) g on c.n = g.n and c.row = g.row

查询表中某些行的求和

2 个答案: