如何在公用表表达式 (CTE) 中执行 SUM() 函数

Question

我已经多次尝试了，但它似乎不起作用。这可能吗？以下是我目前的尝试：

With PrintsPerDayPerEmp(ID,[Month],[Day],[Year], [Number Of Prints])
AS
(Select ID,
    datepart(month,Agent_Local_Time) as [Month],
    datepart(day,Agent_Local_Time) as [Day],
    datepart(year,Agent_Local_Time) as [Year],
    count(RowID) as [Number of Prints]
From Table1
Where Agent_Local_Time >= (getdate()-90)
Group by datepart(day,Agent_Local_Time),
    datepart(month,Agent_Local_Time),
    datepart(year,Agent_Local_Time),
    ID
)
select *
from PrintsPerDayPerEmp
where [Number Of Prints] > (sum([Number Of Prints])/90)*3
order by ID desc

Answer 1

由于getdate() 函数，我假设您正在使用 SQL Server。

我不确定您为什么要尝试使用 CTE。这似乎完全没有必要。

我的猜测是，您期望 sum(count(RowID)) 是总和，但这不是聚合函数的工作方式。 GROUP BY 子句影响所有聚合函数。

我猜这就是你想要的。它使用 OVER() 子句来控制聚合并获得总和：

Select Id,
    datepart(month,Agent_Local_Time) as 'Month',
    datepart(day,Agent_Local_Time) as 'Day',
    datepart(year,Agent_Local_Time) as 'Year',
    count(RowID) as 'Number Of Prints',
    count(RowID) over () as 'Print Sum'
From Table1
Where Agent_Local_Time >= (getdate()-90)
Group by datepart(day,Agent_Local_Time),
    datepart(month,Agent_Local_Time),
    datepart(year,Agent_Local_Time),
    Id
Order by Id desc

或者这个：

Select Id,
    datepart(month,Agent_Local_Time) as 'Month',
    datepart(day,Agent_Local_Time) as 'Day',
    datepart(year,Agent_Local_Time) as 'Year',
    count(RowID) as 'Number Of Prints',
    count(RowID) over (partition by Id) as 'Print Sum'
From Table1
Where Agent_Local_Time >= (getdate()-90)
Group by datepart(day,Agent_Local_Time),
    datepart(month,Agent_Local_Time),
    datepart(year,Agent_Local_Time),
    Id
Order by Id desc

但如果没有一些数据和预期结果，就很难判断。

Answer 2

首先，不要对列别名使用单引号。仅对字符串和日期常量使用单引号。在 SQL Server 中，正确的转义字符是双引号或方括号。

第二，group by全错了。

我认为这基本上是您想要的查询，假设 id 是员工 ID：

With PrintsPerDayPerEmp([Id],[Month],[Day],[Year], [Number Of Prints], [Print Sum])
AS (
      Select Id, datepart(month, Agent_Local_Time) as [Month],
             datepart(day, Agent_Local_Time) as [Day],
             datepart(year, Agent_Local_Time) as [Year],
             count(RowID) as [Number of time per Day],
             sum(count(RowID)) over () as [Print Sum]
      From Table1
      Where Agent_Local_Time >= (getdate() - 90)
      Group by datepart(day, Agent_Local_Time),
               datepart(month, Agent_Local_Time),
               datepart(year, Agent_Local_Time),
               Id
     )
select *
from PrintsPerDayPerEmp 
order by Id desc;