选择最近的日期群组

Question

我有一个T-SQL查询，旨在清除某些产品培训的重复条目，只抓取具有最新DateTaken的条目。例如，如果有人已经进行了三次特定的培训课程，我们只想显示一行，该行是包含最近DateTaken的那一行。以下是我迄今为止，但我收到以下错误：

An expression of non-boolean type specified in a context where a condition is expected, near 'ORDER'.

的ORDER BY是必要的，因为我们希望这组通过查询到期日期的所有结果。下面是完整的查询：

SELECT DISTINCT 
    p.ProductDescription as ProductDesc, 
    c.CourseDescription as CourseDesc, 
    c.Partner, a.DateTaken, a.DateExpired, p.Status 
FROM 
    sNumberToAgentId u, AgentProductTraining a, Course c, Product p 
WHERE 
    @agentId = u.AgentId 
    and u.sNumber = a.sNumber 
    and a.CourseCode = c.CourseCode 
    and (a.DateExpired >= @date or a.DateExpired IS NULL) 
    and a.ProductCode = p.ProductCode 
    and (p.status != 'D' or p.status IS NULL) 
GROUP BY 
    (p.ProductDescription) 
HAVING 
    MIN(a.DateTaken) 
ORDER BY 
    DateExpired ASC 

编辑

我做了以下更改GROUP BY和HAVING子句，但我仍然收到错误：

GROUP BY 
    (p.ProductDescription, c.CourseDescription) 
HAVING 
    MIN(a.DateTaken) > GETUTCDATE() 

在SQL管理Studio，红线错误标记出现在p.ProductDescription后面的'，'，c.CourseDescription后的'）'，.DateTaken中的'a'以及GETUTCDATE（）的右括号'）'下。如果我只是离开GROUP BY语句，只包括p.ProductDescription我收到此错误信息：

Column 'Product.ProductDescription' is invalid in the select list because it is not contained in either an aggregate function or the GROUP BY clause.

我是比较新的SQL，可能有人解释这是怎么回事？谢谢！

Answer 1

我的建议，因为你正在使用SQL Server是执行row_number()并由ProductDescription和CourseDescription分区。这将大大子查询，然后应用过滤器只在该行数等于一个或最近的记录返回那些：

select * 
from 
(
    SELECT p.ProductDescription as ProductDesc, 
     c.CourseDescription as CourseDesc, 
     c.Partner, a.DateTaken, a.DateExpired, p.Status 
     row_number() over(partition by p.ProductDescription, c.CourseDescription order by a.DateTaken desc) rn 
    FROM sNumberToAgentId u 
    INNER JOIN AgentProductTraining a 
     ON u.sNumber = a.sNumber 
     AND (a.DateExpired >= @date or a.DateExpired IS NULL) 
    INNER JOIN Course c 
     ON a.CourseCode = c.CourseCode 
    INNER JOIN Product p 
     ON a.ProductCode = p.ProductCode 
     AND (p.status != 'D' or p.status IS NULL) 
    WHERE u.AgentId = @agentId 
) src 
where rn = 1 
order by DateExpired 

Answer 2

其此行

HAVING MIN(a.DateTaken) 

应该是一个boolean类型，如

HAVING MIN(a.DateTaken) > GETUTCDATE() 

必须返回真或假（布尔）

Answer 3

这里是我结束了使用的最终查询。它类似于上述建议：

SELECT ProductDesc, CourseDesc, Partner, DateTaken, DateExpired, Status 
FROM(
    SELECT 
     p.ProductDescription as ProductDesc, 
     c.CourseDescription as CourseDesc, 
     c.Partner, a.DateTaken, a.DateExpired, p.Status, 
     row_number() OVER (PARTITION BY p.ProductDescription, c.CourseDescription ORDER BY abs(datediff(dd, DateTaken, GETDATE()))) as Ranking 
    FROM 
     sNumberToAgentId u, AgentProductTraining a, Course c, Product p 
    WHERE 
     @agentId = u.AgentId 
     and u.sNumber = a.sNumber 
     and a.CourseCode = c.CourseCode 
     and (a.DateExpired >= @date or a.DateExpired IS NULL) 
     and a.ProductCode = p.ProductCode 
     and (p.status != 'D' or p.status IS NULL) 
    ) aa 
WHERE Ranking = '1'