使用 distinct with stuff/for xml path('')

Posted

技术标签:

【中文标题】使用 distinct with stuff/for xml path(\'\')【英文标题】:Using distinct with stuff/for xml path('')使用 distinct with stuff/for xml path('') 【发布时间】:2017-03-23 11:44:44 【问题描述】:

我只想将连接字符串中的唯一值放在一起。我的代码目前是:

select rc.Routage
    , COUNT(distinct rc.Event)
    , STUFF((select ', ' + cast(rcA.Event as varchar)
            from Receiving rcA
            where rcA.supplier = 'user'
            for xml path(''))
        , 1, 1, '')
from Receiving rc
where rc.supplier = 'user'
group by rc.Routage
order by COUNT(distinct rc.Event)desc

这给了我期望的输出,但我想消除 stuff/for xml 路径字段中的重复值。

我在 stuff/xml 部分尝试了 distinctgroup by 的各种组合,但无法正确组合。

为了澄清,对于COUNT(distinct rc.Event) = 2,我想从 stuff 子句中看到 2 个不同的事件。我该怎么做?

【问题讨论】:

【参考方案1】:

在子查询中使用select distinct

select rc.Routage,
       count(distinct rc.Event),
       stuff((select distinct ', ' + cast(rcA.Event as varchar(max))
              from Receiving rcA
              where rcA.supplier = 'user' and
                    rcA.DATETIME > '20170322' and
                    rc.Routage = rcA.Routage
              for xml path('')
             ), 1, 2, '')
from Receiving rc
where rc.supplier = 'user' and rc.DATETIME > '20170322'
group by rc.Routage;

注意事项:

在 SQL Server 中,从不使用没有长度的 varchar()(或相关类型)。默认值因上下文而异,您(可能)引入了一个很难找到的错误。 您希望 stuff() 删除两个字符,而不是 1,因为您有一个逗号后跟一个空格。 此公式假定Event 没有XML 特殊字符。如果这是一个问题,很容易调整。

此外,如果您消除子查询中的重复项,这种类型的查询通常会更快:

select rc.Routage, rc.numEvents,
       stuff((select distinct ', ' + cast(rcA.Event as varchar(max))
              from Receiving rcA
              where rcA.supplier = 'user' and
                    rcA.DATETIME > '20170322' and
                    rc.Routage = rcA.Routage
              for xml path(''), type
             ).value('.', 'varchar(max)'
                    ), 1, 2, ''
            )
from (select rc.Routage, count(distinct rc.Event) as numEvents
      from Receiving rc
      where rc.supplier = 'user' and rc.DATETIME > '20170322'
      group by rc.Routage
     ) rc;

【讨论】:

【参考方案2】:

在 XML 处理接近它之前,在子查询中执行 distinct

select rc.Routage
    , COUNT(distinct rc.Event)
    , STUFF((select ', ' + cast(rcA.Event as varchar)

            from (select distinct Event from Receiving a
                  where supplier = 'user'
                   and DATETIME > '20170322'
                   and rc.Routage=a.Routage
            ) rcA

            for xml path(''))
        , 1, 1, '')
from Receiving rc
where rc.supplier = 'user'
    and rc.DATETIME > '20170322'
group by rc.Routage
order by COUNT(distinct rc.Event)desc

【讨论】:

以上是关于使用 distinct with stuff/for xml path('')的主要内容,如果未能解决你的问题,请参考以下文章

LeetCode Longest Substring with At Most Two Distinct Characters

SQL Server - INNER JOIN WITH DISTINCT

Microsoft - Union Two Sorted List with Distinct Value

SQL Server count() over() with distinct

Select Distinct Not working in Spring with DSL Context

Leetcode 159: Longest Substring with At Most Two Distinct Characters