CROSS APPLY 与 GROUP BY 一起返回逗号分隔值答案

【问题标题】：CROSS APPLY to return comma separated values along with GROUP BYCROSS APPLY 与 GROUP BY 一起返回逗号分隔值
【发布时间】：2019-01-10 05:01:33
【问题描述】：

我正在处理一个查询以返回一个报告，该报告将对每个业务单位的日志记录进行分组，以及分配给每个业务单位的经理的逗号分隔列表，所以基本上有 4 个表。

[dbo].[Log]
[dbo].[BusinessUnit]
[dbo].[Manager]
[dbo].[BusinessUnitManager]

就报告而言，BusinessUnit、Manager 和 BusinessUnitManager 只是获取基本信息的表，我需要分组的真实信息包含在 Log 表中，我的最终结果集应如下所示：

BusinessUnit.Name,
Log.Code,
Count(*) LogRecordsCount,
Emails (Comma separated list of Managers assigned to the Business Unit)

我一直在尝试使用 CROSS APPLY 获取逗号分隔的列，但无法正常工作，我当前的查询如下所示：

SELECT
L.BusinessUnit,
L.Code, 
COUNT(*) AS RecordsCount,
F.Emails
FROM [dbo].[LOG] AS L
INNER JOIN [dbo].[BusinessUnit] AS D ON L.BusinessUnit = D.ID
INNER JOIN [dbo].[BusinessUnitManager] AS DSO ON D.ID = DSO.BusinessUnit
CROSS APPLY
(
    SELECT STUFF((
    SELECT ',' + O.Email
    FROM [dbo].[Manager] AS O
    WHERE O.Id = DSO.IdManager
    ORDER BY O.Email ASC
    FOR XML PATH ('')), 1, 1, '')
) F(Emails)
GROUP BY L.BusinessUnit, L.Code, F.Emails

这个查询的问题是返回重复的行，每个分配给业务部门的经理电子邮件一行，我的预期结果应该只有一行，其中一列包含所有经理电子邮件的逗号分隔列表。

当前结果：

+--------------+------+--------------+----------------------------+
| BusinessUnit | Code | RecordsCount |           Emails           |
+--------------+------+--------------+----------------------------+
| Americas     |   00 |           21 | americasmanager1@email.com |
| Americas     |   00 |           21 | americasmanager2@email.com |
| Asia         |   10 |            5 | asiamanager1@email.com     |
| Asia         |   10 |            5 | asiamanager2@email.com     |
+--------------+------+--------------+----------------------------+

预期结果：

+--------------+------+--------------+-------------------------------------------------------+
| BusinessUnit | Code | RecordsCount |                        Emails                         |
+--------------+------+--------------+-------------------------------------------------------+
| Americas     |   00 |           21 | americasmanager1@email.com,americasmanager2@email.com |
| Asia         |   10 |            5 | asiamanager1@email.com,asiamanager2@email.com         |
+--------------+------+--------------+-------------------------------------------------------+

感谢您的帮助

【问题讨论】：

一些示例数据在这里可能会有所帮助。
我想你想改变你的东西的逻辑。 SELECT ',' + O.Email FROM [dbo].[Manager] AS O JOIN [dbo].[BusinessUnitManager] AS M ON M.IdManager = O.Id WHERE M.BusinessUnit = DSO.BusinessUnit ORDER BY O.Email FOR XML PATH('') 之类的东西当然，当前的逻辑会为每个经理生成单独的行，但这应该基于业务单位加入......旁注，我认为在常规子查询中交叉应用没有优势，你会如果它只是一个常规子查询，则不必将其包含在 GROUP BY 中...
@TimBiegeleisen 我已经编辑以显示一些示例数据
@ZLK 谢谢，我在 STUFF 查询中添加了 INNER JOIN，现在按预期工作

标签： sql-server tsql

【解决方案1】：

你很接近，只需分两步完成：

1) 将你的东西存储到临时表中

SELECT
L.BusinessUnit,
L.Code, 
O.Emails
into #tmp
FROM [dbo].[LOG] AS L
INNER JOIN [dbo].[BusinessUnit] AS D ON L.BusinessUnit = D.ID
INNER JOIN [dbo].[BusinessUnitManager] AS DSO ON D.ID = DSO.BusinessUnit
JOIN [Manager] AS O ON O.Id = DSO.IdManager
GROUP BY L.BusinessUnit, L.Code, O.Emails

2) 对自己使用外部应用将电子邮件填充在一列中

select BusinessUnit,Code, F.Emails from #tmp t 
outer apply( 
     SELECT STUFF(( SELECT ',' + inner.Email 
     FROM #tmp inner 
     WHERE t.Id = inner. 
     ORDER BY O.Email ASC 
     FOR XML PATH ('')), 1, 1, '') 
) F(Emails) 
group by BusinessUnit,Code

【讨论】：

【解决方案2】：

SELECT
    L.BusinessUnit,
    L.Code, 
    COUNT(*) AS RecordsCount,
    (
        SELECT STUFF((
        SELECT ',' + O.Email
        FROM [dbo].[Manager] AS O
            INNER JOIN [dbo].[BusinessUnitManager] AS DSO ON O.Id = DSO.IdManager 
        WHERE 
            DSO.BusinessUnit = D.ID
        ORDER BY O.Email ASC
        FOR XML PATH ('')), 1, 1, '')
    ) as Emails
FROM [dbo].[LOG] AS L
    INNER JOIN [dbo].[BusinessUnit] AS D ON L.BusinessUnit = D.ID
GROUP BY L.BusinessUnit, L.Code, D.ID

【讨论】：