将列中相似（但不相同）的数据组合成单行

Question

我是 SQL Server 的初学者，正在努力弄清楚如何更新我的查询，以便正确显示我的数据。

我查看了Combine similar column data from two rows into one row 并尝试在 QueryID 上进行分组，但是当我需要它按文本末尾的数字和 id# 分组时，我不确定如何应用它。我还查看了许多关于枢轴/取消枢轴以执行多个枢轴的帖子，但我正在努力应用它。

我当前的查询和示例结果如下。我想看到的是 QueryID 以相同的数字结尾并且员工 ID 相等，日期/级别/原因都返回到同一行而不是 3 个单独的行。我感谢任何指导。

With DisciplineTable as
(
   Select HrEmployees.Number, 
     HrEmployees.FirstName + ' ' + HrEmployees.LastName as Name,    
     HrEmployeeQueries.QueryID, 
     HrEmployeeQueries.Query, 
     HrEmployeeQueries.Response
    From HrEmployeeQueries
    Left Join HrEmployees on HrEmployees.EmployeeID = HrEmployeeQueries.EmployeeID
    Where QueryID Like 'DIS%')
Select * 
From DisciplineTable
Pivot(Max(Response) for Query in ([DATE:],[LEVEL:],[REASON:])) as DisciplinaryAction
Where QueryID Like 'DIS%' 
Order by Name

我的结果如下所示：

+-----+----------+---------+----------+---------+-------------+
|  #  |   Name   | QueryID |  DATE:   | LEVEL:  |   REASON:   |
+-----+----------+---------+----------+---------+-------------+
| 123 | John Doe | Date2   | 10/16/14 | Null    | Null        |
| 123 | John Doe | Level2  | Null     | Serious | Null        |
| 123 | John Doe | Reason2 | Null     | Null    | Attendance  |
| 123 | John Doe | Date3   | 12/13/14 | Null    | Null        |
| 123 | John Doe | Level3  | Null     | Major   | Null        |
| 123 | John Doe | Reason3 | Null     | Null    | Performance |
+-----+----------+---------+----------+---------+-------------+

我正试图让它们看起来像这样。 Action 列应填充 QueryID 末尾的数值，该数值可以从 1 到 10。 QueryID 开头的文本将始终为 Date、Reason 或 Level：

+-----+----------+--------+----------+---------+-------------+
|  #  |   Name   | Action |  DATE:   | LEVEL:  |   REASON:   |
+-----+----------+--------+----------+---------+-------------+
| 123 | John Doe |      2 | 10/16/14 | Serious | Attendance  |
| 123 | John Doe |      3 | 12/13/14 | Major   | Performance |
+-----+----------+--------+----------+---------+-------------+

Answer 1

一些解决此问题的建议。部分问题是QueryID 列中有多个值，因此 PIVOT 函数对生成不同行的每个值进行分组。

首先，要获得Action 值，您需要去掉QueryID 列中的字母字符。考虑到您将获得 1-10 的值，您可以使用 STUFF and PATINDEX 来获得此信息。注意：有几种方法可以做到这一点，包括使用replace 等。一旦你有了Action 的值，你就可以应用 PIVOT。这是一个示例版本：

select Number, Name, Action, [DATE:],[LEVEL:],[REASON:]
from
(
  select number, name, reason, query,
    Action = stuff(QueryID, 1, patindex('%[0-9]%', QueryID)-1, '')
  from yourdata
) d
pivot
(
  max(reason)
  for query in ([DATE:],[LEVEL:],[REASON:])
) p;

见SQL Fiddle with Demo。然后将此添加到您当前的查询中：

With DisciplineTable as
(
    Select 
        HrEmployees.Number, 
        HrEmployees.FirstName + ' ' + HrEmployees.LastName as Name, 
        Action = stuff(HrEmployeeQueries.QueryID, 1, patindex('%[0-9]%', HrEmployeeQueries.QueryID)-1, '') , 
        HrEmployeeQueries.Query, 
        HrEmployeeQueries.Response
    From HrEmployeeQueries
    Left Join HrEmployees on HrEmployees.EmployeeID = HrEmployeeQueries.EmployeeID
    Where QueryID Like 'DIS%'
)
Select Number, Name, Action, [DATE:],[LEVEL:],[REASON:]
From DisciplineTable
Pivot
(
    Max(Response) for Query in ([DATE:],[LEVEL:],[REASON:])
) as DisciplinaryAction
Order by Name

Answer 2

只需添加Max aggregate 和group by

WITH DisciplineTable
     AS (SELECT HrEmployees.Number,
                HrEmployees.FirstName + ' '
                + HrEmployees.LastName AS NAME,
                substring(HrEmployeeQueries.QueryID,patindex('%[0-9]%',HrEmployeeQueries.QueryID),LEN(HrEmployeeQueries.QueryID))QueryID,
                HrEmployeeQueries.Query,
                HrEmployeeQueries.Response
         FROM   HrEmployeeQueries
                LEFT JOIN HrEmployees
                       ON HrEmployees.EmployeeID = HrEmployeeQueries.EmployeeID
         WHERE  QueryID LIKE 'DIS%')
SELECT [#],
       [Name],
       [QueryID] Action,
       [DATE:],
       [LEVEL:],
       [REASON:]
FROM   DisciplineTable
       PIVOT(Max(Response)
            FOR Query IN ([DATE:],
                          [LEVEL:],
                          [REASON:])) AS DisciplinaryAction
ORDER  BY NAME