从行到列的 PIVOT 表答案

【问题标题】：PIVOT table from rows to column从行到列的 PIVOT 表
【发布时间】：2020-01-23 19:17:33
【问题描述】：

您好，我有一张类似于下面的表格，

PId  |  PocType       |   pocDate
---------------------------------------
109  | POC - Start |   2018-09-20
109  | POC - Start |   2018-07-07
109  | POC - Start |   2018-11-15
101  | POC - Start |   2019-07-17
101  | POC - Start |   2019-07-01
100  | POC - Start |   2019-05-07
100  | POC - Start |   2019-05-07
100  | POC - Start |   2019-01-22
100  | POC - Start |   2019-03-20

我想把它改成如下所示：

PId  |  Poc        |  pocDate  | procDate    | procDate   | procDate   | 
------------------------------------------------------------------------
109  | POC - Start | 2018-07-07 | 2018-09-20 | 2018-11-15 |            | 
101  | POC - Start | 2019-07-01 | 2019-07-17 |            |            | 
100  | POC - Start | 2019-01-22 | 2019-03-20 | 2019-05-07 | 2019-05-07 |

我正在尝试使用 SQL PIVOT，但不成功。

select [PId],[pocDate]
from
(
  select [PId],[Poc],[pocDate]
  from #PocTable 
) x
pivot
(
  max(pocDate)
  for SystemProcedure in([pocDate])
)p

感谢您的帮助。

【问题讨论】：

您有 4 个同名的列。另外，Proc 日期的数量是否有上限？其他 Poc 类型呢？
@jk1844 所以你想为每个pocdate创建一个新列？

标签： sql sql-server pivot

【解决方案1】：

我会使用条件聚合，但关键是row_number()：

select pid,
       max(case when seqnum = 1 then pocdate end) as pocdate_1,
       max(case when seqnum = 2 then pocdate end) as pocdate_2,
       max(case when seqnum = 3 then pocdate end) as pocdate_3,
       max(case when seqnum = 4 then pocdate end) as pocdate_4
from (select t.*,
             row_number() over (partition by pid, poc_type order by pocdate) as seqnum
      from t
     ) t
group by pid;

【讨论】：

【解决方案2】：

如果你还是不明白，下面的 sn-p 会帮助你。

您的问题中的架构：

SELECT * INTO #TABLE FROM (
SELECT 109  , 'POC - Start' ,  CAST( '2018-09-20' AS DATE) UNION ALL
SELECT 109  , 'POC - Start' ,  CAST( '2018-07-07' AS DATE) UNION ALL
SELECT 109  , 'POC - Start' ,  CAST( '2018-11-15' AS DATE) UNION ALL
SELECT 101  , 'POC - Start' ,  CAST( '2019-07-17' AS DATE) UNION ALL
SELECT 101  , 'POC - Start' ,  CAST( '2019-07-01' AS DATE) UNION ALL
SELECT 100  , 'POC - Start' ,  CAST( '2019-05-07' AS DATE) UNION ALL
SELECT 100  , 'POC - Start' ,  CAST( '2019-05-07' AS DATE) UNION ALL
SELECT 100  , 'POC - Start' ,  CAST( '2019-01-22' AS DATE) UNION ALL
SELECT 100  , 'POC - Start' ,  CAST( '2019-03-20' AS DATE)
)A(PId,PocType,pocDate)

您需要动态数据透视表和数据透视表列名称的计算列。

DECLARE @COLS VARCHAR(MAX) = '', @SQL VARCHAR(MAX) = ''

SELECT @COLS = @COLS+  QUOTENAME(SNO)+','  FROM (
SELECT DISTINCT 'pocDate'+ 
CAST( ROW_NUMBER() OVER(PARTITION BY PId ORDER BY pocDate ) AS VARCHAR(50)) SNO 
FROM #TABLE
)A

SELECT  @COLS = LEFT(@COLS, LEN(@COLS)-1)

SELECT  @SQL =  
'SELECT * FROM (
SELECT ''pocDate''
+ CAST( ROW_NUMBER() OVER(PARTITION BY PId ORDER BY pocDate ) AS VARCHAR(50)) SNO, * 
FROM #TABLE
)AS A
PIVOT
(
    MAX(pocDate) FOR SNO IN ('+@COLS+')
)PV '


EXEC (@SQL)

结果：

+-----+-------------+------------+------------+------------+------------+
| PId |   PocType   |  pocDate1  |  pocDate2  |  pocDate3  |  pocDate4  |
+-----+-------------+------------+------------+------------+------------+
| 100 | POC - Start | 2019-01-22 | 2019-03-20 | 2019-05-07 | 2019-05-07 |
| 101 | POC - Start | 2019-07-01 | 2019-07-17 | NULL       | NULL       |
| 109 | POC - Start | 2018-07-07 | 2018-09-20 | 2018-11-15 | NULL       |
+-----+-------------+------------+------------+------------+------------+

【讨论】：