Sql 使用派生值可能多次出现的多列的动态透视_Sql_Sql Server_Tsql_Pivot

Sql 使用派生值可能多次出现的多列的动态透视

sql sql-server tsql

Sql 使用派生值可能多次出现的多列的动态透视,sql,sql-server,tsql,pivot,Sql,Sql Server,Tsql,Pivot,我试图找到一种方法，根据两列中的值动态地透视结果集我有一个非常基本的查询，它返回与特定作业关联的事件的信息 SELECT num, dat, [jobevtcls-cde] ,cde FROM jobevt WHERE (num = 3177564) ORDER BY cde 查询结果如下： +=========+============+===============+=====+ | num | dat | jobevtcls-cde | cde | +====

我试图找到一种方法，根据两列中的值动态地透视结果集

我有一个非常基本的查询，它返回与特定作业关联的事件的信息

SELECT num, dat, [jobevtcls-cde] ,cde
FROM jobevt
WHERE   (num = 3177564)
ORDER BY cde

查询结果如下：

+=========+============+===============+=====+
|   num   |    dat     | jobevtcls-cde | cde |
+=========+============+===============+=====+
| 3177564 | 2021-02-24 | SYS           |  10 |
+---------+------------+---------------+-----+
| 3177564 | 2021-02-24 | SYS           |  40 |
+---------+------------+---------------+-----+
| 3177564 | 2021-02-24 | SYS           |  40 |
+---------+------------+---------------+-----+
| 3177564 | 2021-02-24 | SYS           |  40 |
+---------+------------+---------------+-----+
| 3177564 | 2021-02-24 | SYS           |  42 |
+---------+------------+---------------+-----+
| 3177564 | 2021-02-24 | SYS           |  60 |
+---------+------------+---------------+-----+
| 3177564 | 2021-02-24 | SYS           |  60 |
+---------+------------+---------------+-----+
| 3177564 | 2021-02-24 | SYS           |  89 |
+---------+------------+---------------+-----+
| 3177564 | 2021-02-24 | SYS           | 100 |
+---------+------------+---------------+-----+
| 3177564 | 2021-02-24 | SYS           | 115 |
+---------+------------+---------------+-----+
| 3177564 | 2021-02-24 | GEN           | 120 |
+---------+------------+---------------+-----+
| 3177564 | 2021-03-01 | GEN           | 120 |
+---------+------------+---------------+-----+
| 3177564 | 2021-02-26 | GEN           | 160 |
+---------+------------+---------------+-----+
| 3177564 | 2021-02-24 | SYS           | 198 |
+---------+------------+---------------+-----+
| 3177564 | 2021-02-24 | GEN           | 210 |
+---------+------------+---------------+-----+
| 3177564 | 2021-02-26 | GEN           | 220 |
+---------+------------+---------------+-----+
| 3177564 | 2021-02-24 | GEN           | 310 |
+---------+------------+---------------+-----+
| 3177564 | 2021-02-26 | GEN           | 310 |
+---------+------------+---------------+-----+
| 3177564 | 2021-02-24 | SYS           | 422 |
+---------+------------+---------------+-----+

我希望将此数据透视，以将cde+[jobevtcls cde]显示为列标题，并将MAX（dat）显示为行值

下面的代码首先抱怨D中的第四列没有名字，所以我很难找出那个名字。SQL抱怨的另一件事是同一代码的多个实例，例如SYS40

DECLARE @JobEventKey NVARCHAR(MAX) = ''
DECLARE @SQL NVARCHAR(MAX) = ''

SELECT @JobEventKey += QUOTENAME([jobevtcls-cde] + CAST(cde AS VARCHAR(10))) + ','
FROM jobevt

WHERE jobevt.num = '3177564'

SET @JobEventKey = LEFT(@JobEventKey, LEN(@JobEventKey) -1)

SET @SQL = 

'SELECT * FROM
    (SELECT
         num
        ,dat
        ,cde
        ,[jobevtcls-cde] + CAST(cde AS VARCHAR(10))
    FROM
        jobevt
    WHERE
        (num = 3177564)
) AS D

PIVOT (
    MAX(dat)
    FOR cde IN (' + @JobEventKey +
    ')
) AS P'

EXECUTE sp_executesql @SQL

我对这类事情还比较陌生，所以任何指导都非常感谢。

我想知道为什么“相当新”的人总是倾向于从更复杂的东西开始，比如动态SQL？不管怎样，您的解决方案已经接近成功，而且错误很容易修复

动态SQL的一般指导原则：分步构建解决方案，并打印构造的查询（部分），以便在其构造过程中进行验证。这允许您执行构造的查询以验证结果

错误

没有为“D”的第4列指定列名

动态SQL查询中的子查询

缺少列名或别名。添加带有

as

的别名

为“p”多次指定了列“SYS40”

为

@JobEventKey

构建值的方法将生成一个类似

[SYS10]、[SYS40]、[SYS40]、…

的字符串。透视结果中的

pivot

列列表

需要唯一的列名。添加一个

分组依据

，以避免重复

DECLARE @JobEventKey NVARCHAR(MAX) = ''
DECLARE @SQL NVARCHAR(MAX) = ''

SELECT @JobEventKey += QUOTENAME(cls + CAST(cde AS VARCHAR(10))) + ','
FROM jobevt

WHERE jobevt.num = '3177564'
--> num=3177564 contains multiple rows for "[SYS40]"
--> the contatenations contains duplicates!

SET @JobEventKey = LEFT(@JobEventKey, LEN(@JobEventKey) -1)

SET @SQL = 

'SELECT * FROM
    (SELECT
         num
        ,dat
        ,cde
        ,cls + CAST(cde AS VARCHAR(10)) --> 4th column missing a name / alias!
    FROM
        jobevt
    WHERE
        (num = 3177564)
) AS D

PIVOT (
    MAX(dat)
    FOR cde IN (' + @JobEventKey +
    ')
) AS P'

EXECUTE sp_executesql @SQL

改进版

小调整：

将nvarchar（最大值）替换为更合理的内容（取决于您的数据）
尽可能使用表别名
不要在整数值周围加引号（对于整型列）
为数据透视后的排序列添加排序依据

最后，工作查询：

DECLARE @JobEventKey NVARCHAR(1000) = '';

SELECT @JobEventKey += QUOTENAME(je.cls + CONVERT(NVARCHAR(10), je.cde)) + ','
FROM jobevt je
WHERE je.num = 3177564
group by QUOTENAME(je.cls + CONVERT(NVARCHAR(10), je.cde))
order by QUOTENAME(je.cls + CONVERT(NVARCHAR(10), je.cde));

SET @JobEventKey = LEFT(@JobEventKey, LEN(@JobEventKey)-1);

-- validate keys
select @JobEventKey;

DECLARE @sql NVARCHAR(2000) = 
'SELECT P.*
 FROM ( SELECT je.num
              ,je.dat
              ,je.cls + CONVERT(NVARCHAR(10), cde) as clscde
        FROM jobevt je
        WHERE je.num = 3177564 ) AS d
 PIVOT (MAX(d.dat) FOR d.clscde IN (' + @JobEventKey + ')) AS P'

-- validate sql
select @sql;

-- execute
EXECUTE sp_executesql @sql;

结果

用于复制的数据：

create table jobevt
(
  num int,
  dat date,
  cls nvarchar(3),
  cde int
);


insert into jobevt (num, dat, cls, cde) values
(3177564, '2021-02-24', 'SYS',  10),
(3177564, '2021-02-24', 'SYS',  40),
(3177564, '2021-02-24', 'SYS',  40),
(3177564, '2021-02-24', 'SYS',  40),
(3177564, '2021-02-24', 'SYS',  42),
(3177564, '2021-02-24', 'SYS',  60),
(3177564, '2021-02-24', 'SYS',  60),
(3177564, '2021-02-24', 'SYS',  89),
(3177564, '2021-02-24', 'SYS', 100),
(3177564, '2021-02-24', 'SYS', 115),
(3177564, '2021-02-24', 'GEN', 120),
(3177564, '2021-03-01', 'GEN', 120),
(3177564, '2021-02-26', 'GEN', 160),
(3177564, '2021-02-24', 'SYS', 198),
(3177564, '2021-02-24', 'GEN', 210),
(3177564, '2021-02-26', 'GEN', 220),
(3177564, '2021-02-24', 'GEN', 310),
(3177564, '2021-02-26', 'GEN', 310),
(3177564, '2021-02-24', 'SYS', 422);

结果:

num      GEN120      GEN160      GEN210      GEN220      GEN310      SYS10       SYS100      SYS115      SYS198      SYS40       SYS42       SYS422      SYS60       SYS89
-------  ----------  ----------  ----------  ----------  ----------  ----------  ----------  ----------  ----------  ----------  ----------  ----------  ----------  ----------
3177564  2021-03-01  2021-02-26  2021-02-24  2021-02-26  2021-02-26  2021-02-24  2021-02-24  2021-02-24  2021-02-24  2021-02-24  2021-02-24  2021-02-24  2021-02-24  2021-02-24