Question

我很难使多条记录成为oracle中具有多列的一条记录。这是我的查询：

SELECT cnim, ckddos FROM  tjadsidangast j, tsidangast s 
WHERE j.cnojadwal=s.cnojadwal AND cnim='1111500003'

结果：

+------------+--------+
|  CNIM      | CKDDOS |
+------------+--------+
| 1111500003 |  00105 |
| 1111500003 |  00060 |
| 1111500003 |  00126 |
+------------+--------+

我尝试过以下查询：

SELECT cnim,
   LTRIM(MAX(SYS_CONNECT_BY_PATH(ckddos,','))
   KEEP (DENSE_RANK LAST ORDER BY curr),',') AS elements
FROM   (SELECT cnim,
           ckddos,
           ROW_NUMBER() OVER (PARTITION BY cnim ORDER BY cstatus) AS curr,
           ROW_NUMBER() OVER (PARTITION BY cnim ORDER BY cstatus) -1 AS prev
    FROM   tsidangast j, tjadsidangast s WHERE j.cnojadwal=s.cnojadwal AND 
cnim='1111500003')
GROUP BY cnim
CONNECT BY prev = PRIOR curr AND cnim = PRIOR cnim
START WITH curr = 1;

结果是：

+------------+---------------------+
|     CNIM   |  ELEMENTS           |
+------------+---------------------+
| 1111500003 |  00126,00105,00060  |
+------------+---------------------+

我的问题是：如何使结果成为此输出：

+------------+--------+-------+-------+
|     CNIM   |    MD  |  P1   |  P2   |
+------------+--------+-------+-------+
| 1111500003 |  00126 | 00105 | 00060 |
+------------+--------+-------+-------+

谢谢

Answer 1

使用pivot代替缓慢的（这里不需要）分层查询：

^{SQLFiddle demo}

select * 
  from (select row_number() over (order by cstatus) rn, cnim, ckddos
          from your_join
          where cnim = '1111500003')
  pivot (max(ckddos) for rn in (1 md, 2 p1, 3 p2))

如果这是Oracle 10或更低版本，请将pivot替换为max(case ...)：

^{SQLFiddle demo}

select cnim, 
       max(case when rn = 1 then ckddos end) md, 
       max(case when rn = 2 then ckddos end) p1,
       max(case when rn = 3 then ckddos end) p2 
  from (
    select row_number() over (order by cstatus) rn, cnim, ckddos
      from your_join
      where cnim = '1111500003')
  group by cnim

此解决方案类似于@ D-Shih的解决方案，但您不应在查询中对ckddos值进行硬编码，因为这是您要寻找的。您应该并且可以硬编码的是行号（1、2、3）。

查询以将多条记录连接到Oracle中具有多列的一条记录

1 个答案: