0

我想在 oracle 10g 中使用 JOIN 将行转换为列。我不想使用 DECODE 或 XMLAGG 函数。我的表结构如下：

GROUP_ID    ENTITYID    GROUP_ENTITYID
1           A997        A995
2           A997        A993
3           A997        A990
4           A988        A987
5           A988        A982
6           A988        A980
7           A979        A975
8           A979        A974
9           A979        A973

我想获取以下格式的数据：

ENTITYID    GROUP_ENTITYID

A979    A979,A975,A974,A973
A988    A988,A987,A982,A980
A997    A997,A995,A993,A990

你们能建议我怎么做吗？

提前致谢。

4

1 回答 1

1

Tim Hall 列出了Oracle 中可用的各种字符串聚合技术的规范列表。

使用连接来组合数据既不实用也不高效。如果您碰巧知道这EntityID将映射到其他三个Group_EntityID值，则可以使用连接

SELECT a.entity_id, 
       a.entity_id || ','  ||
         a.group_entityID || ',' ||
         b.group_entityID || ',' ||
         c.group_entityID
  FROM table_name a,
       table_name b,
       table_name c
 WHERE a.entityID = b.entityID
   AND a.entityID = c.entityID
   AND a.group_entityID > b.group_entityID
   AND b.group_entityID > c.group_entityID

Group_EntityID但是，如果与特定的值相关的数量发生变化，这种方法既不是特别有效也不是特别灵活EntityID。

我通常更喜欢创建用户定义的聚合函数（蒂姆列表中的选项之一）并执行类似的操作

SELECT entityID, entityID || ',' || string_agg( group_entity_id )
  FROM table_name
 GROUP BY entityID;

如果您想在不创建任何其他对象或函数的情况下执行此操作，可以使用SYS_CONNECT_BY_PATHTim 网站上的方法

SQL> ed
Wrote file afiedt.buf

  1  with x as (
  2    select 1 group_id, 'A997' entityID, 'A995' group_entityID from dual union all
  3    select 2, 'A997', 'A993' from dual union all
  4    select 3, 'A997', 'A990' from dual union all
  5    select 4, 'A998', 'A987' from dual union all
  6    select 5, 'A998', 'A982' from dual union all
  7    select 6, 'A998', 'A980' from dual union all
  8    select 7, 'A979', 'A975' from dual union all
  9    select 8, 'A979', 'A974' from dual union all
 10    select 9, 'A979', 'A973' from dual
 11  )
 12  select entityID,
 13         entityID || ',' ||
 14           ltrim( max( sys_connect_by_path( group_entityID, ','))
 15                     keep(dense_rank last order by group_id),
 16                  ',') group_entityID
 17    from (select group_id,
 18                 entityID,
 19                 group_entityID,
 20                 row_number() over (partition by entityID
 21                                        order by group_id) curr,
 22                 row_number() over (partition by entityID
 23                                        order by group_id) - 1 prev
 24            from x)
 25   group by entityID
 26  connect by prev = PRIOR curr
 27      and entityID = PRIOR entityID
 28*  start with curr = 1
SQL> /

ENTI GROUP_ENTITYID
---- ------------------------------
A979 A979,A975,A974,A973
A997 A997,A995,A993,A990
A998 A998,A987,A982,A980
于 2012-04-04T14:38:19.723 回答