这个问题已经在这里有了答案:




已关闭8年。






假设我有一个这样的表:

NAME          GROUP_NAME
name1         groupA
name2         groupB
name5         groupC
name4         groupA
name3         groupC

我想要这样的结果:
GROUP_NAME     NAMES
groupA         name1,name4
groupB         name2
groupC         name3,name5

如果表中只有一列,我可以通过执行以下操作来将记录连接起来,但是在上下文中进行分组时,我的想法并不多。

合并一列表:
SELECT names
FROM (SELECT SYS_CONNECT_BY_PATH(names,' ') names, level
      FROM name_table

      START WITH names = (SELECT names FROM name_table WHERE rownum = 1)
      CONNECT BY PRIOR names < names
      ORDER BY level DESC)
      WHERE rownum = 1

更新:

我现在有一个解决方案,使用LISTAGG:
SELECT
group_name,
LISTAGG(name, ', ')
WITHIN GROUP (ORDER BY GROUP) "names"
FROM name_table
GROUP BY group_name

对于LISTAGG不可用的情况,仍对更通用的解决方案感兴趣。

最佳答案

如果您使用的是11g,请考虑使用LISTAGG函数:

select grp, listagg(name,',') within group( order by name )
  from name_table group by grp

sqlFiddle

更新:如果不是,请考虑使用分析:
select grp,
    ltrim(max(sys_connect_by_path
       (name, ',' )), ',')
        scbp
  from (select name, grp,
            row_number() over
           (partition by grp
            order by name) rn
         from tab
          )
start with rn = 1
connect by prior rn = rn-1
and prior grp = grp
  group by grp
  order by grp

sqlFiddle

10-04 10:50