使用Oracle的LISTAGG和WM_CONCAT进行行转列操作

wm_concat 和listagg 函数都可以实现对（单个或组合）列的合并，也可以看成是对某一列的“SUM”，这俩个函数功能相同，listagg是在11.2的版本中才首次出现的，wm_concat在12c版本中被取消。

• 常见用法

  select t.job,listagg(t.ename,',')  within group(order by hiredate asc) from scott.emp t group by  t.job;
  
  select t.job,wm_concat(t.ename)   from scott.emp t group by  t.job;
  
  --去重
     with tb as(
       select sys_guid() id,'t1' name, 'CLERK' job from dual 
        union select sys_guid() id,'t1' name, 'SALESMAN' job from dual 
        union select sys_guid() id,'t1' name, 'MANAGER' job from dual 
        union select sys_guid() id,'t1' name, 'CLERK' job from dual 
        union select sys_guid() id,'t2' name, 'MANAGER' job from dual 
        union select sys_guid() id,'t2' name, 'MANAGER' job from dual 
     )
     select t.name,  wm_concat(distinct t.job) jobs  from tb t group by t.name;
  

• 对比

wm_cancat函数行转列后，不会按照原有查询结果排序。listagg函数行转列后，会按照原有查询结果顺序排列。如果考虑到需要行转列，并且保持分组后顺序不变可以使用listagg来完成。

wm_concat的使用（listagg上限varchar4000）

select  dev_id,regexp_replace(regexp_replace(wm,',',';'),'#',',') coords from
(select dev_id ,wm_concat(t.x||'#'||t.y) over(partition by t.dev_id order by gps_time) wm ,
row_number() over(partition by dev_id order by gps_time desc) rm
from t_dev_loc t where t.dev_id in ('37080000000034798','37080000000034488') 
and gps_time between to_date('2018-11-05 00:00:00','yyyy-mm-dd hh24:mi:ss') and sysdate
)where rm=1;

行转列排序2( 突破listagg上限进行行转列排序)

 with tb as(
 select sys_guid() id,'t1' name, 'CLERK' job from dual 
  union select sys_guid() id,'t1' name, 'SALESMAN' job from dual 
  union select sys_guid() id,'t1' name, 'MANAGER' job from dual 
  union select sys_guid() id,'t1' name, 'CLERK' job from dual 
  union select sys_guid() id,'t2' name, 'MANAGER' job from dual 
  union select sys_guid() id,'t2' name, 'MANAGER' job from dual 
  )
  -- 突破listagg上限进行行转列排序
  select t.name,
         xmlagg(xmlparse(content t.job || ',' wellformed) order by t.id asc)
         .getclobval() jobs
    from tb t
   group by t.name