526互联

Oracle listagg如何去重

发布时间 2023-06-29 20:28:26作者: 每天进步多一点

1.去重思路：利用listagg会忽略null值的特点

2.上SQL：

with T_TEST as
 (
  select 'TOOL-001' EQPT_ID, 10493721 ENTITY_GROUP_RRN, 'JITAI-1' STATION_ID from dual union all
  select 'TOOL-003' EQPT_ID, 10493721 ENTITY_GROUP_RRN, 'JITAI-1' STATION_ID from dual union all
  select 'TEST' EQPT_ID, 10493721 ENTITY_GROUP_RRN, 'S1' STATION_ID from dual union all
  select 'TEST' EQPT_ID, 10493721 ENTITY_GROUP_RRN, 'S2' STATION_ID from dual union all
  select 'TEST2' EQPT_ID, 10493721 ENTITY_GROUP_RRN, 'S1' STATION_ID from dual union all
  select 'TEST2' EQPT_ID, 10493721 ENTITY_GROUP_RRN, 'S2' STATION_ID from dual union all
  select 'TOOL-001' EQPT_ID, 123456 ENTITY_GROUP_RRN, 'JITAI-1' STATION_ID from dual union all
  select 'TOOL-003' EQPT_ID, 123456 ENTITY_GROUP_RRN, 'JITAI-1' STATION_ID from dual union all
  select 'TEST' EQPT_ID, 123456 ENTITY_GROUP_RRN, 'S1' STATION_ID from dual union all
  select 'TEST' EQPT_ID, 123456 ENTITY_GROUP_RRN, 'S2' STATION_ID from dual union all
  select 'TEST2' EQPT_ID, 123456 ENTITY_GROUP_RRN, 'S1' STATION_ID from dual union all
  select 'TEST2' EQPT_ID, 123456 ENTITY_GROUP_RRN, 'S2' STATION_ID from dual 
)
select t.entity_group_rrn,
       listagg(decode(rne, 1, t.eqpt_id, null), ',') within group(order by t.eqpt_id) eqpt_ids,
       listagg(decode(rns, 1, t.station_id, null), ',') within group(order by t.station_id) station_ids
from (select row_number() over(partition by t.entity_group_rrn,t.eqpt_id order by rownum) rne,
               row_number() over(partition by t.entity_group_rrn,t.station_id order by rownum) rns,
               t.*
          from T_TEST t) t
group by t.entity_group_rrn;

查询结果

 

总结
1、通常情况可以先分组去重，再合并即可，除非遇到同一分组中，多列分别有重复值，才需要上面那样处理
2、还有另一种去重办法，使用 wm_concat + distinct ，如下：

with T_TEST as
 (select 'TOOL-001' EQPT_ID, 10493721 ENTITY_GROUP_RRN, 'JITAI-1' STATION_ID from dual union all
  select 'TOOL-003' EQPT_ID, 10493721 ENTITY_GROUP_RRN, 'JITAI-1' STATION_ID from dual union all
  select 'TEST' EQPT_ID, 10493721 ENTITY_GROUP_RRN, 'S1' STATION_ID from dual union all
  select 'TEST' EQPT_ID, 10493721 ENTITY_GROUP_RRN, 'S2' STATION_ID from dual union all
  select 'TEST2' EQPT_ID, 10493721 ENTITY_GROUP_RRN, 'S1' STATION_ID from dual union all
  select 'TEST2' EQPT_ID, 10493721 ENTITY_GROUP_RRN, 'S2' STATION_ID from dual 
)
select wm_concat(distinct t.EQPT_ID) ids,entity_group_rrn, wm_concat(distinct t.station_id) ids2
from T_TEST t  group by t.entity_group_rrn

 但是由于官方不推荐使用 wm_concat ，所以尽量不要用此方法