如何从 Oracle LISTAGG 聚合查询的结果中消除重复值 [重复]答案

【问题标题】：How to eliminate repited values from result of Oracle LISTAGG aggregated query [duplicate]如何从 Oracle LISTAGG 聚合查询的结果中消除重复值 [重复]
【发布时间】：2020-04-19 08:01:14
【问题描述】：

我正在尝试消除重复的价值观来自this Oracle's recomendation 的汇总结果：

with depts as (
select 'ACCOUNTING' as department , 'CLERC' as job from dual union all
select 'ACCOUNTING' as department , 'MANAGER' as job from dual union all
select 'ACCOUNTING' as department , 'PRESIDENT' as job from dual union all

select 'RESEARCH' as department , 'ANALYST' as job from dual union all
select 'RESEARCH' as department , 'ANALYST' as job from dual union all
select 'RESEARCH' as department , 'CLERK' as job from dual union all
select 'RESEARCH' as department , 'CLERK' as job from dual union all
select 'RESEARCH' as department , 'MANAGER' as job from dual union all

select 'SALES' as department , 'MANAGER' as job from dual union all
select 'SALES' as department , 'CLERK' as job from dual union all
select 'SALES' as department , 'SALESMAN' as job from dual union all
select 'SALES' as department , 'SALESMAN' as job from dual union all
select 'SALES' as department , 'SALESMAN' as job from dual union all
select 'SALES' as department , 'SALESMAN' as job from dual
)

select d.department,
       listagg (d.job,', ' ) within group (order by d.job) jobs
  from depts d
 group by d.department

正如您所见，外业工作包含重复的价值。

Oracle 建议以这种方式消除这种情况：

with depts as (
select 'ACCOUNTING' as department , 'CLERC' as job from dual union all
select 'ACCOUNTING' as department , 'MANAGER' as job from dual union all
select 'ACCOUNTING' as department , 'PRESIDENT' as job from dual union all

select 'RESEARCH' as department , 'ANALYST' as job from dual union all
select 'RESEARCH' as department , 'ANALYST' as job from dual union all
select 'RESEARCH' as department , 'CLERK' as job from dual union all
select 'RESEARCH' as department , 'CLERK' as job from dual union all
select 'RESEARCH' as department , 'MANAGER' as job from dual union all

select 'SALES' as department , 'MANAGER' as job from dual union all
select 'SALES' as department , 'CLERK' as job from dual union all
select 'SALES' as department , 'SALESMAN' as job from dual union all
select 'SALES' as department , 'SALESMAN' as job from dual union all
select 'SALES' as department , 'SALESMAN' as job from dual union all
select 'SALES' as department , 'SALESMAN' as job from dual
)

SELECT d.department,
       (select LISTAGG(job,', ')
               WITHIN GROUP (ORDER BY job)
          from (select unique job job
                  from depts t
                 where t.department = d.department)) jobs
  FROM depts d

但它不适用于版本。 Oracle Database 11g 企业版 11.2.0.4.0 - 64 位。

这个建议正确吗？据我所知，我们无法将数据发送到嵌套查询到第二级嵌套。

【问题讨论】：

标签： sql string oracle group-by listagg

【解决方案1】：

启动Oracle 19c，LISTAGG()支持DISTINCT：

select department, listagg(distinct job, ',') within group(order by job) jobs
from depts
group by department

在早期版本中，典型的解决方法是在子查询中 SELECT DISTINCT，然后聚合：

select department, listagg(job, ',') within group(order by job) jobs
from (select distinct department, job from depts) t
group by department

Demo on DB Fiddle：

部门 |工作 :--------- | :------------------------ 会计 |文员、经理、总裁研究 |分析师、文员、经理销售 |文员、经理、推销员

【讨论】：

那么，Oracle 网站上的建议是否正确？

【解决方案2】：

另一种可能性是部署LISTAGG 忽略NULL 值的功能。

简单的row_number the duplicates andlistaggonly the values withrn = 1`

with dep as (
select 
  DEPARTMENT, JOB,
  row_number() over (partition by  DEPARTMENT, JOB order by JOB) as rn
from depts)
select 
 department, listagg(case when rn = 1 then job end, ',') within group(order by job) jobs
from dep
group by department

【讨论】：