postgresql中列的不同值

Question

I have the following table with below data

id    description
 1    a,b,a

I need a PostgreSQL script that give me below output

id   description
1    a,b

This what I have tried so far.

create temporary table test
(
  id integer,
  description text
);

insert into test
select 1,'a,b,a';

select id,string_agg(distinct description, ',') as description
from test
group by id;

Answer 1

The best solution for your problem is to normalize your data model and don't store multiple, comma separated values in a single column.

But you can achieve what you want using a combination of unnest and aggregation:

select id, string_agg(distinct c, ',' order by c)
from the_table, unnest(string_to_array(description, ',')) as t(c)
group by id;

With the outdated (and unsupported) version 9.2 you need to use a derived table:

select id, string_agg(distinct c, ',' order by c) as description
from (
  select id, unnest(string_to_array(description, ',')) as c
  from the_table
) t
group by id;

Online example (for 9.6): http://rextester.com/LEE56363