How can I take two rows for one Distinct value in postgresql?

问题

I want to get two rows of distinct column2 from following table.

demo_table 
id  | column1 | column2
 1  |    1    |    3
 2  |    2    |    3
 3  |    3    |    4
 4  |    4    |    3
 5  |    5    |    4
 6  |    6    |    3

When I preform this query

select distinct on (column2) * from demo_table 

it gives me out put as this

id  | column1 | column2
 5  |    5    |    4
 6  |    6    |    3

I am looking for query where I can get output as

id  | column1 | column2
 3  |    3    |    4
 5  |    5    |    4
 6  |    6    |    3
 4  |    4    |    3

it should give two row where column2 is distinct.

回答1:

Use row_number() for that:

select id, column1, column2
from (
  select id, column1, column2, 
         row_number() over (partition by column2 order by id) as rn
  from demo_table
) t 
where rn <= 2;

来源：https://stackoverflow.com/questions/39718851/how-can-i-take-two-rows-for-one-distinct-value-in-postgresql