PostgreSQL选择具有相同列的特定列

时间:2015-01-07 09:32:02

标签: postgresql duplicates greatest-n-per-group

我需要删除仅在特定列中具有相同值的行。例如,在下面的摘录中,我想选择除最后一行之外的所有行,它们等于CODE,START_DATE和TYPE列的倒数第二行(这意味着忽略END_DATE列的值)。

  code         | start_date     | end_date     | type
---------------+----------------+--------------+------
C086000-T10001 | 2014-11-11     | 2014-11-12   | 01
C086000-T10001 | 2014-11-11     | 2014-11-11   | 03
C086000-T10002 | 2014-12-03     | 2014-12-10   | 03
C086000-T10002 | 2014-01-03     | 2014-01-04   | 03
C086000-T10003 | 2012-02-27     | 2014-02-28   | 03
C086000-T10003 | 2014-08-11     | 2014-11-12   | 01
C086000-T10003 | 2014-08-11     | 2014-08-20   | 01

我怎么能这样做?

编辑:以下查询为子查询错误消息返回列太多:

SELECT * FROM my_table WHERE code NOT IN (SELECT DISTINCT code, start_date, type FROM my_table) ;

非常感谢您的帮助!

1 个答案:

答案 0 :(得分:1)

这可以使用Postgres' distinct on运营商:

select distinct on (code, start_date, type) code, start_date, end_date, type
from the_table
order by code, start_date, type;

如果您更喜欢使用标准SQL,也可以使用窗口函数来完成:

select code, start_date, end_date, type
from (
    select code, start_date, end_date, type, 
           row_number() over (partition by code, start_date, type order by end_date) as rn
    from the_table
) t
where rn = 1
order by code, start_date, type;

SQLFiddle示例:http://sqlfiddle.com/#!15/c5044/1