Postgres：獲取不同條目的補充

Question

目前，我可以從像這樣的表中獲得不同的條目：

SELECT DISTINCT ON (abc.entry) abc.* FROM table abc
    JOIN table1 def ON abc.entry = def.entry
WHERE def.created >= 'some_date' AND def.position = 1;

有沒有一種方法可以為每個具有重復項的條目獲取條目（即，在some_date上唯一集的補集？

Answer 1

您可以使用having子句獲取重復的條目列表：

SELECT abc.entry
FROM table abc JOIN
     table1 def
     ON abc.entry = def.entry
WHERE def.created >= 'some_date' AND def.position = 1
group by abc.entry
having count(*) > 1

要獲取行列表，我將使用窗口函數：

select t.*
from (SELECT abc.*, count(*) over (partition by entry) as cnt
      FROM table abc JOIN
           table1 def
           ON abc.entry = def.entry
      WHERE def.created >= 'some_date' AND def.position = 1
     ) t
where cnt > 1;

編輯：

我看到了對術語的困惑。 DISTINCT在SQL中具有非常特殊的含義。 而且， DISTINCT ON在Postgres中具有完全不同的含義。 我認為您必須使用子查詢來執行此操作：

select abc.*
from table abc left outer join
     (SELECT DISTINCT ON (abc.entry) abc.* FROM table abc
      JOIN table1 def ON abc.entry = def.entry
      WHERE def.created >= 'some_date' AND def.position = 1
     ) f
     on abc.id = f.id
where f.id is NULL

不過要小心。 您正在使用不帶order by子句的distinct on 。 數據中返回的特定行是不確定的。 您應在原始數據和此處的子查詢中添加order by 。

select abc.*
from table abc left outer join
     (SELECT abc.entry, max(abc.id) as maxid
      FROM table abc
      JOIN table1 def ON abc.entry = def.entry
      WHERE def.created >= 'some_date' AND def.position = 1
      group by abc.entry
     ) f
     on abc.id = f.maxid
where f.maxid is NULL