Question

我想比较两个数组，忽略元素和重复的顺序。

我可以在两个方向进行收容检查，但有更好/更快的解决方案吗？

select * from my_table where my_table.a1 @> my_table.a2 and
                             my_table.a2 @> my_table.a1

Answer 1

从我的internal testing开始，似乎这两个变体速度最快（甚至比@> + <@检查更快），并且它们都可以处理NULL个太：

where (select array_agg(distinct e order by e) from unnest(arr1) e)
    = (select array_agg(distinct e order by e) from unnest(arr2) e)

where array(select distinct e from unnest(arr1) e order by e)
    = array(select distinct e from unnest(arr2) e order by e)

后者通常要快一点，但有时候不会。这可能取决于很多事情，f.ex。数组的基数等等。

Answer 2

使用不必要且不同的示例：

t=# create or replace function so62(a1 anyarray,a2 anyarray) returns boolean as
$$
declare
 _r boolean := false;
 _a text;
begin
  with p as (select distinct unnest(a1) order by 1) select array_agg(unnest) into _a from p;
  with p as (select distinct unnest(a2) order by 1) select array_agg(unnest)::text = _a into _r from p;
  return _r;
end;
$$ language plpgsql;
CREATE FUNCTION

（@poz后更新的功能请注意，需要两个except进行比较）

<强>试验：

t=# with c as (
  select '{1,null,2,2}'::int[] a1,'{2,1,null}'::int[] a2
) 
select a1 @> a2, a2 @> a1,so62(a1,a2) 
from c;
 ?column? | ?column? | so62
----------+----------+------
 f        | f        | t
(1 row)

在此示例中，

@>和<@根本不起作用。另请阅读Compare arrays for equality, ignoring order of elements

PostgreSQL：比较两个数组，忽略排序

2 个答案: