arrays - Postgres : How to count distinct elements in array columns given a condition

Question

我有如下表格:

ID     array1           array2
7333 | {615593}        | {NULL}
7333 | {2013682}       | {NULL}
7333 | {26573,1508291} | {NULL}
7333 | {1457957}       | {NULL}
7333 | {NULL}          | {1063105}
7333 | {NULL}          | {107978,408431}

我想为每个 edge_id 查询 array1 和 array2 的所有唯一项目。例如

ID   array1_distinct    array2_distinct
7333 5                  3

我试过类似下面的方法

SELECT 
  id, 
  array1_count 
FROM (
  SELECT id,
  COUNT(DISTINCT((CASE WHEN array1 is not null THEN (select unnest(array1)b) END)) as array1_count
  FROM mytable
  GROUP BY id
) totals
limit 1

在尝试使用 unnest 对数组中的元素进行计数时，我总是会遇到诸如“用作表达式的子查询返回的多行”和“在无法接受集合的上下文中调用的集值函数”之类的错误。

Answer 1

select id, count(distinct a1) as a1, count(distinct a2) as a2
from (
    select id, unnest(array1) as a1, unnest(array2) as a2
    from t
) s
group by id