CODEWITHSUNDEEP
sqlgoogle-bigquery
Norah Jones
Norah Jones

Reputation: 467

Checking intersection of arrays

My table looks like this:

id top_id name_important name another_column
111 [AAA, BBB] null
222 111 [AAA] X

What I would like to check is that top_id and name_important are not empty. And that value in array name_important are in array name based on top_id (those arrays might be empty or might not exisit). In this case there is the relation between id 222 and 111 because id 222 has 111 in column top_id. I would like to return true or false. Could you help me?

I would like to use case when clause since I have other cases to include, something like this:

with test as (
  select 111 as id, null as top_id, null as name_important, ["AAA", "BBB"] as name, null as another_column union all
  select 222, 111, ["AAA"], null, X
)

select id,
case when another_column is not null then true
when (question in the topic) then true
else false end result
from test

Upvotes: 0

Views: 618

Answers (2)

Sergey Geron
Sergey Geron

Reputation: 10162

Maybe something like this could fit:

with test as (
  select 111 as id, null as top_id, null as name_important, ["AAA", "BBB"] as name, null as another_column union all
  select 222, 111, ["AAA"], null, "X"
)
select 
  id,
  case 
    when another_column is not null then true
    when intersected then true
    else false
  end result
from (
  select test.*, ifnull(intersected, false) as intersected
  from test
  left join (
    select secondary.id, true as intersected
    from (
      select test.id, name
      from test, test.name as name
    ) as main
    join (
      select test.id, test.top_id, name_important
      from test, test.name_important as name_important
    ) as secondary
    on main.id = secondary.top_id and main.name = secondary.name_important
  ) using (id)
)

enter image description here

Upvotes: 1

Gordon Linoff
Gordon Linoff

Reputation: 1269663

It is not clear from your example if you need all values in the array to match or any values. If you want any value, then:

with test as (
  select 111 as id, null as top_id, null as name_important, ['AAA', 'BBB'] as name union all
  select 222, 111, ['AAA'], null
)
select t2.*, t1.name
from test t1 join
     test t2
     on t1.top_id = t2.id
where t1.name_important is not null and
      exists (select 1
              from unnest(t2.name) n1 join 
                   unnest(t1.name_important) n2
                   on n2 = n1
             )

If all need to match:

with test as (
  select 111 as id, null as top_id, null as name_important, ['AAA', 'BBB'] as name union all
  select 222, 111, ['AAA'], null
)
select t1.*, t2.name
from test t1 join
     test t2
     on t1.top_id = t2.id
where t1.name_important is not null and
      not exists (select 1
                  from unnest(t1.name_important) n1 left join
                       unnest(t2.name) n2  
                       on n2 = n1
                  where n2 is null
                 );

Upvotes: 1

Related Questions