How to compare two arrays and pick only the non matching elements In postgres

Question

How can I pick only the non matching elements between two arrays.

Example:

base_array [12,3,5,7,8]
temp_array [3,7,8]

So here I wan

Answer 1

The contrib/intarray module provides this functionality--for arrays of integers, anyway. For other data types, you may have to write your own functions (or modify the ones provided with intarray).

Answer 2

I'd approach this with the array operator.

select array(select unnest(:arr1) except select unnest(:arr2));

If :arr1 and :arr2 don't intersect, using array_agg() leads to a null.

Answer 3

I've constructed a set of functions to deal specifically with these types of issues: https://github.com/JDBurnZ/anyarray

The greatest thing is these functions work across ALL data-types, not JUST integers, as intarray is limited to.

After loading loading the functions defined in those SQL files from GitHub, all you'd need to do is:

SELECT
  ANYARRAY_DIFF(
    ARRAY[12, 3, 5, 7, 8],
    ARRAY[3, 7, 8]
  )

Returns something similar to: ARRAY[12, 5]

If you also need to return the values sorted:

SELECT
  ANYARRAY_SORT(
    ANYARRAY_DIFF(
      ARRAY[12, 3, 5, 7, 8],
      ARRAY[3, 7, 8]
    )
  )

Returns exactly: ARRAY[5, 12]

Answer 4

An extension to Denis' answer that returns the difference, regardless of which array was entered first. It's not the most concise query, maybe someone has a tidier way.

select array_cat(
   (select array(select unnest(a.b::int[]) except select unnest(a.c::int[]))),
   (select array(select unnest(a.c::int[]) except select unnest(a.b::int[]))))
from (select '{1,2}'::int[] b,'{1,3}'::int[] c) as a;

Returns:

{2,3}

Answer 5

select array_agg(elements)
from (
  select unnest(array[12,3,5,7,8])
  except
  select unnest(array[3,7,8])
) t (elements)

Answer 6

Let's try the unnest() / except :

EXPLAIN ANALYZE SELECT array(select unnest(ARRAY[1,2,3,n]) EXCEPT SELECT unnest(ARRAY[2,3,4,n])) FROM generate_series( 1,10000 ) n;
 Function Scan on generate_series n  (cost=0.00..62.50 rows=1000 width=4) (actual time=1.373..140.969 rows=10000 loops=1)
   SubPlan 1
     ->  HashSetOp Except  (cost=0.00..0.05 rows=1 width=0) (actual time=0.011..0.011 rows=1 loops=10000)
           ->  Append  (cost=0.00..0.04 rows=2 width=0) (actual time=0.002..0.008 rows=8 loops=10000)
                 ->  Subquery Scan "*SELECT* 1"  (cost=0.00..0.02 rows=1 width=0) (actual time=0.002..0.003 rows=4 loops=10000)
                       ->  Result  (cost=0.00..0.01 rows=1 width=0) (actual time=0.001..0.002 rows=4 loops=10000)
                 ->  Subquery Scan "*SELECT* 2"  (cost=0.00..0.02 rows=1 width=0) (actual time=0.001..0.003 rows=4 loops=10000)
                       ->  Result  (cost=0.00..0.01 rows=1 width=0) (actual time=0.001..0.002 rows=4 loops=10000)
 Total runtime: 142.531 ms

And the intarray special operator :

EXPLAIN ANALYZE SELECT ARRAY[1,2,3,n] - ARRAY[2,3,4,n] FROM generate_series( 1,10000 ) n;
 Function Scan on generate_series n  (cost=0.00..15.00 rows=1000 width=4) (actual time=1.338..11.381 rows=10000 loops=1)
 Total runtime: 12.306 ms

Baseline :

EXPLAIN ANALYZE SELECT ARRAY[1,2,3,n], ARRAY[2,3,4,n] FROM generate_series( 1,10000 ) n;
 Function Scan on generate_series n  (cost=0.00..12.50 rows=1000 width=4) (actual time=1.357..7.139 rows=10000 loops=1)
 Total runtime: 8.071 ms

Time per array intersection :

intarray -           :  0.4 µs
unnest() / intersect : 13.4 µs

Of course the intarray way is much faster, but I find it amazing that postgres can zap a dependent subquery (which contains a hash and other stuff) in 13.4 µs...