In PostgreSQL, we can use the ANY()
construct to perform searches against arrays. It returns a Boolean result, and so we can incorporate this into our searches to return just those rows that yield true or false, as the case may be.
There’s also a SOME()
construct, which is a synonym for ANY()
.
Example
Suppose we have the following table:
SELECT * FROM t1;
Result:
c1 | c2 | c3 | c4
----+------------+-------------------------------------------------------+-------------------------------------
1 | Animals | {Zebra,Bird,Mouse,Cow,Pig} | {{1,2,3},{4,5,6},{7,8,9}}
2 | Music | {Rock,Jazz,Blues,Electronica,Metal} | {{32,78,14},{102,99,37},{18,65,29}}
3 | Pets | {Cat,Dog,Zebra,Buffalo} | {{8,9,7},{4,5,3},{9,8,9}}
4 | Sports | {Rugby,Hockey,Soccer,Golf,Cricket,Baseball,Wrestling} | {{1,2,1},{9,10,3},{7,8,5}}
5 | Activities | {Golf,Walking,Running,Climbing} | {{4,32,56},{45,1,56},{37,5,9}}
(5 rows)
Here’s a query that shows the output of the ANY()
construct against the c3
column for a particular expression:
SELECT
c3,
'Zebra' = ANY(c3) AS "True/False"
FROM t1;
Result:
c3 | True/False
-------------------------------------------------------+------------
{Zebra,Bird,Mouse,Cow,Pig} | t
{Rock,Jazz,Blues,Electronica,Metal} | f
{Cat,Dog,Zebra,Buffalo} | t
{Rugby,Hockey,Soccer,Golf,Cricket,Baseball,Wrestling} | f
{Golf,Walking,Running,Climbing} | f
The True/False
column shows the outcome of the expression. If the array in the c3
column contains a Zebra
element, then the result is t
(True
), otherwise it’s f
(False
).
We can therefore filter our queries based on the output of such expressions.
For example:
SELECT c1, c2, c3 FROM t1
WHERE 'Zebra' = ANY(c3);
Result:
c1 | c2 | c3
----+---------+----------------------------
1 | Animals | {Zebra,Bird,Mouse,Cow,Pig}
3 | Pets | {Cat,Dog,Zebra,Buffalo}
Here we returned just those rows where the condition is true.
We can negate the expression with the NOT
operator:
SELECT c1, c2, c3 FROM t1
WHERE NOT 'Zebra' = ANY(c3);
Result:
c1 | c2 | c3
----+------------+-------------------------------------------------------
2 | Music | {Rock,Jazz,Blues,Electronica,Metal}
4 | Sports | {Rugby,Hockey,Soccer,Golf,Cricket,Baseball,Wrestling}
5 | Activities | {Golf,Walking,Running,Climbing}
This time we got all rows where the condition is false.
Passing a NULL
Value
If we pass a NULL value, then NULL
is returned:
SELECT 'Zebra' = ANY(null);
Result:
NULL
Passing a Non Array
Passing a non array results in an error:
SELECT 'Zebra' = ANY(c1) FROM t1;
Result:
ERROR: op ANY/ALL (array) requires array on right side
LINE 1: SELECT 'Zebra' = ANY(c1) FROM t1;
^
In this case I passed column c1
which is an integer column (not an array column), and so an error was returned.
The SOME()
Construct
As mentioned, SOME()
is a synonym for ANY()
. Therefore we can replace ANY()
with SOME()
in any of the previous examples, and we’d get the same result.
Here’s the first example again, but with a column added for SOME()
:
SELECT
c3,
'Zebra' = ANY(c3) AS "ANY()",
'Zebra' = SOME(c3) AS "SOME()"
FROM t1;
Result:
c3 | ANY() | SOME()
-------------------------------------------------------+-------+--------
{Zebra,Bird,Mouse,Cow,Pig} | t | t
{Rock,Jazz,Blues,Electronica,Metal} | f | f
{Cat,Dog,Zebra,Buffalo} | t | t
{Rugby,Hockey,Soccer,Golf,Cricket,Baseball,Wrestling} | f | f
{Golf,Walking,Running,Climbing} | f | f
Both ANY()
and SOME()
return the same result.