Thread: access sub elements using any()
Given an array of arrays ( e.g. v = {{1,'a'},{2,'b'},{3,'c'},{2,'d'}} ) is it possible, in plain sql, to access the first element of the listed arrays using the IN function? Say I wanted just those with 2 as first element. "where 2 = any(v[1])" does not work and not sure it's supported. And I cannot craft a lhs to fussy-match each of the elements in v. Can this be done?
Hi,
On Mon, Oct 31, 2022, at 18:04, Rob Sargent wrote:
Given an array of arrays ( e.g. v = {{1,'a'},{2,'b'},{3,'c'},{2,'d'}} ) is it possible, in plain sql, to access the first element of the listed arrays using the IN function? Say I wanted just those with 2 as first element. "where 2 = any(v[1])" does not work and not sure it's supported. And I cannot craft a lhs to fussy-match each of the elements in v. Can this be done?
if your inner arrays would be row values instead, i.e. if v would read
v = array[(1,'a'),(2,'b'),(3,'c'),(2,'d')] :: t[]
assuming CREATE TYPE t AS (x int, y text);
then your query could simply be
SELECT r.*
FROM unnest(v) AS r
WHERE r.x = 2;
Cheers,
—Torsten
On 10/31/22 16:36, Torsten Grust wrote:
Hi,
On Mon, Oct 31, 2022, at 18:04, Rob Sargent wrote:
Given an array of arrays ( e.g. v = {{1,'a'},{2,'b'},{3,'c'},{2,'d'}} ) is it possible, in plain sql, to access the first element of the listed arrays using the IN function? Say I wanted just those with 2 as first element. "where 2 = any(v[1])" does not work and not sure it's supported. And I cannot craft a lhs to fussy-match each of the elements in v. Can this be done?
if your inner arrays would be row values instead, i.e. if v would read
v = array[(1,'a'),(2,'b'),(3,'c'),(2,'d')] :: t[]
assuming CREATE TYPE t AS (x int, y text);
then your query could simply be
SELECT r.*
FROM unnest(v) AS r
WHERE r.x = 2;
Cheers,
—Torsten
Rob Sargent <robjsargent@gmail.com> writes:
> Given an array of arrays ( e.g. v = {{1,'a'},{2,'b'},{3,'c'},{2,'d'}} )
> is it possible, in plain sql, to access the first element of the listed
> arrays using the IN function?
If it's actually a 2-D array, and not an array-of-composite which is
what your sample data seems to suggest, then array slicing might help:
regression=# select ('{{1,2},{3,4},{5,6}}'::int[])[:][1];
int4
---------------
{{1},{3},{5}}
(1 row)
regression=# select 2 = any (('{{1,2},{3,4},{5,6}}'::int[])[:][1]);
?column?
----------
f
(1 row)
regression=# select 3 = any (('{{1,2},{3,4},{5,6}}'::int[])[:][1]);
?column?
----------
t
(1 row)
regards, tom lane