在Postgresql的数组数组中迭代每个元素

Question

I have a VARCHAR of numbers inside my stored procedure, these numbers are organized as arrays, I will show an example below:

{1,2,3,4,5,6,7,8,9},{1,2,3,4,5},{1,2,3},{9} -- This is a STRING

Now with a help from another guy from here I'm using this to get integer arrays integer[]

SELECT string_to_array(regexp_split_to_table(
      trim('{1,2,3,4,5,6,7,8,9},{1,2,3,4,5},{1,2,3},{9}', '{}')
      , '},{'), ',')::int[]

I will have a set of records, each of them with an array, see below:

{1,2,3,4,5,6,7,8,9}
{1,2,3,4,5}
{1,2,3}
{9}

I was trying but I cannot figure out how can I make a FOR to iterate over each element from these arrays to call another procecure to do with each element from each array.

An example for my array {1,2,3,4,5,6,7,8,9} that I will call my_array :

rec record;
arr integer[];

FOR rec IN SELECT string_to_array(unnest(string_to_array(trim(text_nodes_for_connectivity, '{}'), '},{')), ',')::int[] LOOP
    arr := array_cat(arr, rec);
END LOOP;

I'm getting this error:

function array_cat(integer[], record) does not exist

I need to convert each of my record results to an array, so I can use array_cat or another functions to iterate over array elements

My proc code goes below:

DROP FUNCTION IF EXISTS clustering_nodes();
CREATE OR REPLACE FUNCTION clustering_nodes() RETURNS integer[] AS $$

DECLARE
my_array integer[];
rec record;
arr integer[];
my_var varchar[500];
len integer;

BEGIN

my_var = '{1,2,3,4,5,6,7,8,9},{1,2,3,4,5},{1,2,3},{8}';


FOR rec IN SELECT string_to_array(unnest(string_to_array(trim(my_var, '{}'), '},{')), ',')::int[] LOOP
    len = array_length(rec);
    arr := array_append(arr, len);
END LOOP;

RETURN arr;

END;

$$ LANGUAGE 'plpgsql' STRICT;

select clustering_nodes();

Tips or triks?

Answer 1

CREATE OR REPLACE FUNCTION clustering_nodes()
  RETURNS integer[] AS
$func$
DECLARE
   my_var   text  := '{1,2,3,4,5,6,7,8,9},{1,2,3,4,5},{1,2,3},{8}';
   my_array integer[];
   arr_len  integer[];
BEGIN

FOR my_array IN
   SELECT string_to_array(regexp_split_to_table(
            trim(my_var, '{}'), '},{'), ',')::int[]
LOOP
   arr_len := array_append(arr_len, array_upper(my_array, 1));
END LOOP;

RETURN arr_len;

END
$func$ LANGUAGE plpgsql;

Major points

• array_length() doesn't work on records, only on array.
  What's more, there is no array_length() in Postgres 8.3 . The manual is instrumental in figuring that out yourself. Using the less favorable array_upper() instead.

• Assignment operator in plpgsql is := . Use of = is undocumented.

• plpgsql is an identifier in LANGUAGE plpgsql , not a string. Do not quote it. May lead to sneaky errors.

• You can assign variables at declaration time.

• STRICT modifier is pointless without parameters.

-> SQLfiddle demo for Postgres 8.3.

Simpler with modern Postgres

Again, this could be had in a single (if somewhat complex) call:

SELECT array_agg(array_length(string_to_array(txt, ','), 1))
FROM   unnest(string_to_array(
          trim('{1,2,3,4,5,6,7,8,9},{1,2,3,4,5},{1,2,3},{8}', '{}')
          , '},{')
       ) AS sub(txt);

You need to upgrade to a current version.