轉自 https://blog.csdn.net/victor_ww/article/details/44415895

create type custom_data_type as (
id int,
name varchar(50),
score decimal(5,2),
create_time timestamp
);
 
create or replace function custom_data_type_demo(p_order_unit_array varchar[],p_goods_array int[])
returns custom_data_type[]
 
 as $$
declare
v_order_unit_array varchar[] := array['a','b','c']::varchar[];
v_goods_array int[] := array[60.56,82.12,95.32]::int[];
v_tmp_result custom_data_type;
v_result_array custom_data_type[];
v_index int := 0;
v_order varchar(100);
v_goods int;
begin
if p_order_unit_array is not null then
v_order_unit_array :
 
= p_order_unit_array;
end if;
 
if p_goods_array is not null then
v_goods_array := p_goods_array;
end if;
 
raise notice '-------1---------';
<<order_label>> foreach v_order in array v_order_unit_array loop
<<goods_label>> foreach v_goods in array v_goods_array loop
v_tmp_result.id 
 
= v_index*round(random()*10);
v_tmp_result.name = v_order;
v_tmp_result.score = v_goods;
v_tmp_result.create_time = current_timestamp;
    raise notice '-------goods_label---------';
end loop goods_label;
raise notice '-------order_label---------v_index';
v_result_array[v_index] = v_tmp_result;
v_index := v_index + 1;
end loop order_label;
raise notice '-------2---------';
return v_result_array;
exception when others then
raise exception 'error happen(%)',sqlerrm;
end;
$$ language plpgsql;
 
select custom_data_type_demo(null,null);
exampledb=> select custom_data_type_demo(null,null);
                                                                                    custom_data_type_demo
---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
 [0:3]={"(0,a,95.00,\"2018-10-31 17:43:53.836608\")","(1,b,95.00,\"2018-10-31 17:43:53.836608\")","(2,c,95.00,\"2018-10-31 17:43:53.836608\")","(3,d,95.00,\"2018-10-31 17:43:53.836608\")"}
(1 row)

解析陣列，可用於不定引數時，可使用傳陣列的方式，類似於SQL Server 的Xml；
exampledb=> select T from unnest(custom_data_type_demo(null,null)) as T;         
                    t                     
------------------------------------------
 (0,a,95.00,"2018-10-31 17:35:40.881777")
 (1,b,95.00,"2018-10-31 17:35:40.881777")
 (2,c,95.00,"2018-10-31 17:35:40.881777")
 (3,d,95.00,"2018-10-31 17:35:40.881777")
(4 rows)

exampledb=> select T.id,T.name,T.score,T.create_time from unnest(custom_data_type_demo(null,null)) as T;  
 id | name | score |        create_time         
----+------+-------+----------------------------
  0 | a    | 95.00 | 2018-10-31 17:40:25.939054
  1 | b    | 95.00 | 2018-10-31 17:40:25.939054
  2 | c    | 95.00 | 2018-10-31 17:40:25.939054
  3 | d    | 95.00 | 2018-10-31 17:40:25.939054
(4 rows)