How to write function for optional parameters in postgresql?

My requirement is write optional parameters to a function.Parameters are optional sometimes i will add or i will not pass parameters to function.Can anyone help me how to write function.

I am writing like

select * 
from test 
where field3 in ('value1','value2') 
 and ($1 is null or field1 = $1) 
 and ($2 is null or field2 = $2) 
 and ($3 is null or field3 = $3);

i am passing parameters to Query,But my output is not expected.when i pass all three parameters my output is correct,otherwise it is not expected output.

You can define optional parameters by supplying a default value.

create function foo(p_one integer default null, 
                    p_two integer default 42, 
                    p_three varchar default 'foo')
  returns text
as
$$
begin
    return format('p_one=%s, p_two=%s, p_three=%s', p_one, p_two, p_three);
end;
$$
language plpgsql;

You can "leave out" parameters from the end, so foo(), foo(1) or foo(1,2) are valid. If you want to only supply a parameter that is not the first you have to use the syntax that specifies the parameter names.

select foo(); 

returns: p_one=, p_two=42, p_three=foo

select foo(1); 

returns: p_one=1, p_two=42, p_three=foo

select foo(p_three => 'bar')

returns: p_one=, p_two=42, p_three=bar

Apart of the VARIADIC option pointed by @a_horse_with_no_name, which is only a syntax sugar for passing an array with any number of elements of the same type, you can't define a function with optional parameters because, in postgres, functions are identified not only by its name but also by its arguments and the types of them.

That is: create function foo (int) [...] and create function foo (varchar) [...] will create different functions.

Which is called when you execute, for example, select foo(bar) depends on bar data type itself. That is: if it is an integer, you will call the first one and if it is varchar, then second one will be called.

More than that: if you execute, for example, select foo(now()), then a function not exists exception will be triggered.

So, as I said, you can't implement functions with variable arguments, but you can implement multiple functions with the same name and distinct argument (an/or type) sets returning the same data type.

If you (obviously) doesn't want to implement the function twice, the only thing you need to do is to implement a "master" function with all possible parameters and the others (which have fewer parameters) only calling the "master" one with default values for the non received parameters.

As an option, I got a function i tested with Navicat App: CREATE OR REPLACE FUNCTION "public"."for_loop_through_query"(sponsor_name varchar default 'Save the Children') It generates me this. (Note: Please look at the parameter difference) CREATE OR REPLACE FUNCTION "public"."for_loop_through_query"("sponsor_name" varchar='Save the Children'::character varying)

 CREATE OR REPLACE FUNCTION "public"."for_loop_through_query"("sponsor_name" varchar='Save the Children'::character varying)
  RETURNS "pg_catalog"."void" AS $BODY$
DECLARE
    rec RECORD;
BEGIN
    FOR rec IN SELECT
      companies."name" AS org_name,
      "sponsors"."name" AS sponsor_name
      FROM
      "donor_companies"
      JOIN "sponsors"
      ON "donor_companies"."donor_id" = "sponsors"."id" 
      JOIN companies
      ON "donor_companies"."organization_id" = companies."id"
      WHERE
      "public"."sponsors"."name" = sponsor_name
    LOOP
    RAISE NOTICE '%', rec.org_name;
  END LOOP;
END;
$BODY$
  LANGUAGE plpgsql VOLATILE
  COST 100;