Execute multiple select query based on a query count output - postgresql

Question

I want to execute the query the below "test" table only once. But I'm querying to check for the count and if the count is more than 1, then I need to execute another query.
Any leads on how to store test table result into a variable and then return from that variable to the function output so that I query the "test"table only once?
Below is an example, as my Prod table has millions of records and getting into performance issue when I query multiple times :(
CREATE OR REPLACE FUNCTION lcdm_main.test(IN p_member_crn character varying)
RETURNS TABLE(fname character varying, lname character varying) AS
$BODY$
DECLARE
countRow integer Default 0;
BEGIN
PERFORM
t.fname, t.lastname
from test t where t.email =p_member_crn
;
GET DIAGNOSTICS countRow := ROW_COUNT;
if countRow > 1 then
RETURN QUERY select concat( error_msg_cd , error_msg_nm )::character varying , NULL:: character varying
from lcdm_main.error where error_Cndtn_nm ='SMRF';
ELSE
return query
select
t.fname, t.lastname
from test t where t.email =p_member_crn
;
END IF ;
END;
$BODY$
LANGUAGE plpgsql VOLATILE SECURITY DEFINER
COST 100
ROWS 1000;

Answer 1

A similar question remain without a solution in below thread.
https://dba.stackexchange.com/questions/40214/postgresql-stored-procedure-to-return-rows-or-empty-set
Thanks everyone for different approach provided.

Answer 2

Edit.
I might have misunderstood and miss some things at first.
According to your example, i assume that you don't care about precise record count on "test" query. You only need firstname and lastname if returned one record, or execute and return different query if test returns more than 1 records, right? so why don't you wrap "test" query with another one, limit result set to 2, add column count(*) over() to check whether recordcount is more than 1. Insert result into simple variables and then use IF structure to return variables or other query. Limit 2 might as well improve query performance on "test" query reducing amounts of joins made between rows if no ordering is done at the end of the query. This way you run test query only once, get necessary values and reduce resource footprint used by this function. For example:
CREATE OR REPLACE FUNCTION lcdm_main.test(IN p_member_crn character varying)
RETURNS TABLE(fname character varying, lname character varying) AS
$BODY$
declare
vfname text;
vlastname text;
countRow integer;
BEGIN
select q1.fname,q1.lastname, count(*) over() as cc
from (
select t.fname, t.lastname
from testq t
where t.email = p_member_crn limit 2
)q1
into vfname,vlastname,countRow;
if countRow > 1 then
RETURN QUERY select concat( error_msg_cd , error_msg_nm )::character varying , NULL:: character varying
from lcdm_main.error where error_Cndtn_nm ='SMRF';
ELSE
fname := vfname;
lname := vlastname;
return next;
END IF ;
END;
$BODY$
LANGUAGE plpgsql VOLATILE SECURITY DEFINER
COST 100
ROWS 1000;