Create a function with multiple columns as arguments in Postgresql - postgresql

Question

I am trying to create a function that takes a table and a variable number of columns as arguments, and then returns a table without rows that have duplicates on all of those columns. I am trying to figure out how to have a variable number of columns as arguments, and I have gathered that I will probably need a VARIADIC argument, but I am not sure how to implement it. What I have so far:
CREATE FUNCTION remove_duplicates(orig_table, VARIADIC sel_columns column)
RETURNS table AS $$
SELECT * FROM
(SELECT *,
count(*) over (partition by sel_columns) AS count
FROM orig_table)
WHERE count = 1;
$$ LANGUAGE SQL;
As an example, if I had a table like this:
cola | colb | colc
-------------------
a | b | 1
a | b | 2
a | c | 3
a | d | 4
I would like to run SELECT * FROM remove_duplicates(mytable, cola, colb) and get this result:
cola | colb | colc
-------------------
a | c | 3
a | d | 4
Thank you for the help. I'm using postgresql 9.4.9

Answer 1

You'll cannot get what you want with a simple SQL function, you need the power of a procedural language. A possible solution is:
CREATE OR REPLACE FUNCTION remove_duplicates(orig_table anyelement, VARIADIC sel_columns text[])
RETURNS SETOF anyelement AS $$
DECLARE
orig_table_columns TEXT;
BEGIN
SELECT array_to_string(array_agg(quote_ident(column_name)),',') INTO orig_table_columns FROM information_schema.columns WHERE table_name = CAST(pg_typeof(orig_table) AS TEXT);
RETURN QUERY EXECUTE 'SELECT ' || orig_table_columns || ' FROM '
|| '(SELECT *, '
|| ' count(*) over (partition by ' || array_to_string(sel_columns, ',') || ') AS count '
|| 'FROM ' || pg_typeof(orig_table) || ') AS tmp '
|| ' WHERE count = 1 ';
END
$$ LANGUAGE PLPGSQL;
SELECT * FROM remove_duplicates(NULL::tests, 'cola', 'colb');
Don't forget to do your changes to avoid SQL Injection.
EDIT: For a very good explanation about functions with dynamic return types see Erwin's answer here.