Plpgsql Language Feature?

In the context of a plpgsql script is a possible to access the number of
rows effected by an UPDATE or DELETE statement? It could be used to
simplify, and probably improve performance on, many scripts. I would
like to do something like:

update foo set bar = bar + new.amount where key = new.key;

if ROWS_UPDATED > 0 then
-- This would be less likely to occur
insert into foo (key, bar) values (new.key, new.bar);
end if;

Otherwise I must construct an additional SELECTstatement to see if the
row exists. Something like:

select 1 from foo where key = new.key;

if found then
update foo set bar = bar + new.amount where key = new.key;
else
insert into foo (key, bar) values (new.key, new.bar);
end if;

Might this be a worth while feature if it does not now exist?

TIA.