Nested Table Functions
The nesting of table functions refers to the execution of several table functions in a nested manner: the results of one are sent to the next for further processing, and those results are sent on to the next, and so on. This process is sometimes known as daisy-chaining . Combined with pipelining, the nesting of table functions provides a particularly powerful technique for ETL processing.
I’ll demonstrate by coding a function that accepts the results of my date_parse table function via a cursor and performs an operation on them. To avoid the clutter of explaining a complex ETL transformation, we’ll stick to a simple one by adding up the order_number, year, quarter, and month values queried. Here’s the function.
/* File on web: next_in_line.sql */
CREATE OR REPLACE FUNCTION next_in_line ( p_curs SYS_REFCURSOR )
RETURN next_t
PIPELINED IS
v_ret_val next_t := next_t( );
-- local variables for cursor results
v_on NUMBER;
v_dt VARCHAR2(1);
v_yr NUMBER;
v_qt NUMBER;
v_mt NUMBER;
BEGIN
-- for all date components from the cursor...
LOOP
FETCH p_curs INTO v_on, v_dt, v_yr, v_qt, v_mt;
EXIT WHEN p_curs%NOTFOUND;
-- pipe out the sum of the components
PIPE ROW(next_o(v_on + v_yr + v_qt + v_mt));
END LOOP; -- every date component
RETURN;
END;We’ve already covered most of the syntax in this example. Only the syntax used to nest the two functions together in a query is new:
SELECT * FROM TABLE(next_in_line (CURSOR (SELECT * FROM TABLE(date_parse (CURSOR (SELECT * FROM ...