Table Function return a result set that mimics what we would normally expect from a traditional SQL SELECT statement

Table functions are a new feature in Oracle9i that allow you to define a set of PL/SQL statements that will, when queried, behave just as a regular query to table would. The added benefit to having a table function is that you can perform transformations to the data in question before it is returned in the result set.

CREATE OBJECT
Create our own object type called PERSON_DETAILS.Then we create a table of PERSON_DETAILS called PERSON_DETAILS_TABLE.

CREATE TYPE PERSON_DETAILS AS OBJECT
       (USER_NAME     VARCHAR2(50),
        ADDRESS       VARCHAR2(50),
        LOCATION      VARCHAR2(50));
/
CREATE TYPE PERSON_DETAILS_TABLE AS TABLE OF PERSON_DETAILS;
/

PIPELINED Clause

Within the CREATE FUNCTION clause, there is a new option called PIPELINED. This option tells Oracle to return the results of the function as they are processed, and not wait for a complete execution or completion of the result set. This pipelining of the result set to one row at a time has the immediate advantage of not requiring excessive memory or disk staging resources.
PIPE ROW(out_rec)

The PIPE ROW statement is the interface or mechanism to send a piped row through the PIPELINED option through to the caller of the function.

Working with a simple pipelined function requires 2 things

• collection type
• pipelined function

 CREATE OR REPLACE TYPE number_ntt AS TABLE OF NUMBER;

CREATE FUNCTION row_generator(rows_in IN P_INTEGER) RETURN number_ntt
  PIPELINED IS
BEGIN
  FOR i IN 1 .. rows_in LOOP
    PIPE ROW(i);
  END LOOP;
  RETURN;
END;
/

The CSV_TABLE is a collection which has rows of comma separated value.

CREATE OR REPLACE TYPE "CSV_TABLE" as table of varchar2(32767)
CSV_TABLE

FUNCTION CSV_TO_TABLE(p_delimted_string VARCHAR2,
                         p_delimter        VARCHAR2 := ',')
  RETURN CSV_TABLE
  PIPELINED IS
  indexCount PLS_INTEGER;
  csvString  VARCHAR2(32767) := p_delimted_string;
BEGIN
  LOOP
    indexCount := instr(csvString, p_delimter);
  
    IF indexCount > 0 THEN
      PIPE ROW(substr(csvString, 1, indexCount - 1));
      csvString := substr(csvString, indexCount + length(p_delimter));
    ELSE
      PIPE ROW(csvString);
      EXIT;
    END IF;
  
  END LOOP;
  RETURN;
END CSV_TO_TABLE;

Input

A,B,C,D

Output

A
B
C
D

The above output is a collection.

 SELECT PACKAGE_NAME.CSV_TO_TABLE('A,B,C,D') FROM DUAL;