Oracle Stored Procedure返回表的集合

Question

I want to write a procedure using collections.

I have a table - employee - and I want to return the following:

• TEST1: count of employees whose sal < 10000
• TEST2: count of employees whose dept > 10
• TEST3: count of employees whose hiredate >(SYSDATE-60)
• TEST4: count of employees whose grade = 1

My final recordset or array table should return below values. TEST1, TEST2, TEST3, TEST4 are description values residing in DESCRIPTION TABLE where as count values are from employee table.

Description   COUNT
TEST1         10 
TEST2         15
TEST3         25
TEST4         50

Please help with implementation.

Answer 1

This solution won't use your DESCRIPTION table to calculate counts dynamically, as that would be rather hard to achieve. So, here's a solution with hard-coded TESTn descriptions.

First, create the record type:

CREATE TYPE count_t AS OBJECT (
  description varchar2(100),
  cnt NUMBER(10)
);

Then, create the table type:

CREATE TYPE count_tt AS TABLE OF count_t;

Now, write the function:

CREATE OR REPLACE FUNCTION counts RETURN count_tt
IS
  v_result count_tt;
BEGIN
  SELECT count_t(description, cnt)
  BULK COLLECT INTO v_result
  FROM (
    SELECT 
      count(CASE WHEN sal < 10000 THEN 1 END) TEST1,
      count(CASE WHEN dept > 10 THEN 1 END) TEST2,
      count(CASE WHEN hiredate > SYSDATE - 60 THEN 1 END) TEST3,
      count(CASE WHEN grade = 1 THEN 1 END) TEST4
    FROM employees
  ) t
  -- Behold, the power of unpivot!
  UNPIVOT (
    cnt FOR description IN ("TEST1", "TEST2", "TEST3", "TEST4")
  );

  RETURN v_result;
END counts;
/

Now, call the function, eg from SQL:

SELECT * FROM TABLE(counts)

Enjoy.

By the way, I've written up a blog post comparing this solution from a performance perspective with others that do not use UNPIVOT