Thinking Cap's On Please: How to create a range in PL/SQL similar to C++

Question

Okay, I had to step on go "Whoa!"
Goal: to write a segment of code that will detect and identify a value in a range and then replace it by the following progress on the next element. In other words range A . . Z, and value is current set to D. I want to detect "D" and UPDATE it to "E". I saw an example of using DECODE as follows:
For example:
You could use the decode function in an SQL statement as follows:
SELECT supplier_name,
decode(supplier_id,     10000,     'IBM',
10001,     'Microsoft',
10002,     'Hewlett Packard',
'Gateway') result
FROM suppliers;But, you need to still have logic increment the value within the range to the next higher value. Okay, it can be done, but seems like it could be painful without a build in Oracle function. The code that I must adapt is shown below where the AGL-X, or PGL-X will need to be incremented to AGL-Y, or PGL-Y. It might be easier to just use decode and hard fix 26 values for each if.
declare
cursor c1 is
select * from alumni.amrexrt
where amrexrt_pidm = 2895;
cursor c2 is
select * from alumni.amrexrt
where amrexrt_pidm = 2895;
amrexrt_rec     alumni.amrexrt%rowtype;
r2              alumni.amrexrt%rowtype;
begin
    for amrexrt_rec in c1 loop
    if (amrexrt_rec.amrexrt_exrs_code = 'AGL') then
    update amrexrt
        set amrexrt_exrs_code = 'AGL-X'
        where amrexrt_pidm = 2895
        and amrexrt_exrs_code = 'AGL';
    elsif (amrexrt_rec.amrexrt_exrs_code = 'MGL') then
    update amrexrt
        set amrexrt_exrs_code = 'MGL-X'
        where amrexrt_pidm = 2895
        and amrexrt_exrs_code = 'MGL';
    elsif (amrexrt_rec.amrexrt_exrs_code = 'PGL') then
    update amrexrt
        set amrexrt_exrs_code = 'PGL-X'
        where amrexrt_pidm = 2895
        and amrexrt_exrs_code = 'PGL';
    end if;
    Insert into alumni.amrexrt
        amrexrt_pidm,
        amrexrt_exrs_code,
        amrexrt_ext_value,
        amrexrt_ext_score,
        amrexrt_ext_level,
        amrexrt_activity_date
    Values
        amrexrt_rec.amrexrt_pidm,
        amrexrt_rec.amrexrt_exrs_code,
        amrexrt_rec.amrexrt_ext_value,
        '754',                                 --amrexrt_rec.amrexrt_ext_score,
        amrexrt_rec.amrexrt_ext_level,
        amrexrt_rec.amrexrt_activity_date
    end loop;
    for r2 in C2 loop
    dbms_output.put_line('PIDM: ' || r2.amrexrt_pidm || 'EXRS_CODE: ' || r2.amrexrt_exrs_code);
    end loop;
    --Rollback;
  end;In C++, you are able to specify a range and write it so the compiler can detect the position within the range. I don't or think that PL/SQL does this in the same way. Any thoughts? I am dry out of ideas.
Edited by: Preston on May 3, 2011 12:05 PM

Answer 1

Hi,
Preston wrote:
Very, interesting! Is there another way to utilize the first two lines without an "EXEC"? The security filter sniffs for EXEC outside of production and disables our connections.I used a bind_variable, :letters_in_order, just to make maintenance easier. In this simple example, that string was used twice; in your real application, you may need to use it more times. By setting a bind variable, I guaranteed that, if you needed to change the definition of :letters_in_order, you would only have to change the code in one place. If you have to make the change in several places, there's a chance that you'll miss one of those places, and that kind of mistake can be hard to find and fix.
You don't have to use any kind of variable; you can hard-code the string literal 'KPCOFGS' everywhere I use the bind variable, for example:
...  || SUBSTR ( 'KPCOFGS'
                 , 1 + INSTR ( 'KPCOFGS' ... 
The rest of this message shows different ways to use variables:
(1) Bind Variable (but no EXEC)
(2) Substitution Variable
(3) Column
<h3>(1) Bind Variable (but no EXEC)</h3>
"EXEC stmt;" is equivalent to BEGIN stmt; END;/", so you can try:
{code}
BEGIN
     :letters_in_order := 'KPCOFGS';
END;
{code}
The main query would remain exactly as posted earlier, with :letters_in_order.
I don't see any point in blocking EXEC, but allowing this, and maybe your security mechanism doesn't allow it.
<h3>(2) Substitution Variable</h3>
You could try a substritution variable:
{code}
DEFINE letters_in_order = KPCOFGS
{code}
and, wherever I originally used :lettrs_on_order, use the substitution variable, inside single-quotes:
{code}
... || SUBSTR ( '&letters_in_order'
     , 1 + INSTR ( '&letters_in_order' ...
{code}
Substitution variables are a much bigger security threat than bind variables, but that doesn't mean your system won't allow them.
<h3>(3) Column</h3>
The most promising approach (but perhaps the least efficient) would be to derive a column called letters_in_order right in the query. Here's one way:
{code}
WITH     got_letters_in_order     AS
     SELECT     code          -- and whatever other columns you need
     ,     'KPCOFGS' AS letters_in_order
     FROM     t
--     WHERE     ...          -- if you need any filtering, put it here
SELECT code
,     SUBSTR ( code
     , 1
     , INSTR (code, '-')
|| SUBSTR ( letters_in_order
     , 1 + INSTR ( letters_in_order
               , SUBSTR ( code
                    , 1 + INSTR (code, '-')
                    , 1
     , 1
     )     AS new_code
FROM got_letters_in_order          -- *** CHANGED ***
{code}