Query to translate one string column by referring to another 2 table

Question

I have 3 table which is
1)RATIO_HISTORY
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+ RATIO_ID + RATIO_CRITERIA
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+ 1 + (subjek=2 AND gred<=240) OR (subjek=3 AND gred<=240) OR (subjek=4 AND gred<=567) OR (subjek=5 AND gred<=565)
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+ 2 + (subjek=4 AND gred<=567) OR (subjek=5 AND gred<=565)
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
2)LKP_SUBJEK
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
SUBJEK_ID SUBJEK
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+ 2 + English
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+ 3 + Chemical
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+ 4 + Biology
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+ 5 + Physic
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+ 6 + History
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+ 7 + Mathematics
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
3) LKP_GRED
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
GRED_ID    GRED
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+ 240 + 1
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+ 567 + 2
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+ 565 + 3
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+ 568 + D
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
what i want to do is, i want to create a query than can translate the column of Criteria in table ratio_history to look like this;
example: ratio_id=1
(subjek=English AND gred<=1) OR (subjek=Chemical AND gred<=1) OR (subjek=Biology AND gred<=2) OR (subjek=Physics AND gred<=3)
Can anyone help me?
thank you in advance.
Edited by: user5441678 on Mar 17, 2011 7:12 PM

Answer 1

Hi,
Welcome to the forum!
Whenever you have a question, please post your sample data in a form that people can use to re-create the problem and to test their solutions.
For example:
CREATE TABLE     ratio_history
AS
SELECT      1      AS ratio_id
,     '(subjek=2 AND gred<=240) OR (subjek=3 AND gred<=240) OR (subjek=4 AND gred<=567) OR (subjek=5 AND gred<=565)'
          AS ratio_criteria
FROM     dual     UNION ALL
SELECT     2, '(subjek=4 AND gred<=567) OR (subjek=5 AND gred<=565)'     from dual;
CREATE TABLE     lkp_subjek
AS
SELECT     2 AS subjek_id,     'English' AS subjek     FROM dual     UNION ALL
SELECT      3,          'Chemical'          FROM dual     UNION ALL
SELECT     4,          'Biology'          FROM dual     UNION ALL
SELECT     5,          'Physic'          FROM dual     UNION ALL
SELECT     6,          'History'          FROM dual     UNION ALL
SELECT     7,          'Mathematics'          FROM dual;
CREATE TABLE     lkp_gred
AS
SELECT     240 AS gred_id,     '1' AS gred     FROM dual     UNION ALL
SELECT     567,          '2'          FROM dual     UNION ALL
SELECT     565,          '3'          FROM dual     UNION ALL
SELECT     568,          'D'          FROM dual;Be sure to mention which version of Oracle you're using. The solution below works in Oracle 10 and up. (I tested it in Oracle 10.2.0.1.0, but it should work in Oracle 10.1 as well.
Regular expressions make it easy to split ratio_criteria up into parts. I assume that every row will have from 1 to 6 pairs of numbers (subjek_id and a gred_id), so we can have up to 7 "items" on each row, counting the text that comes after the last number. That's what the "magic number" 7 is in sub-query cntr. Of course, you can increase that if your real data has more numbers.
WITH     cntr     AS
     SELECT     LEVEL     AS n
     FROM     dual
     CONNECT BY     LEVEL <= 7     -- Or whatever you need
,     ratio_items     AS
     SELECT     r.ratio_id
     ,     c.n                                        AS item_num
     ,     REGEXP_SUBSTR ( ratio_criteria, '[^0-9]+', 1, (2 * c.n) - 1)     AS subjek_txt
     ,     REGEXP_SUBSTR ( ratio_criteria, '[0-9]+',  1, (2 * c.n) - 1)     AS subjek_id
     ,     REGEXP_SUBSTR ( ratio_criteria, '[^0-9]+', 1, (2 * c.n)        )     AS gred_txt
     ,     REGEXP_SUBSTR ( ratio_criteria, '[0-9]+',  1, (2 * c.n)        )     AS gred_id
     FROM          ratio_history     r
     CROSS JOIN     cntr          c
,     translated_items     AS
     SELECT     i.ratio_id
     ,     i.item_num
     ,     i.subjek_txt
          || s.subjek
          || i.gred_txt
          || g.gred          AS txt
     FROM          ratio_items     i
     LEFT OUTER JOIN     lkp_subjek     s  ON  s.subjek_id = TO_NUMBER (i.subjek_id)
     LEFT OUTER JOIN     lkp_gred     g  ON  g.gred_id   = TO_NUMBER (i.gred_id)
SELECT       ratio_id
,       REPLACE ( SYS_CONNECT_BY_PATH (txt, '~?')
            ,                      '~?'
            )     AS translated_history
FROM       translated_items
WHERE       CONNECT_BY_ISLEAF     = 1
START WITH     item_num     = 1
CONNECT BY     item_num     = PRIOR item_num + 1
     AND     ratio_id     = PRIOR ratio_id
ORDER BY  ratio_id
,       item_num
;Sub-query ratio_items splits each row into 7 parts, each potentially containing some text that comes before the subjek_id, the subjek_id, some text that comes before the grep_id, and the grep_id.
Sub-query translated_items looks up the translations of the ids in the appropriate tables. Outer-joins are used, because the last item will not have any ids, but we still need to preserve its text. This means that if ratio_criteria has an incorrect id, the query will still run; just the incorrect part will be missing.
The main query re-unites the translated items, in order, using SYS_CONNECT_BY_PATH, which requires a delimiter between items. The delimiter must never appear in the strings to be combined. I assumed that the string '~?' will never occur in your real data. You can change that to any string, of any length, if you want. (The individual characters in the delimiter may appear in the text, so it's okay to have ~'s and ?'s, you just can't have a ~ followed immediately by a ?.) We don't actually want a delimiter in this problem, so REPLACE gets rid of it.
Output:
  RATIO_ID TRANSLATED_HISTORY
         1 (subjek=English AND gred<=1) OR (subjek=Chemical AND gred<=1
           ) OR (subjek=Biology AND gred<=2) OR (subjek=Physic AND gred
           <=3)
         2 (subjek=Biology AND gred<=2) OR (subjek=Physic AND gred<=3)Interesting problem!