Parse string into different column and optimization

Question

We are in process of building an audit process for any changes that occur automatically or manually by the user on some of the table data. To do this we have two options:
1. Have master table to store the audit event summary and a detail table to store each column change with old and new values. Something like:
CREATE TABLE TEST_ADT_DTL
  EVNT_ID      NUMBER,
  COL_NAME     VARCHAR2(1000),
  OLD_COL_VAL     VARCHAR2(1000),
  NEW_COL_VAL     VARCHAR2(1000)
);but this approach has some processing overhead since for the changes to each record there will be multiple records based on number of columns updated. If we are loading 40K transaction twice a month, and the changes are almost 30-40% so the detail table will grow considerably.
2. To have the detail table with one column that will have a concatenated string of changes with field name, old and new values.
CREATE TABLE TEST_ADT_EVNT
  EVNT_ID        NUMBER,
  TBL_NAME       VARCHAR2(100),
  OPER_CD        VARCHAR2(1),
  USR_ID         VARCHAR2(10),
  ACT_DT         DATE,
  PK_STRNG_VAL   VARCHAR2(100),
  CMNT_TXT       VARCHAR2(1000) 
CREATE TABLE TEST_ADT_DTL
  EVNT_ID      NUMBER,
  ADT_LOG     VARCHAR2(1000)
INSERT INTO TEST_ADT_EVNT VALUES (1, 'CUSTOMER', 'A', 'ABC', SYSDATE, 'CUS0001', 'SOME COMMENT');
INSERT INTO TEST_ADT_EVNT VALUES (2, 'CUSTOMER', 'U', 'ABC', SYSDATE, 'CUS0001', 'SOME COMMENT');
INSERT INTO TEST_ADT_EVNT VALUES (3, 'ORDER', 'A', 'XYZ', SYSDATE, 'CUS0002', 'SOME COMMENT');
INSERT INTO TEST_ADT_EVNT VALUES (4, 'ORDER', 'U', 'EFG', SYSDATE, 'CUS0002', 'SOME COMMENT');
INSERT INTO TEST_ADT_EVNT VALUES (5, 'ORDER', 'U', 'XYZ', SYSDATE, 'CUS0002', 'SOME COMMENT');
INSERT INTO TEST_ADT_DTL VALUES (2, 'FIELD:CITY,OLD:AVENEL,NEW:EDISON;FIELD:ZIP,OLD:07001,NEW:07056;');
INSERT INTO TEST_ADT_DTL VALUES (4, 'FIELD:ADDRESS,OLD:234 ROGER ST,NEW:124 WEST FIELD AVE;FIELD:STATE,OLD:NJ,NEW:NY;FIELD:PHONE,OLD:,NEW:2012230912;');
INSERT INTO TEST_ADT_DTL VALUES (5, 'FIELD:MID_NAME,OLD:,NEW:JASON;FIELD:ADDRESS,OLD:,NEW:3 COURT CT;');
COMMIT;I want to know if we want to generate a report for audit log, how can I display the data from detail table in columns. I mean how to parse the ADT_LOG column to show the data in three different columns like:
FIELD     OLD         NEW
CITY     AVENEL    EDISON
ZIP       07001      07056
.along with the columns from EVNT table.
And, want to know which approach would be better.

Answer 1

hey I think I finally got it using the model clause.
not sure if this will be faster or not.
you can increase the number of iterations if you are not hitting them all,
( the lower your iteration number the faster this will run)
select adt_log, field, old, new from
with TEST_ADT_DTL as
(select 2 evnt_id, 'FIELD:CITY,OLD:AVENEL,NEW:EDISON;FIELD:ZIP,OLD:07001,NEW:07056;' ADT_LOG FROM DUAL UNION
select 4, 'FIELD:ADDRESS,OLD:234 ROGER ST,NEW:124 WEST FIELD AVE;FIELD:STATE,OLD:NJ,NEW:NY;FIELD:PHONE,OLD:,NEW:2012230912;' from dual union
select 5, 'FIELD:MID_NAME,OLD:,NEW:JASON;FIELD:ADDRESS,OLD:,NEW:3 COURT CT;' from dual
select evnt_id, adt_log, field, old, new  from test_adt_dtl
model return updated rows
partition by (evnt_id)
dimension by ( 0 d)
measures (adt_log, adt_log field, adt_log old, adt_log new, 0 it_num )
rules iterate (50) -- until ?
adt_log[any] = adt_log[0],
field[0] = substr(adt_log[0], instr(adt_log[0],'FIELD',1,1)+6, instr(adt_log[0],',',1,1) - instr(adt_log[0],'FIELD',1,1)-6),
old[0] =   substr(adt_log[0], instr(adt_log[0],'OLD',1,1)+4, instr(adt_log[0],',',1,2) - instr(adt_log[0],'OLD',1,1)-4),
new[0] =   substr(adt_log[0], instr(adt_log[0],'NEW',1,1)+4, instr(adt_log[0],';',1,1) - instr(adt_log[0],'NEW',1,1)-4),
field[iteration_number ] =    substr(adt_log[0],
                                     instr(adt_log[0],'FIELD:',1,iteration_number + 1 ) + 6,
                                     (instr(adt_log[0],',',( instr(adt_log[0],'FIELD:',1,iteration_number + 1 ) + 6 ),1)
                                     (  instr(adt_log[0],'FIELD:',1,iteration_number + 1 ) + 6))
old[iteration_number ] =    substr(adt_log[0],
                                     instr(adt_log[0],'OLD:',1,iteration_number + 1 ) + 4,
                                     (instr(adt_log[0],',',( instr(adt_log[0],'OLD:',1,iteration_number + 1 ) + 4 ),1)
                                     (  instr(adt_log[0],'OLD:',1,iteration_number + 1 ) + 4))
new[iteration_number]  =    substr(adt_log[0],
                                     instr(adt_log[0],'NEW:',1,iteration_number + 1 ) + 4,
                                       (instr(adt_log[0],';',1,iteration_number + 1)
                                       (instr(adt_log[0],'NEW:',1,iteration_number + 1 ) + 4)
order by evnt_id, it_num
where new is not nullEdited by: pollywog on Apr 13, 2010 10:28 AM