Split range of dates into rows

Question

Hi,
I have a table which basically has an account id and two dates (from and to). Sample data:
Account ID       Balance Amt        Valid From           Valid To
100000524       104.32               29/07/2010         05/08/2010
100000524       104.32               03/11/2010         25/12/2010
100000365       104.32               29/07/2010         05/08/2010What I currently do is that I loop on each record (per Account and Order by valid From). Then each row I processed into this query to split it into multiple rows - final result should be that for each of the month on the range I'll get the last day of it. See below for results:
select account_id, balance_amt,
       last_day(add_months(valid_from_dttm, level - 1)) as valid_to_dttm
FROM d
CONNECT BY LEVEL <= ( SELECT TRUNC( MONTHS_BETWEEN( LAST_DAY(valid_to_dttm),
                                                    TRUNC(valid_from_dttm, 'Month'))
                                  ) + 1 lvl
                      FROM dual )
Account ID       Balance Amt        Valid To
100000524       104.32               31/07/2010        
100000524       104.32               31/08/2010 It's pretty slow since I do it row by row. Is there any other way (faster) to do this? Thanks.
I forgot to say is that one of the trouble I'm getting here is that the dates are overlapping e.g.:
Account ID       Balance Amt        Valid From           Valid To
100000524       104.32               29/07/2010         05/08/2010
100000524       65                     05/08/2010         25/10/2010
100000524       10                     25/10/2010         20/12/2010So in this case the August month should only be displayed once with the value of 65 since it's the "latest".
Edited by: mortonmorton on Apr 25, 2011 2:14 AM
Edited by: mortonmorton on Apr 25, 2011 2:19 AM

Answer 1

Hi Guys,
I've tried both your queries and it's the same issue I'm getting which is duplicate rows of the month end dates. So using the model clause above (just tweaking the data):
with d as
( select 100000524  account_id, 104.32 balance_amt, to_date('29/07/2010','DD/MM/YYYY') valid_from_dttm,  to_date('05/09/2010','DD/MM/YYYY') valid_to_dttm from dual union
  select 100000524, 16.32,  to_date('05/09/2010','DD/MM/YYYY'), to_date('25/12/2010','DD/MM/YYYY') from dual union
  select 100000365 , 104.32,  to_date(' 29/07/2010 ','DD/MM/YYYY'), to_date('05/08/2010','DD/MM/YYYY') from dual
select account_id, balance_amt,  last_day(add_months(valid_to_dttm,-1)) new_valid_to_dttm
from d
model
partition by (account_id, balance_amt, valid_from_dttm, valid_to_dttm vtd)
dimension by (0 d)
measures (valid_to_dttm,  valid_from_dttm vfd )
rules
valid_to_dttm[for d  from 1 to
                      MONTHS_BETWEEN (LAST_DAY (valid_to_dttm[0]), TRUNC (vfd[0], 'Month') )
                      increment 1] =   add_months(valid_to_dttm[cv()-1],1)
order by account_id, valid_to_dttm
ACCOUNT_ID     BALANCE_AMT     NEW_VALID_TO_DTTM
100000365                     55.63     31/07/2010
100000365      55.63     31/08/2010
100000524      104.32     31/08/2010
100000524      104.32     30/09/2010
100000524      104.32     31/10/2010
100000524      16.32     30/11/2010
100000524      16.32     31/12/2010
100000524      16.32     31/01/2011
100000524      16.32     28/02/2011But expected output should be:
ACCOUNT_ID     BALANCE_AMT     NEW_VALID_TO_DTTM
100000524                104.32     31/07/2010
100000524      104.32     31/08/2010
100000524      16.32     30/09/2010
100000524      16.32     31/10/2010
100000524      16.32     30/11/2010
100000524      16.32     31/12/2010
100000365                55.63     31/07/2010
100000365      55.63     31/08/2010One thing to note here is the September month which had a final value of 16.32 - it overlapped the previous entry of 104.32.