Query doubt

Question

Dear Masters,
I have table called price detail with the following structure (with some sample rows)
with price_detail
as
select to_date('1/1/2013' , 'mm/dd/yyyy') sell_date, 100 dealer_price, 10 discount from dual
union all
select to_date('1/2/2013' , 'mm/dd/yyyy') sell_date, 120 dealer_price, 20 discount from dual
union all
select to_date('1/3/2013' , 'mm/dd/yyyy') sell_date, 200 dealer_price, 10 discount from dual
union all
select to_date('1/4/2013' , 'mm/dd/yyyy') sell_date, 100 dealer_price, 20 discount from dual
union all
select to_date('1/5/2013' , 'mm/dd/yyyy') sell_date, 120 dealer_price, 10 discount from dual
union all
select to_date('1/6/2013' , 'mm/dd/yyyy') sell_date, 210 dealer_price, 30 discount from dual
union all
select to_date('1/7/2013' , 'mm/dd/yyyy') sell_date, 140 dealer_price, 40 discount from dual
union all
select to_date('1/8/2013' , 'mm/dd/yyyy') sell_date, 150 dealer_price, 10 discount from dual
union all
select to_date('1/9/2013' , 'mm/dd/yyyy') sell_date, 200 dealer_price, 20 discount from dual
union all
select to_date('1/10/2013', 'mm/dd/yyyy') sell_date, 100 dealer_price, 10 discount from dual
union all
select to_date('1/11/2013', 'mm/dd/yyyy') sell_date, 120 dealer_price, 30 discount from dual
union all
select to_date('1/12/2013', 'mm/dd/yyyy') sell_date, 140 dealer_price, 20 discount from dual
Net_price and Net_price_before_discount are two calculated columns
Net price = Dealer_price – Discount ( from 2rd day after the current day  , so if we are calculating for 1/4/2013 then the discount should come from 1/6/2013)
Net_ Price_ bd = Dealer_price (  From the 2rd day before the current day, so if we are calculating for 1/4/2013 then the Dealer_price should come from 1/2/2013) – Discount
Please refer below table after the highlighted date
Sell_date(mm/dd/yyyy)
Dealer_Price
Discount
Net Price
Net_Price_bd
Ch_flag
1/1/2013
100
10
90
140
1/2/2013
120
20
100
80
1/3/2013
200
10
190
90
1/4/2013
100
20
70
100
1/5/2013
120
10
80
190
1
1/6/2013
210
30
200
70
1
1/7/2013
140
40
120
80
1/8/2013
150
10
140
200
2
1/9/2013
200
20
170
120
1/10/2013
100
10
80
140
2
1/11/2013
120
30
120
170
1
1/12/2013
140
20
140
80
The task is that if  for a row ch_flag  is set  to 1 (ch_ flag is available in the same  table)  which means Discount has been changed for that row and as  Discount is being used to calculate the Net_price  (for sell_date-2)  and Net_Price_bd ( for sell_date) , corresoding records should get changed.
If for a row ch_flag is set to 2 which means Dealer_price has been changed for that row  and as Dealer_price is being used to calculate the net_price ( for sell_date) and Net_Price_bd ( sell_date +2 ), corresponding records should also get changed
e.g.   Say Discount has been modified for sell_date 1/5/2013 the flag is set to 1 , now Net_price_bd for sell_date and Net_price for sell_date -2 should get recalculated.
Please suggest me.
There will be multiple rows with ch_flag 1 and 2 , so i need to update all the corresonding rows in a single shot.
Can it be achived through Merge ?
Ab

Answer 1

Below may be a possible solution to your problem:
alter session set nls_date_format = 'mm/dd/yyyy';
drop table test_table;
create table test_table as
select to_date('1/1/2013' , 'mm/dd/yyyy') sell_date, 100 dealer_price, 10 discount from dual union all
select to_date('1/2/2013' , 'mm/dd/yyyy') sell_date, 120 dealer_price, 20 discount from dual union all
select to_date('1/3/2013' , 'mm/dd/yyyy') sell_date, 200 dealer_price, 10 discount from dual union all
select to_date('1/4/2013' , 'mm/dd/yyyy') sell_date, 100 dealer_price, 20 discount from dual union all
select to_date('1/5/2013' , 'mm/dd/yyyy') sell_date, 120 dealer_price, 10 discount from dual union all
select to_date('1/6/2013' , 'mm/dd/yyyy') sell_date, 210 dealer_price, 30 discount from dual union all
select to_date('1/7/2013' , 'mm/dd/yyyy') sell_date, 140 dealer_price, 40 discount from dual union all
select to_date('1/8/2013' , 'mm/dd/yyyy') sell_date, 150 dealer_price, 10 discount from dual union all
select to_date('1/9/2013' , 'mm/dd/yyyy') sell_date, 200 dealer_price, 20 discount from dual union all
select to_date('1/10/2013', 'mm/dd/yyyy') sell_date, 100 dealer_price, 10 discount from dual union all
select to_date('1/11/2013', 'mm/dd/yyyy') sell_date, 120 dealer_price, 30 discount from dual union all
select to_date('1/12/2013', 'mm/dd/yyyy') sell_date, 140 dealer_price, 20 discount from dual;
table test_table created.
alter table test_table add net_price number;
alter table test_table add net_price_bd number;
alter table test_table add ch_flag number;
table test_table altered.
--Calculate and update the Net Price and Net Price BD
merge into test_table t
using (
select sell_date,
       dealer_price,
       discount,
       dealer_price - lead(discount, 2, 0) over(order by sell_date) net_price,
       lag(dealer_price, 2, 0) over (order by sell_date) - discount net_price_bd,
       ch_flag
  from test_table
) s
on (t.sell_date = s.sell_date)
when matched then
  update set net_price = s.net_price, net_price_bd = s.net_price_bd;
12 rows merged.
Now, Lets see how the data looks:
select *
  from test_table;
SELL_DATE                 DEALER_PRICE           DISCOUNT               NET_PRICE              NET_PRICE_BD           CH_FLAG               
01/01/2013                100                    10                     90                     -10                                          
01/02/2013                120                    20                     100                    -20                                          
01/03/2013                200                    10                     190                    90                                           
01/04/2013                100                    20                     70                     100                                          
01/05/2013                120                    10                     80                     190                                          
01/06/2013                210                    30                     200                    70                                           
01/07/2013                140                    40                     120                    80                                           
01/08/2013                150                    10                     140                    200                                          
01/09/2013                200                    20                     170                    120                                          
01/10/2013                100                    10                     80                     140                                          
01/11/2013                120                    30                     120                    170                                          
01/12/2013                140                    20                     140                    80                                           
12 rows selected
Now, update few discounts and dealer price and set corresponding CH_FLAG.
update test_table set discount = 15, ch_flag = 1 where sell_date =  to_date('01/05/2013');
update test_table set discount = 25, ch_flag = 1 where sell_date =  to_date('01/06/2013');
update test_table set discount = 35, ch_flag = 1 where sell_date =  to_date('01/11/2013');
update test_table set dealer_price = 175, ch_flag = 2 where sell_date =  to_date('01/08/2013');
update test_table set dealer_price = 110, ch_flag = 2 where sell_date =  to_date('01/10/2013');
commit;
Update the TEST_TABLE for the rows dependent on previous/later dates, where the discount and/or dealer price is modified:
merge into test_table t
using
  select sell_date,
         discount,
         dealer_price,
         decode(a.prev_ch_flag, 1, dealer_price - next_discount, net_price) upd_net_price,
         decode(a.prev_ch_flag, 1, 'Modified', 'Same') is_net_price_modified,
         decode(a.next_ch_flag, 2, next_dealer - discount, net_price_bd) upd_net_price_bd,
         decode(a.next_ch_flag, 2, 'Modified', 'Same') is_net_price_bd_modified,
         ch_flag,
         next_ch_flag,
         prev_ch_flag
    from (
          select sell_date,
                 discount,
                 lead(discount, 2, 0) over(order by sell_date) next_discount,
                 dealer_price,
                 lag(dealer_price, 2, 0) over (order by sell_date) next_dealer,
                 ch_flag,
                 lead(ch_flag, 2, 0) over (order by sell_date) next_ch_flag,
                 lag(ch_flag, 2, 0) over (order by sell_date) prev_ch_flag,
                 net_price,
                 net_price_bd
            from test_table
         ) a
) s
on (t.sell_date = s.sell_date and (s.is_net_price_modified = 'Modified' or s.is_net_price_bd_modified = 'Modified'))
when matched then
  update set net_price = s.upd_net_price, net_price_bd = s.upd_net_price_bd;
3 rows merged.
Now, lets see the final data:
select *
  from test_table;
SELL_DATE                 DEALER_PRICE           DISCOUNT               NET_PRICE              NET_PRICE_BD           CH_FLAG               
01/01/2013                100                    10                     90                     -10                                          
01/02/2013                120                    20                     100                    -20                                          
01/03/2013                200                    10                     190                    90                                           
01/04/2013                100                    20                     70                     100                                          
01/05/2013                120                    15                     80                     190                    1                     
01/06/2013                210                    25                     200                    75                     1                     
01/07/2013                140                    40                     120                    80                                           
01/08/2013                175                    10                     165                    200                    2                     
01/09/2013                200                    20                     170                    120                                          
01/10/2013                110                    10                     80                     140                    2                     
01/11/2013                120                    35                     120                    170                    1                     
01/12/2013                140                    20                     140                    80                                           
12 rows selected
This operation is possible using a regular Update, but not sure why I did not use it; perhaps I was thinking weird while arriving at solution