Tek-Tips is the largest IT community on the Internet today!

Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!

  • Congratulations SkipVought on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!

Merge records to get sequence in TIME with condition, Please help...

Status
Not open for further replies.

peac3

Technical User
Jan 17, 2009
226
0
0
AU
Hello all,

I am trying to extract minimum maximum time from terminal ID until it changes the business according to our flag.

1 = means changing business
0 = same business

and I need to merge into one record from multiple records in multiple records in the sequence of time.
as per below codes

sample data

Code:
DECLARE @Test TABLE (TERMINAL_ID int,
                     TERMINAL_NAME varchar(200),
                     START_DATE datetime,
                     END_DATE datetime,
                     CHANGE_FLAG bit)
INSERT INTO @Test VALUES(1128565, 'JAPANPOST BANK(705910) OKINAWA JP','20160501','20161231',1)
INSERT INTO @Test VALUES(1128565, 'JAPANPOST BANK(003500) TOKYO JP 1','20170101','20170531',0)
INSERT INTO @Test VALUES(1128565, 'JAPANPOST BANK(003500) TOKYO JP 2','20170601','20170630',0)
INSERT INTO @Test VALUES(1128565, 'JAPAN RAMEN','20170701','20170831',1)
INSERT INTO @Test VALUES(1128565, 'JAPAN RAMEN AU','20170901','20170930',0)
INSERT INTO @Test VALUES(1128565, 'JAPAN RAMEN AUS','20171001','99991231',0)
INSERT INTO @Test VALUES(4184558, 'POST Diamond Creek F VicAU','20160501','20161031',1)
INSERT INTO @Test VALUES(4184558, 'BRIDGE CLINIC MURRAY BRIDGESA AU','20170501','20170630',1)
INSERT INTO @Test VALUES(4184558, 'BRIDGE CLINIC MURRAY BRIDGESA AU','20170701','20171231',0)
INSERT INTO @Test VALUES(4184558, 'BRIDGE CLINIC MURRAY BRIDGESA AU','20180101','99991231',0)

Expected result:

Code:
1128565	JAPANPOST BANK(705910) OKINAWA JP	2016-05-01 00:00:00.000	2017-06-30 00:00:00.000	1
1128565	JAPAN RAMEN	                        2017-07-01 00:00:00.000	9999-12-31 00:00:00.000	1
4184558	POST Diamond Creek F VicAU	        2016-05-01 00:00:00.000	2016-10-31 00:00:00.000	1
4184558	BRIDGE CLINIC MURRAY BRIDGESA AU	2017-05-01 00:00:00.000	9999-12-31 00:00:00.000	1

any help would be much appreciated!
 
one way I was thinking to create additional column sequence first to handle within the same terminal I D.
So, if I see the Change Flag = 1, I will add number but if I see Change Flag = 0 I populate the same number as before

Code:
TERM_ID TERM_NAME                               STRT_D                   END_D                  CHANGE_F  SEQ
1128565	JAPANPOST BANK(705910) OKINAWA JP	2016-05-01 00:00:00.000	2016-12-31 00:00:00.000	1        1
1128565	JAPANPOST BANK(003500) TOKYO JP 1	2017-01-01 00:00:00.000	2017-05-31 00:00:00.000	0        1
1128565	JAPANPOST BANK(003500) TOKYO JP 2	2017-06-01 00:00:00.000	2017-06-30 00:00:00.000	0        1
1128565	JAPAN RAMEN	                        2017-07-01 00:00:00.000	2017-08-31 00:00:00.000	1        2
1128565	JAPAN RAMEN AU	                        2017-09-01 00:00:00.000	2017-09-30 00:00:00.000	0        2
1128565	JAPAN RAMEN AUS	                        2017-10-01 00:00:00.000	9999-12-31 00:00:00.000	0        2
4184558	POST Diamond Creek F VicAU	        2016-05-01 00:00:00.000	2016-10-31 00:00:00.000	1        3
4184558	BRIDGE CLINIC MURRAY BRIDGESA AU	2017-05-01 00:00:00.000	2017-06-30 00:00:00.000	1        4
4184558	BRIDGE CLINIC MURRAY BRIDGESA AU	2017-07-01 00:00:00.000	2017-12-31 00:00:00.000	0        4
4184558	BRIDGE CLINIC MURRAY BRIDGESA AU	2018-01-01 00:00:00.000	9999-12-31 00:00:00.000	0        4

Any idea to do this without using loop?
Thanks.
 
what version of SQL server you going to use for this?

for 2008 R2 + the following will work - depending on your volumes spliting the 2 CTE's onto temp tables with support indexes may be better on performance
Code:
;
with rownumbers
as
(select *
      , row_number() over (partition by TERMINAL_ID order by START_DATE) as rownum
       from @Test
),
groupnumbers
as
(select *
      , row_number() over (partition by TERMINAL_ID order by START_DATE) as grp

       from rownumbers w1
       where w1.CHANGE_FLAG = 1
)
select w1.TERMINAL_ID
     , w1.TERMINAL_NAME
     , w1.START_DATE
     --, w1.END_DATE
     , w1.CHANGE_FLAG
     --, w1.rownum
     --, w1.grp
     --, endgroup.TERMINAL_ID
     --, endgroup.TERMINAL_NAME
     --, endgroup.START_DATE
     , endgroup.END_DATE
     --, endgroup.CHANGE_FLAG
     --, endgroup.rownum
from groupnumbers w1
outer apply (select top 1 *
       from groupnumbers g2
       where g2.TERMINAL_ID = w1.TERMINAL_ID
           and g2.grp = w1.grp + 1
) nextgroup

outer apply (select top 1 *
       from rownumbers rn1
       where rn1.TERMINAL_ID = w1.TERMINAL_ID
           and rn1.rownum < coalesce(nextgroup.rownum, 9999999999)
   order by rn1.rownum desc
) endgroup
order by w1.TERMINAL_ID
       , w1.START_DATE



Regards

Frederico Fonseca
SysSoft Integrated Ltd

FAQ219-2884
FAQ181-2886
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor

Back
Top