|
|
| Author |
Message |
rbowers
Member
Joined: 16 Jan 2010
Posts: 4
|
 Posted: Sat Jan 16, 2010 3:21 am Post subject: Need to keep only the last records for a certain group |
 |
|
Data Example
# date seq
1234567 2009-06-18 1
1234567 2009-06-18 2
1234567 2009-06-25 1
1234567 2009-06-25 2
1234567 2009-06-26 1
1234567 2009-06-26 2
8888888 2009-06-05 1
8888888 2009-06-05 2
8888888 2009-06-05 3
8888888 2009-06-12 1
8888888 2009-06-12 2
8888888 2009-06-12 3
I want to keep only the latest dated records for each #
1234567 2009-06-26 1
1234567 2009-06-26 2
8888888 2009-06-12 1
8888888 2009-06-12 2
8888888 2009-06-12 3 |
|
| Back to top |
|
|
|
Frank Yaeger
Moderator
Joined: 18 Feb 2006
Posts: 443
Location: San Jose, CA
|
| Posted: Sat Jan 16, 2010 4:19 am Post subject: |
|
|
Here's a DFSORT/ICETOOL job that will do what you asked for. I assumed your input file has RECFM=FB and LRECL=80, but the job can be changed appropriately for other attributes. I assumed your input records are already in order by the # and date as shown in your example.
| Code: |
//S1 EXEC PGM=ICETOOL
//TOOLMSG DD SYSOUT=*
//DFSMSG DD SYSOUT=*
//IN DD DSN=... input file (FB/80)
//T1 DD DSN=&&T1,UNIT=SYSDA,SPACE=(CYL,(5,5)),DISP=(MOD,PASS)
//OUT DD DSN=... output file (FB/80)
//TOOLIN DD *
COPY FROM(IN) USING(CTL1)
COPY FROM(IN) TO(T1) USING(CTL2)
SPLICE FROM(T1) TO(OUT) ON(1,7,CH) ON(9,10,CH) WITHALL -
WITH(1,81) USING(CTL3)
/*
//CTL1CNTL DD *
INREC OVERLAY=(81:C'BB')
OUTFIL FNAMES=T1,REMOVECC,NODETAIL,
SECTIONS=(1,7,
TRAILER3=(1,82))
/*
//CTL2CNTL DD *
INREC OVERLAY=(81:C'VV')
/*
//CTL3CNTL DD *
OUTFIL FNAMES=OUT,INCLUDE=(81,2,CH,EQ,C'VB'),
BUILD=(1,80)
/*
|
_________________
Frank Yaeger - DFSORT Development Team (IBM) - yaeger@us.ibm.com
Specialties: JOINKEYS, FINDREP, WHEN=GROUP, ICETOOL, Symbols, Migration
=> DFSORT/MVS is on the Web at http://www.ibm.com/storage/dfsort/ |
|
| Back to top |
|
|
|
rbowers
Member
Joined: 16 Jan 2010
Posts: 4
|
| Posted: Mon Jan 18, 2010 7:52 pm Post subject: |
|
|
| Thank you for your quick response. I should have mentioned my input file is VB and the records are not always in order. |
|
| Back to top |
|
|
|
Frank Yaeger
Moderator
Joined: 18 Feb 2006
Posts: 443
Location: San Jose, CA
|
| Posted: Mon Jan 18, 2010 10:35 pm Post subject: |
|
|
| Quote: | | I should have mentioned my input file is VB and the records are not always in order. |
That does make a difference. Here's a DFSORT/ICETOOL for that requirement:
| Code: |
//S1 EXEC PGM=ICETOOL
//TOOLMSG DD SYSOUT=*
//DFSMSG DD SYSOUT=*
//IN DD DSN=... input file (VB/24)
//T1 DD DSN=&&T1,UNIT=SYSDA,SPACE=(CYL,(5,5)),DISP=(MOD,PASS)
//OUT DD DSN=... output file (VB/24)
//TOOLIN DD *
SELECT FROM(IN) TO(T1) ON(5,7,CH) LAST USING(CTL1)
COPY FROM(IN) TO(T1) USING(CTL2)
SPLICE FROM(T1) TO(OUT) ON(7,7,CH) ON(15,10,CH) WITHALL VLENOVLY-
WITH(5,1) WITH(7,20) USING(CTL3)
/*
//CTL1CNTL DD *
SORT FIELDS=(5,7,CH,A,13,10,CH,A)
OUTFIL FNAMES=T1,BUILD=(1,4,5:C'BB',7:5)
/*
//CTL2CNTL DD *
OUTFIL FNAMES=T1,BUILD=(1,4,5:C'VV',7:5)
/*
//CTL3CNTL DD *
OUTFIL FNAMES=OUT,INCLUDE=(5,2,CH,EQ,C'VB'),
BUILD=(1,4,5:7)
/*
|
_________________
Frank Yaeger - DFSORT Development Team (IBM) - yaeger@us.ibm.com
Specialties: JOINKEYS, FINDREP, WHEN=GROUP, ICETOOL, Symbols, Migration
=> DFSORT/MVS is on the Web at http://www.ibm.com/storage/dfsort/ |
|
| Back to top |
|
|
|
rbowers
Member
Joined: 16 Jan 2010
Posts: 4
|
| Posted: Tue Jan 19, 2010 2:07 am Post subject: |
|
|
Thank you again! This works great. Can you point me to link where I can read what this code means so I can understand what it's doing? Or can you explain it? I don't want to implement this until I can understand it completely and document it well.
Thank you so much for your time! |
|
| Back to top |
|
|
|
Frank Yaeger
Moderator
Joined: 18 Feb 2006
Posts: 443
Location: San Jose, CA
|
| Posted: Tue Jan 19, 2010 2:41 am Post subject: |
|
|
For explanations of how this kind of DFSORT/ICETOOL code works, see the "Create files with matching and non-matching records " Smart DFSORT Trick at:
http://www.ibm.com/support/docview.wss?rs=114&uid=isg3T7000094
In your case, we only have one input file, but we break it into two input files before we do the splice. The first input file will have the "last" records with the 'BB' id in positions 5-6. The second input will have all of the records with the 'VV' id in positions 5-6.
For complete information on DFSORT's ICETOOL, see:
http://publibz.boulder.ibm.com/cgi-bin/bookmgr_OS390/BOOKS/ICE1CA40/6.0?DT=20090527161936
_________________
Frank Yaeger - DFSORT Development Team (IBM) - yaeger@us.ibm.com
Specialties: JOINKEYS, FINDREP, WHEN=GROUP, ICETOOL, Symbols, Migration
=> DFSORT/MVS is on the Web at http://www.ibm.com/storage/dfsort/ |
|
| Back to top |
|
|
|
rbowers
Member
Joined: 16 Jan 2010
Posts: 4
|
| Posted: Mon Feb 01, 2010 8:37 pm Post subject: SPLICE Question |
|
|
| On the SPLICE WITH(7,20)...I have MANY files that have the same first 2 records and then have different data for the rest of the file. The LRECL varies for each file. Is there any negative to using WITH(7,2000) or some extra large # that greatly exceeds the longest LRECL of all my files so I can use this one sort card for all the files without needing to maintain individual files. Your advice is greatly appreciated. |
|
| Back to top |
|
|
|
Frank Yaeger
Moderator
Joined: 18 Feb 2006
Posts: 443
Location: San Jose, CA
|
| Posted: Mon Feb 01, 2010 10:26 pm Post subject: |
|
|
Yes, you should be able to use the longest LRECL. There might be a performance degradation, but it probably won't be anything to worry about. The only way to find out for sure is to try it and evaluate the results by your own criteria.
_________________
Frank Yaeger - DFSORT Development Team (IBM) - yaeger@us.ibm.com
Specialties: JOINKEYS, FINDREP, WHEN=GROUP, ICETOOL, Symbols, Migration
=> DFSORT/MVS is on the Web at http://www.ibm.com/storage/dfsort/ |
|
| Back to top |
|
|
|
|
|
|
FAQ 
Search 
Register
Profile 
Log in to check your private messages 
Log in 
