IBM Mainframe Forum

by **neverfail2try** » Sun Mar 31, 2013 2:20 am

Hi all, please help me with this. I have written a JCL which uses DFSORT and does the following function
1) Input file one : Consists of 50 rows with records length 9.
2) Input file two: Consists of 100 rows with records length 3.

Problem : I have to take one record every time from input file 2 and paste it in the position 4 to 6 in all the rows of input file 1 which means there will be 5000 records in the output file.
file 1:
aaabbbccc
dddeeefff
ggghhhiii

file 2:
111
222
333

Desired Output file :
aaa111ccc
ddd111fff
ggg111iii
aaa222ccc
ddd222fff
ggg222iii
aaa333ccc
ddd333fff
ggg333iii

The JCL which I have written is allowing only the first record number provided in the SYSIN DD statement. I am just getting the output as :
aaa111ccc
ddd111fff
ggg111iii

I have used the above example just for explanation purpose, the real JCL looks like below:

Select all

//STEP010 EXEC PGM=SORT
//SYSOUT DD SYSOUT=*
//SORTIN DD *
//SORTOUT DD DSN=**.****.OUTPUT1,
// DISP=(NEW,CATLG,DELETE),SPACE=(CYL,(5,5),RLSE)
//SYSIN DD *
SORT FIELDS=COPY
OUTREC FIELDS=(1,9,50:SEQNUM,5,ZD,START=00001,INCR=1)
/*
//STEP020 EXEC PGM=SORT
//SYSOUT DD SYSOUT=*
//SORTIN DD *
111
222
333
//*SORTOUT DD SYSOUT=*
//SORTOUT DD DSN=**.****.OUTPUT2,
// DISP=(NEW,CATLG,DELETE),SPACE=(CYL,(5,5),RLSE)
//SYSIN DD *
OPTION COPY
OUTFIL REPEAT=3,OVERLAY=(50:SEQNUM,5,ZD)
/*
//STEP030 EXEC PGM=SORT
//SYSOUT DD SYSOUT=*
//SORTJNF1 DD DISP=SHR,DSN=**.****.OUTPUT1
//SORTJNF2 DD DISP=SHR,DSN=**.****.OUTPUT2
//SORTOUT DD DSN=**.****.OUTPUT3,DISP=(NEW,CATLG,DELETE),
// SPACE=(CYL,(10,10),RLSE)
//SYSIN DD *
JOINKEYS F1=SORTJNF1,FIELDS=(50,5,A),SORTED
JOINKEYS F2=SORTJNF2,FIELDS=(50,5,A),SORTED
REFORMAT FIELDS=(F1:1,3,F2:1,3,F1:7,9)
SORT FIELDS=COPY
/*

Please tell me how should I proceed with this. I want to know of something which picks one record from the second input file and executes the job. After that, again it picks the second record and executes it. I am also open to any new methods of achieving this.

by **NicC** » Sun Mar 31, 2013 3:19 am

The JCL which I have written is allowing only the first record

is rubbish as JCL has nothing to do with individual records. JCL tells the OS what programs to run and what resources those programs need. In sort it is the sort control records that control what sort does - not JCL. (JCL is not even executed - it is read by the OS and then tossed away - maybe days before the job executes.)

As you are having problems with DFSort I have moved the topic to the DFSort part of the forum.

by **BillyBoyo** » Sun Mar 31, 2013 5:33 am

If you want to match every record on one file to every record on the other file, you need to have a key field to match on which is the same value for all records on both files.

Since you don't have this, you need to add it, with JNFnCNTL files. Once that is done, your JOINKEYS step is about there:

Select all

//STEP030 EXEC PGM=SORT
//SYSOUT DD SYSOUT=*
//SORTJNF1 DD DISP=SHR,DSN=your input file of nine bytes
//SORTJNF2 DD DISP=SHR,DSN=your input file of three bytes
//SORTOUT DD DSN=**.****.OUTPUT3,DISP=(NEW,CATLG,DELETE),
// SPACE=(TRK,1)
//SYSIN DD *
JOINKEYS F1=SORTJNF1,FIELDS=(10,1,A),SORTED,NOSEQCK
JOINKEYS F2=SORTJNF2,FIELDS=(04,1,A),SORTED,NOSEQCK
REFORMAT FIELDS=(F1:1,3,F2:1,3,F1:7,9)
SORT FIELDS=COPY
/*
//JNF1CNTL DD *
INREC OVERLAY=(10:C'X')
//JNF2CNTL DD *
INREC OVERLAY=(04:C'X')

by **neverfail2try** » Sun Mar 31, 2013 12:16 pm

NicC wrote:
The JCL which I have written is allowing only the first record

is rubbish as JCL has nothing to do with individual records. JCL tells the OS what programs to run and what resources those programs need. In sort it is the sort control records that control what sort does - not JCL. (JCL is not even executed - it is read by the OS and then tossed away - maybe days before the job executes.)

As you are having problems with DFSort I have moved the topic to the DFSort part of the forum.

I beg your pardon for my ignorance and thanks for explaining.

by **neverfail2try** » Sun Mar 31, 2013 12:19 pm

BillyBoyo wrote:If you want to match every record on one file to every record on the other file, you need to have a key field to match on which is the same value for all records on both files.

Since you don't have this, you need to add it, with JNFnCNTL files. Once that is done, your JOINKEYS step is about there:

Thank you very much Sir. I made some changes according to my requirement in the lines which you pasted and also added the SORT for removing the duplicates and it ran completely fine. Thanks again.

by **BillyBoyo** » Sun Mar 31, 2013 5:07 pm

If you can post your Control Cards, sample input and expected output, we may be able to "get rid of" the SORT.

This is with your 7,9 definition, which may be a typo.

Putting your three-byte file as the "first" joinkeys file will cause the order of that file to be used for the order of your REFORMAT records, and thus the order of your SORTOUT. Duplicates, on your nine- or 15-byte file can be dealt with as below.

Select all

//STEP0100 EXEC PGM=SORT
//SYSOUT DD SYSOUT=*
//SORTOUT DD SYSOUT=*
//SYSIN DD *
JOINKEYS F1=SORTJNF1,FIELDS=(04,1,A),SORTED,NOSEQCK
JOINKEYS F2=SORTJNF2,FIELDS=(16,1,A),SORTED,NOSEQCK
REFORMAT FIELDS=(F2:1,3,F1:1,3,F2:7,9)
SORT FIELDS=COPY
OUTFIL SECTIONS=(1,3,3,3,HEADER3=(1,15)),REMOVECC,NODETAIL
//JNF1CNTL DD *
INREC OVERLAY=(04:C'X')
//JNF2CNTL DD *
INREC OVERLAY=(16:C'X')
//SORTJNF1 DD *
111
222
333
//SORTJNF2 DD *
AAABBBCCC123456
DDDEEEFFF123455
DDDEEEFFF123456
GGGHHHIII123456

The output is:

Select all

AAA111CCC123456
DDD111FFF123455
GGG111III123456
AAA222CCC123456
DDD222FFF123455
GGG222III123456
AAA333CCC123456
DDD333FFF123455
GGG333III123456

If you want to keep the "last" record of any duplicate key, you change the HEADER3 to TRAILER3.

For your three-byte file, the code (to avoid a SORT) is a little different:

Select all

//STEP0100 EXEC PGM=SORT
//SYSOUT DD SYSOUT=*
//SORTOUT DD SYSOUT=*
//SYSIN DD *
JOINKEYS F1=SORTJNF1,FIELDS=(04,1,A),SORTED,NOSEQCK
JOINKEYS F2=SORTJNF2,FIELDS=(16,1,A),SORTED,NOSEQCK
REFORMAT FIELDS=(F2:1,3,F1:1,3,F2:7,9)
SORT FIELDS=COPY
OUTFIL SECTIONS=(1,3,3,3,TRAILER3=(1,15)),REMOVECC,NODETAIL
//JNF1CNTL DD *
INREC IFOUTLEN=4,
IFTHEN=(WHEN=INIT,
OVERLAY=(04:C'X')),
IFTHEN=(WHEN=GROUP,
KEYBEGIN=(1,3),
PUSH=(5:SEQ=1)),
IFTHEN=(WHEN=(5,1,CH,NE,C'1'),
OVERLAY=(4:Z))
//JNF2CNTL DD *
INREC OVERLAY=(16:C'X')
//SORTJNF1 DD *
111
222
333
333
//SORTJNF2 DD *
AAABBBCCC123456
DDDEEEFFF123455
DDDEEEFFF123456
GGGHHHIII123456

Produces:

Select all

AAA111CCC123456
DDD111FFF123456
GGG111III123456
AAA222CCC123456
DDD222FFF123456
GGG222III123456
AAA333CCC123456
DDD333FFF123456
GGG333III123456

by **neverfail2try** » Mon Apr 01, 2013 12:16 pm

Thanks a lot Sir, it works.

by **BillyBoyo** » Mon Apr 01, 2013 12:47 pm

It is very important that you understand it, so you can change it when necessary and apply the solution to other requirements. If you have doubts, post them here.

IBM Mainframe Forum

How to take one record every time and pass it to a step

How to take one record every time and pass it to a step

Re: How to take one record every time and pass it to a step

Re: How to take one record every time and pass it to a step

Re: How to take one record every time and pass it to a step

Re: How to take one record every time and pass it to a step

Re: How to take one record every time and pass it to a step

Re: How to take one record every time and pass it to a step

Re: How to take one record every time and pass it to a step