View previous topic :: View next topic
Author
Message
Kalyani karuppiah New User Joined: 24 Mar 2023Posts: 2 Location: Chennai
My requirement is to split one input file to multiple files (I don't know how many as it can vary for each run) and also the endrec is not constant first file I might need to send 100 records second file I can send 230 records based on a header.
Sample input:
Header1
Aaa
Baa
Can
Header2
Daa
Eaa
Faaa
F
Gaa
Header3
Haha
Iaaa
Jaaa
Kaaa
I used the below code and it dynamically allotted number of output files based on my input record count but it equally splits the file. How to split the file using header and modify the endrec number based on my header keyword position in my file.
//STEP1 EXEC PGM=SYNCTOOL
//TOOLMSG DD SYSOUT=*
//DFSMSG DD SYSOUT=*
//IN DD DSN= Your.input.file
//T1 DD DSN=&&T1,DISP=(,PASS)
//OUT DD SYSOUT=*
//TOOLIN DD *
COPY FROM(IN) TO(T1) USING(CTL1)
SORT FROM(T1) TO(OUT) USING(CTL2)
//CTL1CNTL DD *
INREC OVERLAY=(81:SEQNUM,8,ZD,89:81,8,ZD,MOD,+5,M11,LENGTH=8)
OUTFIL INCLUDE=(89,8,ZD,EQ,1),
BUILD=(C' OUTFIL FILES=',SEQNUM,2,ZD,C',STARTREC=',81,8,C',',
C'ENDREC=',81,8,ZD,ADD,+4,M11,LENGTH=8,/,
C'//SORTOF',SEQNUM,2,ZD,C' DD DSN=xxxx.SPLTFILE.F',
SEQNUM,2,ZD,C',UNIT=SYSDA,',/,
C'//',12X,C'SPACE=(CYL,(100,100),RLSE),DISP=(,CATLG)',80:X)
//CTL2CNTL DD *
INREC IFTHEN=(WHEN=(1,1,CH,EQ,C' '),OVERLAY=(81:C'0')),
IFTHEN=(WHEN=NONE,OVERLAY=(81:C'9'))
SORT FIELDS=(81,1,CH,A)
OUTFIL REMOVECC,BUILD=(1,80),
HEADER1=(C'//xxxxxxxx JOB (YYYY),''SPLIT'',CLASS=A,MSGCLASS=X,',/,
C'// REGION=2048K,NOTIFY=&SYSUID',/,
C'//STEP1 EXEC PGM=SORT',/,
C'//SYSOUT DD SYSOUT=*',/,
C'//SORTIN DD DSN=Your.input.file,DISP=SHR',/,
C'//SYSIN DD *',/,
C' OPTION COPY',80:X)
The above job is for writing 5 records each into each of the split files.
I have given the final output filenames as
Code:
xxxx.SPLTFILE.F01
xxxx.SPLTFILE.F02
..............
..............
.[/quote]
Back to top
sergeyken Senior Member Joined: 29 Apr 2008Posts: 2140 Location: USA
Unless you learned how to use the Code button to present your samples of code and data, nobody would dig into your pile of garbage.
Back to top
sergeyken Senior Member Joined: 29 Apr 2008Posts: 2140 Location: USA
P.S.
0) SYNCTOOL is using //SSMSG DD for log messages, not //DFSMSG DD
1) Before "creating a JCL to generate another JCL" you need to learn by yourself: how to create that final JCL manually ?
2) The way to detect variable number of records between your "headers", or whatever else, may be: try using IFTHEN=(WHEN=GROUP,...) parameters
3) This question is related mainly to SYNCSORT, rather than to JCL section
Back to top
sergeyken Senior Member Joined: 29 Apr 2008Posts: 2140 Location: USA
The simple solution, without using the "bend over backwards" methods, might be like this one below.
The only question to be resolved is: how to deal with those output datasets which left empty?
Code:
//SPLIT EXEC PGM=SYNCSORT
//SYSOUT DD SYSOUT=*
//*
//SORTIN DD *
Header1
Aaa
Baa
Can
Header2
Daa
Eaa
Faaa
F
Gaa
Header3
Haha
Iaaa
Jaaa
Kaaa
//*
//SORTOF01 DD SYSOUT=*
//SORTOF02 DD SYSOUT=*
//SORTOF03 DD SYSOUT=*
//SORTOF04 DD SYSOUT=*
//SORTOF05 DD SYSOUT=*
//SORTOF06 DD SYSOUT=*
//SORTOF07 DD SYSOUT=*
//SORTOF08 DD SYSOUT=*
//SORTOF09 DD SYSOUT=*
//SORTOF10 DD SYSOUT=*
//*
//SYSIN DD *
INREC IFTHEN=(WHEN=GROUP,
BEGIN=(1,6,CH,EQ,C'Header'),
PUSH=(81:ID=2))
SORT FIELDS=COPY
OUTFIL FNAMES=SORTOF01,INCLUDE=(81,2,CH,EQ,C'01'),BUILD=(1,80)
OUTFIL FNAMES=SORTOF02,INCLUDE=(81,2,CH,EQ,C'02'),BUILD=(1,80)
OUTFIL FNAMES=SORTOF03,INCLUDE=(81,2,CH,EQ,C'03'),BUILD=(1,80)
OUTFIL FNAMES=SORTOF04,INCLUDE=(81,2,CH,EQ,C'04'),BUILD=(1,80)
OUTFIL FNAMES=SORTOF05,INCLUDE=(81,2,CH,EQ,C'05'),BUILD=(1,80)
OUTFIL FNAMES=SORTOF06,INCLUDE=(81,2,CH,EQ,C'06'),BUILD=(1,80)
OUTFIL FNAMES=SORTOF07,INCLUDE=(81,2,CH,EQ,C'07'),BUILD=(1,80)
OUTFIL FNAMES=SORTOF08,INCLUDE=(81,2,CH,EQ,C'08'),BUILD=(1,80)
OUTFIL FNAMES=SORTOF09,INCLUDE=(81,2,CH,EQ,C'09'),BUILD=(1,80)
OUTFIL FNAMES=SORTOF10,INCLUDE=(81,2,CH,EQ,C'10'),BUILD=(1,80)
//*
//*=====================================================================
Code:
********************************* TOP OF DATA *********
Header1
Aaa
Baa
Can
******************************** BOTTOM OF DATA *******
Code:
********************************* TOP OF DATA ***********
Header2
Daa
Eaa
Faaa
F
Gaa
******************************** BOTTOM OF DATA *********
Code:
********************************* TOP OF DATA ********
Header3
Haha
Iaaa
Jaaa
Kaaa
******************************** BOTTOM OF DATA ******
Back to top
sergeyken Senior Member Joined: 29 Apr 2008Posts: 2140 Location: USA
dup ...
Back to top
Kalyani karuppiah New User Joined: 24 Mar 2023Posts: 2 Location: Chennai
sergeyken wrote:
The simple solution, without using the "bend over backwards" methods, might be like this one below.
The only question to be resolved is: how to deal with those output datasets which left empty?
Code:
//SPLIT EXEC PGM=SYNCSORT
//SYSOUT DD SYSOUT=*
//*
//SORTIN DD *
Header1
Aaa
Baa
Can
Header2
Daa
Eaa
Faaa
F
Gaa
Header3
Haha
Iaaa
Jaaa
Kaaa
//*
//SORTOF01 DD SYSOUT=*
//SORTOF02 DD SYSOUT=*
//SORTOF03 DD SYSOUT=*
//SORTOF04 DD SYSOUT=*
//SORTOF05 DD SYSOUT=*
//SORTOF06 DD SYSOUT=*
//SORTOF07 DD SYSOUT=*
//SORTOF08 DD SYSOUT=*
//SORTOF09 DD SYSOUT=*
//SORTOF10 DD SYSOUT=*
//*
//SYSIN DD *
INREC IFTHEN=(WHEN=GROUP,
BEGIN=(1,6,CH,EQ,C'Header'),
PUSH=(81:ID=2))
SORT FIELDS=COPY
OUTFIL FNAMES=SORTOF01,INCLUDE=(81,2,CH,EQ,C'01'),BUILD=(1,80)
OUTFIL FNAMES=SORTOF02,INCLUDE=(81,2,CH,EQ,C'02'),BUILD=(1,80)
OUTFIL FNAMES=SORTOF03,INCLUDE=(81,2,CH,EQ,C'03'),BUILD=(1,80)
OUTFIL FNAMES=SORTOF04,INCLUDE=(81,2,CH,EQ,C'04'),BUILD=(1,80)
OUTFIL FNAMES=SORTOF05,INCLUDE=(81,2,CH,EQ,C'05'),BUILD=(1,80)
OUTFIL FNAMES=SORTOF06,INCLUDE=(81,2,CH,EQ,C'06'),BUILD=(1,80)
OUTFIL FNAMES=SORTOF07,INCLUDE=(81,2,CH,EQ,C'07'),BUILD=(1,80)
OUTFIL FNAMES=SORTOF08,INCLUDE=(81,2,CH,EQ,C'08'),BUILD=(1,80)
OUTFIL FNAMES=SORTOF09,INCLUDE=(81,2,CH,EQ,C'09'),BUILD=(1,80)
OUTFIL FNAMES=SORTOF10,INCLUDE=(81,2,CH,EQ,C'10'),BUILD=(1,80)
//*
//*=====================================================================
Code:
********************************* TOP OF DATA *********
Header1
Aaa
Baa
Can
******************************** BOTTOM OF DATA *******
Code:
********************************* TOP OF DATA ***********
Header2
Daa
Eaa
Faaa
F
Gaa
******************************** BOTTOM OF DATA *********
Code:
********************************* TOP OF DATA ********
Header3
Haha
Iaaa
Jaaa
Kaaa
******************************** BOTTOM OF DATA ******
Yes and also I might need more than 10 files too due to the input file size, so cannot define it in a straightforward way
Back to top
Joerg.Findeisen Senior Member Joined: 15 Aug 2015Posts: 1334 Location: Bamberg, Germany
If you have HDR/TRL, IFTHEN=(GROUP,.. PUSH is the best option. When splitting by numbers only, see SPLITBY * keywords.
It is left to the TS to solve the problem of too few or too many DSNs being allocated.
Back to top
sergeyken Senior Member Joined: 29 Apr 2008Posts: 2140 Location: USA
Kalyani karuppiah wrote:
Yes and also I might need more than 10 files too due to the input file size, so cannot define it in a straightforward way
1. What prevents you from defining, let's say, 100 datasets in a straightforward way (not files, @#$%&$#!!!)?
2. As I already mentioned: this is the first step you need to do, and to understand deeply: HOW TO CREATE THE REQUIRED JCL MANUALLY. Next step might be: how to create another JCL, to dynamically generate the first one?
3. There are other tools available, to really dynamically allocate exactly the required number of datasets. But you don't need a musket to shoot a butterfly, I believe so.
Back to top
Please enable JavaScript!