IBM Mainframe Forum

[plopoff]

Hello

I'm trying to code a DFSORT procedure that would merge 2 files into one with the following model :
<header1>
<detail1>
..
<detail1>
<trailer1>
<header2>
<detail2>
..
etc.

The first input file contains every headers and trailers while the second input file only contains the details records.
The records look like that :
Header : 01XXXXXXCCCCCXXXXXXXXX..
Detail : 02XXXXXXCCCCCXXXXXXXXXXXXXX....
Header : 09XXXXXXCCCCC0000000

The X's are random irrelevant characters
The C's represent the key on which both files are sorted
The counter of the trailer is initialy at 0000000

What I've been able to do so far is just a simple sort merging both files using :

Select all

//SORT01    EXEC PGM=SORT             
//SYSOUT  DD   SYSOUT=*               
//SORTIN  DD   DISP=SHR,DSN=dsn of input1
//        DD   DISP=SHR,DSN=dsn of input2
//SORTOUT DD   DISP=SHR,DSN=dsn of output
//SYSIN   DD   *                     
 SORT FIELDS=(9,5,CH,A,1,2,CH,A)     
/*

but I then have to manually count the detail records to change the trailer for each header/details/trailer bloc.

So I need your help to add a counter to each trailer, that would reset back to 0 everytime the sorting key changes
The output file should look like this

Select all

01XXXXXX12135XXXXXXXXX                       
02XXXXXX12135XXXXXXXXXXXXXXXXXXXXXXX           
09XXXXXX121350000001                               
01XXXXXX13135XXXXXXXXX                       
0200000013135XXXXXXXXXXXXXXXXXXXXXXX             
09XXXXXX131350000001                               
01XXXXXX13335XXXXXXXXX                       
0200000013335XXXXXXXXXXXXXXXXXXXXXXX
0200000013335XXXXXXXXXXXXXXXXXXXXXXX
09XXXXXX133350000002                               
etc.

I tried a few things using Frank Yaeger's documentation but nothing succesful so far.

I'm sorry for any english mistake and/or if my post isn't clear enough, but english is not my first langage
Thanks in advance for any help.

[BillyBoyo]

Temporarily extend your records to include a sequence number, and RESTART that on your key (Use INREC, with IFTHEN=(WHEN=INIT). Use IFTHEN=(WHEN=(logicalexpression) to identify the trailer, and by subtracting two (for the header and trailer) from the extended sequence number you will be able to calculate the number of data-records in a header/trailer group, and store the calculated value on the trailer for that group.

If you have a file trailer, do a similar thing with a second sequence number, which is not RESTARTed, and do a similar calculation after identifying the file trailer.

[plopoff]

Hello BillyBoyo

Thanks a lot for the idea. I have managed to add a seqnum that reset every time the key changes with the simple code :

Select all

OUTFIL BUILD=(1:1,1493, -                               
              SEQNUM,7,ZD,START=1,INCR=1,RESTART=(9,5))

But I'm actually having trouble substracting 2 to the seqnum.
I have reformated the trailer to look like it is supposed to with an INREC IFTHEN :

Select all

09XXXXXX113150000008XXXXXXXX...

And i have tried to substract 2 by using this :

Select all

OUTREC IFTHEN=(WHEN=(1,2,CH,EQ,C'09'), -
             BUILD=(1:1,13, -           
                    14:14,7,ZD,SUB,+2,TO=ZD,LENGTH=7, - 
                    21:21,1485))

But I am getting the error :

Select all

CE126A 2 INCONSISTENT REFORMATTING FOR *OUTREC : REASON CODE 04, IFTHEN 1

So I am guessing syntax error, somewhere in my OUTREC but I couldn't find it.
Hopefully you'll be able to help me one more time.
Thanks again.

[BillyBoyo]

Are your record variable-length? Can you show the full code that you have? It will be easier to explain that way.

[plopoff]

No my record are 1500 char long and not variable.
But nevermind, it was a stupid syntax mistake that I corrected and it acutally works as I want now !

Thanks a lot BillyBoyo, this will probably save me many hours