Split a file to two files retaining Header n Trailer on each

Jayalakshmi · New User Joined: 02 Sep 2005 Posts: 26 Location: Hyderabad

Hi all,

I have a file (GDG) containing records like: (The file in real scenario will
have millions of records)

Hfilenamedate
D131434add1
D123434add2
T002

I would want to split this into two files and want the Header and
Trailer records in each of the splitted files. (Trailer needs the count generated)

File1:
Hfilenamedate
D131434add1
T001

File2:
Hfilenamedate
D123434add2
T001

Could anyone help me on this?

Thanks,
Jaya.

dick scherrer · Posted: Fri Jun 15, 2007 11:53 am

Hello,

Please post your dcb info (dsorg/lrecl/etc) for both the input and output files.

Why is the trailer not the same in the output as the input (T002/T001)?

Will there be only one header and trailer in the input data?

Your sample has only 4 records and no "rules" on how to split them. Please post a bit more data (showing multiple headers/trailers if this may occur or clarifying that it may not) and more than just 2 data records. By chance id it the "add1" and "add2" that determines which output file the record should be placed in? Will all of the milliions of records have "addn" in them?

The more clear you present your requirement, the better/quicker responses you will receive.

Jayalakshmi · New User Joined: 02 Sep 2005 Posts: 26 Location: Hyderabad

The input file is a GDG like file.G0079V00
dsorg: sequential
rlecl: 336
27888.

I would want the two ouput files to be created as (+1) of input GDG
file.G0080V00
file.G0081V00

The input file will have a single header and trailer record. No multiple header n trailer records.
The Trailer record will have the count/number of detailed records in the file.
Hence, in a file if there are 2 D records then the trailer record will be T002
The Header, detail and trailer identified by H, D and T in the first position of each record.
There is no criteria for the split up. Just want the file to splitted into two (may not be exactly equal if the total number of D records are odd) with H and T records placed at the top and bottom of the splitted files as shown above.
The splitted file's trailer record should reflect the no. of D records in that file.

A input file with 1 H , 400 D, 1T (Trailer will be T400)
2 Output files with (1) -> 1H 200D, 1T (Trailer : T200)
and (2) -> 1H 200D, 1T (Trailer : T200)

Thanks again,
Jaya.

Devzee · Posted: Fri Jun 15, 2007 12:22 pm

Here is the link to SORT tricks, go thru it and definitely you can come up with solution.

www-304.ibm.com/jct01004c/systems/support/storage/software/sort/mvs/tricks/pdf/sorttrck.pdf

krisprems · Posted: Fri Jun 15, 2007 1:02 pm

Jayalakshmi

Is the header copied as it is from the i/p file, or
It is dynamically decided based on the file name of the new version of file created and also clarify about the date filed in the header record.

Jayalakshmi · New User Joined: 02 Sep 2005 Posts: 26 Location: Hyderabad

The Header record just needs to be copied from the input file.
No significance here on date field

Thanks,
Jaya.

Frank Yaeger · Posted: Fri Jun 15, 2007 9:08 pm

Jaya,

Here's a DFSORT job that will do what you asked for:

krisprems · Posted: Sat Jun 16, 2007 6:46 pm

Jayalakshmi · New User Joined: 02 Sep 2005 Posts: 26 Location: Hyderabad

hi Frank,

Great Thanks!

I haven't worked much on DFSORT though my rusted brain
couldn't hit on such a trick. Many Thanks!

Jaya.

ibmmainframesyntel · Posted: Mon Jun 25, 2007 11:31 am

Hi Jaya,
According to Frank's JCL,
If input file like this

Frank Yaeger · Posted: Mon Jun 25, 2007 8:08 pm