Reformat the variable pipe delimited fields to fixed field

devmisra · New User Joined: 29 Jan 2009 Posts: 39 Location: India

Hi All,

I have a variable block input file and data are like,

Garry Carroll · Posted: Mon Aug 15, 2011 11:56 am

Look at ftp.software.ibm.com/storage/dfsort/mvs/sorttrck.pdf for the PARSE function.

Garry.

devmisra · New User Joined: 29 Jan 2009 Posts: 39 Location: India

I coded below sort card

Garry Carroll · Posted: Mon Aug 15, 2011 1:02 pm

Line 5 of your JCL has SYSNAMES - the DD should read SYMNAMES

Garry.

Bill Woodger · Posted: Mon Aug 15, 2011 1:06 pm

WER message as SYNCSORT.

Should have been posted to the JCL forum where the SYNCSORT queries live.

devmisra · New User Joined: 29 Jan 2009 Posts: 39 Location: India

Oh sorry,, yes Garry, after posting the code into this formun ,I noticed that one and changed to //SYMNAMES and ran it but still getting the syntax error.

Bill Woodger · Posted: Mon Aug 15, 2011 1:13 pm

Hi misra,

I would if I could :-) I was just highlighting it for a moderator (who can) and for you for next time, and for you to know where it is gone if you can't find it in DFSORT later.

devmisra · New User Joined: 29 Jan 2009 Posts: 39 Location: India

Thanks Bill,

Hi Garry,

i corrected the syntax and ran the sort card and it seems sort card is fine now. but still its throwing error because both the files are VB.

Garry Carroll · Posted: Mon Aug 15, 2011 1:57 pm

Have you specified LRECL for SORTOUT in JCL?

devmisra · New User Joined: 29 Jan 2009 Posts: 39 Location: India

yes. and both are having same LRECL=219 and RECFM=VB

Bill Woodger · Posted: Mon Aug 15, 2011 2:22 pm

You have a "short" record on your input file, which means that the record finishes before the end of one of the parts you are using for your output record.

If you try to copy 20 bytes to some position in the output but the input record is only 10 bytes long, you'll get this message.

Bill Woodger · Posted: Mon Aug 15, 2011 2:35 pm

Misra,

Nice to see you using SYMNAMES, by the way.

Fld1 could be more meaningful.

It also, for a VB, should start at 5, not 1, which is in the start of the RDW, or you should not include the 1,4 for the RDW (which you didn't give a SYMNAME).

This is in addition to your short record problem.

Ed Goodman · Active Member Joined: 08 Jun 2011 Posts: 556 Location: USA

Isn't the output going to be the same length for each record now? The sort product is going to write it as variable because the input is variable, but is that what you really want?

devmisra · New User Joined: 29 Jan 2009 Posts: 39 Location: India

Hi Ed,

The output record length will not be same as input reccord. After reformatting, the output field will have greater length than the input field.

for e,g Field-4 is Invoice-No and value is 'ERS-37164-2177914' which is 17 char length and after reformatting it will be fixed one with 35 char length.rest will be be spaces.

But input file will always have shorter record than 219(VB) which is file lengfth.

Bill Woodger · Posted: Tue Aug 16, 2011 11:21 am

Ed means, "aren't all output records going to be the same length as each other" - with the implied question, "so can they be recfm fb"?

devmisra · New User Joined: 29 Jan 2009 Posts: 39 Location: India

yes, all the output records will be off same length as each other. They can be recfm FB.

Arun Raj · Posted: Tue Aug 16, 2011 5:48 pm

devmisra,

Your header record itself is in fact a "short record" since your parse starts at ABSPOS=27.
If you don't have any more short records other than the header record, you can have a condition check like this.

Ed Goodman · Active Member Joined: 08 Jun 2011 Posts: 556 Location: USA

exactly...

Is your goal to have a fixed length file? Will the receiving program/utility be expecting it as fixed length?

It's the type of thing that will sneak up on you later, when you think you've got it figured out. Then you have to come back to this step and try to fix it.

devmisra · New User Joined: 29 Jan 2009 Posts: 39 Location: India

Thanks to Everyone to help me on this.

For the time being, I am copying that VB file into FB file. Then using FB file, i am Skipping the Header record and reformatting the other records.

And its working fine but here i need your help.I reformat the amount fields like below

dick scherrer · Posted: Wed Aug 17, 2011 2:24 pm

Hello,

Why was it decided that this exercise is even needed

The data shown in the first post is perfectly usable as is (unless this is to accommodate unqualified programmers). If this is the case, i suggest the programmers be brought up to speed rather than waste machine resources "forever".

If there is some real business requirement it would be nteresting to know what it is.

Keep in mind this could be done in a very simple COBOL program using UNSTRING. . . A very simple program.
read the record
if the header - whatever
if not the header unstring and write the new record

Assuming this is even needed. . . .

devmisra · New User Joined: 29 Jan 2009 Posts: 39 Location: India

Hi d.sch,

You are absolutely correct on this approach.

Actually, I am generating a VB file which will contain around 200k records. To review the file data with table data, i need to know the Sum of the amount fields. I thought SORT will be an easier way to achieve the Total amounts. The whole process is not for production, only to review Test/QA data.

When i went thru the SORT cards I learnt PARSE function, so it got more interesting when i tested and got the result.... :-)

dick scherrer · Posted: Wed Aug 17, 2011 2:38 pm

Hello,

When PARSEing you might be able to sum the amount fields by using UFF if they will always be positive or SFF if they are allowed to be negative also.

I'm not able to try this just now, but thought i'd at least pass it along.

Good luck