Consolidating and summing file contents

Kevin Lindsley · New User Joined: 02 Jun 2016 Posts: 5 Location: us

My input file has the following format with multiple records starting with 0, 1 and 9. Input data below:

Bill Woodger · Posted: Thu Jun 09, 2016 10:50 am

From what you've shown, it would be possible to INCLUDE= for the "1"s, and use OUTFIL reporting featers, HEADER1,REMOVECC and create the header and trailers. There is a COUNT field available for TRAILERn, you can even modify it if you don't want to count header or trailer or both.

The date on your header depends what you want. One way would be a JPn symbol to take a PARM from the JCL if your SyncSORT supports it.

Another is to make a symbol file (dictionary SyncSORT calls them) from one step (OPTION STOPAFT=1) and use that in the second step. Other ways are hard-coding and more, depending on your actual requirement.

Rohit Umarjikar · Posted: Fri Jun 10, 2016 12:51 am

Isn't INCLUDE for '1's will lose all '9's for the TOT? also TS may need sum on the values associated with '9's instead of counting all '1's. please correct me.

Bill Woodger · Posted: Fri Jun 10, 2016 3:00 am

The header and trailer on the file can be ignored if they can be recreated. The trailer is easy. A 9, some blanks, and a TOT/TOTAL, adjusted if necessary. TRAILER1 can do that.

HEADER1 can do a header, it does need a source for the date-like bit of data.

Rohit Umarjikar · Posted: Fri Jun 10, 2016 5:43 am

Understood, But wouldn't INCLUDE take precedence over trailer1 at the time it does TOT and have no '9's left because of INCLUDE?

Bill Woodger · Posted: Fri Jun 10, 2016 10:39 am

Yes. That is the point of the INCLUDE. The original 9-type records are redundant. Only one of the three is required, and the data required for one is constant (according to the sample) plus a count of the records.

Count of the records in an OUTFIL group is what TOT/TOTAL does. So the entire trailer record can be generated by TRAILERn.

The header is constant plus-some-data. The only issuer for HEADERn is the source of plus-some-data.

Perhaps I should read the question again when I have time :-)

sergeyken · Posted: Thu Aug 11, 2016 2:55 am

Three differently handled group of records do require three separate output files, within the same SYNCSORT step.

Abid Hasan · New User Joined: 25 Mar 2013 Posts: 88 Location: India

Hello Kevin,

A simple method can be (assuming LRECL=80); tested on DFSORT, should work with SyncSORT too:

enrico-sorichetti · Posted: Thu Aug 11, 2016 5:17 pm

sergeyken · Posted: Thu Aug 11, 2016 5:47 pm

enrico-sorichetti · Posted: Thu Aug 11, 2016 5:59 pm

instead of adding one consolidating step

why not check if the DFSORT approach works with SYSNCORT
and post the result ?

sergeyken · Posted: Thu Aug 11, 2016 6:05 pm

Abid Hasan · New User Joined: 25 Mar 2013 Posts: 88 Location: India

Hello Sergeyken,

Indeed, you're spot-on.
Yet, going by the data shared, it seemed that the 'number' present in the trailer was the count of records minus the header present. If you notice, there are 3 'xxx1' records, 2 'xxx2' records and 1 'xxx3' record. TS has not specified the significance of the number in trailer, since it matched my assumption, hence the usage of COUNT operator; though, it CAN be further trimmed with a bit more logic; kindly refer Mr. Woodger's earlier post which enunciates this.

sergeyken · Posted: Thu Aug 11, 2016 6:26 pm

Rohit Umarjikar · Posted: Thu Aug 11, 2016 8:53 pm

I think,all that matter is # of passes and the least in either case is better otherwise both the solution should work, Unless TS is really serious and wants to confirm how trailer is to be determined COUNT/TOT choice would remain open point.