sort/rexx job to replicate records in file

Saini19 · New User Joined: 12 Apr 2012 Posts: 23 Location: Singapore

I have a file (VB 32191 length) with 6 records ... where each record actually spans in 2 lines ... now I want to replicate each record 500 times ... making total as 3000.

The problem is, there are two fields referring them as "A" and "B". "A" field occurs at static 191'th byte in both rows of a single record, as well as between tags <TAGONE> and </TAGONE> that can occur at any position ... I want that for all 500 records, this field (a 10byte numeric field) should have unique values or else my logic will abend ...

Also, the field "B" appears between tags <TAGTWO> and </TAGTWO>, again at varying positions 9 byte alphanumeric, which again has to be unique for all records ...

This needs something in REXX I believe to replicate the records but I am totally unaware of REXX ... can anybody help me here please.

Gnanas N · Posted: Thu Nov 15, 2012 1:41 pm

Please post sample input and output records, for clarification. Kindly use code tags.

Bill Woodger · Posted: Thu Nov 15, 2012 1:43 pm

If you're considering SORT, look at your documentation for OUTFIL REPEAT.

You can use a SEQNUM to make A in fixed position unique.

You need to check on PARSE for limits to the size of a field, but basically break it up by PARSEing it, then put it back together with two "tagged" values, (the B can also be a SEQNUM).

If you want to go that way, tell us which SORT product you use so that the topic can be moved.

You do want 500 copies of record one, followed by 500 copies of record two, etc...?

Saini19 · New User Joined: 12 Apr 2012 Posts: 23 Location: Singapore

Saini19 · New User Joined: 12 Apr 2012 Posts: 23 Location: Singapore

Saini19 · New User Joined: 12 Apr 2012 Posts: 23 Location: Singapore

Bill Woodger · Posted: Thu Nov 15, 2012 3:01 pm

OK, before moving it we need to know if it may fly.

I can't see a maximum for FIXLEN in PARSE, so how about you have an experiment to see.

With your records, try a simple PARSE with FIXLEN=32000

Let us know.

enrico-sorichetti · Posted: Thu Nov 15, 2012 3:39 pm

Saini19 · New User Joined: 12 Apr 2012 Posts: 23 Location: Singapore

enrico-sorichetti · Posted: Thu Nov 15, 2012 5:50 pm

and ... apart the <language>/<utility> used
what are the rules to create unique A and B values ?

Saini19 · New User Joined: 12 Apr 2012 Posts: 23 Location: Singapore

enrico-sorichetti · Posted: Thu Nov 15, 2012 7:08 pm

Your description is inconsistent with the data you provided

since at the end You will have 1500 different couples of records
the requirement about equality of TAGONE filed with the data at 6,3 cannot be satisfied

here is a REXX snippet that will do it
( the whole first 8 bytes will be considered as the <key>

enrico-sorichetti · Posted: Thu Nov 15, 2012 7:15 pm

follow on ...
changing the k2 seed to

Bill Woodger · Posted: Fri Nov 16, 2012 2:58 am

With the size of your records, I'm sure PARSE will not work, so it is good you have some Rexx now.

There is a simple solution, though it may not suit your requirement. You could use FINDREP to destroy the original TAGONE and TAGTWO and then insert them in the first variable position on the relevant record. This makes them "fixed" in position, so simple to insert sequence numbers. However, TAGONE and TAGTWO will no longer be in their original positions, although it presumably will not affect the processing...

enrico-sorichetti · Posted: Fri Nov 16, 2012 3:54 am

I hope that the TS will be able to understand it..
the code is pretty tight

disregard the previous snippets ,
( cut and paste syndrome struck again )
I pasted the increment of the values in the wrong place - outer loop instead of inner loop
( giving duplicate values for the <keys> )

here is the fixed code

Saini19 · New User Joined: 12 Apr 2012 Posts: 23 Location: Singapore

Thanks to Bill and Enrico, and everybody who gave me certain ideas. I will execute the code and keep you posted on output.

Saini19 · New User Joined: 12 Apr 2012 Posts: 23 Location: Singapore

Well, thanks Enrico for the solution, however it is having two impacts.

First, you have used RIGHT function to change the bytes, which is forcing my first 5 bytes also to be "0". I wanted to retain the original data. I am able to figure out it's solution. So this issue exists no longer.

Second, the RANDOM function used is changing the length of my B field, which is having an impact. I am not able to resolve this issue. Can you please help here with a solution, that should make it more dynamic. In the sense that it should consume the same number of bytes as is present in the input record.

enrico-sorichetti · Posted: Mon Nov 19, 2012 1:33 am

before complaining with stupidities You should learn to post properly
( reread the whole topic from the beginning )

did You care to read my previous comments about the poor description of the requirements

as I already said there is a mismatch between the data shown and the description in words of the requirement ...

You said to repeat 500 hundred times to get 1500 <couples> of records
where the TAGONE must be numeric unique and must match <some part> of the <prefix key>

You defined as the unique part 3 three bytes,
with that assumption You cannot have uniqueness for 1500 <things>

second You told that
TAGONE content would be numeric up to 9 digits
TAGTWO content would be alphanumeric up to 10 chars

Bill Woodger · Posted: Mon Nov 19, 2012 1:23 pm

Before complaining with stupidities you should learn to post properly. Reread the whole topic from the beginning.

Such good words, I thought I'd use them again after your latest stupidity (was deleted).

If you are unable to take benefit from such good words, then I suggest you go to someone at your site for assistance. It is up to you how you go about that, but if you genuinely expect someone to help you, then learn how to take advice when seeking it, even if the advice is not what you want, or to ignore the advice with grace. The advice was well-intended and whether you know it or not was the advice you needed at that point.