How to insert a delimiter after every 5 bytes using sort

spamidighantam · New User Joined: 30 Aug 2006 Posts: 8

Hi,

Would you please help to insert a delimiter '+' in between every 5 bytes in a file. The record length is 6000. It should insert '+' after every 5 bytes till it reaches of end of record or a blank space. For example,

input

Bill Woodger · Posted: Fri Mar 16, 2012 1:00 pm

Well, what is wrong with typing-up a BUILD with 1,200 different elements? :-)

We know the LRECL of your file, what is the RECFM? How many records will be processed?

vasanthz · Posted: Fri Mar 16, 2012 1:03 pm

If file size is small, REXX can do this 'with less amount of code' not necessarily resource-effecient than SORT.

enrico-sorichetti · Posted: Fri Mar 16, 2012 1:09 pm

as usual there is something untold here ...

if the input record length is 6000 and all the columns are filled,
then the output record length will be 7199

unless each 5 bytes chunk must be + terminated and in this case,
the output record length will be 7200

meditate and consider describing better the whole requirement

dbzTHEdinosauer · Posted: Fri Mar 16, 2012 1:54 pm

if the delimiters were to be inserted at irregular intervals,
this requirement could make sense.

but, since every 5 bytes is to be delimited,
one asks: Why is this necessary?

A REXX Script could generate the BUILD control cards very easily.

sqlcode1 · Active Member Joined: 08 Apr 2010 Posts: 577 Location: USA

spamidighantam,
May be a crude way but see if below works... (UNTESTED)....

dbzTHEdinosauer · Posted: Fri Mar 16, 2012 8:12 pm

sqlcode1,
a nice way to start the weekend......

spamidighantam · New User Joined: 30 Aug 2006 Posts: 8

Thank you all for your responses.

Hi enrico-sorichetti,

The input file record length is 6000. Of course, when we insert '+' after every 5 bytes the length will be increased as you said. Can this be achieved through SORT ? Please advice.

Hi dbzTHEdinosauer,

Yes, you got my requirement. If REXX can create BUILD script. Could you please let me know the necessary BUILD script. So that I can write REXX program to repeat the BUILD script. Thank you.

Thanks,
Sreedhar

dbzTHEdinosauer · Posted: Fri Mar 16, 2012 8:22 pm

for f*** sake,

look at sqlcode1's solution..................................

Garry Carroll · Posted: Fri Mar 16, 2012 8:41 pm

Bill Woodger · Posted: Fri Mar 16, 2012 8:45 pm

I asked how many records. I asked about your RECFM.

If you have a lot of records, you may not want to chop them and put them together.

If we go with RECFM F, you have explain the bit about the blanks arriving. If one record has 50 of your groups and another 1200 and another one or three, what do you want to do? Get rid of the blanks? Or, if RECFM V, are there still blanks, etc.

If you are complete in your description, input, sample output, rules from how to get there, anything you can think of that we can otherwise only guess, then you'll get answers to match.

dbzTHEdinosauer · Posted: Fri Mar 16, 2012 8:52 pm

though i would not worry too much about number of records,
since the temp file is only 5 char,
Bill you are correct about the space stuff.

Sqlcode1,

would a
SQZ spaces to nothing and then a REP '++' ' ' after the BUILD work?

anything would be better than 1200 BUILD statements

spamidighantam · New User Joined: 30 Aug 2006 Posts: 8

Record length is 6000. Its is variable format.. I have attached the screen shot of actual input file. I need to insert '+' after every 5 bytes till I reach a space or end of record.

spamidighantam · New User Joined: 30 Aug 2006 Posts: 8

sorry here is the input file
ugly image deleted

dbzTHEdinosauer · Posted: Fri Mar 16, 2012 9:40 pm

is it a fixed length or variable length record?

enrico-sorichetti · Posted: Fri Mar 16, 2012 9:48 pm

do not post screen shots they just clutter the forum, nothing that a text cut and paste would not do.

apart that the jpeg posted is unreadable because of the ugly colors.

image deleted

spamidighantam · New User Joined: 30 Aug 2006 Posts: 8

It is variable length like below.

enrico-sorichetti · Posted: Fri Mar 16, 2012 10:20 pm

the code tags ...

Your post was edited to provide them

dbzTHEdinosauer · Posted: Fri Mar 16, 2012 10:35 pm

i would like to see the 3.2 I for the input file,
to insure that it is a variable and not fixed length file,
as the control cards will be different.

Bill Woodger · Posted: Fri Mar 16, 2012 10:45 pm

Some of that data only has nine bytes before the first space. How is that supposed to work?

EDIT: In fact, only one of the sample records shown is a multiple of five. What was that? Mass typo?

dbzTHEdinosauer · Posted: Fri Mar 16, 2012 11:00 pm

no, the TS did stipulate in his first post,
end of record or space.
he never indicated
that there would be an evenly divisible by 5
number of non/space characters in each record.

spamidighantam · New User Joined: 30 Aug 2006 Posts: 8

Just came to know from user that we need to add delimiter after every 9 bytes instead of 5 bytes..

Bill Woodger · Posted: Fri Mar 16, 2012 11:15 pm

Yes, dbz, but the only example given showed five characters as the final element, implying (to me) that that was the only possibility.

Where it might matter for a solution is if the records are variable, can you end up with a final element that is, if treated as being five bytes, beyond the record-length? Or will there always be a "balancing" number of spaces, or is there some other reason why the data is variable-length (if indeed it is) yet there are some trailing blanks which are vital enough to be included until they are excluded.

I now notice the records won't even fit exactly any more - or has the lrecl changed as well? was the lrecl ever correct, at 6000...

If you want a solution, tell everything you know...

Bill Woodger · Posted: Mon Apr 02, 2012 3:41 pm

TS/OP has done a bunk, but developing sqlcode1's solution with variable file...

Just a step to make the test data...

Bill Woodger · Posted: Wed Apr 04, 2012 5:59 am

In a single step. TS/OP still missing. Only tested with 80-byte records.

Despite the title, the requirement changed to 9-byte fields along the way. Therefore original LRECL is extended by 5. Final LRECL is extended by a further 689.

The work is done with a BUILD containing 690 elements from the input record (including RDW) and 689 delimiters (including after the final field).

The two other important bits are the extension of the record to its maximum size with OVERLAY and the FINDREP at the end to remove the "empty" delimited fields. NB. If there is a "genuine" empty field, it will disappear too :-)

This is going to be less resource-hungry than multiple passes of the file.

The BUILD has not been coded by hand. Here is how it was generated, then it was copied and the final part of it was edited for the closing of the IFTHEN.

689 records are generated, with a sequence number starting at zero.

Using each sequence number the start position is calculated for each group of 9 bytes, and adjusting for the start position of the first field (5) after the RDW. The word DELIM is included in the element being built, which will later be resolved on the SYMNAMES of the step that is doing the work.

After the records are generated, they are "stacked up" into groups of five, to make things more compact.