MVSFORUMS.com

mf_user · Intermediate Joined: 01 Jun 2003 Posts: 372 Topics: 105

Hi,

I have unpacked around 98 COMP-3 fields into a dataset and finding below given special characters in the unpacked data (output LRECL=900).

kolusu

mf_user · Intermediate Joined: 01 Jun 2003 Posts: 372 Topics: 105

Hi Kolusu,

I have unpacked them using TO=ZD,LENGTH=n.
the downstream system wants the junk to be replaced with zeroes.

Thanks.
_________________
MF
==
Any training that does not include the emotions, mind and body is incomplete; knowledge fades without feeling.
==

kolusu · Posted: Sat Feb 25, 2023 10:07 am Post subject:

mf_user · Intermediate Joined: 01 Jun 2003 Posts: 372 Topics: 105

Hi Kolusu,

After further analysis, and due to frequent S0C7 abend on certain fields (invalid input data from input file), we had to go for program approach. Declared an input filed of X(105) with all these characters and an output filed with 105 zeroes and used INSPECT to replace all such characters with zeroes.

since they asked us to ignore sign field, we did not consider it at all. Initial approach was SORT where I used OUTREC BUILD and tried TO=ZD, TO=ZDF, TO=ZDC and EDIT=() options.

However, I saw a post (either by you or Frank) few days ago in some forum or in google, where you provided a FINDREP solution (a bit lengthy though) and wanted to tell the possibility to client because its a onetime ask. I am not able to find that link /not able to recall where exactly I saw it. I will search for it again. Embarassed

Unfortunately, I can't copy, paste the code as I could not copy it from RDP.

Please advise.
_________________
MF
==
Any training that does not include the emotions, mind and body is incomplete; knowledge fades without feeling.
==

kolusu · Posted: Sun Feb 26, 2023 10:20 am Post subject:

mf_user,

I guess you are stuck on special characters and looking for the solution in that direction. Unfortunately nothing makes sense. Lets do something simple. lets just unpack 1 field.
Assuming 1 packed decimal filed starts at 10 for a length of 4 bytes. what is the output of this?

mf_user · Intermediate Joined: 01 Jun 2003 Posts: 372 Topics: 105

Hi Kolusu,

Example with HEXON:
i/p:

mf_user · Intermediate Joined: 01 Jun 2003 Posts: 372 Topics: 105

sorry, I hit enter too soon... bonk

sort card:

kolusu · Posted: Mon Feb 27, 2023 5:40 pm Post subject:

mf_user,

X'015C3F7C' is not a valid packed decimal number. Did you receive this file from an ASCII(windows/linux/mac) source? When converting ASCII code to EBCDIC code, all ASCII code not having an EBCDIC equivalent is converted to X'3F'.

That explains as to why you are getting X'3F' in the middle.

As I said you need to fix the data rather than trying to change the value.

I believe that changing it X'3F' to X'00' will make the values incorrect. Find out what the right value is supposed to be.
_________________
Kolusu - DFSORT Development Team (IBM)
DFSORT is on the Web at:
www.ibm.com/storage/dfsort

www.linkedin.com/in/kolusu

mf_user · Intermediate Joined: 01 Jun 2003 Posts: 372 Topics: 105

Kolusu,

Thanks for the suggestion.
We had that discussion and the expectation is to replace it with zeroes.
so, how do I replace bulk of it with FINDREP, please let me know.

Thanks.
_________________
MF
==
Any training that does not include the emotions, mind and body is incomplete; knowledge fades without feeling.
==

kolusu · Posted: Tue Feb 28, 2023 8:15 am Post subject:

mf_user,

You did NOT answer my question about the source of the data. If your intention is to replace the values , list all the invalid values as IN list and have X'00' as OUT List. So what is the problem. Look up the syntax of FINDREP in Syncsort manuals
_________________
Kolusu - DFSORT Development Team (IBM)
DFSORT is on the Web at:
www.ibm.com/storage/dfsort

www.linkedin.com/in/kolusu