Remove junk characters by keeping the 'good' characters

Suresh Shankarakrishnan · New User Joined: 11 Jul 2008 Posts: 42 Location: USA

Found a few examples, but did not help me.

Input file - lrecl = 1358.

1. It contains the following characters that I want to retain ( and maintain the same column position ) -

'A' through 'Z'
'a' through 'z'
0 through 9
Special characters like '!', '@', '#', '$' etc. ( did not want to list the entire set here, as I can add them once I know the correct jcl)

2. Want to remove junk characters like X'AD', X'00', X'15' etc. from the record and replace EACH of them with space ( X'40').

I know the 'good' characters, but I do not know all the 'bad' characters.

Can someone help me?

Frank Yaeger · Posted: Tue Jul 22, 2008 12:54 am

You can use the TRAN=ALTSEQ technique discussed in the "Change all zeros in your records to spaces" Smart DFSORT Trick at:

Use [URL] BBCode for External Links

You just need to set up the ALTSEQ statement with xx40 pairs for each bad character you want to replace with a blank.

Suresh Shankarakrishnan · New User Joined: 11 Jul 2008 Posts: 42 Location: USA

Thanks Frank, I checked that technique, but I do not know the bad characters before hand. All I know are the good characters.

Frank Yaeger · Posted: Tue Jul 22, 2008 1:02 am

Wouldn't the "bad" characters be the set of characters that are NOT the good characters? If you really don't know the bad characters, how could you specify them using any method? Are you expecting a "program" to figure out the "bad" characters by magic rather than by logic?

dick scherrer · Posted: Tue Jul 22, 2008 1:04 am

Hello,

FWIW, if you know the "good" characters, you also know the "bad" characters.

One byte may have a value of x'00' thru x'ff'. All of the values not represented in the set of "good" characters is a "bad" character.

Suresh Shankarakrishnan · New User Joined: 11 Jul 2008 Posts: 42 Location: USA

Thanks Frank and Dick.

Looking at Frank work wonders, I thought I might ask here.

If I 'define' the set of good characters, then obviously it follows that the rest of the character set has junk data. Hence, this list might be longer, but given the range of value - from x'00' to x'ff', yes it can be done.

Sagar_mainframe · Posted: Wed Jul 23, 2008 10:42 pm

Thanks a lot, all of you.

Frank Yaeger · Posted: Wed Jul 23, 2008 11:05 pm

You seem to want to "remove" some characters and "overlay" other characters with blank, but it isn't clear exactly what you want to do. Are you trying to "remove" or "overlay" characters only in certain columns or everywhere in the record or what? You need to do a better job of explaining the "rules" for getting from input to output.

Sagar_mainframe · Posted: Wed Jul 23, 2008 11:25 pm

1) In above two records, from column 23 to column 62 , I have to overlay \ (fwd slash) and "(double quotes) and to remove , (comma).

Overall if the special charater is present at the starting position of the field (means column 23 for last name and column 48 for first name), I have to overlay ( means shift the rest of the characters in the fields to left by removing that special character) and it should not disturb other fields in the record (means other than laat name and first name).

Last name = column 23 to column 47
First name = column 48 to column 62

2)Also if any special characters like `~!@#$%^&*()_-+=\|:";'{}[]?/>.<, present at the end or in between (means column 24 to column 47 for last name and column 49 to column 62 for first name) of the first name or last name that also i have to replace it with space X'40'. For this I used the following code, and it
was working fine.

Frank Yaeger · Posted: Thu Jul 24, 2008 3:24 am

Sagar,

I think this DFSORT job will do what you asked for. I added 7F40 to your ALTSEQ list to cover " (double quote). You can add whatever other characters you need to replace with blank to the ALTSEQ list. You can add whatever other characters you need to remove and shift left to the PREBLANK lists.

Sagar_mainframe · Posted: Thu Jul 24, 2008 10:28 pm

I am getting following error after incorporaing the above code:

dick scherrer · Posted: Thu Jul 24, 2008 10:35 pm

Hello,

You are using Syncsort rather than DFSORT.

You are also not using the current release of Syncsort.

Sagar_mainframe · Posted: Thu Jul 24, 2008 10:41 pm

Hi Dick,

So it not possible for me to use the above code under this version (release) of SYNCSORT..

?

Which release of SYNCSORT will be helpful execute ?

Thanks,

dick scherrer · Posted: Thu Jul 24, 2008 11:32 pm

Hello,

JFY is available in the current release of Syncsort (1.3).

Sagar_mainframe · Posted: Thu Jul 24, 2008 11:41 pm

OK...Thanks a lot!!! Frank and Dick... Atlast I think i need to do it with COBOL batch program only...

dick scherrer · Posted: Thu Jul 24, 2008 11:53 pm

Hello,

The good news is that if you use COBOL, the entire determination of good versus bad character becomes a single IF statement

You'd need to define the GOOD-CHAR as a condition name, but that can be done with 'A' THRU 'I', etc rather than naming every value.

Sagar_mainframe · Posted: Fri Jul 25, 2008 12:12 am

Dick,

I have implemented part 2) in COBOL using INSPECT verb as follows

part 2) "Also if any special characters like `~!@#$%^&*()_-+=\|:";'{}[]?/>.<, present at the end or in between (means column 24 to column 47 for last name and column 49 to column 62 for first name) of the first name or last name that also i have to replace it with space X'40'. "

Frank Yaeger · Posted: Fri Jul 25, 2008 1:34 am

Sagar_mainframe · Posted: Fri Jul 25, 2008 1:43 am

OK Frank, I think mine is 2005....if you look at the top-line of the snapshot (code posted) of error encountered ...

dick scherrer · Posted: Fri Jul 25, 2008 2:10 am

Hello Sagar,

You need to pay closer attention. . .

To repeat - You are not using DFSORT - you are using Syncsort. . . . They are entirely different products.

dick scherrer · Posted: Fri Jul 25, 2008 2:29 am

Hello Sagar,

To do what you want for part 1 (removing unwanted characters), i'd suggest defining a field to "receive" the corrected string and initialize it to spaces. Then positon yourself at the beginning of the input field and the receiving field. If the current character is "good", move it to the receiving field and increment the position in both fields. If the current character is not "good" only increment the input field positon.

To define the good charactgers, i'd use something like:

Frank Yaeger · Posted: Fri Jul 25, 2008 2:42 am