topic Scan for non-printable EBCDIC Characters in SAS Programming

Scan for non-printable EBCDIC Characters

SASuser28 — Wed, 19 Oct 2016 20:28:15 GMT

Hi Community,

I work with SAS on z/OS, and I was wondering if anyone had a recommendation on setting up a scan for non-printable EBCDIC characters from an incoming dataset.

Thanks for any input.

Re: Scan for non-printable EBCDIC Characters

ChrisNZ — Wed, 19 Oct 2016 22:22:49 GMT

Like this?

data T;
  length STR1 STR2 STR3 $256;
  do I=0 to 255;
    STR1=cats(STR1,byte(I));
  end;
  STR2=prxchange('s/[^[:graph:]]//',-1,STR1);   
  STR3=prxchange("s/[^ a-i j-r ~-z {-I }-R S-Z 0-9 \x4A-\x4E \x5A-\x5F \x6A-\x6F \x79-\x7F &\/\-\\]//",-1,STR1);   
  putlog STR1= / STR2= / STR3=;
run;

The posix character class can't deal with EBCDIC it seems, but spelling out the characters to keep works.

STR1= € ‚ƒ„
…†　-‡ˆ‰Š‹ŒŽ‘’""•–—˜™š›œž âäàáãåçñ¢.<(+|&éêëèíîïìß!$*);^-/ÂÄÀÁÃÅÇÑ¦,%_>?øÉÊËÈÍÎÏÌ`:#@'="Øabcdefg

hi«»ðýþ±°jklmnopqrªºæ¸Æ¤µ~stuvwxyz¡¿Ð[Þ®¬£¥·©§¶¼½¾Ý¨¯]´×{ABCDEFGHIôöòóõ}JKLMNOPQR¹ûüùúÿ\÷STUVWXYZ²ÔÖÒÓÕ0123456789³ÛÜÙÚŸ

STR2=âäàáãåçñ¢.<(+|&éêëèíîïìß!$*);^-/ÂÄÀÁÃÅÇÑ¦,%_>?øÉÊËÈÍÎÏÌ`:#@'="Øabcdefghi«»ðýþ±°jklmnopqrªºæ¸Æ¤µ~stuvwxyz¡¿Ð[Þ®¬£¥·©§¶¼½¾Ý¨¯]´×{ABCDEFGH

Iôöòóõ}JKLMNOPQR¹ûüùúÿ\÷STUVWXYZ²ÔÖÒÓÕ0123456789³ÛÜÙÚ

STR3=¢.<(+&!$*);^-/¦,%_>?`:#@'="abcdefghijklmnopqr~stuvwxyz{ABCDEFGHI}JKLMNOPQR\STUVWXYZ0123456789

Re: Scan for non-printable EBCDIC Characters

SASuser28 — Wed, 19 Oct 2016 22:30:20 GMT

That's great. I was hoping someone would reply back using a PRX function.

Re: Scan for non-printable EBCDIC Characters

data_null__ — Thu, 20 Oct 2016 17:25:45 GMT

If you want to remove everything that is non-printable COMPRESS would be easy enough.

compress(str1,,'kw')

Re: Scan for non-printable EBCDIC Characters

data_null__ — Thu, 20 Oct 2016 19:59:47 GMT

@ChrisNZ you're missing one character from STR1 do you know why? Which one?

This is a good example for using SUBSTR on the LEFT or better still COLLATE to gen the entire string

substr(str1,i+1,1)=byte(I);

collate(0,256);

Re: Scan for non-printable EBCDIC Characters

SASuser28 — Thu, 20 Oct 2016 20:02:27 GMT

Thanks data_null_!

Re: Scan for non-printable EBCDIC Characters

ChrisNZ — Thu, 20 Oct 2016 20:43:02 GMT

@data_null__ Thanks for your input.

1- It doesnt matter a space is missing from the example output.That's not the point.
I must use collate more though, Thanks for the reminder.

2- STR4=compress(STR1,,'kw'); gives the same resut as the posix expression in STR2. Not good.

Re: Scan for non-printable EBCDIC Characters

data_null__ — Thu, 20 Oct 2016 20:52:01 GMT

@ChrisNZ wrote:

@data_null__ Thanks for your input.

1- It doesnt matter a space is missing from the example output.That's not the point.
I must use collate more though, Thanks for the reminder.

2- STR4=compress(STR1,,'kw'); gives the same resut as the posix expression in STR2. Not good.

1. I didn't say it was the point I just ask if you which and why.

2. I guess it depends on your definition of printable. Seems like your are defining printable characters as characters on the keyboard "type-able" and I could understand where that might be useful but I don't think that's what is being asked.

Re: Scan for non-printable EBCDIC Characters

ChrisNZ — Thu, 20 Oct 2016 21:42:01 GMT

@data_null__ EBCDIC has 95 printable characters and that's it. Not the 190 that compress() lets through. If not, I want to learn more. 🙂

Re: Scan for non-printable EBCDIC Characters

ChrisNZ — Mon, 24 Oct 2016 21:13:02 GMT

I would call this a defect actually. Why on earth does the compress function on Z/OS use the posix character list?