Need help with TRANSLATE()

Harold Barnes · Posted: Thu Sep 17, 2020 8:28 am

I've built 2 tables, one for EBCDIC to ASCII and the other for ASCII to EBCDIC

strin = "A"
estr = translate(strin,ASC,EBC)
say "estr:" c2x(estr)

I'm expecting:
estr: 41

Instead I get :
estr: C1

So translate is not finding 'C1'x in the EBCDIC table and does not change it.

Thinking that my translate table might be bad (never happens) I manually extract the 'C1'x position from the EBCDIC table.

epos = x2d('c1'x) + 1
say "epos: " epos

epos: 194

aval = substr(EBC,epos,1)
say "aval: c2x(aval)

aval: 41

So my EBCDIC table is good. EBCDIC 'A' is ASCII '41'x

apos = x2d('41'x) + 1
apos: 66

aval = substr(ASC,apos,1)
say "aval:" c2x(aval)

aval: C1

So my ASCII table is good. ASCII '41'x is EBCDIC 'C1'x' (letter A)

Each table is 256 bytes of hex values. All values have corresponding values.

Why doesn't translate find the value?

Harold Barnes · Posted: Thu Sep 17, 2020 8:59 am

Uploading code. (Attached)

prino · Posted: Thu Sep 17, 2020 3:25 pm

Nic Clouston · Posted: Thu Sep 17, 2020 5:51 pm

Please do not post attachments - some people are not allowed to download and some do not do it as a security precaution. They are also an interference with rapid browsing of the problem. A simple cut and paste is good enough and more economical. Use the code tags to present code and data.

Harold Barnes · Posted: Thu Sep 17, 2020 6:09 pm

I was being sarcastic about the possibility of me making a mistake.

I found the problem. I expected it to work like an ALC translate but it doesn't.

Assumptions will get you every time.

The ASC table just needed to be generated like this:

ASC = ""
do i = 0 to 255
val = x2c(d2x(i)
ASC = ASC || val
end

So basically I generate a table entry for every ascii value. Seems stupid but that's the way translate() works.

Coding:
atran = transalate("A",ASC,EBC)

..will scan the EBC table and find x'C1' in the x'41' position.

Then it will go to the x'41' position in the ASC table and find x'41' (surprising)

The reverse works as well.

atran = transalate('41'x,EBC,ASC)

..will scan the ASC table and find x'41' in the x'41' position. (surprising)

Then it will go to the x'41' position in the EBC table and find x'C1'

Problem solved.

enrico-sorichetti · Posted: Thu Sep 17, 2020 6:28 pm

the tables might be good , the translate invocation is wrong

when translating from ascii to ebcdic and/or the other way around it is enough to use the output table
the input translate table when omitted defaults to XRANGE('00'x,'FF'x)

Harold Barnes · Posted: Thu Sep 17, 2020 9:49 pm

enrico-sorichetti

Thanks for the reminder to use XRANGE() and for mentioning the default input of TRANSLATE().

Your table coding style is a lot cleaner than mine as well. Thanks for sharing.

Nic Clouston:

I attached the code thinking that it might make it easier for someone to play. I'll avoid in the future.

Pedro · Posted: Fri Sep 18, 2020 7:57 am

Enrico's table has '40'x for non-display characters, but I vaguely recall that some displays (maybe it was IPCS) show non-display characters as 'periods'. as place holders so that you can tell there is something there other than spaces.