reading/writing <CR>s and <FF>s in DCL.

[Russel Wilmes]

Hi,

I've got a (not very elegant) script that strips formfeeds from a text file and I also need to strip crlfs but can't figure out how to insert the cr in the dcl procedure.

$ open/read infile file1.dat
$ open/write outfile file2.dat
$ read infile line
$ if line .ne. "ctrlL" (meaning a ff) then write outfile line

Our application dumps files with trailing crlfs at the end of every file. ctrl M doesn't insert the cr. I'm using edt but can try another as well. I saw a reference using tpu and some function key sequences but am unfamiliar with tpu.

We're sending these text files to a fax server and they're reciving lots of blank pages on the far end.

Thanks in advance,
Rusty

[Russel Wilmes]

also, this is on an alpha running vms 7.3-2.

[Peter Quodling]

I would definitely consider something a little more powerful the EDT or DCL for attacking this problem (I'll be chastised for that comment, for sure). TPU is far more powerful, Perl (free) is worth considering, and there was a text manipulation language called Scan, which I think may have moved to the freeware archives.

Of course, the most elegant way to do it,is probably in Teco, but those neurons are failing...

It's also not clear what you are looking for. Do you a) want to turn multiple CRLF's into FF's, or b) delete extraneous FFs or c) delete final FF's.

PeterQ

[Russel Wilmes]

I actually want to find them and strip them. I can find the form feeds no problem, that's a control l into the dcl command procedure that does the reading and writing. i just can't figure out how to get a CR in there. some of the stuff I saw said a ctrl M but that just does a "delete to begnning of line" or somesuch.

I'd love to use perl, however, time is short and my learning curve is steep!

[Russel Wilmes]

Or maybe this would be more clear.

When using EDT to create a dcl command procedure...

ctrl [[ =
ctrl l =
? =

Hope that helps...

Thanks again!

[Doug Phillips]

To insert any special character with EDT:

PF1 decimal-value-of-charcter PF1 KP3

so for a CR,
press the PF1 key (aka: the gold key)
type 13
press PF1
press the 3 on the keypad.

However, this makes the DCL file look strange when you type it, so you might want to consider instead using a value assignment like:

$ CR[0,8] = %X0D !hex zero D = dec 13

which places the hex value 0D into a symbol named CR.

However, however (;-) there's probably a easier and better way to make the file more fax-friendly.

Doug

[Dale A. Marcy · Accepted solution]

$ CR[0,8] = 13 ! Carriage return
$ open/read infile file1.dat
$ open/write outfile file2.dat
$ read infile line
$ line = f$extract(0, f$locate(CR,line), line)
$ if line .ne. "ctrlL" (meaning a ff) then write outfile line


I have just scribbled the above out and have not tested it, but it should truncate the line at the first carriage return in the line. If there is no carriage return in the line, it should leave the line unchanged.

[Jan van den Ende]

Rusty,

in DCL, you define a symbol to be any ASCI char, by NUMERIC assign of 8 bits.

eg the Escape char (ASCI 27), using ther recognisable symbolname esc:

$ esc[0,8]=27

means "assign value 27 to 8 bits, starting from bit 0, of the symbol esc"

$ bell[0,8]=7 : bell signal

Synyax:
symbol name - (NO spaces!) - left square bracket - offset of first bit - comma - length in bits - right square bracket - equals sihn -
Now to insert it into a string, simply use string concatenation.

$ String = symbol1 + "Litteral" + string2 + string3 (etc...)
or
$ Strng := "''symbol1'litteral''string2'''string3' ..."
(single and double quotes unclear in presentation. Copy / paste into plain-text window for clarity)


Left to do: find ASCI numeric value of desired char.

hth

Proost.

Have one on me.

jpe

[Russel Wilmes]

Success! Many thanks to all who replied.

I took a roundabout way and am not looking for any carriage returns.

Dales note about using f$locate helped imensely. I'd forgotten about that.

I wanted to strip s and s.

so rather than checking for either two strings just check for existence of

open/read infile file1.dat
open/write outfile file2.dat
top:
read/end=cya infile line
a = f$locate("ctrll",line)
b = f$length(line)
if a .eq. b then write outfile line
goto top
cya:
close infile
close outfile
send file2.dat to fax server!

Still not elegant but it works. At least until our apps folks change the report layout again...

Thanks again!

[Eberhard Wacker]

Hi, maybe not "elegant" but also a working procedure using the line editor of EDT removing all CRFFs:
$ edit/edt replace.com
then enter therein:

$ edit/edt file1.dat
s///%wh
s///%wh
exit
$ exit

where within edit/edt replace.com
is created with PF1-13-PF1-3
PF1-12-PF1-3

and
PF1-10-PF1-3
if this should be replaced.

Cheers,
EW

[David Jones_21]

I can never remember the syntax for bit insertion in DCL, so I use F$FAO("!^") to get a form feed and F$EXTRACT(0,1,F$FAO("!/")) to get a CR (first character of CRLF sequence).

[William Webb_2]

You could also websearch for a TPU routine (which I believe is attributable to Ken Fairfield) that strips not only and , but all control characters from a text file.

It's especially useful if you have to take the results of a SET HOST/LOG file and turn it into plain text.

If you can't find it, drop me an email first name period middle initial (all my initials are identical) period last name at gmail period com
and I'll send it to tomorrow.

[Antoniov.]

Hi Rusty,
If you want to remove all CR/LF/FF you can:
$ CR[0,8]=13
$ LF[0,8]=10
$ FF[0,8]=12
$ OPEN/REA SRC INFILE.DAT
$ OPEN/WRITE TGT OUTFILE.DAT
$MLOOP:
$ READ SRC LINE/END=XIT
$ LINE = LINE - CR - LF - FF
$ WRITE TGT LINE
$ GOTO MLOOP
$XIT:
$ CLOSE TGT
$ CLOSE SRC

Antonio Vigliotti

[Ken Robinson]

The above procedure will only work if there is no more that one of each of the characters in each line.

I would write the procedure to use a subroutine to do the actual removing. Something like

$ set noon
$ cr[0,8]=13
$ lf[0,8]=10
$ ff[0,8]=12
$ open/err=nofile inp input_file.txt
$ copy nl: output_file.txt ! create an empty file with better attributes than open/write
$ open/app outp output_file.txt
$rl:
$ read/end=done inp line
$ test_line == line
$ call remove_chars "''cr'"
$ call remove_chars "''lf'"
$ call remove_chars "''ff'"
$ write/sym outp test_line
$ goto rl:
$done:
$ close inp
$ close outp
$ exit
$nofile:
$ write sys$Output "Input file does not exist"
$ exit
$!
$!
$remove_chars: subroutine
$rmloop:
$ if f$loc(p1,test_line) .eq. f$len(test_line) then exit
$ test_line = test_line - p1
$ goto rmloop
$ exit
$endsubroutine
$!

Note: this script is untested and not proofed for errors. I just wrote it.

Ken

[William Webb_2]

Here's the TPU procedure- STRIP_TPU.TPU

Apologies to whoever wrote it, as they didn't put their name in comments.

WWWebb

[Robert_Boyd]

Here is yet another great trick for generating CR and LF:

$ CRLF = F$FAO("!/")
$ CR = F$EXTRACT(0,1,CRLF)
$ LF = F$EXTRACT(1,2,CRLF)

Robert

[Robert_Boyd]

Woops -- I got distracted while writing that last line... should be

$ LF = F$EXTRACT(1,1,CRLF)