HP-UX newline character with sed

jhwilliams · 07-26-2007, 11:36 AM

On GNU/Linux:

Code:

echo "absolut balogne." | sed "s/\ /\n/"
absolut
balogne.

On HP-UX:

Code:

echo "absolut balogne." | sed "s/\ /\n/"
absolutnbalogne.

Why doesn't it appreciate my newline character, and what do I do so that I get the desired effect (a newline in the output) ?

Thank you!

- Jameson

stress_junkie · 07-27-2007, 04:37 PM

I figured I'd try responding without having a good answer.

My guess is that since Linux does not have a true Bourne Shell then this may be a difference between the true Bourne Shell and bash.

I would look for a common syntax for the command. For instance the following code is different than yours and it works on Linux. I don't have an HP-UX machine to test it on.

Code:

echo "absolut balogne." | sed "s/ /\n/"

jschiwal · 07-27-2007, 04:54 PM

From the sed manpage (on a Mac):

Code:

Sed Regular Expressions
     The regular expressions used in sed, by default, are basic regular
     expressions (BREs, see re_format(7) for more information), but extended
     (modern) regular expressions can be used instead if the -E flag is given.
     In addition, sed has the following two additions to regular expressions:

     1.   In a context address, any character other than a backslash (``\'')
          or newline character may be used to delimit the regular expression.
          Also, putting a backslash character before the delimiting character
          causes the character to be treated literally.  For example, in the
          context address \xabc\xdefx, the RE delimiter is an ``x'' and the
          second ``x'' stands for itself, so that the regular expression is
          ``abcxdef''.

     2.   The escape sequence \n matches a newline character embedded in the
          pattern space.  You can't, however, use a literal newline character
          in an address or in the substitute command.

Try your first example in Linux using single quotes or use a sed script. I'm not at my Linux computer now so I can't try it.

jhwilliams · 07-27-2007, 05:08 PM

stress_junkie -

Code:

hp-ux-on-bash-3.00$ echo "asbolut balogne." | sed "s/ /\n/"
asbolutnbalogne.

same stuff :-(
--

Quote:

Try your first example in Linux using single quotes

Yea, that usually works pretty good.

Code:

/home/jameson% exec /bin/bash
bash-3.00$ echo 'absolut balgogne.' | sed 's/\ /\n/'
absolutnbalgogne.
bash-3.00$ echo 'absolut balgogne.' | sed 's/ /\n/'
absolutnbalgogne.

makes no different on this ux box though :-/

jschiwal · 07-27-2007, 05:31 PM

The second example is using Linux, isn't it? It is behaving now the same as the the UX example did. I think that the "\n" is being changed to an actual return character before being read in by sed.

jschiwal · 07-27-2007, 05:43 PM

OK, I tried it both ways in Linux. A newline was inserted. The difference is in the sed command itself. The paragraphs I quoted don't exist in the GNU sed manpage.

jhwilliams · 07-27-2007, 07:57 PM

Quote:

The second example is using Linux, isn't it?

No, sorry. I should have made that clear as in the first example. All of the commands in my previous post were on UX.

choogendyk · 08-06-2007, 03:21 PM

Solaris treats the \n in sed the same as HP/UX.

I learned sed and have always used it as an editor that works within individual lines within files. ^ and $ delimit the beginning and end of lines; but, for example, 's/^18//g' just removes the 18 from the beginning of the line, it does not remove the beginning of the line.

The Gnu/Linux example is probably an extension of the tradional functionality.

I just jump to other tools when one doesn't quite do it -- grep, tr, sed, awk, sort, uniq, etc. and then on to perl snippets. Although I agree it can be annoying when a really simple solution turns out not to be.

jhwilliams · 08-06-2007, 05:13 PM

I have embarrasing news to report. The whole time I said "HP-UX," I was actually on an IRIX6 machine. (blushes)