Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!usc!snorkelwacker.mit.edu!bloom-beacon!eru!hagbard!sunic!mcsun!i2unix!inria!seti!minos!deschamp
From: deschamp@minos.inria.fr (Philippe Deschamp)
Newsgroups: comp.unix.shell
Subject: Re: How to strip A NEWLINE
Message-ID: <1781@seti.inria.fr>
Date: 30 Nov 90 17:25:01 GMT
References: <25665@uflorida.cis.ufl.EDU>
Sender: news@seti.inria.fr
Reply-To: deschamp@seti.inria.fr
Organization: INRIA
Lines: 35


In article <25665@uflorida.cis.ufl.EDU>, bjs@reef.cis.ufl.edu (Brian J. Smith)
writes:
|> I have searched high and low for the answer on how to strip a NEWLINE
|> charactor from a file.  Well not really a NEWLINE char, but a ":\\\n"
|> (colon,back-slash,newline).  I have tried useing sed with:
|> 
|> 	sed 's/:\\\n//' file
|> 
|> and using the RS variable in awk, but to no avail.  sed will remove
|> the colon and the back-slash, but not the newline.

   The key is that "\n Matches a NEWLINE embedded in the pattern space." (from
sed(1)).  So you must embed a NL in the pattern space!  With the "N" command,
for example.  The following does what you asked for, I should think:

	sed -e ': more
		/:\\$/{
		N
		s/:\\\n	*:*/:/
		t more
		}'

You can read this as
	while the current line ends with ":\", do
		append next line to current, with NL in between,
		get rid of "\", the NL and some TABs and superfluous ":"s
		end while

Have fun! Sed is a wonderful tool, as long as you don't overuse it...

					Philippe Deschamp.
Tlx: 697033F   Fax: +33 (1) 39-63-53-30   Tel: +33 (1) 39-63-58-58
Email: Philippe.Deschamp@nuri.inria.fr   ||   ...!inria!deschamp
Smail: INRIA, Rocquencourt, BP 105, 78153 Le Chesnay Cedex, France