From: Robert Klemme <shortcutter@...>
Date: 2009-04-11T06:45:59+09:00
Subject: Re: regex woes with breaklines

On 10.04.2009 14:52, Aa Wilson wrote:
> Andrew Timberlake wrote:
> 
>> s = "test\r\ntest<br>\r\ntest\ntest\rtest<br/>\ntest"
>> s.gsub(/(?:<br *\/*>)*(?:\r\n|\n|\r)/, "<br />\n")
>> #=> "test<br />\ntest<br />\ntest<br />\ntest<br />\ntest<br />\ntest"
> 
> Thanks a million.  It feels like a waste replacing something that's 
> already there, but I suppose that's just my conceptions of real life 
> getting in the way of my abstractions.

Not necessarily: with 1.9's regular expression engine there is negative 
lookbehind as well.  You could use that to prevent substitution of 
newlines which are preceeded by a <br/> already.

http://www.geocities.jp/kosako3/oniguruma/doc/RE.txt

> I was actually hoping that '$' (or possibly '$$?') would cover all the 
> CR/LF/CRLF cases for me, was I mistaken about that?

I believe it does not cover \r alone.  In every case it is safer to be 
explicit IMHO.  Btw, I believe you can simplify to (?:\r\n?|\n).

Kind regards

	robert