RE: sed to translate

Thanks all,
I think I agree.
Concatenating 2 regexes should suffice.
But it would be nice to tuck away a more arcane single expression for
future reference.

Peter

-----Original Message-----
From: rpg400-l-bounces@xxxxxxxxxxxx
[mailto:rpg400-l-bounces@xxxxxxxxxxxx] On Behalf Of Adam Glauser
Sent: Thursday, 13 August 2009 1:07 a.m.
To: rpg400-l@xxxxxxxxxxxx
Subject: Re: sed to translate

Peter Connell wrote:

For example, if my xml file named test1.xml contains the following -
<company>smith & brown not smith &amp; jones but F&P </company>

[...] when I run this from QShell
sed 's/\&[^amp]/\&amp;/g' test1.xml

I get
<person>smith &amp;brown not smith &amp; jones but F&amp; </person>

I think your regular expression '\&[^amp]', means 'an ampersand,
followed by one character that is not a, m or p.'. You can make a
regular expression that means "not followed by this sequence of
characters", but the only examples I've found either use more advanced
regex dialects, or are really ugly[1].

It seems like the KISS principle applies here, in which case Joe's
suggestion wins by a long shot.

[1] http://www.kleenecode.net/2008/10/10/regex-fu-not-followed-by/