Path: utzoo!utgpu!news-server.csri.toronto.edu!bonnie.concordia.ca!uunet!convex!usenet
From: tchrist@convex.COM (Tom Christiansen)
Newsgroups: comp.lang.perl
Subject: Re: Pattern matching
Message-ID: <1991Jun16.042730.5266@convex.com>
Date: 16 Jun 91 04:27:30 GMT
References: <2799@moscom.UUCP>
Sender: usenet@convex.com (news access account)
Reply-To: tchrist@convex.COM (Tom Christiansen)
Organization: CONVEX Software Development, Richardson, TX
Lines: 43
Nntp-Posting-Host: concave.convex.com

From the keyboard of wjb@moscom.UUCP (Bill de Beaubien):
:While working on a Perl script a little while back, I found myself trying
:to substitute occurances of the contents of a variable with nothing.  This
:is a straight-forward thing, except that the string I was trying to match
:on contained parentheses, which Perl took as RE parens rather the text 
:parens.  Easy enough to solve, but the solution I came up with just didn't
:appeal to me esthetically;  it seems kind of bulky, and I have to wonder
:if there isn't a better way.  So, tell me... is there a better way?
:
:This is the code I came up with;  The list I'm trying to remove from is
:in $nodes{$elt}, and the pattern I'm trying to remove is in $elt. 
:
:...
:	$nodes{$elt} .= join(" ",@list);
:	$ptrn=$elt;
:	$ptrn=~s/([\(\)])/\\$1/g;
:	($nodes{$elt}) =~ s/(.*)$ptrn(.*)/$1$2/
From the FAQ:

    18) How can I quote a variable to use in a regexp?

	From the manual:

	    $pattern =~ s/(\W)/\\$1/g;

	Now you can freely use /$pattern/ without fear of any unexpected
	meta-characters in it throwing off the search.  If you don't know
	whether a pattern is valid or not, enclose it in an eval to avoid
	a fatal run-time error.


I think I might have written it this way:

    # (apparently in some kind of loop?)
    $nodes{$elt}  .= " @list";

    # then once and once only:
    ($ptrn = $elt) =~ s/(\W)/\\$1/g;
    $nodes{$elt}   =~ s/$ptrn//g;



--tom