pattern matching question

jxia has asked for the wisdom of the Perl Monks concerning the following question:

I am trying to remove the first '/' in all file names in a string. But my first operator only removes the first occurrence when I use '.*' in my pattern. The second one works when I use '\w*'. I just couldn't figure out what is the difference.

Following is my code:

$_ = 'A /gif/cart.gif is a /gif/cart.gif';
s#/(gif/.*\.gif)#$1#g;     # Only replaced first one
print "NEW1: $_\n";

$_ = 'A /gif/cart.gif is a /gif/cart.gif';
s#/(gif/\w*\.gif)#$1#g;    # This one works 
print "NEW2: $_\n";
[download]

Comment on pattern matching question Download Code

Replies are listed 'Best First'.
Re: pattern matching question by hardburn (Abbot) on Jun 20, 2003 at 15:48 UTC
You're problem is that .* is greedy (i.e., it matches everything it possibly can). See also Death to Dot Star!. ---- I wanted to explore how Perl's closures can be manipulated, and ended up creating an object system by accident. -- Schemer Note: All code is untested, unless otherwise stated	[reply]
Re: pattern matching question by KPeter0314 (Deacon) on Jun 20, 2003 at 16:02 UTC
The biggest thing is the difference in the definintion between . and \w. . matches anything except \n \w matches a word character [a-zA-Z_0-9] Quite a big difference in my book. -Kurt Doh! Update: the second item was misstyped as \n and is fixed to \w. Note to self, slow down and read what you type.	[reply]
Re: pattern matching question by DigitalKitty (Parson) on Jun 20, 2003 at 17:19 UTC
Hi jxia. The '*' quantifier matches as many characters as it can (zero or more). The '.' implies any character except a newline. You can disable greediness by including a '?' to the end of your regex pattern. Suggestion: Don't use '#' in a pattern match or replacement. It looks like a comment and can frustrate debugging efforts. Hope this helps, -Katie.	[reply]
Re: pattern matching question by monsieur_champs (Curate) on Jun 20, 2003 at 18:38 UTC
Dear fellow Maybe you would like to use `s,\B/\b,,og` in place of the polluted `s#/(gif/\w\.gif)#$1#g`. Its more generic and will run a little faster. The main difference between using `.` and `\w` in your code is that `.` allways try to "eat" as many characters it can, including spaces and everything else in your way(1). In your case, it will match "gif/cart.gif is a /gif/cart" (yes, all those). On the other hand, `\w` can't match that much characters, and stops matching when it finds a space character (e.g.: one of the " ", "\n", "\t", "\f"). That way, it will match "gif/cart", because the "gif/cart.gif is a /gif/cart" have spaces before the "\.gif" that you asked your regexp to match before that. Hope that helps. =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= Just Another Perl Monk Note #1:* Actually, not exactly. But "What you don't know don't hurts you".	[reply] [d/l] [select]