You really want File::Basename, which is made for file/directory manipulation.

But for regular expressions, you want what is called an anchor. There are two (well, technically four) of them: ^ anchors the match to the start of the string, and $ anchors it to the end (see perlre). So your problem could be solved by:

my $file = "c:/progra~1/apache~1/apache2/cgi-bin/test/new/Feb23-2006_0
+_test_error.txt";
if ($file =~ m!^(.*)/(\w+\.txt)$!) {
  print "Found $1 - $2\n";
} else {
  print "No match found for '$file'\n";
};
[download]

There are five anchors:

	without m	with m
Start of string	\A and ^	\A
Start of line		^
End of string or before \n at end of string	\Z and $	\Z
End of line or before \n at end of line		$
End of string	\z	\z

You might consider using File::Basename.

Update: Since [id://Corion] beat me by seconds in answering your question, I'll have to redeem myself by adding a little more detail.

The reason that [id://Corion]'s solution works is because of regular expression greed. Specifically, the '.*' part of the expression is greedy (since it doesn't have a '?' after it) and it will expand to consume the largest string possible while still leaving a few crumbs for the rest of the regex -- that is, the /(\w+\.txt) part.

There is probably a better way to describe regular expression greed, but that is the way it has made sense to me.

I try to think of what defines the thing I'm looking for. In this case, I'd think it's the last thing in the string, and the string is separated by '/' marks, that would give me a regex like this:

$file =~ /\/([^\/]+)$/;
[download]

This says, find a '/' followed by one or more non-'/' up until the end of the string. Capture the non-'/' stuff.

Update: More of a breakdown. Because '/' itself is used by perl to delimit the regex, in order to search for a '/' in a string I escape it in the regex, hence '\/'.

The bit that looks like '\/([^\/]+)' is saying find a '/' followed by some non-'/' chars. '\/' means find a '/', '([^\/]+)' means get a bunch of non-'/' chars. '[^\/]' matches any character that isn't a '/'. Putting a '+' at the end of '[^\/]' means match 1 or more times, and putting that inside of parenthesis means capture it for a back reference.

The '$' anchors that to the end of the string.

Once you look at how a regex works, it can be fun to think about strings that could break it. What would $1 look like if $file were any of these?

• foo.txt
• /my/head/hurts/

Another Update: Great point kwaping! (see below)

You can use different delimiters for your match, which removes the need to escape your slashes. This may or may not make that line less confusing. :)

$file =~ m|/([^/]+)$|;
[download]

This one worked great, but can you explain a little how it is doing it? Thanks a lot you all!

It reads as /, followed by one or more characters which aren't /, followed by the end of string. It works because of backtracking. When it can't find a match with the first slash (because there are other slashes between it and the end of string), it tries starting at a different slash. Rinse and repeat until a match is found, or there are no more slashes at which to start.

my $ file = "c:/progra~1/apache~1/apache2/cgi-bin/test/new/Feb23-2006_
+0_test_error.txt";
print "$'\n", if($file=~/(.*)\//g);
[download]

(my $file = "c:/progra~1/apache~1/apache2/cgi-bin/test/new/Feb23-2006_
+0_test_error.txt")=~ s/.*\///g;
print $file;
[download]