Perl s /../../无法正常工作

Question

The code is a bit convoluted, but I simplified it a bit. I know I could easily do this:

$svn_module s#.*/##;

And pull off just the end of the module. But something strange is going on here:

#! /usr/bin/env perl
use strict;
use warnings;

my $svn_module = "http://svn.vegicorp.net/svn/trunk/SessionController";
print qq(DEBUG: svn_module = "$svn_module"\n);
$svn_module =~ s#^.*(branches/.+?/)|(trunk)/##;
print qq(DEBUG: svn_module = "$svn_module"\n);

This prints:

DEBUG: svn_module = "http://svn.vegicorp.net/svn/trunk/SessionController"
DEBUG: svn_module = "http://svn.vegicorp.net/svn/SessionController"

However, I was expecting:

DEBUG: svn_module = "http://svn.vegicorp.net/svn/trunk/SessionController"
DEBUG: svn_module = "SessionController"

Why does my substitution expression remove trunk/ , but not the rest of the string up to trunk/ ?

By the way, adding an extra set of parentheses helps:

$svn_module =~ s#^.*((branches/.+?/)|(trunk))/##;

will work.

By the way, this is Perl 5.8.8. This is a server, and pretty much the version I'm stuck on.

Answer 1

Probably it should be this

s#^.*(?:branches/.+?/|trunk)/##;

Because, the other way it was a single alternation where the ^.* are not part
of the second alternation (the one that matched).

   ^ .* 
   ( branches/ .+? / )
|  
   ( trunk )

Edit: Expanded new regex explained

 ^                       # Beginning of string anchor
 .*                      # Optional match as many as possible non-newline character until ..
 (?:                     # Start non-capture grouping
      branches/ .+? /        # 'branches' plus '/' plus 1 or more chars plus '/'
   |  trunk                  # Or, 'trunk'
 )                       # End grouping

The equivalent in terms of your original regex is this

   ^ .* 
   ( branches/ .+? / )
|  
   ^ .* 
   ( trunk )

Answer 2

I think you answered the question by yourself. The | operator has lowest precedence, so putting the additional parentheses is the solution.

Answer 3

Perhaps I'm missing something here, David, but instead of substituting everything up to the last part that you want to get what you want, why not capture just that last part?

use strict;
use warnings;

my $svn_module = "http://svn.vegicorp.net/svn/trunk/SessionController";
my ($end_module) = $svn_module =~ /([^\/]+)$/;
print $end_module;

Output:

SessionController