$_='ABC AB AB12/83';
tr / \//_/;
print
[download]

Why's the space before the backslash?

Update: ah, I forgot. Perl repeats the last character of the right side of the tr if there are too few. Great.

Because that way both spaces and slashes get translated into underscores. This strips out the brackets as well:

$_='[ABC AB AB12/83]';
tr# /[]#__#d;
print;
[download]

---

The PerlMonk tr/// Advocate

#!perl -wl

use strict;

$_ = "[ABC AB AB12/83]";
print "I have a string like this\n$_";
s{\[([A-Z_]+)\s([A-Z_]+)\s([A-Z_]+)(\d+)/(\d+)\]}{$1_$2_$3$4_$5};
print "I would like to end up with\n$_";
[download]

I didn't explain myself quite right.


The "/" can be in any part of the string, so a new regexp won't fix it. What I want is to capture four clusters and then, when substituting, replace the "/" inside the clusters.

Right now, I did this:


$do_the_change_mon = sub { 
   $thingy = shift;
   $thingy =~ tr|/|_|;
   return $thingy;
}

$expr = s|
\[([A-Z/]+)\s([A-Z/]+)\s([A-Z/]+)\s([A-Z/]+)\s\]
|&$do_the_change_mon("$1_$2_$3_$4)|xe;
[download]

Thank you all for your kind answers, monks!

That's not bad. Consider this. I rolled your A-Z/ loop up and then did the substitution on a copy inside the RHS.

$expr =~ s(\[((?:[A-Z/]+)\s){4,4})\]){
    my $match = $_;
    $match =~ s([/\s])(_)g;
    $match;
}ge;
[download]

s/^\[([A-Z_]+)\s([A-Z_]+)\s([A-Z0-9]+)\/([0-9]+)\]$/$1_$2_$3_$4/g
[download]

I'd probably use duff's solution, but here's another way to do it:

  my $str = join "_", split m([ /]), "[ABC AB AB12/83]";
  substr($str,  0, 1) = "";
  substr($str, -1, 1) = "";
[download]

Update: here's a [very ugly] way to do it all in one line:

  my $str = substr ${\join "_", split m([ /]), "[ABC AB AB12/83]"},
                   1, -1;
[download]