while ($dir =~ /,\d{3} ((.*?)\.avi)/gi) {
   # work with $1 here
}
[download]

Since I have no MS DOS or descendent operation systems available here, I don't know exactly how the data looks, so I don't know if the regexes can be simplified or merged.

A perlish way to transverse directories recursively is to use the core module File::Find, which eliminates the use for parsing dir /s output entirely, and makes the script more portable across operation systems.

This is not a direct answer, just some points you might want to know about.

• There's nothing wrong with using two regexen. Don't try to be too clever.
• However, you can probably unify the two regexes with something like (UNTESTED)

     if ($temp =~ /(.+?)(\d{1,2}x\d{1,2}|s\d{1,2}e\d{1,2})/ix)
          {
          $inventory{$1}=$2;
          }
  [download]

• You could also try substitutions like

     if (($temp=~s/(\d{1,2}x\d{1,2})//i)
          or
         ($temp=~s/(s\d{1,2}e\d{1,2})//i)
        )
          {
          $inventory{$_}=$1;
          }
  [download]

• It seems that the second regex in your code is missing an opening parenthesis.
• You may want to include a \z anchor to your regexen, because the way they are they could match an 1x1 marker inside the string and lose the part after it.
• The same applies to the regex that removes .avi: it could match .avi in the middle of the string.
• If you want to unify two regexen but they're not so simple, you may want to use named captures, such as

  $ perl -we 'for (qw"14/3 03-14") { m"^(?:(?<m>\d\d)-(?<d>\d\d)|(?<d>\d
  +{1,2})/(?<m>\d{1,2}))$" or die; print "$_ => month=$+{m}, day=$+{d}.\
  +n"; }'
  14/3 => month=3, day=14.
  03-14 => month=03, day=14.
  [download]

  If you didn't have named captures (such as before perl 5.10), you would have to write this in an uglier way:

  perl -we 'for (qw"14/3 03-14") { m"^(?:(\d\d)-(\d\d)|(\d{1,2})/(\d{1,2
  +}))$" or die; my $m = $1||$4; my $d = $2||$3; print "$_ => month=$m, 
  +day=$d.\n"; }'
  [download]

  which is prone to errors if you have to change the regex later so the captures are renumbered.
• Similarly sometimes the $+ special variable can help.

If I understand your first two lines correctly, you're trying to isolate just the filename, without '.avi'. You can use the /b switch, along with /s, to get a bare listing, and then parse out the filename on the fly.

File::Find would also be an excellent choice, but I'm just commenting on the code you have.

The two patterns are almost identical, so just anchor the second one, use a character class ([xe]), and you should be good.

And lastly, you can put most of that into the for loop, and use the regexes on $_ instead of creating temporary variables.

for( map{ s/\\*([^\\]+)\.avi\n/$1/; $_ } `cmd /c dir /s /b *.avi` ) {
   $inventory{$1} = $2 if /^(.*?)s\d{1,2}[xe]\d{1,2})/i);
}

# BROKEN OUT -- follow the numbers
for( map{ s/\\*([^\\]+)\.avi\n/$1/; # 2: isolate filename between last
                                    #     backslash (if any) and newli
+ne
          $_ }                      # 3: pass on $_, not result of s//
+/, which is the
                                    #     number of replacements made
     `cmd /c dir /s /b *.avi`       # 1: get avi files, including sub-
+dirs.
                                    #     dir is a shell (cmd) command
+, so
                                    #     you might need to call it th
+is way.
)
{
   $inventory{$1} = $2 if /^(.*?)s\d{1,2}[xe]\d{1,2})/i);
}
[download]

our $dir = `dir $movies\\*.avi /s`;
our $inv;

@_ = $dir =~ /,\d{3} ((.*?)\.avi)/gi;

for (@_) {
 $inv->{$1} = $2 if /(.*?)s?(\d{1,2}[ex]\d{1,2})/i
}
[download]