Can you show us the values that are equal but your code shows as different?

Please also show us the relevant part of the code you've written and tell us the exact error message/output you get.

As it is, it is difficult to understand for me where exactly you are having problems. You seem to have code that detects identity when there are no spaces in the product names, but it seems to fail when there is whitespace in the names.

A potential cause for this might be that in the YAML or in the Excel data, there is whitespace at the end of the values.

You can remove whitespace at the end of your values by using:

$value =~ s/\s+$//;
[download]

I agree with Corion and haukex that your problem is very vaguely stated. However, I never let my ignorance keep me from offering advice. Based on several WAGs (Wild-Ass Guesses) about your actual data and your actual problem, here's a possible (?) approach to developing a framework for creating a solution:

c:\@Work\Perl\monks>perl -wMstrict -le
"my @data = (
   'NABCv-ABC : NABCv-ABC',
   'BC8200 3*AB25 Products : BC8200 3*AB25 Products',
   '  BC8200 3*AB25 Products    :     BC8200 3*AB25 Products     ',
   '  Q   :      Q       ',
   'something : else',
   'U:U',  ' : V',  'W : ',  'X',  '',
   );
 ;;
 DATUM:
 for my $datum (@data, @ARGV) {
   my $parsed =
   my ($ya, $ex) =
   $datum =~ m{ \A \s* (\S .*?) \s+ : \s+ (\S .*?) \s* \z }xms;
   ;;
   if (not $parsed) {
     print qq{nothing parsed from '$datum'};
     next DATUM;
     }
   s{ \A \s+ | \s+ \z }{}xmsg for $ya, $ex;
   print qq{'$ya' and '$ex' are }, $ya eq $ex ? '' : 'NOT ', 'equal';
   }
"  "what : ever"
'NABCv-ABC' and 'NABCv-ABC' are equal
'BC8200 3*AB25 Products' and 'BC8200 3*AB25 Products' are equal
'BC8200 3*AB25 Products' and 'BC8200 3*AB25 Products' are equal
'Q' and 'Q' are equal
'something' and 'else' are NOT equal
nothing parsed from 'U:U'
nothing parsed from ' : V'
nothing parsed from 'W : '
nothing parsed from 'X'
nothing parsed from ''
'what' and 'ever' are NOT equal
[download]

Update 1: Any need to strip leading/trailing whitespace can be eliminated by proper design of the field components of the field extraction regex (tested):

my $rx_ya  = qr{ \S (?: \s* \S+)* }xms;
my $rx_ex  = $rx_ya;
my $rx_sep = qr{ \s+ : \s+ }xms;
...
$datum =~ m{ \A \s* ($rx_ya) $rx_sep ($rx_ex) \s* \z }xms;
[download]

Update 2: Also note that it's easy to write the extraction regex so that a single, whitespace-trimmed field is extracted only if both fields are equal.

Please see How do I post a question effectively? - you haven't shown which error message you're getting, and please use <code> tags to format your code, sample input, and expected output. Also, when asking questions about regexes, please provide as much sample input (both what should match and what shouldn't) as you can - see Re: How to ask better questions using Test::More and sample data.

I'm just wondering why you can't just use YAML to solve this problem.