There's no need to backslash a space in regular expressions. Moreover, there's no need for a substitution in this case, transliteration does the same job (and is simpler and faster).

There's a pattern in the computation of the sum which could be expressed by a for loop.

See my version below:

sub check_abn {
    my ($number) = @_;
    $number =~ tr/ //d;
    return if 11 != length $number;

    my $sum = 0;
    $sum += (2 * $_ - 1) * substr $number, $_, 1 for 1 .. 10;
    $sum += 10 * (substr($number, 0, 1) - 1);
    return $sum % 89 == 0
}
[download]

Thanks for your suggestions, especially to reverse the logic and reduce surprise.

I like the code from choroba, he posted that code while I was still writing mine.

Some notation comments: Instead of "ChkABN", I would use "isABN". That is a common convention that makes it clear that this is a yes/no question, usually >0 or 0 for return value (any non zero value is "true").

I would reserve any variable name that starts with a capital letter for a Class.

I personally like the style that choroba use for naming (all lower case and _ for separation). is_ABN() would be a good name for me.

I'm needing a procedure to check Australian Business Numbers (aka ABN's).

Might I suggest Business::AU::ABN and the validate_abn function?

Thanks for pointing me in that direction. I'd tried googling for this module, but should have searched metacpan directly. package Business::AU::TFN will be a handy package too. Thanks for sharing your wisdom.

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


my $x = isABN("51 824 753 556");
print "$x\n";
$x = isABN("51 824 755 556");
print "$x\n";


sub isABN  # returns True if ABN, False if not an ABN
{
   my $abn = shift;
   $abn =~ tr/ //d;  #delete spaces
   
   return "False" unless (length $abn == 11);
   
   my @digits = split '',$abn;
   $digits[0] -= 1;
   my @weights = (10,1,3,5,7,9,11,13,15,17,19);
   
   my $sum =0;

   for my $i (0..10)
   {
      $sum = $sum + ($digits[$i] * $weights[$i]);
   }
   print "sum = $sum\n";  #for debugging...
   return ($sum % 89 == 0? "True": "False");
}

__END__
sum = 534
True
sum = 560
False
[download]

Hello bgroper

I thought it might be useful to construct a separate weighting array, and then apply that as part of the algorithm to the submitted ABN to check. It's a little rough but I am interested to see what the thoughts are.

#!/usr/bin/perl

use v5.20;
use strict;
use warnings;

my $Num = Prepare_Num( q'51 824 753 556' );

my $verified = check_calc($Num);

if( $verified == 1){
 say "Valid ABN"
}else{
 say "ABN check has been unsuccessful"
}

sub check_calc{

 my $Numset = shift;
 my $w_oset = weighting_oset();
 
 --$Numset->[0][0]; # -[[0],[1]] ?
 print_list($Numset);   

 my $sum;
 for my $index( 0 .. 10 ){

 $sum += $Numset->[$index][0] *= $w_oset->[$index][0]; 

 }

 print_list($Numset);

 $sum %89 == 0;
 
}

sub Prepare_Num{

my $NumString = shift;
say "arg Num: $NumString";

$NumString =~ s/[^0-9]//g;
say "s Num: $NumString";

my $count = $NumString =~ tr/0-9/0-9/;
say "count: $count";
$count == 11 or die "non-conforming ABN. Please check.";

say "tr Num: $NumString";

my $Numset = [];
while( $NumString =~ m/([0-9])/g ){
  push @$Numset, [$1]; 
}

print_list( $Numset, 1 );

return( $Numset );

}

sub weighting_oset{

 #my $woset = [[[0],[1]],[10]];
 my $w_oset = [[10]];
 push @$w_oset, map { $_ %2 == 1 ? [$_] : () } ( 1..19 );
 print_oset( $w_oset );
 return $w_oset;

}


sub print_list{

 my $Numlist = shift;
 say "list: ",@$Numlist if @_; # $_[0] == 1;
 say "list: \[", map("[@$_],", @$Numlist), "\]";

}

sub print_oset{

 my $Oset = shift;
 say "oset: ",@$Oset if @_; # $_[0] == 1;
 say "oset: \{", map("[@$_],", @$Oset), "\}";

}
[download]