A liberal use of map and grep and this is quite simple

use File::Slurp;

my @cats = map {
  chomp;
  map [reverse split ','], split /\|/;
} read_file('your_category_file');

my %desc = map {
  chomp;
  split ',', $_, 2;
} read_file('your_description_file');

my @res = map {
  my $d = $_;
  join '|',
       "$d,$desc{$d}",
       map { join ',', @$_ }
       grep { $_->[0] =~ /^$d?/ } @cats
} keys %desc;

print "$_\n" for @res;

__output__

'BB','Drinks Industry'|'BB100','Coffee Suppliers'|'BB106','Tea Supplie
+rs'
'BL','Construction Industry'|'BL100','Plasterers'|'BL102','Fencing Com
+panies'
[download]

So we build an array of categories (with fields reveresed), then a hash of descriptions, then create an array of strings for every description, prime for outputting to a file.