I don't fully understand what you're after, but here's an approach that might be interesting. It parses the template elements in order through a string. (Actually, a full-blown parser might be what you really need.)

c:\@Work\Perl\monks>perl -wMstrict -le
"my @lines = (
   'my-test 10:11:23 blah yada',
   'test 11:22:33',
   '   taste-test   01:02:03   fie',
   '10:20:30 toast',
   '999 hooha yip yap',
   'another-test 999 foo fee',
   );
 ;;
 my %parse = (
   '$(long_awkward.string)' =>
     qr{ (?<! [[:alpha:]]) [[:alpha:]]+ (?: - [[:alpha:]]+)* (?! [[:al
+pha:]]) }xms,
   '%t' =>
     qr{ (?<! \d) \d\d : \d\d : \d\d (?! \d) }xms,
   );
 ;;
 my $template = '$(long_awkward.string) %t xxx yyy zzz';
 ;;
 for my $line (@lines) {
   my @sub_tmpls = (split ' ', $template)[0, 1];
   my $match = validate($line, \%parse, @sub_tmpls);
   print qq{'$line' }, $match ? 'matches' : 'NO match';
   }
 ;;
 sub validate {
   my ($string, $hr_parser, @sub_templates) = @_;
   ;;
   for my $st (@sub_templates) {
     return unless $string =~ m{ \G \s* $hr_parser->{$st} }xmsg;
     }
   return 1;
   }
"
'my-test 10:11:23 blah yada' matches
'test 11:22:33' matches
'   taste-test   01:02:03   fie' matches
'10:20:30 toast' NO match
'999 hooha yip yap' NO match
'another-test 999 foo fee' NO match
[download]