grammar G { token TOP { <sentRefEnd> } token sentFrag { <char>+? <?before <endPunc>> } token sentRefEnd { <sentFrag> <endPunc> <ref> } # token sentRefEnd { <char>+? <endPunc> <ref> } token endPunc { \. } token char { \w || \s } token ref { \(\d+\) } } my $book = "Hello.(1)"; my $match = G.parse($book); say $match;