menz.pl

gistfile1.txt

#!/home/ben/software/install/bin/perl
use warnings;
use strict;
use Lingua::JA::Moji ':all';
binmode STDIN, ":encoding(UTF-8)";
binmode STDOUT, ":encoding(UTF-8)";
while (<STDIN>) {
    if (/^(\S+)\s+\[(\p{InKana}+)\]/) {
	print "$2\t$1\n";
    }
    elsif (/^(\p{InHiragana}+)\s+/) {
	print "$1\t$1\n";
    }
    elsif (/^(\p{InKatakana}+)\s+/) {
	next;
    }
    else {
	chomp;
	print STDERR "$.: $_ unmatched\n";
    }
}