#!/usr/bin/perl
# Created On      : 2010
# Last Modified On: Fri Apr 16 02:05:56 2010
# Update Count    : 151

=encoding utf8

=head1 NAME

tongyong2hanyu -- Tongyong Pinyin to Hanyu Pinyin converter

=head1 USAGE EXAMPLES

 $ tongyong2hanyu file
 jhuci => zhuqi
 Lioujiao => liujiao

=head1 DESCRIPTION

Input: Tongyong Pinyin multi-syllable strings and dashes (-).
Output: lower case Hanyu Pinyin appended to each line.
Empty lines are passed through unaltered.

=head1 COPYRIGHT

http://www.fsf.org/copyleft/gpl.html

=head1 AUTHOR

積丹尼 Dan Jacobson http://jidanni.org/lang/pinyin/
...whose goal is to help clean up the Taiwan Pinyin mess.

=cut

use strict;
use warnings FATAL => 'all';
while (<>) {
    if (/^$/) { print; next; }
    chomp;
    my $original = $_;
    $_ = lc;
    s/-/'/g;
    s/([wf])ong/$1eng/g;
    s/wun/wen/;
    s/jh/zh/g;
    s/uei/ui/g;
    s/iou/iu/g;
    s/(\bn|[ljcs])yu/$1ü/g;
    s/[jz]ü/ju/g;
    s/[qc]ü/qu/g;
    s/[xs]ü/xu/g;
    s/([zcsjqx])y/$1i/g;
    s/ihu/i'hu/g;
    s/ci(?!h)/qi/g;
    s/si(?!h)/xi/g;
    s/ih/i/g;
    s/i'hu/ihu/g;    #P.S.: siha si'ha sih'a: sia xi'ha si'a...
    print "$original => $_\n";
}