File Coverage

blib/lib/Lingua/Han/CanonicalPinYin.pm

Criterion	Covered	Total	%
statement	30	31	96.7
branch	9	12	75.0
condition	2	6	33.3
subroutine	6	6	100.0
pod	0	1	0.0
total	47	56	83.9

package Lingua::Han::CanonicalPinYin;
use strict;
use warnings;

our $VERSION = 0.04;

use base 'Exporter';

our @EXPORT_OK = 'canonicalize_pinyin';
use utf8;
use Encode;
my @tones = ( "\x{304}", "\x{301}", "\x{30c}", "\x{300}" );
sub canonicalize_pinyin {
    my $pinyin = lc shift;
    $pinyin = decode_utf8( $pinyin ) unless utf8::is_utf8( $pinyin );
    my $tone;
    ($pinyin, $tone) = ( $1, $2 ) if $pinyin =~ /^(.*)(\d)$/;
    return $pinyin if ! defined $tone || $tone == 5;

    if ( $tone < 1 || $tone > 5 ) {
        die "invalid pinyin $pinyin: tone $tone doesn't exist";
    }

    $tone = $tones[$tone-1];
    for my $vowel (qw/a o e iu i u v ü/) {
        if ( $pinyin =~ /$vowel/ ) {
            if ( $vowel eq 'v' ) {
                $pinyin =~ s/v/u\x{308}$tone/;
            }
            else {
                $pinyin =~ s/$vowel/$vowel$tone/;
            }
            last;
        }
    }
    $pinyin =~ s/v/ü/g;

    return $pinyin;
};

1;

__END__