File Coverage

blib/lib/Lingua/JA/Summarize/Extract/Plugin/Parser/Ngram.pm

Criterion	Covered	Total	%
statement	6	23	26.0
branch	0	2	0.0
condition	0	6	0.0
subroutine	2	4	50.0
pod	0	1	0.0
total	8	36	22.2

package Lingua::JA::Summarize::Extract::Plugin::Parser::Ngram;

use strict;
use base qw( Lingua::JA::Summarize::Extract::Plugin );
__PACKAGE__->mk_accessors(qw/ latin_gram kana_gram han_gram /);

sub parse {
    my ($self) = @_;
    my $latin_gram = $self->latin_gram || 2;
    my $kana_gram = $self->kana_gram || 3;
    my $han_gram = $self->han_gram || 2;

    my $term_list = {};
    $self->_gram($term_list, 'Latin', $latin_gram);
    $self->_gram($term_list, 'Katakana', $kana_gram);
    $self->_gram($term_list, 'Han', $han_gram);

    $term_list;
}

sub _gram {
    my($self, $list, $block, $gram) = @_;

    my $text = $self->text;
    while ($text =~ /(\p{$block}+)/g) {
        my $word = $1;
        my @part;
        for (my $i = 0;$i + $gram <= length $word;$i++) {
            push @part, substr $word, $i, $gram;
        }
        $list->{join ' ', @part}++ if @part;
    }
}

1;

__END__