line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package Lingua::JA::Summarize::Extract::Plugin::Parser::Trim; |
2
|
|
|
|
|
|
|
|
3
|
1
|
|
|
1
|
|
6
|
use strict; |
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
41
|
|
4
|
1
|
|
|
1
|
|
5
|
use base qw( Lingua::JA::Summarize::Extract::Plugin ); |
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
215
|
|
5
|
|
|
|
|
|
|
__PACKAGE__->mk_accessors(qw/ han_size kana_size latin_size /); |
6
|
|
|
|
|
|
|
|
7
|
|
|
|
|
|
|
sub parse { |
8
|
0
|
|
|
0
|
0
|
|
my ($self) = @_; |
9
|
0
|
|
0
|
|
|
|
my $han_size = $self->han_size || 2; |
10
|
0
|
|
0
|
|
|
|
my $kana_size = $self->kana_size || 3; |
11
|
0
|
|
0
|
|
|
|
my $latin_size = $self->latin_size || 3; |
12
|
|
|
|
|
|
|
|
13
|
0
|
|
|
|
|
|
my $term_list = {}; |
14
|
0
|
|
|
|
|
|
my $text = $self->text; |
15
|
0
|
|
|
|
|
|
while ($text =~ /(\p{Katakana}{$kana_size,}|\p{Han}{$han_size,}|\p{Latin}{$latin_size,})/g) { |
16
|
0
|
|
|
|
|
|
$term_list->{$1}++; |
17
|
|
|
|
|
|
|
} |
18
|
|
|
|
|
|
|
|
19
|
0
|
|
|
|
|
|
$term_list; |
20
|
|
|
|
|
|
|
} |
21
|
|
|
|
|
|
|
|
22
|
|
|
|
|
|
|
1; |
23
|
|
|
|
|
|
|
__END__ |