line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package Plucene::Plugin::Analyzer::MetaphoneAnalyzer;
|
2
|
|
|
|
|
|
|
|
3
|
1
|
|
|
1
|
|
41751
|
use base 'Plucene::Analysis::Analyzer';
|
|
1
|
|
|
|
|
1
|
|
|
1
|
|
|
|
|
1006
|
|
4
|
1
|
|
|
1
|
|
203
|
use 5.006;
|
|
1
|
|
|
|
|
4
|
|
|
1
|
|
|
|
|
39
|
|
5
|
1
|
|
|
1
|
|
5
|
use strict;
|
|
1
|
|
|
|
|
6
|
|
|
1
|
|
|
|
|
28
|
|
6
|
1
|
|
|
1
|
|
5
|
use warnings;
|
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
38
|
|
7
|
|
|
|
|
|
|
our $VERSION = '1.02';
|
8
|
1
|
|
|
1
|
|
1004
|
use Plucene::Analysis::Standard::StandardTokenizer;
|
|
1
|
|
|
|
|
48768
|
|
|
1
|
|
|
|
|
21
|
|
9
|
1
|
|
|
1
|
|
852
|
use Plucene::Plugin::Analyzer::MetaphoneFilter;
|
|
1
|
|
|
|
|
5
|
|
|
1
|
|
|
|
|
15
|
|
10
|
|
|
|
|
|
|
|
11
|
|
|
|
|
|
|
our $LANG = 'en';
|
12
|
|
|
|
|
|
|
|
13
|
|
|
|
|
|
|
=head1 NAME
|
14
|
|
|
|
|
|
|
|
15
|
|
|
|
|
|
|
Plucene::Plugin::Analyzer::MetaphoneAnalyzer - Metaphone analyzer
|
16
|
|
|
|
|
|
|
|
17
|
|
|
|
|
|
|
=head1 DESCRIPTION
|
18
|
|
|
|
|
|
|
|
19
|
|
|
|
|
|
|
Filters StandardTokenizer with MetaphoneFilter
|
20
|
|
|
|
|
|
|
|
21
|
|
|
|
|
|
|
=head1 EXAMPLE
|
22
|
|
|
|
|
|
|
|
23
|
|
|
|
|
|
|
#!/usr/bin/perl
|
24
|
|
|
|
|
|
|
|
25
|
|
|
|
|
|
|
use strict;
|
26
|
|
|
|
|
|
|
use Plucene;
|
27
|
|
|
|
|
|
|
use Plucene::Index::Writer;
|
28
|
|
|
|
|
|
|
use Plucene::Plugin::Analyzer::MetaphoneAnalyzer;
|
29
|
|
|
|
|
|
|
use Plucene::Search::IndexSearcher;
|
30
|
|
|
|
|
|
|
use Plucene::QueryParser;
|
31
|
|
|
|
|
|
|
|
32
|
|
|
|
|
|
|
my $db = "plucene_index";
|
33
|
|
|
|
|
|
|
|
34
|
|
|
|
|
|
|
my $writer = Plucene::Index::Writer->new($db, Plucene::Plugin::Analyzer::MetaphoneAnalyzer->new(), 1);
|
35
|
|
|
|
|
|
|
|
36
|
|
|
|
|
|
|
$doc = Plucene::Document->new();
|
37
|
|
|
|
|
|
|
$doc->add(Plucene::Document::Field->Keyword(filename => 'test2.html'));
|
38
|
|
|
|
|
|
|
$doc->add(Plucene::Document::Field->Text(title => 'Another file title'));
|
39
|
|
|
|
|
|
|
$doc->add(Plucene::Document::Field->UnStored(content => 'Nothing HERE.'));
|
40
|
|
|
|
|
|
|
$writer->add_document($doc);
|
41
|
|
|
|
|
|
|
|
42
|
|
|
|
|
|
|
$writer->optimize;
|
43
|
|
|
|
|
|
|
|
44
|
|
|
|
|
|
|
my $searcher = Plucene::Search::IndexSearcher->new($db);
|
45
|
|
|
|
|
|
|
my $parser = Plucene::QueryParser->new({
|
46
|
|
|
|
|
|
|
analyzer => Plucene::Plugin::Analyzer::MetaphoneAnalyzer->new(),
|
47
|
|
|
|
|
|
|
default => 'content',
|
48
|
|
|
|
|
|
|
});
|
49
|
|
|
|
|
|
|
my $parsed = $parser->parse("nothink");
|
50
|
|
|
|
|
|
|
|
51
|
|
|
|
|
|
|
my @docs;
|
52
|
|
|
|
|
|
|
my $hc = Plucene::Search::HitCollector->new(
|
53
|
|
|
|
|
|
|
collect => sub {
|
54
|
|
|
|
|
|
|
my ($self, $doc, $score) = @_;
|
55
|
|
|
|
|
|
|
my $res = eval { $searcher->doc($doc); };
|
56
|
|
|
|
|
|
|
push @docs, $res if res;
|
57
|
|
|
|
|
|
|
},
|
58
|
|
|
|
|
|
|
);
|
59
|
|
|
|
|
|
|
|
60
|
|
|
|
|
|
|
$searcher->search_hc($parsed, $hc);
|
61
|
|
|
|
|
|
|
my @results = map {{
|
62
|
|
|
|
|
|
|
filename => $_->get('filename')->string,
|
63
|
|
|
|
|
|
|
title => $_->get('title')->string,
|
64
|
|
|
|
|
|
|
}} @docs;
|
65
|
|
|
|
|
|
|
|
66
|
|
|
|
|
|
|
print "Results:\n";
|
67
|
|
|
|
|
|
|
foreach my $result (@results) {
|
68
|
|
|
|
|
|
|
print "\t$result->{title} ($result->{filename})\n";
|
69
|
|
|
|
|
|
|
}
|
70
|
|
|
|
|
|
|
|
71
|
|
|
|
|
|
|
=cut
|
72
|
|
|
|
|
|
|
|
73
|
|
|
|
|
|
|
sub tokenstream {
|
74
|
1
|
|
|
1
|
1
|
234
|
my $self = shift;
|
75
|
|
|
|
|
|
|
|
76
|
1
|
|
|
|
|
24
|
return Plucene::Plugin::Analyzer::MetaphoneFilter->new({
|
77
|
|
|
|
|
|
|
input => Plucene::Analysis::Standard::StandardTokenizer->new(@_)
|
78
|
|
|
|
|
|
|
});
|
79
|
|
|
|
|
|
|
}
|
80
|
|
|
|
|
|
|
|
81
|
|
|
|
|
|
|
=head1 AUTHOR
|
82
|
|
|
|
|
|
|
|
83
|
|
|
|
|
|
|
Alan Schwartz C
|
84
|
|
|
|
|
|
|
|
85
|
|
|
|
|
|
|
=head1 LICENSE
|
86
|
|
|
|
|
|
|
|
87
|
|
|
|
|
|
|
You may distribute this code under the same terms as Plucene himself.
|