1 |
dpavlin |
431 |
package WebPAC::Output::KinoSearch; |
2 |
|
|
|
3 |
|
|
use warnings; |
4 |
|
|
use strict; |
5 |
|
|
|
6 |
dpavlin |
914 |
use base qw/WebPAC::Common WebPAC::Output Class::Accessor/; |
7 |
|
|
__PACKAGE__->mk_accessors(qw( |
8 |
|
|
path |
9 |
|
|
database |
10 |
|
|
encoding |
11 |
|
|
clean |
12 |
dpavlin |
431 |
|
13 |
dpavlin |
914 |
index |
14 |
|
|
)); |
15 |
|
|
|
16 |
|
|
use KinoSearch::Simple; |
17 |
|
|
use File::Path; |
18 |
dpavlin |
431 |
use Encode qw/from_to/; |
19 |
dpavlin |
887 |
use Data::Dump qw/dump/; |
20 |
dpavlin |
536 |
use Storable; |
21 |
dpavlin |
431 |
|
22 |
|
|
=head1 NAME |
23 |
|
|
|
24 |
|
|
WebPAC::Output::KinoSearch - Create KinoSearch full text index |
25 |
|
|
|
26 |
|
|
=head1 VERSION |
27 |
|
|
|
28 |
dpavlin |
919 |
Version 0.05 |
29 |
dpavlin |
431 |
|
30 |
|
|
=cut |
31 |
|
|
|
32 |
dpavlin |
919 |
our $VERSION = '0.05'; |
33 |
dpavlin |
431 |
|
34 |
|
|
=head1 SYNOPSIS |
35 |
|
|
|
36 |
|
|
Create full text index using KinoSearch index from data with |
37 |
|
|
type C<search>. |
38 |
|
|
|
39 |
|
|
=head1 FUNCTIONS |
40 |
|
|
|
41 |
|
|
=head2 new |
42 |
|
|
|
43 |
|
|
Open KinoSearch index |
44 |
|
|
|
45 |
dpavlin |
917 |
my $out = new WebPAC::Output::KinoSearch({ |
46 |
dpavlin |
914 |
path => '/path/to/invindex', |
47 |
dpavlin |
431 |
database => 'demo', |
48 |
|
|
encoding => 'iso-8859-2', |
49 |
|
|
clean => 1, |
50 |
dpavlin |
914 |
}); |
51 |
dpavlin |
431 |
|
52 |
|
|
Options are: |
53 |
|
|
|
54 |
|
|
=over 4 |
55 |
|
|
|
56 |
dpavlin |
914 |
=item path |
57 |
dpavlin |
431 |
|
58 |
|
|
path to KinoSearch index to use |
59 |
|
|
|
60 |
|
|
=item database |
61 |
|
|
|
62 |
|
|
name of database from which data comes |
63 |
|
|
|
64 |
|
|
=item encoding |
65 |
|
|
|
66 |
|
|
character encoding of C<data_structure> if it's differenet than C<ISO-8859-2> |
67 |
|
|
(and it probably is). This encoding will be converted to C<UTF-8> for |
68 |
|
|
index. |
69 |
|
|
|
70 |
|
|
=back |
71 |
|
|
|
72 |
dpavlin |
917 |
=head2 init |
73 |
|
|
|
74 |
|
|
$out->init; |
75 |
|
|
|
76 |
dpavlin |
431 |
=cut |
77 |
|
|
|
78 |
dpavlin |
914 |
sub init { |
79 |
|
|
my $self = shift; |
80 |
dpavlin |
431 |
|
81 |
|
|
my $log = $self->_get_logger; |
82 |
|
|
|
83 |
dpavlin |
887 |
#$log->debug("self: ", sub { dump($self) }); |
84 |
dpavlin |
431 |
|
85 |
dpavlin |
914 |
foreach my $p (qw/path database/) { |
86 |
|
|
$log->logdie("need $p") unless ($self->$p); |
87 |
dpavlin |
431 |
} |
88 |
|
|
|
89 |
dpavlin |
914 |
# $log->logdie("fields is not ARRAY") unless (ref($self->{fields}) eq 'ARRAY'); |
90 |
dpavlin |
431 |
|
91 |
dpavlin |
914 |
$self->encoding( 'ISO-8859-2' ) unless $self->encoding; |
92 |
dpavlin |
431 |
|
93 |
dpavlin |
914 |
if ( ! -e $self->path ) { |
94 |
|
|
mkpath $self->path || $log->logdie("can't create ", $self->path,": $!"); |
95 |
|
|
$log->info("created ", $self->path); |
96 |
dpavlin |
919 |
} elsif ( $self->clean ) { |
97 |
|
|
$log->info("removing existing ", $self->path); |
98 |
|
|
rmtree $self->path || $log->logdie("can't remove ", $self->path,": $!"); |
99 |
|
|
mkpath $self->path || $log->logdie("can't create ", $self->path,": $!"); |
100 |
dpavlin |
914 |
} |
101 |
dpavlin |
610 |
|
102 |
dpavlin |
914 |
my $path = $self->path . '/' . $self->database; |
103 |
dpavlin |
431 |
|
104 |
dpavlin |
914 |
$log->info("using index $path with encoding ", $self->encoding); |
105 |
dpavlin |
609 |
|
106 |
dpavlin |
914 |
my $index = KinoSearch::Simple->new( |
107 |
|
|
path => $path, |
108 |
|
|
language => 'en', |
109 |
dpavlin |
431 |
); |
110 |
|
|
|
111 |
dpavlin |
914 |
$log->logdie("can't open $path: $!") unless $index; |
112 |
dpavlin |
536 |
|
113 |
dpavlin |
914 |
$self->index( $index ); |
114 |
dpavlin |
431 |
|
115 |
|
|
} |
116 |
|
|
|
117 |
|
|
|
118 |
|
|
=head2 add |
119 |
|
|
|
120 |
dpavlin |
914 |
Adds one entry |
121 |
dpavlin |
431 |
|
122 |
dpavlin |
917 |
$out->add( 42, $ds ); |
123 |
dpavlin |
431 |
|
124 |
|
|
=cut |
125 |
|
|
|
126 |
|
|
sub add { |
127 |
|
|
my $self = shift; |
128 |
|
|
|
129 |
dpavlin |
914 |
my ( $id, $ds ) = @_; |
130 |
dpavlin |
431 |
|
131 |
|
|
my $log = $self->_get_logger; |
132 |
dpavlin |
914 |
$log->logdie("need id") unless defined $id; |
133 |
|
|
$log->logdie("need ds") unless $ds; |
134 |
dpavlin |
431 |
|
135 |
dpavlin |
914 |
$log->debug("id: $id ds = ",dump($ds)); |
136 |
dpavlin |
431 |
|
137 |
dpavlin |
914 |
my $hash = $self->ds_to_hash( $ds, 'search' ) || return; |
138 |
dpavlin |
431 |
|
139 |
dpavlin |
919 |
$hash->{database} ||= $self->database; |
140 |
|
|
$hash->{id} ||= $id; |
141 |
dpavlin |
431 |
|
142 |
dpavlin |
919 |
$log->debug("add( $id, ", sub { dump($ds) }," ) => ", sub { dump( $hash ) }); |
143 |
|
|
|
144 |
dpavlin |
914 |
$self->index->add_doc( $hash ); |
145 |
dpavlin |
431 |
|
146 |
dpavlin |
922 |
$self->{count}++; |
147 |
|
|
|
148 |
dpavlin |
431 |
return 1; |
149 |
|
|
} |
150 |
|
|
|
151 |
dpavlin |
434 |
=head2 finish |
152 |
dpavlin |
431 |
|
153 |
dpavlin |
434 |
Close index |
154 |
|
|
|
155 |
dpavlin |
917 |
$out->finish; |
156 |
dpavlin |
434 |
|
157 |
|
|
=cut |
158 |
|
|
|
159 |
|
|
sub finish { |
160 |
|
|
my $self = shift; |
161 |
|
|
|
162 |
dpavlin |
536 |
my $log = $self->_get_logger(); |
163 |
|
|
|
164 |
dpavlin |
922 |
$log->info("indexed ", $self->{count}, " records"); |
165 |
dpavlin |
536 |
|
166 |
dpavlin |
434 |
} |
167 |
|
|
|
168 |
dpavlin |
431 |
=head1 AUTHOR |
169 |
|
|
|
170 |
|
|
Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >> |
171 |
|
|
|
172 |
|
|
=head1 COPYRIGHT & LICENSE |
173 |
|
|
|
174 |
dpavlin |
914 |
Copyright 2005-2007 Dobrica Pavlinusic, All Rights Reserved. |
175 |
dpavlin |
431 |
|
176 |
|
|
This program is free software; you can redistribute it and/or modify it |
177 |
|
|
under the same terms as Perl itself. |
178 |
|
|
|
179 |
|
|
=cut |
180 |
|
|
|
181 |
|
|
1; # End of WebPAC::Output::Estraier |