| 1 |
package WebPAC::Output::KinoSearch; |
| 2 |
|
| 3 |
use warnings; |
| 4 |
use strict; |
| 5 |
|
| 6 |
use base qw/WebPAC::Common WebPAC::Output Class::Accessor/; |
| 7 |
__PACKAGE__->mk_accessors(qw( |
| 8 |
path |
| 9 |
database |
| 10 |
encoding |
| 11 |
clean |
| 12 |
|
| 13 |
index |
| 14 |
)); |
| 15 |
|
| 16 |
use KinoSearch::Simple; |
| 17 |
use File::Path; |
| 18 |
use Encode qw/from_to/; |
| 19 |
use Data::Dump qw/dump/; |
| 20 |
use Storable; |
| 21 |
|
| 22 |
=head1 NAME |
| 23 |
|
| 24 |
WebPAC::Output::KinoSearch - Create KinoSearch full text index |
| 25 |
|
| 26 |
=head1 VERSION |
| 27 |
|
| 28 |
Version 0.04 |
| 29 |
|
| 30 |
=cut |
| 31 |
|
| 32 |
our $VERSION = '0.04'; |
| 33 |
|
| 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 |
my $est = new WebPAC::Output::KinoSearch({ |
| 46 |
path => '/path/to/invindex', |
| 47 |
database => 'demo', |
| 48 |
encoding => 'iso-8859-2', |
| 49 |
clean => 1, |
| 50 |
}); |
| 51 |
|
| 52 |
Options are: |
| 53 |
|
| 54 |
=over 4 |
| 55 |
|
| 56 |
=item path |
| 57 |
|
| 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 |
=cut |
| 73 |
|
| 74 |
sub init { |
| 75 |
my $self = shift; |
| 76 |
|
| 77 |
my $log = $self->_get_logger; |
| 78 |
|
| 79 |
#$log->debug("self: ", sub { dump($self) }); |
| 80 |
|
| 81 |
foreach my $p (qw/path database/) { |
| 82 |
$log->logdie("need $p") unless ($self->$p); |
| 83 |
} |
| 84 |
|
| 85 |
# $log->logdie("fields is not ARRAY") unless (ref($self->{fields}) eq 'ARRAY'); |
| 86 |
|
| 87 |
$self->encoding( 'ISO-8859-2' ) unless $self->encoding; |
| 88 |
|
| 89 |
if ( ! -e $self->path ) { |
| 90 |
mkpath $self->path || $log->logdie("can't create ", $self->path,": $!"); |
| 91 |
$log->info("created ", $self->path); |
| 92 |
} |
| 93 |
|
| 94 |
my $path = $self->path . '/' . $self->database; |
| 95 |
|
| 96 |
$log->info("using index $path with encoding ", $self->encoding); |
| 97 |
|
| 98 |
my $index = KinoSearch::Simple->new( |
| 99 |
path => $path, |
| 100 |
language => 'en', |
| 101 |
); |
| 102 |
|
| 103 |
$log->logdie("can't open $path: $!") unless $index; |
| 104 |
|
| 105 |
$self->index( $index ); |
| 106 |
|
| 107 |
} |
| 108 |
|
| 109 |
|
| 110 |
=head2 add |
| 111 |
|
| 112 |
Adds one entry |
| 113 |
|
| 114 |
$est->add( 42, $ds ); |
| 115 |
|
| 116 |
=cut |
| 117 |
|
| 118 |
sub add { |
| 119 |
my $self = shift; |
| 120 |
|
| 121 |
my ( $id, $ds ) = @_; |
| 122 |
|
| 123 |
my $log = $self->_get_logger; |
| 124 |
$log->logdie("need id") unless defined $id; |
| 125 |
$log->logdie("need ds") unless $ds; |
| 126 |
|
| 127 |
$log->debug("id: $id ds = ",dump($ds)); |
| 128 |
|
| 129 |
my $hash = $self->ds_to_hash( $ds, 'search' ) || return; |
| 130 |
|
| 131 |
warn "add( $id, ",dump($ds)," ) => ", dump( $hash ); |
| 132 |
|
| 133 |
$self->index->add_doc( $hash ); |
| 134 |
|
| 135 |
return 1; |
| 136 |
} |
| 137 |
|
| 138 |
=head2 finish |
| 139 |
|
| 140 |
Close index |
| 141 |
|
| 142 |
$index->finish; |
| 143 |
|
| 144 |
=cut |
| 145 |
|
| 146 |
sub finish { |
| 147 |
my $self = shift; |
| 148 |
|
| 149 |
my $log = $self->_get_logger(); |
| 150 |
|
| 151 |
$log->info("dummy finish"); |
| 152 |
|
| 153 |
} |
| 154 |
|
| 155 |
=head2 convert |
| 156 |
|
| 157 |
my $utf8_string = $self->convert('string in codepage'); |
| 158 |
|
| 159 |
=cut |
| 160 |
|
| 161 |
sub convert { |
| 162 |
my $self = shift; |
| 163 |
|
| 164 |
my $text = shift || return; |
| 165 |
from_to($text, $self->{encoding}, 'UTF-8'); |
| 166 |
return $text; |
| 167 |
} |
| 168 |
|
| 169 |
=head1 AUTHOR |
| 170 |
|
| 171 |
Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >> |
| 172 |
|
| 173 |
=head1 COPYRIGHT & LICENSE |
| 174 |
|
| 175 |
Copyright 2005-2007 Dobrica Pavlinusic, All Rights Reserved. |
| 176 |
|
| 177 |
This program is free software; you can redistribute it and/or modify it |
| 178 |
under the same terms as Perl itself. |
| 179 |
|
| 180 |
=cut |
| 181 |
|
| 182 |
1; # End of WebPAC::Output::Estraier |
Parent Directory
| 
Revision Log