WebPAC/Output/KinoSearch.pm

package WebPAC::Output::KinoSearch;

use warnings;
use strict;

use base qw/WebPAC::Common/;

use KinoSearch::InvIndexer;
use KinoSearch::Analysis::PolyAnalyzer;
use Encode qw/from_to/;
use Data::Dumper;
use Storable;

=head1 NAME

WebPAC::Output::KinoSearch - Create KinoSearch full text index

=head1 VERSION

Version 0.03

=cut

our $VERSION = '0.03';

=head1 SYNOPSIS

Create full text index using KinoSearch index from data with
type C<search>.

=head1 FUNCTIONS

=head2 new

Open KinoSearch index

 my $est = new WebPAC::Output::KinoSearch(
        index_path => '/path/to/invindex',
        fields => qw/name of all filelds used/,
        database => 'demo',
        label => 'node label',
        encoding => 'iso-8859-2',
        clean => 1,
 );

Options are:

=over 4

=item index_path

path to KinoSearch index to use

=item fields

name of all fields used in this index

=item database

name of database from which data comes

=item label

label for node (optional)

=item encoding

character encoding of C<data_structure> if it's differenet than C<ISO-8859-2>
(and it probably is). This encoding will be converted to C<UTF-8> for
index.

=back

=cut

sub new {
        my $class = shift;
        my $self = {@_};
        bless($self, $class);

        my $log = $self->_get_logger;

        #$log->debug("self: ", sub { Dumper($self) });

        foreach my $p (qw/index_path fields database/) {
                $log->logdie("need $p") unless ($self->{$p});
        }

        $log->logdie("fields is not ARRAY") unless (ref($self->{fields}) eq 'ARRAY');

        $self->{encoding} ||= 'ISO-8859-2';

        $self->{clean} = 1 if (! -e $self->{index_path} . '/segments');

        $log->info("using", $self->{clean} ? ' new' : '', " index $self->{index_path} with encoding $self->{encoding}");

        my $analyzer = KinoSearch::Analysis::PolyAnalyzer->new( language => 'en' );

        $self->{invindex} = KinoSearch::InvIndexer->new(
                invindex => $self->{index_path},
                create   => $self->{clean},
                analyzer => $analyzer,
        );

        my $fields_path = $self->{index_path} . '/fields.storable';
        $fields_path =~ s#//#/#g;
        if (-e $fields_path) {
                $self->{fields} = retrieve($fields_path) ||
                        $log->warn("can't open $fields_path: $!");
        } else {
                $log->error("This will be dummy run since no fields statistics are found!");
                $log->error("You will have to re-run indexing to get search results!");
                $self->{dummy_run} = 1;
        }
        $self->{fields_path} = $fields_path;

        foreach my $f (@{ $self->{fields} }) {
                $self->{invindex}->spec_field( 
                        name  => $f,
#                       boost => 10,
                        stored => 1,
                        indexed => 1,
                        vectorized => 0,
                );
        }

        $self ? return $self : return undef;
}


=head2 add

Adds one entry to database.

  $est->add(
        id => 42,
        ds => $ds,
        type => 'display',
        text => 'optional text from which snippet is created',
  );

This function will create  entries in index using following URI format:

  C<file:///type/database%20name/000>

Each tag in C<data_structure> with specified C<type> will create one
attribute and corresponding hidden text (used for search).

=cut

sub add {
        my $self = shift;

        my $args = {@_};

        my $log = $self->_get_logger;

        my $database = $self->{'database'} || $log->logconfess('no database in $self');
        $log->logconfess('need invindex in object') unless ($self->{'invindex'});

        foreach my $p (qw/id ds type/) {
                $log->logdie("need $p") unless ($args->{$p});
        }

        my $type = $args->{'type'};
        my $id = $args->{'id'};

        my $uri = "file:///$type/$database/$id";
        $log->debug("creating $uri");

        my $doc = $self->{invindex}->new_doc( $uri ) || $log->logdie("can't create new_doc( $uri )");

        sub add_value($$$$$) {
                my ($self,$log,$doc,$n,$v) = @_;
                return unless ($v);

                $self->{value_usage}->{$n}++;
                return if ($self->{dummy_run});

                eval { $doc->set_value($n, $self->convert($v) ) };
                $log->warn("can't insert: $n = $v") if ($@);
        }

        add_value($self,$log,$doc, 'uri', $uri);

        $log->debug("ds = ", sub { Dumper($args->{'ds'}) } );

        # filter all tags which have type defined
        my @tags = grep {
                ref($args->{'ds'}->{$_}) eq 'HASH' && defined( $args->{'ds'}->{$_}->{$type} )
        } keys %{ $args->{'ds'} };

        $log->debug("tags = ", join(",", @tags));

        return unless (@tags);

        foreach my $tag (@tags) {

                my $vals = join(" ", @{ $args->{'ds'}->{$tag}->{$type} });

                next if (! $vals);

                $vals = $self->convert( $vals ) or
                        $log->logdie("can't convert '$vals' to UTF-8");

                add_value($self, $log, $doc, $tag, $vals );
        }

        if (my $text = $args->{'text'}) {
                add_value($self, $log, $doc, 'bodytext', $text );
        }

        #$log->debug("adding ", sub { $doc->dump_draft } );
        $self->{invindex}->add_doc($doc) || $log->warn("can't add document $uri");

        return 1;
}

=head2 finish

Close index

 $index->finish;

=cut

sub finish {
        my $self = shift;

        my $log = $self->_get_logger();

        $log->info("finish index writing to disk");
        $self->{invindex}->finish;

        $log->info("writing value usage file");

        # add fields from last run
        map { $self->{value_usage}->{$_}++ } @{ $self->{fields} };

        my @fields = keys %{ $self->{value_usage} };
        store \@fields, $self->{fields_path} ||
                $log->warn("can't write $self->{fields_path}: $!");

}

=head2 convert

 my $utf8_string = $self->convert('string in codepage');

=cut

sub convert {
        my $self = shift;

        my $text = shift || return;
        from_to($text, $self->{encoding}, 'UTF-8');
        return $text;
}

=head1 AUTHOR

Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >>

=head1 COPYRIGHT & LICENSE

Copyright 2005 Dobrica Pavlinusic, All Rights Reserved.

This program is free software; you can redistribute it and/or modify it
under the same terms as Perl itself.

=cut

1; # End of WebPAC::Output::Estraier
1	dpavlin	431	package WebPAC::Output::KinoSearch;
2
3			use warnings;
4			use strict;
5
6			use base qw/WebPAC::Common/;
7
8			use KinoSearch::InvIndexer;
9			use KinoSearch::Analysis::PolyAnalyzer;
10			use Encode qw/from_to/;
11			use Data::Dumper;
12	dpavlin	536	use Storable;
13	dpavlin	431
14			=head1 NAME
15
16			WebPAC::Output::KinoSearch - Create KinoSearch full text index
17
18			=head1 VERSION
19
20	dpavlin	609	Version 0.03
21	dpavlin	431
22			=cut
23
24	dpavlin	609	our $VERSION = '0.03';
25	dpavlin	431
26			=head1 SYNOPSIS
27
28			Create full text index using KinoSearch index from data with
29			type C<search>.
30
31			=head1 FUNCTIONS
32
33			=head2 new
34
35			Open KinoSearch index
36
37			my $est = new WebPAC::Output::KinoSearch(
38			index_path => '/path/to/invindex',
39			fields => qw/name of all filelds used/,
40			database => 'demo',
41			label => 'node label',
42			encoding => 'iso-8859-2',
43			clean => 1,
44			);
45
46			Options are:
47
48			=over 4
49
50			=item index_path
51
52			path to KinoSearch index to use
53
54			=item fields
55
56			name of all fields used in this index
57
58			=item database
59
60			name of database from which data comes
61
62			=item label
63
64			label for node (optional)
65
66			=item encoding
67
68			character encoding of C<data_structure> if it's differenet than C<ISO-8859-2>
69			(and it probably is). This encoding will be converted to C<UTF-8> for
70			index.
71
72			=back
73
74			=cut
75
76			sub new {
77			my $class = shift;
78			my $self = {@_};
79			bless($self, $class);
80
81			my $log = $self->_get_logger;
82
83			#$log->debug("self: ", sub { Dumper($self) });
84
85			foreach my $p (qw/index_path fields database/) {
86			$log->logdie("need $p") unless ($self->{$p});
87			}
88
89			$log->logdie("fields is not ARRAY") unless (ref($self->{fields}) eq 'ARRAY');
90
91			$self->{encoding} \|\|= 'ISO-8859-2';
92
93	dpavlin	609	$self->{clean} = 1 if (! -e $self->{index_path} . '/segments');
94	dpavlin	431
95	dpavlin	609	$log->info("using", $self->{clean} ? ' new' : '', " index $self->{index_path} with encoding $self->{encoding}");
96
97	dpavlin	431	my $analyzer = KinoSearch::Analysis::PolyAnalyzer->new( language => 'en' );
98
99			$self->{invindex} = KinoSearch::InvIndexer->new(
100			invindex => $self->{index_path},
101			create => $self->{clean},
102			analyzer => $analyzer,
103			);
104
105	dpavlin	536	my $fields_path = $self->{index_path} . '/fields.storable';
106			$fields_path =~ s#//#/#g;
107			if (-e $fields_path) {
108			$self->{fields} = retrieve($fields_path) \|\|
109			$log->warn("can't open $fields_path: $!");
110			} else {
111			$log->error("This will be dummy run since no fields statistics are found!");
112			$log->error("You will have to re-run indexing to get search results!");
113			$self->{dummy_run} = 1;
114			}
115			$self->{fields_path} = $fields_path;
116
117	dpavlin	431	foreach my $f (@{ $self->{fields} }) {
118			$self->{invindex}->spec_field(
119			name => $f,
120			# boost => 10,
121			stored => 1,
122			indexed => 1,
123			vectorized => 0,
124			);
125			}
126
127			$self ? return $self : return undef;
128			}
129
130
131			=head2 add
132
133			Adds one entry to database.
134
135			$est->add(
136			id => 42,
137			ds => $ds,
138			type => 'display',
139			text => 'optional text from which snippet is created',
140			);
141
142			This function will create entries in index using following URI format:
143
144			C<file:///type/database%20name/000>
145
146			Each tag in C<data_structure> with specified C<type> will create one
147			attribute and corresponding hidden text (used for search).
148
149			=cut
150
151			sub add {
152			my $self = shift;
153
154			my $args = {@_};
155
156			my $log = $self->_get_logger;
157
158			my $database = $self->{'database'} \|\| $log->logconfess('no database in $self');
159			$log->logconfess('need invindex in object') unless ($self->{'invindex'});
160
161			foreach my $p (qw/id ds type/) {
162			$log->logdie("need $p") unless ($args->{$p});
163			}
164
165			my $type = $args->{'type'};
166			my $id = $args->{'id'};
167
168			my $uri = "file:///$type/$database/$id";
169			$log->debug("creating $uri");
170
171			my $doc = $self->{invindex}->new_doc( $uri ) \|\| $log->logdie("can't create new_doc( $uri )");
172
173	dpavlin	437	sub add_value($$$$$) {
174			my ($self,$log,$doc,$n,$v) = @_;
175			return unless ($v);
176	dpavlin	536
177			$self->{value_usage}->{$n}++;
178			return if ($self->{dummy_run});
179
180	dpavlin	432	eval { $doc->set_value($n, $self->convert($v) ) };
181			$log->warn("can't insert: $n = $v") if ($@);
182			}
183
184	dpavlin	437	add_value($self,$log,$doc, 'uri', $uri);
185	dpavlin	432
186	dpavlin	431	$log->debug("ds = ", sub { Dumper($args->{'ds'}) } );
187
188			# filter all tags which have type defined
189			my @tags = grep {
190			ref($args->{'ds'}->{$_}) eq 'HASH' && defined( $args->{'ds'}->{$_}->{$type} )
191			} keys %{ $args->{'ds'} };
192
193			$log->debug("tags = ", join(",", @tags));
194
195			return unless (@tags);
196
197			foreach my $tag (@tags) {
198
199			my $vals = join(" ", @{ $args->{'ds'}->{$tag}->{$type} });
200
201			next if (! $vals);
202
203			$vals = $self->convert( $vals ) or
204			$log->logdie("can't convert '$vals' to UTF-8");
205
206	dpavlin	437	add_value($self, $log, $doc, $tag, $vals );
207	dpavlin	431	}
208
209	dpavlin	432	if (my $text = $args->{'text'}) {
210	dpavlin	437	add_value($self, $log, $doc, 'bodytext', $text );
211	dpavlin	431	}
212
213			#$log->debug("adding ", sub { $doc->dump_draft } );
214			$self->{invindex}->add_doc($doc) \|\| $log->warn("can't add document $uri");
215
216			return 1;
217			}
218
219	dpavlin	434	=head2 finish
220	dpavlin	431
221	dpavlin	434	Close index
222
223			$index->finish;
224
225			=cut
226
227			sub finish {
228			my $self = shift;
229
230	dpavlin	536	my $log = $self->_get_logger();
231
232			$log->info("finish index writing to disk");
233	dpavlin	434	$self->{invindex}->finish;
234	dpavlin	536
235			$log->info("writing value usage file");
236
237			# add fields from last run
238			map { $self->{value_usage}->{$_}++ } @{ $self->{fields} };
239
240			my @fields = keys %{ $self->{value_usage} };
241			store \@fields, $self->{fields_path} \|\|
242			$log->warn("can't write $self->{fields_path}: $!");
243
244	dpavlin	434	}
245
246	dpavlin	431	=head2 convert
247
248			my $utf8_string = $self->convert('string in codepage');
249
250			=cut
251
252			sub convert {
253			my $self = shift;
254
255			my $text = shift \|\| return;
256			from_to($text, $self->{encoding}, 'UTF-8');
257			return $text;
258			}
259
260			=head1 AUTHOR
261
262			Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >>
263
264			=head1 COPYRIGHT & LICENSE
265
266			Copyright 2005 Dobrica Pavlinusic, All Rights Reserved.
267
268			This program is free software; you can redistribute it and/or modify it
269			under the same terms as Perl itself.
270
271			=cut
272
273			1; # End of WebPAC::Output::Estraier