WebPAC/Output/EstraierNative.pm

package WebPAC::Output::EstraierNative;

use warnings;
use strict;

use base qw/WebPAC::Common/;

use Estraier;
use Encode qw/from_to/;
use Data::Dumper;
use LWP;
use URI::Escape;
use List::Util qw/first/;
use File::Path;

$Estraier::DEBUG = 1;


=head1 NAME

WebPAC::Output::EstraierNative - Create Hyper Estraier full text index using native bindings

=head1 VERSION

Version 0.2

=cut

our $VERSION = '0.2';

=head1 SYNOPSIS

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

=head1 FUNCTIONS

=head2 new

Connect to Hyper Estraier index using HTTP

 my $est = new WebPAC::Output::Estraier(
        path => 'casket/',
        database => 'demo',
        label => 'node label',
        encoding => 'iso-8859-2',
        clean => 1,
 );

Options are:

=over 4

=item path

full or relative path to Hyper Estraier database

=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
Hyper Estraier.

=back

Name of database will be used to form URI of documents in index.

=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/path database/) {
                $log->logdie("need $p") unless ($self->{$p});
        }

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

        $self->{label} ||= "WebPAC $self->{database}";

        my $path = 'casket';
        $path =~ s!/+$!!;

        $path .= '/' . $self->{database};

        $self->{path} = $path;

        $path .= '.tmp';
        if (-e $path) {
                rmtree($path) || $log->logdie("can't remove old temporary directory $path: $!");
        }
        mkpath($path) || $log->logdie("can't create new temporary directory $path: $!");

        my $db = new Database();
        unless($db->open($path, Database::DBWRITER | Database::DBCREAT)) {
                $log->logdie("can't open $path: ", $db->err_msg($db->error()) );
        }

        $self->{db} = $db;

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

        $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');
        $log->logconfess('need db in object') unless ($self->{'db'});

        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 = new Document();
        $doc->add_attr('@uri', $self->convert($uri) );

        # store type and database name
        $doc->add_attr('_database', $database );
        $doc->add_attr('_type', $type );

        $log->debug("ds = ", sub { Dumper($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) {

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

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

                next if (! $vals);

                $vals = join(" ") if (ref($vals) eq 'ARRAY');

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

                $doc->add_attr( $tag, $vals );
                $doc->add_hidden_text( $vals );
        }

        my $text = $args->{'text'};
        if ( $text ) {
                $text = $self->convert( $text ) or
                        $log->logdie("can't convert '$text' to UTF-8");
                $doc->add_text( $text );
        }

        $log->debug("adding ", sub { $doc->dump_draft } );
        $self->{'db'}->put_doc($doc, Database::PDCLEAN) || $log->warn("can't add document $uri with draft " . $doc->dump_draft . " to node " . $self->{path} . " status: " . $self->{db}->status());

        return 1;
}

=head2 add_link

  $est->add_link(
        from => 'ps',
        to => 'webpac2',
        credit => 10000,
  );

=cut

sub add_link {
        my $self = shift;

        my $args = {@_};
        my $log = $self->_get_logger;

        $log->warn("add_link is not implemented");
        return;

        foreach my $p (qw/from to credit/) {
                $log->logdie("need $p") unless ($args->{$p});
        }

        my $node = first { $_->{name} eq $args->{to} } $self->master( action => 'nodelist' );

        if (! $node) {
                $log->warn("can't find node $args->{to}, skipping link creaton");
                return;
        }

        my $label = $node->{label};

        if (! $label) {
                $log->warn("can't find label for $args->{to}, skipping link creaton");
                return;
        }

        $log->debug("using label $label for $args->{to}");

        return $self->{db}->set_link(
                $self->{masterurl} . '/node/' . $args->{to},
                $label,
                $args->{credit},
        );
}


=head2 finish

Close index and rename of to final path

        $est->finish;

=cut

sub finish {
        my $self = shift;

        my $log = $self->_get_logger;
        $log->info("closing Hyper Estraier index make it current...");

        $self->{db}->close || $log->logdie("can't close index");

        my $path = $self->{path} || $log->logdie("no path?");

        if (-e $path) {
                $log->warn("removing old $path");
                rmtree($path) || $log->logdie("can't remove old temporary directory $path: $!");
        }

        rename $path . '.tmp', $path || $log->logdie("can't rename ${path}.tmp -> $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	627	package WebPAC::Output::EstraierNative;
2
3			use warnings;
4			use strict;
5
6			use base qw/WebPAC::Common/;
7
8			use Estraier;
9			use Encode qw/from_to/;
10			use Data::Dumper;
11			use LWP;
12			use URI::Escape;
13			use List::Util qw/first/;
14	dpavlin	673	use File::Path;
15	dpavlin	627
16			$Estraier::DEBUG = 1;
17
18
19			=head1 NAME
20
21			WebPAC::Output::EstraierNative - Create Hyper Estraier full text index using native bindings
22
23			=head1 VERSION
24
25	dpavlin	673	Version 0.2
26	dpavlin	627
27			=cut
28
29	dpavlin	673	our $VERSION = '0.2';
30	dpavlin	627
31			=head1 SYNOPSIS
32
33			Create full text index using Hyper Estraier index from data with
34			type C<search>.
35
36			=head1 FUNCTIONS
37
38			=head2 new
39
40			Connect to Hyper Estraier index using HTTP
41
42			my $est = new WebPAC::Output::Estraier(
43			path => 'casket/',
44			database => 'demo',
45			label => 'node label',
46			encoding => 'iso-8859-2',
47			clean => 1,
48			);
49
50			Options are:
51
52			=over 4
53
54			=item path
55
56			full or relative path to Hyper Estraier database
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			Hyper Estraier.
71
72			=back
73
74			Name of database will be used to form URI of documents in index.
75
76			=cut
77
78			sub new {
79			my $class = shift;
80			my $self = {@_};
81			bless($self, $class);
82
83			my $log = $self->_get_logger;
84
85			#$log->debug("self: ", sub { Dumper($self) });
86
87			foreach my $p (qw/path database/) {
88			$log->logdie("need $p") unless ($self->{$p});
89			}
90
91			$self->{encoding} \|\|= 'ISO-8859-2';
92
93			$self->{label} \|\|= "WebPAC $self->{database}";
94
95			my $path = 'casket';
96	dpavlin	673	$path =~ s!/+$!!;
97	dpavlin	627
98	dpavlin	674	$path .= '/' . $self->{database};
99	dpavlin	673
100	dpavlin	674	$self->{path} = $path;
101
102	dpavlin	673	$path .= '.tmp';
103			if (-e $path) {
104			rmtree($path) \|\| $log->logdie("can't remove old temporary directory $path: $!");
105			}
106			mkpath($path) \|\| $log->logdie("can't create new temporary directory $path: $!");
107
108	dpavlin	627	my $db = new Database();
109			unless($db->open($path, Database::DBWRITER \| Database::DBCREAT)) {
110			$log->logdie("can't open $path: ", $db->err_msg($db->error()) );
111			}
112
113			$self->{db} = $db;
114
115			$log->info("using ", $self->{clean} ? "new " : "", "index $self->{path} '$self->{label}' with encoding $self->{encoding}");
116
117			$self ? return $self : return undef;
118			}
119
120
121			=head2 add
122
123			Adds one entry to database.
124
125			$est->add(
126			id => 42,
127			ds => $ds,
128			type => 'display',
129			text => 'optional text from which snippet is created',
130			);
131
132			This function will create entries in index using following URI format:
133
134			C<file:///type/database%20name/000>
135
136			Each tag in C<data_structure> with specified C<type> will create one
137			attribute and corresponding hidden text (used for search).
138
139			=cut
140
141			sub add {
142			my $self = shift;
143
144			my $args = {@_};
145
146			my $log = $self->_get_logger;
147
148			my $database = $self->{'database'} \|\| $log->logconfess('no database in $self');
149			$log->logconfess('need db in object') unless ($self->{'db'});
150
151			foreach my $p (qw/id ds type/) {
152			$log->logdie("need $p") unless ($args->{$p});
153			}
154
155			my $type = $args->{'type'};
156			my $id = $args->{'id'};
157
158			my $uri = "file:///$type/$database/$id";
159			$log->debug("creating $uri");
160
161			my $doc = new Document();
162			$doc->add_attr('@uri', $self->convert($uri) );
163
164			# store type and database name
165			$doc->add_attr('_database', $database );
166			$doc->add_attr('_type', $type );
167
168			$log->debug("ds = ", sub { Dumper($args->{'ds'}) } );
169
170			# filter all tags which have type defined
171			my @tags = grep {
172			ref($args->{'ds'}->{$_}) eq 'HASH' && defined( $args->{'ds'}->{$_}->{$type} )
173			} keys %{ $args->{'ds'} };
174
175			$log->debug("tags = ", join(",", @tags));
176
177			return unless (@tags);
178
179			foreach my $tag (@tags) {
180
181	dpavlin	630	$log->debug("$tag :: $type == ",Dumper( $args->{'ds'}->{$tag}->{$type} ) );
182
183	dpavlin	627	my $vals = join(" ", @{ $args->{'ds'}->{$tag}->{$type} });
184
185			next if (! $vals);
186
187			$vals = join(" ") if (ref($vals) eq 'ARRAY');
188
189			$vals = $self->convert( $vals ) or
190			$log->logdie("can't convert '$vals' to UTF-8");
191
192			$doc->add_attr( $tag, $vals );
193			$doc->add_hidden_text( $vals );
194			}
195
196			my $text = $args->{'text'};
197			if ( $text ) {
198			$text = $self->convert( $text ) or
199			$log->logdie("can't convert '$text' to UTF-8");
200			$doc->add_text( $text );
201			}
202
203			$log->debug("adding ", sub { $doc->dump_draft } );
204			$self->{'db'}->put_doc($doc, Database::PDCLEAN) \|\| $log->warn("can't add document $uri with draft " . $doc->dump_draft . " to node " . $self->{path} . " status: " . $self->{db}->status());
205
206			return 1;
207			}
208
209			=head2 add_link
210
211			$est->add_link(
212			from => 'ps',
213			to => 'webpac2',
214			credit => 10000,
215			);
216
217			=cut
218
219			sub add_link {
220			my $self = shift;
221
222			my $args = {@_};
223			my $log = $self->_get_logger;
224
225			$log->warn("add_link is not implemented");
226			return;
227
228			foreach my $p (qw/from to credit/) {
229			$log->logdie("need $p") unless ($args->{$p});
230			}
231
232			my $node = first { $_->{name} eq $args->{to} } $self->master( action => 'nodelist' );
233
234			if (! $node) {
235			$log->warn("can't find node $args->{to}, skipping link creaton");
236			return;
237			}
238
239			my $label = $node->{label};
240
241			if (! $label) {
242			$log->warn("can't find label for $args->{to}, skipping link creaton");
243			return;
244			}
245
246			$log->debug("using label $label for $args->{to}");
247
248			return $self->{db}->set_link(
249			$self->{masterurl} . '/node/' . $args->{to},
250			$label,
251			$args->{credit},
252			);
253			}
254
255
256	dpavlin	673	=head2 finish
257
258			Close index and rename of to final path
259
260			$est->finish;
261
262			=cut
263
264			sub finish {
265			my $self = shift;
266
267			my $log = $self->_get_logger;
268			$log->info("closing Hyper Estraier index make it current...");
269
270			$self->{db}->close \|\| $log->logdie("can't close index");
271
272	dpavlin	674	my $path = $self->{path} \|\| $log->logdie("no path?");
273	dpavlin	673
274			if (-e $path) {
275			$log->warn("removing old $path");
276			rmtree($path) \|\| $log->logdie("can't remove old temporary directory $path: $!");
277			}
278
279			rename $path . '.tmp', $path \|\| $log->logdie("can't rename ${path}.tmp -> $path: $!");
280
281			}
282
283
284	dpavlin	627	=head2 convert
285
286			my $utf8_string = $self->convert('string in codepage');
287
288			=cut
289
290			sub convert {
291			my $self = shift;
292
293			my $text = shift \|\| return;
294			from_to($text, $self->{encoding}, 'UTF-8');
295			return $text;
296			}
297
298			=head1 AUTHOR
299
300			Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >>
301
302			=head1 COPYRIGHT & LICENSE
303
304			Copyright 2005 Dobrica Pavlinusic, All Rights Reserved.
305
306			This program is free software; you can redistribute it and/or modify it
307			under the same terms as Perl itself.
308
309			=cut
310
311			1; # End of WebPAC::Output::Estraier