WebPAC/Output/Estraier.pm

package WebPAC::Output::Estraier;

use warnings;
use strict;

use base qw/WebPAC::Common/;

use HyperEstraier;
use Text::Iconv;
use Data::Dumper;
use LWP;
use URI::Escape;

=head1 NAME

WebPAC::Output::Estraier - Create Hyper Estraier full text index

=head1 VERSION

Version 0.03

=cut

our $VERSION = '0.03';

=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(
        masterurl => 'http://localhost:1978/',
        user => 'admin',
        passwd => 'admin',
        database => 'demo',
        encoding => 'iso-8859-2',
 );

Options are:

=over 4

=item masterurl

URI to C<estmaster> node

=item user

C<estmaster> user with sufficient rights

=item passwd

password for user

=item database

name of database from which data comes

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

        my $url = $self->{masterurl} . '/node/' . $self->{database};
        $self->{url} = $url;

        $log->info("opening Hyper Estraier index $self->{url}");

        my $nodes = $self->est_master( action => 'nodelist' );

        $log->debug("nodes found: $nodes");

        if ($nodes !~ m/^$self->{database}\t/sm) {
                $log->info("creating index $url");
                $self->est_master(
                        action => 'nodeadd',
                        name => $self->{database},
                        label => "WebPAC $self->{database}",
                ) || $log->logdie("can't create Hyper Estraier node $self->{database}");
        }

        $self->{'db'} = HyperEstraier::Node->new($self->{url});
        $self->{'db'}->set_auth($self->{'user'}, $self->{passwd});

        my $encoding = $self->{'encoding'} || 'ISO-8859-2';
        $log->info("using encoding $encoding");

        $self->{'iconv'} = new Text::Iconv($encoding, 'UTF-8') or
                $log->logdie("can't create conversion from $encoding to UTF-8");

        $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 = HyperEstraier::Document->new;
        $doc->add_attr('@uri', $self->{'iconv'}->convert($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} });

                $log->logconfess("no values for $tag/$type") unless ($vals);

                $vals = $self->{'iconv'}->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->{'iconv'}->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) || $log->logdie("can't add document $uri to node " . $self->{url} . " status: " . $self->{db}->status());

        return 1;
}

=head2 est_master

Issue administrative commands to C<estmaster> process and receive response
as array of lines

  my $nodelist = $self->est_master( action => nodelist );

=cut

my $estmaster_actions = {
        userdel => [ qw/name/ ],
        nodelist => [],
        nodeadd => [ qw/name label/ ],
        nodedel => [ qw/name/ ],
};

sub est_ua {
        my $self = shift;

        return $self->{_master_ua} if ($self->{_master_ua});

        $self->{_master_ua} = LWP::UserAgent->new( ) || sub {
                my $log = $self->_get_logger;
                $log->logdie("can't create LWP::UserAgent: $!");
        };

        $self->{_master_ua}->credentials('localhost:1978','Super User', $self->{user} => $self->{passwd});

        return $self->{_master_ua};
}

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

        $log->debug(Dumper($args));

        my $action = $args->{action} || $log->logconfess("no action specified");

        $log->logdie("action '$action' isn't supported") unless ($estmaster_actions->{$action});

        my $url = $self->{masterurl} . '/master?action=' . $action;

        foreach my $arg (@{ $estmaster_actions->{$action} }) {
                $log->logdie("missing parametar $arg for action $action") unless ($args->{$arg});
                $url .= '&' . $arg . '=' . uri_escape( $args->{$arg} );
        }

        $log->debug("calling $url");

        my $res = $self->est_ua()->get($url);

        if ($res->is_success) {
                #$log->debug( $res->content );
                return split(/\n/, $res->content) if wantarray;
                return $res->content;
        } else {
                $log->warn("unable to call $url: " . $res->status_line);
                return;
        }

}

=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	package WebPAC::Output::Estraier;
2
3	use warnings;
4	use strict;
5
6	use base qw/WebPAC::Common/;
7
8	use HyperEstraier;
9	use Text::Iconv;
10	use Data::Dumper;
11	use LWP;
12	use URI::Escape;
13
14	=head1 NAME
15
16	WebPAC::Output::Estraier - Create Hyper Estraier full text index
17
18	=head1 VERSION
19
20	Version 0.03
21
22	=cut
23
24	our $VERSION = '0.03';
25
26	=head1 SYNOPSIS
27
28	Create full text index using Hyper Estraier index from data with
29	type C<search>.
30
31	=head1 FUNCTIONS
32
33	=head2 new
34
35	Connect to Hyper Estraier index using HTTP
36
37	my $est = new WebPAC::Output::Estraier(
38	masterurl => 'http://localhost:1978/',
39	user => 'admin',
40	passwd => 'admin',
41	database => 'demo',
42	encoding => 'iso-8859-2',
43	);
44
45	Options are:
46
47	=over 4
48
49	=item masterurl
50
51	URI to C<estmaster> node
52
53	=item user
54
55	C<estmaster> user with sufficient rights
56
57	=item passwd
58
59	password for user
60
61	=item database
62
63	name of database from which data comes
64
65	=item encoding
66
67	character encoding of C<data_structure> if it's differenet than C<ISO-8859-2>
68	(and it probably is). This encoding will be converted to C<UTF-8> for
69	Hyper Estraier.
70
71	=back
72
73	Name of database will be used to form URI of documents in index.
74
75	=cut
76
77	sub new {
78	my $class = shift;
79	my $self = {@_};
80	bless($self, $class);
81
82	my $log = $self->_get_logger;
83
84	#$log->debug("self: ", sub { Dumper($self) });
85
86	foreach my $p (qw/masterurl user passwd database/) {
87	$log->logdie("need $p") unless ($self->{$p});
88	}
89
90	my $url = $self->{masterurl} . '/node/' . $self->{database};
91	$self->{url} = $url;
92
93	$log->info("opening Hyper Estraier index $self->{url}");
94
95	my $nodes = $self->est_master( action => 'nodelist' );
96
97	$log->debug("nodes found: $nodes");
98
99	if ($nodes !~ m/^$self->{database}\t/sm) {
100	$log->info("creating index $url");
101	$self->est_master(
102	action => 'nodeadd',
103	name => $self->{database},
104	label => "WebPAC $self->{database}",
105	) \|\| $log->logdie("can't create Hyper Estraier node $self->{database}");
106	}
107
108	$self->{'db'} = HyperEstraier::Node->new($self->{url});
109	$self->{'db'}->set_auth($self->{'user'}, $self->{passwd});
110
111	my $encoding = $self->{'encoding'} \|\| 'ISO-8859-2';
112	$log->info("using encoding $encoding");
113
114	$self->{'iconv'} = new Text::Iconv($encoding, 'UTF-8') or
115	$log->logdie("can't create conversion from $encoding to UTF-8");
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 = HyperEstraier::Document->new;
162	$doc->add_attr('@uri', $self->{'iconv'}->convert($uri) );
163
164	$log->debug("ds = ", sub { Dumper($args->{'ds'}) } );
165
166	# filter all tags which have type defined
167	my @tags = grep {
168	ref($args->{'ds'}->{$_}) eq 'HASH' && defined( $args->{'ds'}->{$_}->{$type} )
169	} keys %{ $args->{'ds'} };
170
171	$log->debug("tags = ", join(",", @tags));
172
173	return unless (@tags);
174
175	foreach my $tag (@tags) {
176
177	my $vals = join(" ", @{ $args->{'ds'}->{$tag}->{$type} });
178
179	$log->logconfess("no values for $tag/$type") unless ($vals);
180
181	$vals = $self->{'iconv'}->convert( $vals ) or
182	$log->logdie("can't convert '$vals' to UTF-8");
183
184	$doc->add_attr( $tag, $vals );
185	$doc->add_hidden_text( $vals );
186	}
187
188	my $text = $args->{'text'};
189	if ( $text ) {
190	$text = $self->{'iconv'}->convert( $text ) or
191	$log->logdie("can't convert '$text' to UTF-8");
192	$doc->add_text( $text );
193	}
194
195	$log->debug("adding ", sub { $doc->dump_draft } );
196	$self->{'db'}->put_doc($doc) \|\| $log->logdie("can't add document $uri to node " . $self->{url} . " status: " . $self->{db}->status());
197
198	return 1;
199	}
200
201	=head2 est_master
202
203	Issue administrative commands to C<estmaster> process and receive response
204	as array of lines
205
206	my $nodelist = $self->est_master( action => nodelist );
207
208	=cut
209
210	my $estmaster_actions = {
211	userdel => [ qw/name/ ],
212	nodelist => [],
213	nodeadd => [ qw/name label/ ],
214	nodedel => [ qw/name/ ],
215	};
216
217	sub est_ua {
218	my $self = shift;
219
220	return $self->{_master_ua} if ($self->{_master_ua});
221
222	$self->{_master_ua} = LWP::UserAgent->new( ) \|\| sub {
223	my $log = $self->_get_logger;
224	$log->logdie("can't create LWP::UserAgent: $!");
225	};
226
227	$self->{_master_ua}->credentials('localhost:1978','Super User', $self->{user} => $self->{passwd});
228
229	return $self->{_master_ua};
230	}
231
232	sub est_master {
233	my $self = shift;
234	my $args = {@_};
235	my $log = $self->_get_logger;
236
237	$log->debug(Dumper($args));
238
239	my $action = $args->{action} \|\| $log->logconfess("no action specified");
240
241	$log->logdie("action '$action' isn't supported") unless ($estmaster_actions->{$action});
242
243	my $url = $self->{masterurl} . '/master?action=' . $action;
244
245	foreach my $arg (@{ $estmaster_actions->{$action} }) {
246	$log->logdie("missing parametar $arg for action $action") unless ($args->{$arg});
247	$url .= '&' . $arg . '=' . uri_escape( $args->{$arg} );
248	}
249
250	$log->debug("calling $url");
251
252	my $res = $self->est_ua()->get($url);
253
254	if ($res->is_success) {
255	#$log->debug( $res->content );
256	return split(/\n/, $res->content) if wantarray;
257	return $res->content;
258	} else {
259	$log->warn("unable to call $url: " . $res->status_line);
260	return;
261	}
262
263	}
264
265	=head1 AUTHOR
266
267	Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >>
268
269	=head1 COPYRIGHT & LICENSE
270
271	Copyright 2005 Dobrica Pavlinusic, All Rights Reserved.
272
273	This program is free software; you can redistribute it and/or modify it
274	under the same terms as Perl itself.
275
276	=cut
277
278	1; # End of WebPAC::Output::Estraier