1 package WebPAC::Output::Estraier;
6 use base qw/WebPAC::Common/;
16 WebPAC::Output::Estraier - Create Hyper Estraier full text index
24 our $VERSION = '0.02';
28 Create full text index using Hyper Estraier index from data with
35 Connect to Hyper Estraier index using HTTP
37 my $est = new WebPAC::Output::Estraier(
38 masterurl => 'http://localhost:1978/',
42 encoding => 'iso-8859-2',
51 URI to C<estmaster> node
55 C<estmaster> user with sufficient rights
63 name of database from which data comes
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
73 Name of database will be used to form URI of documents in index.
82 my $log = $self->_get_logger;
84 $log->debug("self: ", sub { Dumper($self) });
86 foreach my $p (qw/masterurl user passwd database/) {
87 $log->logdie("need $p") unless ($self->{$p});
90 my $url = $self->{masterurl} . '/node/' . $self->{database};
94 $log->info("opening Hyper Estraier index $self->{url}");
96 my @nodes = $self->est_master( action => 'nodelist' );
98 if (! grep(/$self->{database}/, @nodes)) {
99 $log->info("creating index $url");
102 name => $self->{database},
103 label => "WebPAC $self->{database}",
104 ) || $log->logdie("can't create Hyper Estraier node $self->{database}");
107 $self->{'db'} = HyperEstraier::Node->new($self->{url});
108 $self->{'db'}->set_auth($self->{'user'}, $self->{passwd});
110 my $encoding = $self->{'encoding'} || 'ISO-8859-2';
111 $log->info("using encoding $encoding");
113 $self->{'iconv'} = new Text::Iconv($encoding, 'UTF-8') or
114 $log->logdie("can't create conversion from $encoding to UTF-8");
116 $self ? return $self : return undef;
122 Adds one entry to database.
128 url_prefix => 'database name',
129 text => 'optional text from which snippet is created',
132 This function will create entries in index using following URI format:
134 C<file:///database%20name/000>
136 Each tag in C<data_structure> with specified C<type> will create one
137 attribute and corresponding hidden text (used for search).
146 my $log = $self->_get_logger;
148 my $database = $self->{'database'} || $log->logconfess('no database in $self');
149 $log->logconfess('need db in object') unless ($self->{'db'});
151 foreach my $p (qw/id ds type/) {
152 $log->logdie("need $p") unless ($args->{$p});
155 my $type = $args->{'type'};
156 my $mfn = $args->{'id'};
158 my $uri = "file:///$type/$database/$mfn";
159 $log->debug("creating $uri");
161 my $doc = HyperEstraier::Document->new;
162 $doc->add_attr('@uri', $self->{'iconv'}->convert($uri) );
164 $log->debug("ds = ", sub { Dumper($args->{'ds'}) } );
166 # filter all tags which have type defined
168 ref($args->{'ds'}->{$_}) eq 'HASH' && defined( $args->{'ds'}->{$_}->{$type} )
169 } keys %{ $args->{'ds'} };
171 $log->debug("tags = ", join(",", @tags));
173 return unless (@tags);
175 foreach my $tag (@tags) {
177 my $vals = join(" ", @{ $args->{'ds'}->{$tag}->{$type} });
179 $log->logconfess("no values for $tag/$type") unless ($vals);
181 $vals = $self->{'iconv'}->convert( $vals ) or
182 $log->logdie("can't convert '$vals' to UTF-8");
184 $doc->add_attr( $tag, $vals );
185 $doc->add_hidden_text( $vals );
188 my $text = $args->{'text'};
190 $text = $self->{'iconv'}->convert( $text ) or
191 $log->logdie("can't convert '$text' to UTF-8");
192 $doc->add_text( $text );
195 $log->debug("adding ", sub { $doc->dump_draft } );
196 $self->{'db'}->put_doc($doc) || $log->logdie("can't add document $uri to index");
203 Issue administrative commands to C<estmaster> process and receive response
206 my $nodelist = $self->est_master( action => nodelist );
210 my $estmaster_actions = {
211 userdel => [ qw/name/ ],
213 nodeadd => [ qw/name label/ ],
214 nodedel => [ qw/name/ ],
220 my $log = $self->_get_logger;
222 $log->debug(Dumper($args));
224 my $action = $args->{action} || $log->logconfess("no action specified");
226 $log->logdie("action '$action' isn't supported") unless ($estmaster_actions->{$action});
228 my $url = $self->{masterurl} . '/master?action=' . $action;
230 foreach my $arg (@{ $estmaster_actions->{$action} }) {
231 $log->logdie("missing parametar $arg for action $action") unless ($args->{$arg});
232 $url .= '&' . $arg . '=' . uri_escape( $args->{$arg} );
235 $log->debug("calling $url");
240 $log->warn("unable to call $url");
244 return split(/\n/, $tsv);
249 Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >>
251 =head1 COPYRIGHT & LICENSE
253 Copyright 2005 Dobrica Pavlinusic, All Rights Reserved.
255 This program is free software; you can redistribute it and/or modify it
256 under the same terms as Perl itself.
260 1; # End of WebPAC::Output::Estraier