1 package WebPAC::Search::Estraier;
12 WebPAC::Search::Estraier - search Hyper Estraier full text index
20 our $VERSION = '0.03';
24 Search WebPAC data using Hyper Estraier full text index created with
25 L<WebPAC::Output::Estraier>.
31 Connect to Hyper Estraier index using HTTP
33 my $est = new WebPAC::Search::Estraier(
34 url => 'http://localhost:1978/node/webpac2',
37 encoding => 'iso-8859-2',
38 log => $Log::Log4perl->log_object,
47 URI to C<estmaster> node
51 C<estmaster> user with read rights
59 character encoding of C<data_structure> if it's differenet than C<ISO-8859-2>
60 (and it probably is). This encoding will be converted to C<UTF-8> for
65 L<Log::Log4perl> object or equivalent (C<< $c->log >> can be used in
66 L<Catalyst> and there is support for it).
77 my $log = $self->_get_logger;
79 foreach my $p (qw/url user passwd/) {
80 $log->logdie("need $p") unless ($self->{$p});
83 $log->info("opening Hyper Estraier index $self->{'url'} as $self->{'user'}");
85 $self->{'db'} = HyperEstraier::Node->new($self->{'url'});
86 $self->{'db'}->set_auth($self->{'user'}, $self->{'passwd'});
88 my $encoding = $self->{'encoding'} || 'ISO-8859-2';
89 $log->info("using encoding $encoding");
91 $self->{'iconv'} = new Text::Iconv('UTF-8', $encoding) or
92 $log->die("can't create conversion from UTF-8 to $encoding");
94 $self ? return $self : return undef;
100 Locate items in index
102 my @results = $est->search(
103 phrase => 'name of book or novel',
105 "filepath ISTRINC $q",
108 get_attr => qw/PersonalName TitleProper/,
111 options => $HyperEstraier::Condition::SURE,
115 Options are close match to Hyper Estraier API, except C<get_attr> which defines
116 attributes which will be returned in hash for each record.
118 Results are returned as hash array with keys named by attributes
120 Pages are numbered C< 1 ... hits/max >.
129 my $log = $self->_get_logger;
131 #$log->debug( 'search args: ' . Dumper($args) );
133 $self->confess('need db in object') unless ($self->{db});
134 $self->confess('need get_attr') unless ($args->{get_attr});
136 $self->confess("need get_attr as array not " . ref($args->{get_attr}) ) unless (ref($args->{get_attr}) eq 'ARRAY');
138 my $q = $args->{phrase};
140 $log->debug("args: " . Dumper( $args ));
142 my $cond = HyperEstraier::Condition->new();
143 if ( ref($args->{add_attr}) eq 'ARRAY' ) {
144 $log->debug("adding search attributes: " . join(", ", @{ $args->{add_attr} }) );
146 $cond->add_attr( $self->{iconv}->convert( $_ ) );
147 $log->debug(" + $_");
148 } @{ $args->{add_attr} };
151 $cond->set_phrase( $self->{iconv}->convert($q) ) if ($q);
152 $cond->set_options( $args->{options} ) if ($args->{options});
153 $cond->set_order( $args->{order} ) if ($args->{order});
155 my $max = $args->{max} || 7;
156 my $page = $args->{page} || 1;
158 $log->warn("page number $page < 1");
162 $cond->set_max( $page * $max );
164 my $result = $self->{db}->search($cond, 0) ||
165 $log->die("can't search for ", sub { Dumper( $args ) });
167 my $hits = $result->doc_num;
168 $log->debug("found $hits hits for '$q'");
172 for my $i ( (($page - 1) * $max) .. ( $hits - 1 ) ) {
174 #$log->debug("get_doc($i)");
175 my $doc = $result->get_doc( $i );
177 $log->warn("can't find result $i");
183 foreach my $attr (@{ $args->{get_attr} }) {
184 my $val = $doc->attr( $attr );
185 #$log->debug("attr $attr = ", $val || 'undef');
186 $hash->{$attr} = $self->{iconv}->convert( $val ) if (defined($val));
190 push @results, $hash;
195 # $log->debug("results " . Dumper( \@results ));
197 $self->confess("expected to return array") unless (wantarray);
204 wrapper around L<Log::Log4perl> C<confess> or C<< $log->fatal >> or
205 C<< $log->error >> if they exists (like in L<Catalyst>), else plain
212 if (my $log = $self->{'log'}) {
213 if ($log->can('logconfess')) {
214 $log->logconfess(@_);
215 } elsif ($log->can('fatal')) {
218 } elsif ($log->can('error')) {
230 For compatibility with same method from L<WebPAC::Common>, but without
238 return $self->{'log'} || die "really need log!";
243 Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >>
245 =head1 COPYRIGHT & LICENSE
247 Copyright 2005 Dobrica Pavlinusic, All Rights Reserved.
249 This program is free software; you can redistribute it and/or modify it
250 under the same terms as Perl itself.
254 1; # End of WebPAC::Search::Estraier