1 package WebPAC::Output::Estraier;
6 use base qw/WebPAC::Common/;
9 use Encode qw/from_to/;
16 WebPAC::Output::Estraier - Create Hyper Estraier full text index
24 our $VERSION = '0.10';
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 label => 'node label',
43 encoding => 'iso-8859-2',
53 URI to C<estmaster> node
57 C<estmaster> user with sufficient rights
65 name of database from which data comes
69 label for node (optional)
73 character encoding of C<data_structure> if it's differenet than C<ISO-8859-2>
74 (and it probably is). This encoding will be converted to C<UTF-8> for
79 Name of database will be used to form URI of documents in index.
88 my $log = $self->_get_logger;
90 #$log->debug("self: ", sub { Dumper($self) });
92 foreach my $p (qw/masterurl user passwd database/) {
93 $log->logdie("need $p") unless ($self->{$p});
96 $self->{encoding} ||= 'ISO-8859-2';
98 my $url = $self->{masterurl} . '/node/' . $self->{database};
101 if ($self->{clean}) {
102 $log->debug("nodedel $self->{database}");
103 $self->master( action => 'nodedel', name => $self->{database} );
105 $log->debug("opening index $self->{url}");
108 my $nodes = $self->master( action => 'nodelist' );
110 $log->debug("nodes found: $nodes");
112 if ($nodes !~ m/^$self->{database}\t/sm) {
113 my $label = $self->{label} || 'WebPAC ' . $self->{database};
114 $log->warn("creating index $url ($label)");
117 name => $self->{database},
118 label => $self->convert( $label ),
119 ) || $log->logdie("can't create Hyper Estraier node $self->{database}");
122 $self->{db} = Search::Estraier::Node->new( debug => $self->{debug} );
123 $self->{db}->set_url($self->{url});
124 $self->{db}->set_auth($self->{user}, $self->{passwd});
126 $log->info("using index $self->{url} with encoding $self->{encoding}");
128 $self ? return $self : return undef;
134 Adds one entry to database.
140 text => 'optional text from which snippet is created',
143 This function will create entries in index using following URI format:
145 C<file:///type/database%20name/000>
147 Each tag in C<data_structure> with specified C<type> will create one
148 attribute and corresponding hidden text (used for search).
157 my $log = $self->_get_logger;
159 my $database = $self->{'database'} || $log->logconfess('no database in $self');
160 $log->logconfess('need db in object') unless ($self->{'db'});
162 foreach my $p (qw/id ds type/) {
163 $log->logdie("need $p") unless ($args->{$p});
166 my $type = $args->{'type'};
167 my $id = $args->{'id'};
169 my $uri = "file:///$type/$database/$id";
170 $log->debug("creating $uri");
172 my $doc = Search::Estraier::Document->new;
173 $doc->add_attr('@uri', $self->convert($uri) );
175 $log->debug("ds = ", sub { Dumper($args->{'ds'}) } );
177 # filter all tags which have type defined
179 ref($args->{'ds'}->{$_}) eq 'HASH' && defined( $args->{'ds'}->{$_}->{$type} )
180 } keys %{ $args->{'ds'} };
182 $log->debug("tags = ", join(",", @tags));
184 return unless (@tags);
186 foreach my $tag (@tags) {
188 my $vals = join(" ", @{ $args->{'ds'}->{$tag}->{$type} });
192 $vals = $self->convert( $vals ) or
193 $log->logdie("can't convert '$vals' to UTF-8");
195 $doc->add_attr( $tag, $vals );
196 $doc->add_hidden_text( $vals );
199 my $text = $args->{'text'};
201 $text = $self->convert( $text ) or
202 $log->logdie("can't convert '$text' to UTF-8");
203 $doc->add_text( $text );
206 $log->debug("adding ", sub { $doc->dump_draft } );
207 $self->{'db'}->put_doc($doc) || $log->warn("can't add document $uri with draft " . $doc->dump_draft . " to node " . $self->{url} . " status: " . $self->{db}->status());
213 # REST parametars validation data
216 my $estraier_rest = {
218 userdel => [ qw/name/ ],
220 nodeadd => [ qw/name label/ ],
221 nodedel => [ qw/name/ ],
224 _set_link => [ qw/url label credit/ ],
230 Issue administrative commands to C<estmaster> process and receive response
233 my $nodelist = $est->master( action => 'nodelist' );
241 my $log = $self->_get_logger;
243 my $action = $args->{action} || $log->logconfess("no action specified");
245 $log->logdie("action '$action' isn't supported") unless ($estraier_rest->{master}->{$action});
247 $log->debug("master action: $action");
249 return $self->estcall(
250 validate => 'master',
251 rest_url => $self->{masterurl} . '/master?action=' . $action ,
271 my $log = $self->_get_logger;
273 my @labels = $self->master( action => 'nodelist' );
275 $log->debug("got labels: ", join("|", @labels));
277 @labels = grep(/^$args->{to}\t/, @labels);
278 my $label = shift @labels;
279 (undef,$label) = split(/\t/, $label) if ($label);
282 $log->warn("can't find label for $args->{to}, skipping link creaton");
286 $log->debug("using label $label for $args->{to}");
288 return $self->estcall(
290 action => '_set_link',
291 rest_url => $self->{masterurl} . '/node/' . $args->{from} . '/_set_link' ,
292 url => $self->{masterurl} . '/node/' . $args->{to},
294 credit => $args->{credit},
300 Workhourse which does actual calls to Hyper Estraier
303 rest_url => '/master?action=' . $action,
304 validate => 'master',
308 C<rest_url> is relative URL to C<estmaster> and C<validate> is entry into
309 internal hash which will check if all parametars are available before
317 my $log = $self->_get_logger;
319 $log->debug("estcall: ",Dumper($args));
321 foreach my $p (qw/rest_url validate action/) {
322 $log->die("ectcall needs $p parametar") unless ($args->{$p});
325 my $url = $args->{rest_url};
327 $del = '&' if ($url =~ m#\?#);
331 foreach my $arg (@{ $estraier_rest->{ $args->{validate} }->{ $args->{action} } }) {
332 $log->logdie("missing parametar $arg for action $args->{action}") unless ($args->{$arg});
333 $url_args .= $del . $arg . '=' . uri_escape( $args->{$arg} );
337 $url .= $url_args if ($url_args);
339 $log->debug("calling $url");
341 my $res = $self->est_ua()->get($url);
343 if ($res->is_success) {
344 #$log->debug( $res->content );
345 return split(/\n/, $res->content) if wantarray;
346 return $res->content || 0E0;
348 $log->warn("unable to call $url: " . $res->status_line);
356 This is helper function to create C<LWP::UserAgent> object with Super User
359 my $ua = $self->est_ua( user => 'admin', passwd => 'admin' );
368 return $self->{_master_ua} if ($self->{_master_ua});
371 package AdminUserAgent;
372 use base qw/LWP::UserAgent/;
374 my $self = LWP::UserAgent::new(@_);
375 $self->agent("webpac/$VERSION");
378 sub get_basic_credentials {
379 my($self, $realm, $uri) = @_;
380 return ($self->{user}, $self->{passwd});
382 sub set_basic_credentials {
383 my ($self, $user, $passwd) = @_;
384 $self->{user} = $user;
385 $self->{passwd} = $passwd;
389 $self->{_master_ua} = AdminUserAgent->new( ) || sub {
390 my $log = $self->_get_logger;
391 $log->logdie("can't create LWP::UserAgent: $!");
394 $self->{_master_ua}->set_basic_credentials($self->{user}, $self->{passwd});
396 return $self->{_master_ua};
401 my $utf8_string = $self->convert('string in codepage');
408 my $text = shift || return;
409 from_to($text, $self->{encoding}, 'UTF-8');
415 Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >>
417 =head1 COPYRIGHT & LICENSE
419 Copyright 2005 Dobrica Pavlinusic, All Rights Reserved.
421 This program is free software; you can redistribute it and/or modify it
422 under the same terms as Perl itself.
426 1; # End of WebPAC::Output::Estraier