/[webpac2]/trunk/lib/WebPAC/Output/Estraier.pm
This is repository of my old source code which isn't updated any more. Go to git.rot13.org for current projects!
ViewVC logotype

Annotation of /trunk/lib/WebPAC/Output/Estraier.pm

Parent Directory Parent Directory | Revision Log Revision Log


Revision 459 - (hide annotations)
Wed May 10 14:48:25 2006 UTC (18 years ago) by dpavlin
File size: 5493 byte(s)
 r577@llin:  dpavlin | 2006-05-10 16:51:07 +0200
 fixed add_link and wrote tests for it

1 dpavlin 1 package WebPAC::Output::Estraier;
2    
3     use warnings;
4     use strict;
5    
6 dpavlin 74 use base qw/WebPAC::Common/;
7    
8 dpavlin 458 use Search::Estraier 0.06;
9 dpavlin 401 use Encode qw/from_to/;
10 dpavlin 74 use Data::Dumper;
11 dpavlin 211 use LWP;
12 dpavlin 210 use URI::Escape;
13 dpavlin 459 use List::Util qw/first/;
14 dpavlin 74
15 dpavlin 1 =head1 NAME
16    
17 dpavlin 74 WebPAC::Output::Estraier - Create Hyper Estraier full text index
18 dpavlin 1
19     =head1 VERSION
20    
21 dpavlin 458 Version 0.11
22 dpavlin 1
23     =cut
24    
25 dpavlin 458 our $VERSION = '0.11';
26 dpavlin 1
27     =head1 SYNOPSIS
28    
29 dpavlin 74 Create full text index using Hyper Estraier index from data with
30     type C<search>.
31 dpavlin 1
32 dpavlin 74 =head1 FUNCTIONS
33 dpavlin 1
34 dpavlin 74 =head2 new
35 dpavlin 1
36 dpavlin 74 Connect to Hyper Estraier index using HTTP
37 dpavlin 1
38 dpavlin 74 my $est = new WebPAC::Output::Estraier(
39 dpavlin 210 masterurl => 'http://localhost:1978/',
40 dpavlin 74 user => 'admin',
41     passwd => 'admin',
42     database => 'demo',
43 dpavlin 401 label => 'node label',
44 dpavlin 74 encoding => 'iso-8859-2',
45 dpavlin 307 clean => 1,
46 dpavlin 74 );
47 dpavlin 1
48 dpavlin 74 Options are:
49 dpavlin 1
50 dpavlin 74 =over 4
51 dpavlin 1
52 dpavlin 210 =item masterurl
53 dpavlin 1
54 dpavlin 74 URI to C<estmaster> node
55    
56     =item user
57    
58     C<estmaster> user with sufficient rights
59    
60     =item passwd
61    
62     password for user
63    
64     =item database
65    
66     name of database from which data comes
67    
68 dpavlin 401 =item label
69    
70     label for node (optional)
71    
72 dpavlin 74 =item encoding
73    
74     character encoding of C<data_structure> if it's differenet than C<ISO-8859-2>
75     (and it probably is). This encoding will be converted to C<UTF-8> for
76     Hyper Estraier.
77    
78     =back
79    
80     Name of database will be used to form URI of documents in index.
81    
82 dpavlin 1 =cut
83    
84 dpavlin 74 sub new {
85     my $class = shift;
86 dpavlin 343 my $self = {@_};
87     bless($self, $class);
88 dpavlin 74
89     my $log = $self->_get_logger;
90    
91 dpavlin 211 #$log->debug("self: ", sub { Dumper($self) });
92 dpavlin 210
93     foreach my $p (qw/masterurl user passwd database/) {
94 dpavlin 74 $log->logdie("need $p") unless ($self->{$p});
95     }
96    
97 dpavlin 401 $self->{encoding} ||= 'ISO-8859-2';
98    
99 dpavlin 210 my $url = $self->{masterurl} . '/node/' . $self->{database};
100     $self->{url} = $url;
101 dpavlin 74
102 dpavlin 458 $self->{db} = Search::Estraier::Node->new(
103     url => $url,
104     user => $self->{user},
105     passwd => $self->{passwd},
106     debug => $self->{debug},
107     create => 1,
108     label => "WebPAC $self->{database}",
109     );
110    
111     $log->info("using index $self->{url} with encoding $self->{encoding}");
112    
113 dpavlin 307 if ($self->{clean}) {
114 dpavlin 458 $log->debug("clean $self->{database}");
115     $self->master( action => 'nodeclr', name => $self->{database} );
116 dpavlin 307 } else {
117     $log->debug("opening index $self->{url}");
118     }
119 dpavlin 74
120     $self ? return $self : return undef;
121 dpavlin 1 }
122    
123 dpavlin 75
124 dpavlin 74 =head2 add
125 dpavlin 1
126 dpavlin 74 Adds one entry to database.
127    
128     $est->add(
129     id => 42,
130     ds => $ds,
131     type => 'display',
132     text => 'optional text from which snippet is created',
133     );
134    
135     This function will create entries in index using following URI format:
136    
137 dpavlin 212 C<file:///type/database%20name/000>
138 dpavlin 74
139     Each tag in C<data_structure> with specified C<type> will create one
140     attribute and corresponding hidden text (used for search).
141    
142 dpavlin 1 =cut
143    
144 dpavlin 74 sub add {
145     my $self = shift;
146    
147 dpavlin 456 my $args = {@_};
148 dpavlin 74
149     my $log = $self->_get_logger;
150    
151     my $database = $self->{'database'} || $log->logconfess('no database in $self');
152     $log->logconfess('need db in object') unless ($self->{'db'});
153    
154     foreach my $p (qw/id ds type/) {
155     $log->logdie("need $p") unless ($args->{$p});
156     }
157    
158     my $type = $args->{'type'};
159 dpavlin 213 my $id = $args->{'id'};
160 dpavlin 74
161 dpavlin 213 my $uri = "file:///$type/$database/$id";
162 dpavlin 74 $log->debug("creating $uri");
163    
164 dpavlin 343 my $doc = Search::Estraier::Document->new;
165 dpavlin 401 $doc->add_attr('@uri', $self->convert($uri) );
166 dpavlin 74
167     $log->debug("ds = ", sub { Dumper($args->{'ds'}) } );
168    
169     # filter all tags which have type defined
170     my @tags = grep {
171 dpavlin 113 ref($args->{'ds'}->{$_}) eq 'HASH' && defined( $args->{'ds'}->{$_}->{$type} )
172 dpavlin 74 } keys %{ $args->{'ds'} };
173    
174     $log->debug("tags = ", join(",", @tags));
175    
176     return unless (@tags);
177    
178     foreach my $tag (@tags) {
179    
180     my $vals = join(" ", @{ $args->{'ds'}->{$tag}->{$type} });
181    
182 dpavlin 374 next if (! $vals);
183 dpavlin 74
184 dpavlin 401 $vals = $self->convert( $vals ) or
185 dpavlin 85 $log->logdie("can't convert '$vals' to UTF-8");
186 dpavlin 75
187     $doc->add_attr( $tag, $vals );
188     $doc->add_hidden_text( $vals );
189 dpavlin 74 }
190    
191     my $text = $args->{'text'};
192 dpavlin 75 if ( $text ) {
193 dpavlin 401 $text = $self->convert( $text ) or
194 dpavlin 85 $log->logdie("can't convert '$text' to UTF-8");
195 dpavlin 75 $doc->add_text( $text );
196     }
197 dpavlin 74
198     $log->debug("adding ", sub { $doc->dump_draft } );
199 dpavlin 319 $self->{'db'}->put_doc($doc) || $log->warn("can't add document $uri with draft " . $doc->dump_draft . " to node " . $self->{url} . " status: " . $self->{db}->status());
200 dpavlin 74
201     return 1;
202 dpavlin 1 }
203    
204 dpavlin 255 =head2 add_link
205    
206     $est->add_link(
207     from => 'ps',
208     to => 'webpac2',
209     credit => 10000,
210     );
211    
212 dpavlin 246 =cut
213    
214 dpavlin 255 sub add_link {
215 dpavlin 238 my $self = shift;
216    
217 dpavlin 255 my $args = {@_};
218     my $log = $self->_get_logger;
219 dpavlin 238
220 dpavlin 459 my $node = first { $_->{name} eq $args->{to} } $self->master( action => 'nodelist' );
221 dpavlin 238
222 dpavlin 459 if (! $node) {
223     $log->warn("can't find node $args->{to}, skipping link creaton");
224     return;
225     }
226 dpavlin 238
227 dpavlin 459 my $label = $node->{label};
228 dpavlin 255
229     if (! $label) {
230     $log->warn("can't find label for $args->{to}, skipping link creaton");
231     return;
232     }
233    
234     $log->debug("using label $label for $args->{to}");
235    
236 dpavlin 459 return $self->{db}->set_link(
237     $self->{masterurl} . '/node/' . $args->{to},
238     $label,
239     $args->{credit},
240 dpavlin 255 );
241 dpavlin 238 }
242    
243 dpavlin 255
244 dpavlin 458 =head2 master
245 dpavlin 255
246 dpavlin 458 Issue administrative commands to C<estmaster> process. See documentation for
247     C<master> in L<Search::Estraier>::Node.
248    
249     $self->master(
250     action => 'nodeclr',
251     name => 'foobar',
252 dpavlin 255 );
253    
254     =cut
255    
256 dpavlin 458 sub master {
257 dpavlin 210 my $self = shift;
258 dpavlin 458 $self->{db}->master( @_ );
259 dpavlin 210 }
260    
261 dpavlin 255
262 dpavlin 401 =head2 convert
263    
264     my $utf8_string = $self->convert('string in codepage');
265    
266     =cut
267    
268     sub convert {
269     my $self = shift;
270    
271     my $text = shift || return;
272     from_to($text, $self->{encoding}, 'UTF-8');
273     return $text;
274     }
275    
276 dpavlin 1 =head1 AUTHOR
277    
278     Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >>
279    
280     =head1 COPYRIGHT & LICENSE
281    
282     Copyright 2005 Dobrica Pavlinusic, All Rights Reserved.
283    
284     This program is free software; you can redistribute it and/or modify it
285     under the same terms as Perl itself.
286    
287     =cut
288    
289     1; # End of WebPAC::Output::Estraier

  ViewVC Help
Powered by ViewVC 1.1.26