/[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 1148 - (hide annotations)
Fri Apr 24 09:37:08 2009 UTC (15 years, 1 month ago) by dpavlin
File size: 6150 byte(s)
create var/estraier if it doesn't exist

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 887 use Data::Dump qw/dump/;
11 dpavlin 211 use LWP;
12 dpavlin 210 use URI::Escape;
13 dpavlin 459 use List::Util qw/first/;
14 dpavlin 1148 use File::Path qw/mkpath/;
15 dpavlin 1131 use YAML;
16 dpavlin 74
17 dpavlin 1 =head1 NAME
18    
19 dpavlin 74 WebPAC::Output::Estraier - Create Hyper Estraier full text index
20 dpavlin 1
21     =head1 VERSION
22    
23 dpavlin 525 Version 0.12
24 dpavlin 1
25     =cut
26    
27 dpavlin 525 our $VERSION = '0.12';
28 dpavlin 1
29     =head1 SYNOPSIS
30    
31 dpavlin 74 Create full text index using Hyper Estraier index from data with
32     type C<search>.
33 dpavlin 1
34 dpavlin 74 =head1 FUNCTIONS
35 dpavlin 1
36 dpavlin 74 =head2 new
37 dpavlin 1
38 dpavlin 74 Connect to Hyper Estraier index using HTTP
39 dpavlin 1
40 dpavlin 74 my $est = new WebPAC::Output::Estraier(
41 dpavlin 210 masterurl => 'http://localhost:1978/',
42 dpavlin 74 user => 'admin',
43     passwd => 'admin',
44     database => 'demo',
45 dpavlin 401 label => 'node label',
46 dpavlin 74 encoding => 'iso-8859-2',
47 dpavlin 307 clean => 1,
48 dpavlin 74 );
49 dpavlin 1
50 dpavlin 74 Options are:
51 dpavlin 1
52 dpavlin 74 =over 4
53 dpavlin 1
54 dpavlin 210 =item masterurl
55 dpavlin 1
56 dpavlin 74 URI to C<estmaster> node
57    
58     =item user
59    
60     C<estmaster> user with sufficient rights
61    
62     =item passwd
63    
64     password for user
65    
66     =item database
67    
68     name of database from which data comes
69    
70 dpavlin 401 =item label
71    
72     label for node (optional)
73    
74 dpavlin 74 =item encoding
75    
76     character encoding of C<data_structure> if it's differenet than C<ISO-8859-2>
77     (and it probably is). This encoding will be converted to C<UTF-8> for
78     Hyper Estraier.
79    
80     =back
81    
82     Name of database will be used to form URI of documents in index.
83    
84 dpavlin 1 =cut
85    
86 dpavlin 74 sub new {
87     my $class = shift;
88 dpavlin 343 my $self = {@_};
89     bless($self, $class);
90 dpavlin 74
91     my $log = $self->_get_logger;
92    
93 dpavlin 887 #$log->debug("self: ", sub { dump($self) });
94 dpavlin 210
95     foreach my $p (qw/masterurl user passwd database/) {
96 dpavlin 74 $log->logdie("need $p") unless ($self->{$p});
97     }
98    
99 dpavlin 401 $self->{encoding} ||= 'ISO-8859-2';
100    
101 dpavlin 210 my $url = $self->{masterurl} . '/node/' . $self->{database};
102     $self->{url} = $url;
103 dpavlin 74
104 dpavlin 525 $self->{label} ||= "WebPAC $self->{database}";
105    
106 dpavlin 458 $self->{db} = Search::Estraier::Node->new(
107     url => $url,
108     user => $self->{user},
109     passwd => $self->{passwd},
110     debug => $self->{debug},
111     create => 1,
112 dpavlin 525 label => $self->convert( $self->{label} ),
113 dpavlin 458 );
114    
115 dpavlin 525 $log->info("using ", $self->{clean} ? "new " : "", "index $self->{url} '$self->{label}' with encoding $self->{encoding}");
116 dpavlin 458
117 dpavlin 307 if ($self->{clean}) {
118 dpavlin 458 $log->debug("clean $self->{database}");
119     $self->master( action => 'nodeclr', name => $self->{database} );
120 dpavlin 307 } else {
121     $log->debug("opening index $self->{url}");
122     }
123 dpavlin 74
124 dpavlin 1131 $self->{stats} = {};
125    
126 dpavlin 74 $self ? return $self : return undef;
127 dpavlin 1 }
128    
129 dpavlin 75
130 dpavlin 74 =head2 add
131 dpavlin 1
132 dpavlin 74 Adds one entry to database.
133    
134     $est->add(
135     id => 42,
136     ds => $ds,
137     type => 'display',
138     text => 'optional text from which snippet is created',
139     );
140    
141     This function will create entries in index using following URI format:
142    
143 dpavlin 212 C<file:///type/database%20name/000>
144 dpavlin 74
145     Each tag in C<data_structure> with specified C<type> will create one
146     attribute and corresponding hidden text (used for search).
147    
148 dpavlin 1 =cut
149    
150 dpavlin 74 sub add {
151     my $self = shift;
152    
153 dpavlin 456 my $args = {@_};
154 dpavlin 74
155     my $log = $self->_get_logger;
156    
157     my $database = $self->{'database'} || $log->logconfess('no database in $self');
158     $log->logconfess('need db in object') unless ($self->{'db'});
159    
160     foreach my $p (qw/id ds type/) {
161     $log->logdie("need $p") unless ($args->{$p});
162     }
163    
164     my $type = $args->{'type'};
165 dpavlin 213 my $id = $args->{'id'};
166 dpavlin 74
167 dpavlin 213 my $uri = "file:///$type/$database/$id";
168 dpavlin 74 $log->debug("creating $uri");
169    
170 dpavlin 343 my $doc = Search::Estraier::Document->new;
171 dpavlin 401 $doc->add_attr('@uri', $self->convert($uri) );
172 dpavlin 74
173 dpavlin 887 $log->debug("ds = ", sub { dump($args->{'ds'}) } );
174 dpavlin 74
175     # filter all tags which have type defined
176     my @tags = grep {
177 dpavlin 113 ref($args->{'ds'}->{$_}) eq 'HASH' && defined( $args->{'ds'}->{$_}->{$type} )
178 dpavlin 74 } keys %{ $args->{'ds'} };
179    
180     $log->debug("tags = ", join(",", @tags));
181    
182     return unless (@tags);
183    
184     foreach my $tag (@tags) {
185    
186     my $vals = join(" ", @{ $args->{'ds'}->{$tag}->{$type} });
187    
188 dpavlin 374 next if (! $vals);
189 dpavlin 74
190 dpavlin 401 $vals = $self->convert( $vals ) or
191 dpavlin 85 $log->logdie("can't convert '$vals' to UTF-8");
192 dpavlin 75
193     $doc->add_attr( $tag, $vals );
194     $doc->add_hidden_text( $vals );
195 dpavlin 1131
196     $self->{stats}->{attr}->{$tag}++;
197 dpavlin 74 }
198    
199     my $text = $args->{'text'};
200 dpavlin 75 if ( $text ) {
201 dpavlin 401 $text = $self->convert( $text ) or
202 dpavlin 85 $log->logdie("can't convert '$text' to UTF-8");
203 dpavlin 75 $doc->add_text( $text );
204     }
205 dpavlin 74
206     $log->debug("adding ", sub { $doc->dump_draft } );
207 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());
208 dpavlin 74
209     return 1;
210 dpavlin 1 }
211    
212 dpavlin 255 =head2 add_link
213    
214     $est->add_link(
215     from => 'ps',
216     to => 'webpac2',
217     credit => 10000,
218     );
219    
220 dpavlin 246 =cut
221    
222 dpavlin 255 sub add_link {
223 dpavlin 238 my $self = shift;
224    
225 dpavlin 255 my $args = {@_};
226     my $log = $self->_get_logger;
227 dpavlin 238
228 dpavlin 527 foreach my $p (qw/from to credit/) {
229     $log->logdie("need $p") unless ($args->{$p});
230     }
231    
232 dpavlin 459 my $node = first { $_->{name} eq $args->{to} } $self->master( action => 'nodelist' );
233 dpavlin 238
234 dpavlin 459 if (! $node) {
235     $log->warn("can't find node $args->{to}, skipping link creaton");
236     return;
237     }
238 dpavlin 238
239 dpavlin 459 my $label = $node->{label};
240 dpavlin 255
241     if (! $label) {
242     $log->warn("can't find label for $args->{to}, skipping link creaton");
243     return;
244     }
245    
246     $log->debug("using label $label for $args->{to}");
247    
248 dpavlin 459 return $self->{db}->set_link(
249     $self->{masterurl} . '/node/' . $args->{to},
250     $label,
251     $args->{credit},
252 dpavlin 255 );
253 dpavlin 238 }
254    
255 dpavlin 255
256 dpavlin 458 =head2 master
257 dpavlin 255
258 dpavlin 458 Issue administrative commands to C<estmaster> process. See documentation for
259     C<master> in L<Search::Estraier>::Node.
260    
261     $self->master(
262     action => 'nodeclr',
263     name => 'foobar',
264 dpavlin 255 );
265    
266     =cut
267    
268 dpavlin 458 sub master {
269 dpavlin 210 my $self = shift;
270 dpavlin 458 $self->{db}->master( @_ );
271 dpavlin 210 }
272    
273 dpavlin 255
274 dpavlin 401 =head2 convert
275    
276     my $utf8_string = $self->convert('string in codepage');
277    
278     =cut
279    
280     sub convert {
281     my $self = shift;
282    
283     my $text = shift || return;
284     from_to($text, $self->{encoding}, 'UTF-8');
285     return $text;
286     }
287    
288 dpavlin 1131 =head2 finish
289    
290     Dump attributes used on disk
291    
292     =cut
293    
294     sub finish {
295     my $self = shift;
296     my $log = $self->_get_logger();
297    
298 dpavlin 1148 my $path = 'var/estraier';
299    
300     mkpath $path unless -e $path;
301    
302     $path .= '/' . $self->{database} . '.yaml';
303 dpavlin 1131 YAML::DumpFile( $path, $self->{stats} );
304     $log->info("created $path ", -s $path, " bytes");
305     $log->debug( dump( $self->{stats} ) );
306     }
307    
308 dpavlin 1 =head1 AUTHOR
309    
310     Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >>
311    
312     =head1 COPYRIGHT & LICENSE
313    
314     Copyright 2005 Dobrica Pavlinusic, All Rights Reserved.
315    
316     This program is free software; you can redistribute it and/or modify it
317     under the same terms as Perl itself.
318    
319     =cut
320    
321     1; # End of WebPAC::Output::Estraier

  ViewVC Help
Powered by ViewVC 1.1.26