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