1 |
package WebPAC::Output::Estraier; |
2 |
|
3 |
use warnings; |
4 |
use strict; |
5 |
|
6 |
use base qw/WebPAC::Common/; |
7 |
|
8 |
use Search::Estraier 0.06; |
9 |
use Encode qw/from_to/; |
10 |
use Data::Dump qw/dump/; |
11 |
use LWP; |
12 |
use URI::Escape; |
13 |
use List::Util qw/first/; |
14 |
use File::Path qw/mkpath/; |
15 |
use YAML; |
16 |
|
17 |
=head1 NAME |
18 |
|
19 |
WebPAC::Output::Estraier - Create Hyper Estraier full text index |
20 |
|
21 |
=head1 VERSION |
22 |
|
23 |
Version 0.12 |
24 |
|
25 |
=cut |
26 |
|
27 |
our $VERSION = '0.12'; |
28 |
|
29 |
=head1 SYNOPSIS |
30 |
|
31 |
Create full text index using Hyper Estraier index from data with |
32 |
type C<search>. |
33 |
|
34 |
=head1 FUNCTIONS |
35 |
|
36 |
=head2 new |
37 |
|
38 |
Connect to Hyper Estraier index using HTTP |
39 |
|
40 |
my $est = new WebPAC::Output::Estraier( |
41 |
masterurl => 'http://localhost:1978/', |
42 |
user => 'admin', |
43 |
passwd => 'admin', |
44 |
database => 'demo', |
45 |
label => 'node label', |
46 |
encoding => 'iso-8859-2', |
47 |
clean => 1, |
48 |
); |
49 |
|
50 |
Options are: |
51 |
|
52 |
=over 4 |
53 |
|
54 |
=item masterurl |
55 |
|
56 |
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 |
=item label |
71 |
|
72 |
label for node (optional) |
73 |
|
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 |
=cut |
85 |
|
86 |
sub new { |
87 |
my $class = shift; |
88 |
my $self = {@_}; |
89 |
bless($self, $class); |
90 |
|
91 |
my $log = $self->_get_logger; |
92 |
|
93 |
#$log->debug("self: ", sub { dump($self) }); |
94 |
|
95 |
foreach my $p (qw/masterurl user passwd database/) { |
96 |
$log->logdie("need $p") unless ($self->{$p}); |
97 |
} |
98 |
|
99 |
$self->{encoding} ||= 'ISO-8859-2'; |
100 |
|
101 |
my $url = $self->{masterurl} . '/node/' . $self->{database}; |
102 |
$self->{url} = $url; |
103 |
|
104 |
$self->{label} ||= "WebPAC $self->{database}"; |
105 |
|
106 |
$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 |
label => $self->convert( $self->{label} ), |
113 |
); |
114 |
|
115 |
$log->info("using ", $self->{clean} ? "new " : "", "index $self->{url} '$self->{label}' with encoding $self->{encoding}"); |
116 |
|
117 |
if ($self->{clean}) { |
118 |
$log->debug("clean $self->{database}"); |
119 |
$self->master( action => 'nodeclr', name => $self->{database} ); |
120 |
} else { |
121 |
$log->debug("opening index $self->{url}"); |
122 |
} |
123 |
|
124 |
$self->{stats} = {}; |
125 |
|
126 |
$self ? return $self : return undef; |
127 |
} |
128 |
|
129 |
|
130 |
=head2 add |
131 |
|
132 |
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 |
C<file:///type/database%20name/000> |
144 |
|
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 |
=cut |
149 |
|
150 |
sub add { |
151 |
my $self = shift; |
152 |
|
153 |
my $args = {@_}; |
154 |
|
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 |
my $id = $args->{'id'}; |
166 |
|
167 |
my $uri = "file:///$type/$database/$id"; |
168 |
$log->debug("creating $uri"); |
169 |
|
170 |
my $doc = Search::Estraier::Document->new; |
171 |
$doc->add_attr('@uri', $self->convert($uri) ); |
172 |
|
173 |
$log->debug("ds = ", sub { dump($args->{'ds'}) } ); |
174 |
|
175 |
# filter all tags which have type defined |
176 |
my @tags = grep { |
177 |
ref($args->{'ds'}->{$_}) eq 'HASH' && defined( $args->{'ds'}->{$_}->{$type} ) |
178 |
} 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 |
next if (! $vals); |
189 |
|
190 |
$vals = $self->convert( $vals ) or |
191 |
$log->logdie("can't convert '$vals' to UTF-8"); |
192 |
|
193 |
$doc->add_attr( $tag, $vals ); |
194 |
$doc->add_hidden_text( $vals ); |
195 |
|
196 |
$self->{stats}->{attr}->{$tag}++; |
197 |
} |
198 |
|
199 |
my $text = $args->{'text'}; |
200 |
if ( $text ) { |
201 |
$text = $self->convert( $text ) or |
202 |
$log->logdie("can't convert '$text' to UTF-8"); |
203 |
$doc->add_text( $text ); |
204 |
} |
205 |
|
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()); |
208 |
|
209 |
return 1; |
210 |
} |
211 |
|
212 |
=head2 add_link |
213 |
|
214 |
$est->add_link( |
215 |
from => 'ps', |
216 |
to => 'webpac2', |
217 |
credit => 10000, |
218 |
); |
219 |
|
220 |
=cut |
221 |
|
222 |
sub add_link { |
223 |
my $self = shift; |
224 |
|
225 |
my $args = {@_}; |
226 |
my $log = $self->_get_logger; |
227 |
|
228 |
foreach my $p (qw/from to credit/) { |
229 |
$log->logdie("need $p") unless ($args->{$p}); |
230 |
} |
231 |
|
232 |
my $node = first { $_->{name} eq $args->{to} } $self->master( action => 'nodelist' ); |
233 |
|
234 |
if (! $node) { |
235 |
$log->warn("can't find node $args->{to}, skipping link creaton"); |
236 |
return; |
237 |
} |
238 |
|
239 |
my $label = $node->{label}; |
240 |
|
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 |
return $self->{db}->set_link( |
249 |
$self->{masterurl} . '/node/' . $args->{to}, |
250 |
$label, |
251 |
$args->{credit}, |
252 |
); |
253 |
} |
254 |
|
255 |
|
256 |
=head2 master |
257 |
|
258 |
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 |
); |
265 |
|
266 |
=cut |
267 |
|
268 |
sub master { |
269 |
my $self = shift; |
270 |
$self->{db}->master( @_ ); |
271 |
} |
272 |
|
273 |
|
274 |
=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 |
=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 |
my $path = 'var/estraier'; |
299 |
|
300 |
mkpath $path unless -e $path; |
301 |
|
302 |
$path .= '/' . $self->{database} . '.yaml'; |
303 |
YAML::DumpFile( $path, $self->{stats} ); |
304 |
$log->info("created $path ", -s $path, " bytes"); |
305 |
$log->debug( dump( $self->{stats} ) ); |
306 |
} |
307 |
|
308 |
=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 |