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 |
525 |
Version 0.12 |
22 |
dpavlin |
1 |
|
23 |
|
|
=cut |
24 |
|
|
|
25 |
dpavlin |
525 |
our $VERSION = '0.12'; |
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 |
525 |
$self->{label} ||= "WebPAC $self->{database}"; |
103 |
|
|
|
104 |
dpavlin |
458 |
$self->{db} = Search::Estraier::Node->new( |
105 |
|
|
url => $url, |
106 |
|
|
user => $self->{user}, |
107 |
|
|
passwd => $self->{passwd}, |
108 |
|
|
debug => $self->{debug}, |
109 |
|
|
create => 1, |
110 |
dpavlin |
525 |
label => $self->convert( $self->{label} ), |
111 |
dpavlin |
458 |
); |
112 |
|
|
|
113 |
dpavlin |
525 |
$log->info("using ", $self->{clean} ? "new " : "", "index $self->{url} '$self->{label}' with encoding $self->{encoding}"); |
114 |
dpavlin |
458 |
|
115 |
dpavlin |
307 |
if ($self->{clean}) { |
116 |
dpavlin |
458 |
$log->debug("clean $self->{database}"); |
117 |
|
|
$self->master( action => 'nodeclr', name => $self->{database} ); |
118 |
dpavlin |
307 |
} else { |
119 |
|
|
$log->debug("opening index $self->{url}"); |
120 |
|
|
} |
121 |
dpavlin |
74 |
|
122 |
|
|
$self ? return $self : return undef; |
123 |
dpavlin |
1 |
} |
124 |
|
|
|
125 |
dpavlin |
75 |
|
126 |
dpavlin |
74 |
=head2 add |
127 |
dpavlin |
1 |
|
128 |
dpavlin |
74 |
Adds one entry to database. |
129 |
|
|
|
130 |
|
|
$est->add( |
131 |
|
|
id => 42, |
132 |
|
|
ds => $ds, |
133 |
|
|
type => 'display', |
134 |
|
|
text => 'optional text from which snippet is created', |
135 |
|
|
); |
136 |
|
|
|
137 |
|
|
This function will create entries in index using following URI format: |
138 |
|
|
|
139 |
dpavlin |
212 |
C<file:///type/database%20name/000> |
140 |
dpavlin |
74 |
|
141 |
|
|
Each tag in C<data_structure> with specified C<type> will create one |
142 |
|
|
attribute and corresponding hidden text (used for search). |
143 |
|
|
|
144 |
dpavlin |
1 |
=cut |
145 |
|
|
|
146 |
dpavlin |
74 |
sub add { |
147 |
|
|
my $self = shift; |
148 |
|
|
|
149 |
dpavlin |
456 |
my $args = {@_}; |
150 |
dpavlin |
74 |
|
151 |
|
|
my $log = $self->_get_logger; |
152 |
|
|
|
153 |
|
|
my $database = $self->{'database'} || $log->logconfess('no database in $self'); |
154 |
|
|
$log->logconfess('need db in object') unless ($self->{'db'}); |
155 |
|
|
|
156 |
|
|
foreach my $p (qw/id ds type/) { |
157 |
|
|
$log->logdie("need $p") unless ($args->{$p}); |
158 |
|
|
} |
159 |
|
|
|
160 |
|
|
my $type = $args->{'type'}; |
161 |
dpavlin |
213 |
my $id = $args->{'id'}; |
162 |
dpavlin |
74 |
|
163 |
dpavlin |
213 |
my $uri = "file:///$type/$database/$id"; |
164 |
dpavlin |
74 |
$log->debug("creating $uri"); |
165 |
|
|
|
166 |
dpavlin |
343 |
my $doc = Search::Estraier::Document->new; |
167 |
dpavlin |
401 |
$doc->add_attr('@uri', $self->convert($uri) ); |
168 |
dpavlin |
74 |
|
169 |
|
|
$log->debug("ds = ", sub { Dumper($args->{'ds'}) } ); |
170 |
|
|
|
171 |
|
|
# filter all tags which have type defined |
172 |
|
|
my @tags = grep { |
173 |
dpavlin |
113 |
ref($args->{'ds'}->{$_}) eq 'HASH' && defined( $args->{'ds'}->{$_}->{$type} ) |
174 |
dpavlin |
74 |
} keys %{ $args->{'ds'} }; |
175 |
|
|
|
176 |
|
|
$log->debug("tags = ", join(",", @tags)); |
177 |
|
|
|
178 |
|
|
return unless (@tags); |
179 |
|
|
|
180 |
|
|
foreach my $tag (@tags) { |
181 |
|
|
|
182 |
|
|
my $vals = join(" ", @{ $args->{'ds'}->{$tag}->{$type} }); |
183 |
|
|
|
184 |
dpavlin |
374 |
next if (! $vals); |
185 |
dpavlin |
74 |
|
186 |
dpavlin |
401 |
$vals = $self->convert( $vals ) or |
187 |
dpavlin |
85 |
$log->logdie("can't convert '$vals' to UTF-8"); |
188 |
dpavlin |
75 |
|
189 |
|
|
$doc->add_attr( $tag, $vals ); |
190 |
|
|
$doc->add_hidden_text( $vals ); |
191 |
dpavlin |
74 |
} |
192 |
|
|
|
193 |
|
|
my $text = $args->{'text'}; |
194 |
dpavlin |
75 |
if ( $text ) { |
195 |
dpavlin |
401 |
$text = $self->convert( $text ) or |
196 |
dpavlin |
85 |
$log->logdie("can't convert '$text' to UTF-8"); |
197 |
dpavlin |
75 |
$doc->add_text( $text ); |
198 |
|
|
} |
199 |
dpavlin |
74 |
|
200 |
|
|
$log->debug("adding ", sub { $doc->dump_draft } ); |
201 |
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()); |
202 |
dpavlin |
74 |
|
203 |
|
|
return 1; |
204 |
dpavlin |
1 |
} |
205 |
|
|
|
206 |
dpavlin |
255 |
=head2 add_link |
207 |
|
|
|
208 |
|
|
$est->add_link( |
209 |
|
|
from => 'ps', |
210 |
|
|
to => 'webpac2', |
211 |
|
|
credit => 10000, |
212 |
|
|
); |
213 |
|
|
|
214 |
dpavlin |
246 |
=cut |
215 |
|
|
|
216 |
dpavlin |
255 |
sub add_link { |
217 |
dpavlin |
238 |
my $self = shift; |
218 |
|
|
|
219 |
dpavlin |
255 |
my $args = {@_}; |
220 |
|
|
my $log = $self->_get_logger; |
221 |
dpavlin |
238 |
|
222 |
dpavlin |
459 |
my $node = first { $_->{name} eq $args->{to} } $self->master( action => 'nodelist' ); |
223 |
dpavlin |
238 |
|
224 |
dpavlin |
459 |
if (! $node) { |
225 |
|
|
$log->warn("can't find node $args->{to}, skipping link creaton"); |
226 |
|
|
return; |
227 |
|
|
} |
228 |
dpavlin |
238 |
|
229 |
dpavlin |
459 |
my $label = $node->{label}; |
230 |
dpavlin |
255 |
|
231 |
|
|
if (! $label) { |
232 |
|
|
$log->warn("can't find label for $args->{to}, skipping link creaton"); |
233 |
|
|
return; |
234 |
|
|
} |
235 |
|
|
|
236 |
|
|
$log->debug("using label $label for $args->{to}"); |
237 |
|
|
|
238 |
dpavlin |
459 |
return $self->{db}->set_link( |
239 |
|
|
$self->{masterurl} . '/node/' . $args->{to}, |
240 |
|
|
$label, |
241 |
|
|
$args->{credit}, |
242 |
dpavlin |
255 |
); |
243 |
dpavlin |
238 |
} |
244 |
|
|
|
245 |
dpavlin |
255 |
|
246 |
dpavlin |
458 |
=head2 master |
247 |
dpavlin |
255 |
|
248 |
dpavlin |
458 |
Issue administrative commands to C<estmaster> process. See documentation for |
249 |
|
|
C<master> in L<Search::Estraier>::Node. |
250 |
|
|
|
251 |
|
|
$self->master( |
252 |
|
|
action => 'nodeclr', |
253 |
|
|
name => 'foobar', |
254 |
dpavlin |
255 |
); |
255 |
|
|
|
256 |
|
|
=cut |
257 |
|
|
|
258 |
dpavlin |
458 |
sub master { |
259 |
dpavlin |
210 |
my $self = shift; |
260 |
dpavlin |
458 |
$self->{db}->master( @_ ); |
261 |
dpavlin |
210 |
} |
262 |
|
|
|
263 |
dpavlin |
255 |
|
264 |
dpavlin |
401 |
=head2 convert |
265 |
|
|
|
266 |
|
|
my $utf8_string = $self->convert('string in codepage'); |
267 |
|
|
|
268 |
|
|
=cut |
269 |
|
|
|
270 |
|
|
sub convert { |
271 |
|
|
my $self = shift; |
272 |
|
|
|
273 |
|
|
my $text = shift || return; |
274 |
|
|
from_to($text, $self->{encoding}, 'UTF-8'); |
275 |
|
|
return $text; |
276 |
|
|
} |
277 |
|
|
|
278 |
dpavlin |
1 |
=head1 AUTHOR |
279 |
|
|
|
280 |
|
|
Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >> |
281 |
|
|
|
282 |
|
|
=head1 COPYRIGHT & LICENSE |
283 |
|
|
|
284 |
|
|
Copyright 2005 Dobrica Pavlinusic, All Rights Reserved. |
285 |
|
|
|
286 |
|
|
This program is free software; you can redistribute it and/or modify it |
287 |
|
|
under the same terms as Perl itself. |
288 |
|
|
|
289 |
|
|
=cut |
290 |
|
|
|
291 |
|
|
1; # End of WebPAC::Output::Estraier |