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 |
343 |
use Search::Estraier; |
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 |
74 |
|
14 |
dpavlin |
1 |
=head1 NAME |
15 |
|
|
|
16 |
dpavlin |
74 |
WebPAC::Output::Estraier - Create Hyper Estraier full text index |
17 |
dpavlin |
1 |
|
18 |
|
|
=head1 VERSION |
19 |
|
|
|
20 |
dpavlin |
401 |
Version 0.10 |
21 |
dpavlin |
1 |
|
22 |
|
|
=cut |
23 |
|
|
|
24 |
dpavlin |
401 |
our $VERSION = '0.10'; |
25 |
dpavlin |
1 |
|
26 |
|
|
=head1 SYNOPSIS |
27 |
|
|
|
28 |
dpavlin |
74 |
Create full text index using Hyper Estraier index from data with |
29 |
|
|
type C<search>. |
30 |
dpavlin |
1 |
|
31 |
dpavlin |
74 |
=head1 FUNCTIONS |
32 |
dpavlin |
1 |
|
33 |
dpavlin |
74 |
=head2 new |
34 |
dpavlin |
1 |
|
35 |
dpavlin |
74 |
Connect to Hyper Estraier index using HTTP |
36 |
dpavlin |
1 |
|
37 |
dpavlin |
74 |
my $est = new WebPAC::Output::Estraier( |
38 |
dpavlin |
210 |
masterurl => 'http://localhost:1978/', |
39 |
dpavlin |
74 |
user => 'admin', |
40 |
|
|
passwd => 'admin', |
41 |
|
|
database => 'demo', |
42 |
dpavlin |
401 |
label => 'node label', |
43 |
dpavlin |
74 |
encoding => 'iso-8859-2', |
44 |
dpavlin |
307 |
clean => 1, |
45 |
dpavlin |
74 |
); |
46 |
dpavlin |
1 |
|
47 |
dpavlin |
74 |
Options are: |
48 |
dpavlin |
1 |
|
49 |
dpavlin |
74 |
=over 4 |
50 |
dpavlin |
1 |
|
51 |
dpavlin |
210 |
=item masterurl |
52 |
dpavlin |
1 |
|
53 |
dpavlin |
74 |
URI to C<estmaster> node |
54 |
|
|
|
55 |
|
|
=item user |
56 |
|
|
|
57 |
|
|
C<estmaster> user with sufficient rights |
58 |
|
|
|
59 |
|
|
=item passwd |
60 |
|
|
|
61 |
|
|
password for user |
62 |
|
|
|
63 |
|
|
=item database |
64 |
|
|
|
65 |
|
|
name of database from which data comes |
66 |
|
|
|
67 |
dpavlin |
401 |
=item label |
68 |
|
|
|
69 |
|
|
label for node (optional) |
70 |
|
|
|
71 |
dpavlin |
74 |
=item encoding |
72 |
|
|
|
73 |
|
|
character encoding of C<data_structure> if it's differenet than C<ISO-8859-2> |
74 |
|
|
(and it probably is). This encoding will be converted to C<UTF-8> for |
75 |
|
|
Hyper Estraier. |
76 |
|
|
|
77 |
|
|
=back |
78 |
|
|
|
79 |
|
|
Name of database will be used to form URI of documents in index. |
80 |
|
|
|
81 |
dpavlin |
1 |
=cut |
82 |
|
|
|
83 |
dpavlin |
74 |
sub new { |
84 |
|
|
my $class = shift; |
85 |
dpavlin |
343 |
my $self = {@_}; |
86 |
|
|
bless($self, $class); |
87 |
dpavlin |
74 |
|
88 |
|
|
my $log = $self->_get_logger; |
89 |
|
|
|
90 |
dpavlin |
211 |
#$log->debug("self: ", sub { Dumper($self) }); |
91 |
dpavlin |
210 |
|
92 |
|
|
foreach my $p (qw/masterurl user passwd database/) { |
93 |
dpavlin |
74 |
$log->logdie("need $p") unless ($self->{$p}); |
94 |
|
|
} |
95 |
|
|
|
96 |
dpavlin |
401 |
$self->{encoding} ||= 'ISO-8859-2'; |
97 |
|
|
|
98 |
dpavlin |
210 |
my $url = $self->{masterurl} . '/node/' . $self->{database}; |
99 |
|
|
$self->{url} = $url; |
100 |
dpavlin |
74 |
|
101 |
dpavlin |
307 |
if ($self->{clean}) { |
102 |
|
|
$log->debug("nodedel $self->{database}"); |
103 |
|
|
$self->master( action => 'nodedel', name => $self->{database} ); |
104 |
|
|
} else { |
105 |
|
|
$log->debug("opening index $self->{url}"); |
106 |
|
|
} |
107 |
dpavlin |
74 |
|
108 |
dpavlin |
255 |
my $nodes = $self->master( action => 'nodelist' ); |
109 |
dpavlin |
210 |
|
110 |
dpavlin |
211 |
$log->debug("nodes found: $nodes"); |
111 |
|
|
|
112 |
|
|
if ($nodes !~ m/^$self->{database}\t/sm) { |
113 |
dpavlin |
401 |
my $label = $self->{label} || 'WebPAC ' . $self->{database}; |
114 |
|
|
$log->warn("creating index $url ($label)"); |
115 |
dpavlin |
255 |
$self->master( |
116 |
dpavlin |
210 |
action => 'nodeadd', |
117 |
|
|
name => $self->{database}, |
118 |
dpavlin |
401 |
label => $self->convert( $label ), |
119 |
dpavlin |
210 |
) || $log->logdie("can't create Hyper Estraier node $self->{database}"); |
120 |
|
|
} |
121 |
|
|
|
122 |
dpavlin |
345 |
$self->{db} = Search::Estraier::Node->new( debug => $self->{debug} ); |
123 |
dpavlin |
343 |
$self->{db}->set_url($self->{url}); |
124 |
|
|
$self->{db}->set_auth($self->{user}, $self->{passwd}); |
125 |
dpavlin |
210 |
|
126 |
dpavlin |
401 |
$log->info("using index $self->{url} with encoding $self->{encoding}"); |
127 |
dpavlin |
74 |
|
128 |
|
|
$self ? return $self : return undef; |
129 |
dpavlin |
1 |
} |
130 |
|
|
|
131 |
dpavlin |
75 |
|
132 |
dpavlin |
74 |
=head2 add |
133 |
dpavlin |
1 |
|
134 |
dpavlin |
74 |
Adds one entry to database. |
135 |
|
|
|
136 |
|
|
$est->add( |
137 |
|
|
id => 42, |
138 |
|
|
ds => $ds, |
139 |
|
|
type => 'display', |
140 |
|
|
text => 'optional text from which snippet is created', |
141 |
|
|
); |
142 |
|
|
|
143 |
|
|
This function will create entries in index using following URI format: |
144 |
|
|
|
145 |
dpavlin |
212 |
C<file:///type/database%20name/000> |
146 |
dpavlin |
74 |
|
147 |
|
|
Each tag in C<data_structure> with specified C<type> will create one |
148 |
|
|
attribute and corresponding hidden text (used for search). |
149 |
|
|
|
150 |
dpavlin |
1 |
=cut |
151 |
|
|
|
152 |
dpavlin |
74 |
sub add { |
153 |
|
|
my $self = shift; |
154 |
|
|
|
155 |
dpavlin |
456 |
my $args = {@_}; |
156 |
dpavlin |
74 |
|
157 |
|
|
my $log = $self->_get_logger; |
158 |
|
|
|
159 |
|
|
my $database = $self->{'database'} || $log->logconfess('no database in $self'); |
160 |
|
|
$log->logconfess('need db in object') unless ($self->{'db'}); |
161 |
|
|
|
162 |
|
|
foreach my $p (qw/id ds type/) { |
163 |
|
|
$log->logdie("need $p") unless ($args->{$p}); |
164 |
|
|
} |
165 |
|
|
|
166 |
|
|
my $type = $args->{'type'}; |
167 |
dpavlin |
213 |
my $id = $args->{'id'}; |
168 |
dpavlin |
74 |
|
169 |
dpavlin |
213 |
my $uri = "file:///$type/$database/$id"; |
170 |
dpavlin |
74 |
$log->debug("creating $uri"); |
171 |
|
|
|
172 |
dpavlin |
343 |
my $doc = Search::Estraier::Document->new; |
173 |
dpavlin |
401 |
$doc->add_attr('@uri', $self->convert($uri) ); |
174 |
dpavlin |
74 |
|
175 |
|
|
$log->debug("ds = ", sub { Dumper($args->{'ds'}) } ); |
176 |
|
|
|
177 |
|
|
# filter all tags which have type defined |
178 |
|
|
my @tags = grep { |
179 |
dpavlin |
113 |
ref($args->{'ds'}->{$_}) eq 'HASH' && defined( $args->{'ds'}->{$_}->{$type} ) |
180 |
dpavlin |
74 |
} keys %{ $args->{'ds'} }; |
181 |
|
|
|
182 |
|
|
$log->debug("tags = ", join(",", @tags)); |
183 |
|
|
|
184 |
|
|
return unless (@tags); |
185 |
|
|
|
186 |
|
|
foreach my $tag (@tags) { |
187 |
|
|
|
188 |
|
|
my $vals = join(" ", @{ $args->{'ds'}->{$tag}->{$type} }); |
189 |
|
|
|
190 |
dpavlin |
374 |
next if (! $vals); |
191 |
dpavlin |
74 |
|
192 |
dpavlin |
401 |
$vals = $self->convert( $vals ) or |
193 |
dpavlin |
85 |
$log->logdie("can't convert '$vals' to UTF-8"); |
194 |
dpavlin |
75 |
|
195 |
|
|
$doc->add_attr( $tag, $vals ); |
196 |
|
|
$doc->add_hidden_text( $vals ); |
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 |
# |
213 |
|
|
# REST parametars validation data |
214 |
|
|
# |
215 |
dpavlin |
210 |
|
216 |
dpavlin |
255 |
my $estraier_rest = { |
217 |
|
|
master => { |
218 |
|
|
userdel => [ qw/name/ ], |
219 |
|
|
nodelist => [], |
220 |
|
|
nodeadd => [ qw/name label/ ], |
221 |
|
|
nodedel => [ qw/name/ ], |
222 |
|
|
}, |
223 |
|
|
node => { |
224 |
|
|
_set_link => [ qw/url label credit/ ], |
225 |
|
|
}, |
226 |
|
|
}; |
227 |
|
|
|
228 |
|
|
=head2 master |
229 |
|
|
|
230 |
dpavlin |
210 |
Issue administrative commands to C<estmaster> process and receive response |
231 |
|
|
as array of lines |
232 |
|
|
|
233 |
dpavlin |
255 |
my $nodelist = $est->master( action => 'nodelist' ); |
234 |
dpavlin |
210 |
|
235 |
|
|
=cut |
236 |
|
|
|
237 |
dpavlin |
255 |
sub master { |
238 |
|
|
my $self = shift; |
239 |
dpavlin |
210 |
|
240 |
dpavlin |
255 |
my $args = {@_}; |
241 |
|
|
my $log = $self->_get_logger; |
242 |
dpavlin |
246 |
|
243 |
dpavlin |
255 |
my $action = $args->{action} || $log->logconfess("no action specified"); |
244 |
dpavlin |
246 |
|
245 |
dpavlin |
255 |
$log->logdie("action '$action' isn't supported") unless ($estraier_rest->{master}->{$action}); |
246 |
dpavlin |
246 |
|
247 |
dpavlin |
255 |
$log->debug("master action: $action"); |
248 |
|
|
|
249 |
|
|
return $self->estcall( |
250 |
|
|
validate => 'master', |
251 |
|
|
rest_url => $self->{masterurl} . '/master?action=' . $action , |
252 |
|
|
action => $action, |
253 |
dpavlin |
284 |
%{ $args }, |
254 |
dpavlin |
255 |
); |
255 |
|
|
} |
256 |
|
|
|
257 |
|
|
=head2 add_link |
258 |
|
|
|
259 |
|
|
$est->add_link( |
260 |
|
|
from => 'ps', |
261 |
|
|
to => 'webpac2', |
262 |
|
|
credit => 10000, |
263 |
|
|
); |
264 |
|
|
|
265 |
dpavlin |
246 |
=cut |
266 |
|
|
|
267 |
dpavlin |
255 |
sub add_link { |
268 |
dpavlin |
238 |
my $self = shift; |
269 |
|
|
|
270 |
dpavlin |
255 |
my $args = {@_}; |
271 |
|
|
my $log = $self->_get_logger; |
272 |
dpavlin |
238 |
|
273 |
dpavlin |
255 |
my @labels = $self->master( action => 'nodelist' ); |
274 |
dpavlin |
238 |
|
275 |
dpavlin |
255 |
$log->debug("got labels: ", join("|", @labels)); |
276 |
dpavlin |
238 |
|
277 |
dpavlin |
303 |
@labels = grep(/^$args->{to}\t/, @labels); |
278 |
|
|
my $label = shift @labels; |
279 |
dpavlin |
307 |
(undef,$label) = split(/\t/, $label) if ($label); |
280 |
dpavlin |
255 |
|
281 |
|
|
if (! $label) { |
282 |
|
|
$log->warn("can't find label for $args->{to}, skipping link creaton"); |
283 |
|
|
return; |
284 |
|
|
} |
285 |
|
|
|
286 |
|
|
$log->debug("using label $label for $args->{to}"); |
287 |
|
|
|
288 |
|
|
return $self->estcall( |
289 |
|
|
validate => 'node', |
290 |
|
|
action => '_set_link', |
291 |
|
|
rest_url => $self->{masterurl} . '/node/' . $args->{from} . '/_set_link' , |
292 |
|
|
url => $self->{masterurl} . '/node/' . $args->{to}, |
293 |
|
|
label => $label, |
294 |
|
|
credit => $args->{credit}, |
295 |
|
|
); |
296 |
dpavlin |
238 |
} |
297 |
|
|
|
298 |
dpavlin |
255 |
=head2 estcall |
299 |
|
|
|
300 |
|
|
Workhourse which does actual calls to Hyper Estraier |
301 |
|
|
|
302 |
|
|
$self->estcall( |
303 |
|
|
rest_url => '/master?action=' . $action, |
304 |
|
|
validate => 'master', |
305 |
|
|
# ... |
306 |
|
|
); |
307 |
|
|
|
308 |
|
|
C<rest_url> is relative URL to C<estmaster> and C<validate> is entry into |
309 |
|
|
internal hash which will check if all parametars are available before |
310 |
|
|
calling function. |
311 |
|
|
|
312 |
|
|
=cut |
313 |
|
|
|
314 |
|
|
sub estcall { |
315 |
dpavlin |
210 |
my $self = shift; |
316 |
|
|
my $args = {@_}; |
317 |
|
|
my $log = $self->_get_logger; |
318 |
|
|
|
319 |
dpavlin |
255 |
$log->debug("estcall: ",Dumper($args)); |
320 |
dpavlin |
210 |
|
321 |
dpavlin |
255 |
foreach my $p (qw/rest_url validate action/) { |
322 |
|
|
$log->die("ectcall needs $p parametar") unless ($args->{$p}); |
323 |
|
|
} |
324 |
dpavlin |
210 |
|
325 |
dpavlin |
255 |
my $url = $args->{rest_url}; |
326 |
|
|
my $del = '?'; |
327 |
|
|
$del = '&' if ($url =~ m#\?#); |
328 |
dpavlin |
210 |
|
329 |
dpavlin |
255 |
my $url_args; |
330 |
dpavlin |
210 |
|
331 |
dpavlin |
255 |
foreach my $arg (@{ $estraier_rest->{ $args->{validate} }->{ $args->{action} } }) { |
332 |
|
|
$log->logdie("missing parametar $arg for action $args->{action}") unless ($args->{$arg}); |
333 |
|
|
$url_args .= $del . $arg . '=' . uri_escape( $args->{$arg} ); |
334 |
|
|
$del = '&'; |
335 |
dpavlin |
210 |
} |
336 |
|
|
|
337 |
dpavlin |
284 |
$url .= $url_args if ($url_args); |
338 |
dpavlin |
255 |
|
339 |
dpavlin |
210 |
$log->debug("calling $url"); |
340 |
|
|
|
341 |
dpavlin |
238 |
my $res = $self->est_ua()->get($url); |
342 |
dpavlin |
210 |
|
343 |
dpavlin |
211 |
if ($res->is_success) { |
344 |
|
|
#$log->debug( $res->content ); |
345 |
|
|
return split(/\n/, $res->content) if wantarray; |
346 |
dpavlin |
255 |
return $res->content || 0E0; |
347 |
dpavlin |
211 |
} else { |
348 |
|
|
$log->warn("unable to call $url: " . $res->status_line); |
349 |
dpavlin |
210 |
return; |
350 |
|
|
} |
351 |
|
|
|
352 |
|
|
} |
353 |
|
|
|
354 |
dpavlin |
255 |
=head2 est_ua |
355 |
|
|
|
356 |
|
|
This is helper function to create C<LWP::UserAgent> object with Super User |
357 |
|
|
priviledges |
358 |
|
|
|
359 |
|
|
my $ua = $self->est_ua( user => 'admin', passwd => 'admin' ); |
360 |
|
|
|
361 |
|
|
=cut |
362 |
|
|
|
363 |
dpavlin |
256 |
|
364 |
|
|
|
365 |
dpavlin |
255 |
sub est_ua { |
366 |
|
|
my $self = shift; |
367 |
|
|
|
368 |
|
|
return $self->{_master_ua} if ($self->{_master_ua}); |
369 |
|
|
|
370 |
dpavlin |
256 |
{ |
371 |
|
|
package AdminUserAgent; |
372 |
|
|
use base qw/LWP::UserAgent/; |
373 |
|
|
sub new { |
374 |
|
|
my $self = LWP::UserAgent::new(@_); |
375 |
|
|
$self->agent("webpac/$VERSION"); |
376 |
|
|
$self; |
377 |
|
|
} |
378 |
|
|
sub get_basic_credentials { |
379 |
|
|
my($self, $realm, $uri) = @_; |
380 |
|
|
return ($self->{user}, $self->{passwd}); |
381 |
|
|
} |
382 |
|
|
sub set_basic_credentials { |
383 |
|
|
my ($self, $user, $passwd) = @_; |
384 |
|
|
$self->{user} = $user; |
385 |
|
|
$self->{passwd} = $passwd; |
386 |
|
|
} |
387 |
|
|
}; |
388 |
|
|
|
389 |
|
|
$self->{_master_ua} = AdminUserAgent->new( ) || sub { |
390 |
dpavlin |
255 |
my $log = $self->_get_logger; |
391 |
|
|
$log->logdie("can't create LWP::UserAgent: $!"); |
392 |
|
|
}; |
393 |
|
|
|
394 |
dpavlin |
256 |
$self->{_master_ua}->set_basic_credentials($self->{user}, $self->{passwd}); |
395 |
dpavlin |
255 |
|
396 |
|
|
return $self->{_master_ua}; |
397 |
|
|
} |
398 |
|
|
|
399 |
dpavlin |
401 |
=head2 convert |
400 |
|
|
|
401 |
|
|
my $utf8_string = $self->convert('string in codepage'); |
402 |
|
|
|
403 |
|
|
=cut |
404 |
|
|
|
405 |
|
|
sub convert { |
406 |
|
|
my $self = shift; |
407 |
|
|
|
408 |
|
|
my $text = shift || return; |
409 |
|
|
from_to($text, $self->{encoding}, 'UTF-8'); |
410 |
|
|
return $text; |
411 |
|
|
} |
412 |
|
|
|
413 |
dpavlin |
1 |
=head1 AUTHOR |
414 |
|
|
|
415 |
|
|
Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >> |
416 |
|
|
|
417 |
|
|
=head1 COPYRIGHT & LICENSE |
418 |
|
|
|
419 |
|
|
Copyright 2005 Dobrica Pavlinusic, All Rights Reserved. |
420 |
|
|
|
421 |
|
|
This program is free software; you can redistribute it and/or modify it |
422 |
|
|
under the same terms as Perl itself. |
423 |
|
|
|
424 |
|
|
=cut |
425 |
|
|
|
426 |
|
|
1; # End of WebPAC::Output::Estraier |