/[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 319 - (hide annotations)
Sat Dec 24 11:23:17 2005 UTC (18 years, 4 months ago) by dpavlin
File size: 8638 byte(s)
warn if add_doc fails

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     use HyperEstraier;
9     use Text::Iconv;
10     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 307 Version 0.08
21 dpavlin 1
22     =cut
23    
24 dpavlin 307 our $VERSION = '0.08';
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     encoding => 'iso-8859-2',
43 dpavlin 307 clean => 1,
44 dpavlin 74 );
45 dpavlin 1
46 dpavlin 74 Options are:
47 dpavlin 1
48 dpavlin 74 =over 4
49 dpavlin 1
50 dpavlin 210 =item masterurl
51 dpavlin 1
52 dpavlin 74 URI to C<estmaster> node
53    
54     =item user
55    
56     C<estmaster> user with sufficient rights
57    
58     =item passwd
59    
60     password for user
61    
62     =item database
63    
64     name of database from which data comes
65    
66     =item encoding
67    
68     character encoding of C<data_structure> if it's differenet than C<ISO-8859-2>
69     (and it probably is). This encoding will be converted to C<UTF-8> for
70     Hyper Estraier.
71    
72     =back
73    
74     Name of database will be used to form URI of documents in index.
75    
76 dpavlin 1 =cut
77    
78 dpavlin 74 sub new {
79     my $class = shift;
80     my $self = {@_};
81     bless($self, $class);
82    
83     my $log = $self->_get_logger;
84    
85 dpavlin 211 #$log->debug("self: ", sub { Dumper($self) });
86 dpavlin 210
87     foreach my $p (qw/masterurl user passwd database/) {
88 dpavlin 74 $log->logdie("need $p") unless ($self->{$p});
89     }
90    
91 dpavlin 210 my $url = $self->{masterurl} . '/node/' . $self->{database};
92     $self->{url} = $url;
93 dpavlin 74
94 dpavlin 307 if ($self->{clean}) {
95     $log->debug("nodedel $self->{database}");
96     $self->master( action => 'nodedel', name => $self->{database} );
97     } else {
98     $log->debug("opening index $self->{url}");
99     }
100 dpavlin 74
101 dpavlin 255 my $nodes = $self->master( action => 'nodelist' );
102 dpavlin 210
103 dpavlin 211 $log->debug("nodes found: $nodes");
104    
105     if ($nodes !~ m/^$self->{database}\t/sm) {
106 dpavlin 301 $log->warn("creating index $url");
107 dpavlin 255 $self->master(
108 dpavlin 210 action => 'nodeadd',
109     name => $self->{database},
110     label => "WebPAC $self->{database}",
111     ) || $log->logdie("can't create Hyper Estraier node $self->{database}");
112     }
113    
114     $self->{'db'} = HyperEstraier::Node->new($self->{url});
115     $self->{'db'}->set_auth($self->{'user'}, $self->{passwd});
116    
117 dpavlin 74 my $encoding = $self->{'encoding'} || 'ISO-8859-2';
118 dpavlin 301 $log->info("using index $self->{url} with encoding $encoding");
119 dpavlin 74
120 dpavlin 75 $self->{'iconv'} = new Text::Iconv($encoding, 'UTF-8') or
121 dpavlin 85 $log->logdie("can't create conversion from $encoding to UTF-8");
122 dpavlin 74
123     $self ? return $self : return undef;
124 dpavlin 1 }
125    
126 dpavlin 75
127 dpavlin 74 =head2 add
128 dpavlin 1
129 dpavlin 74 Adds one entry to database.
130    
131     $est->add(
132     id => 42,
133     ds => $ds,
134     type => 'display',
135     text => 'optional text from which snippet is created',
136     );
137    
138     This function will create entries in index using following URI format:
139    
140 dpavlin 212 C<file:///type/database%20name/000>
141 dpavlin 74
142     Each tag in C<data_structure> with specified C<type> will create one
143     attribute and corresponding hidden text (used for search).
144    
145 dpavlin 1 =cut
146    
147 dpavlin 74 sub add {
148     my $self = shift;
149    
150     my $args = {@_};
151    
152     my $log = $self->_get_logger;
153    
154     my $database = $self->{'database'} || $log->logconfess('no database in $self');
155     $log->logconfess('need db in object') unless ($self->{'db'});
156    
157     foreach my $p (qw/id ds type/) {
158     $log->logdie("need $p") unless ($args->{$p});
159     }
160    
161     my $type = $args->{'type'};
162 dpavlin 213 my $id = $args->{'id'};
163 dpavlin 74
164 dpavlin 213 my $uri = "file:///$type/$database/$id";
165 dpavlin 74 $log->debug("creating $uri");
166    
167     my $doc = HyperEstraier::Document->new;
168 dpavlin 75 $doc->add_attr('@uri', $self->{'iconv'}->convert($uri) );
169 dpavlin 74
170     $log->debug("ds = ", sub { Dumper($args->{'ds'}) } );
171    
172     # filter all tags which have type defined
173     my @tags = grep {
174 dpavlin 113 ref($args->{'ds'}->{$_}) eq 'HASH' && defined( $args->{'ds'}->{$_}->{$type} )
175 dpavlin 74 } keys %{ $args->{'ds'} };
176    
177     $log->debug("tags = ", join(",", @tags));
178    
179     return unless (@tags);
180    
181     foreach my $tag (@tags) {
182    
183     my $vals = join(" ", @{ $args->{'ds'}->{$tag}->{$type} });
184    
185     $log->logconfess("no values for $tag/$type") unless ($vals);
186    
187 dpavlin 75 $vals = $self->{'iconv'}->convert( $vals ) or
188 dpavlin 85 $log->logdie("can't convert '$vals' to UTF-8");
189 dpavlin 75
190     $doc->add_attr( $tag, $vals );
191     $doc->add_hidden_text( $vals );
192 dpavlin 74 }
193    
194     my $text = $args->{'text'};
195 dpavlin 75 if ( $text ) {
196     $text = $self->{'iconv'}->convert( $text ) or
197 dpavlin 85 $log->logdie("can't convert '$text' to UTF-8");
198 dpavlin 75 $doc->add_text( $text );
199     }
200 dpavlin 74
201     $log->debug("adding ", sub { $doc->dump_draft } );
202 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());
203 dpavlin 74
204     return 1;
205 dpavlin 1 }
206    
207 dpavlin 255 #
208     # REST parametars validation data
209     #
210 dpavlin 210
211 dpavlin 255 my $estraier_rest = {
212     master => {
213     userdel => [ qw/name/ ],
214     nodelist => [],
215     nodeadd => [ qw/name label/ ],
216     nodedel => [ qw/name/ ],
217     },
218     node => {
219     _set_link => [ qw/url label credit/ ],
220     },
221     };
222    
223     =head2 master
224    
225 dpavlin 210 Issue administrative commands to C<estmaster> process and receive response
226     as array of lines
227    
228 dpavlin 255 my $nodelist = $est->master( action => 'nodelist' );
229 dpavlin 210
230     =cut
231    
232 dpavlin 255 sub master {
233     my $self = shift;
234 dpavlin 210
235 dpavlin 255 my $args = {@_};
236     my $log = $self->_get_logger;
237 dpavlin 246
238 dpavlin 255 my $action = $args->{action} || $log->logconfess("no action specified");
239 dpavlin 246
240 dpavlin 255 $log->logdie("action '$action' isn't supported") unless ($estraier_rest->{master}->{$action});
241 dpavlin 246
242 dpavlin 255 $log->debug("master action: $action");
243    
244     return $self->estcall(
245     validate => 'master',
246     rest_url => $self->{masterurl} . '/master?action=' . $action ,
247     action => $action,
248 dpavlin 284 %{ $args },
249 dpavlin 255 );
250     }
251    
252     =head2 add_link
253    
254     $est->add_link(
255     from => 'ps',
256     to => 'webpac2',
257     credit => 10000,
258     );
259    
260 dpavlin 246 =cut
261    
262 dpavlin 255 sub add_link {
263 dpavlin 238 my $self = shift;
264    
265 dpavlin 255 my $args = {@_};
266     my $log = $self->_get_logger;
267 dpavlin 238
268 dpavlin 255 my @labels = $self->master( action => 'nodelist' );
269 dpavlin 238
270 dpavlin 255 $log->debug("got labels: ", join("|", @labels));
271 dpavlin 238
272 dpavlin 303 @labels = grep(/^$args->{to}\t/, @labels);
273     my $label = shift @labels;
274 dpavlin 307 (undef,$label) = split(/\t/, $label) if ($label);
275 dpavlin 255
276     if (! $label) {
277     $log->warn("can't find label for $args->{to}, skipping link creaton");
278     return;
279     }
280    
281     $log->debug("using label $label for $args->{to}");
282    
283     return $self->estcall(
284     validate => 'node',
285     action => '_set_link',
286     rest_url => $self->{masterurl} . '/node/' . $args->{from} . '/_set_link' ,
287     url => $self->{masterurl} . '/node/' . $args->{to},
288     label => $label,
289     credit => $args->{credit},
290     );
291 dpavlin 238 }
292    
293 dpavlin 255 =head2 estcall
294    
295     Workhourse which does actual calls to Hyper Estraier
296    
297     $self->estcall(
298     rest_url => '/master?action=' . $action,
299     validate => 'master',
300     # ...
301     );
302    
303     C<rest_url> is relative URL to C<estmaster> and C<validate> is entry into
304     internal hash which will check if all parametars are available before
305     calling function.
306    
307     =cut
308    
309     sub estcall {
310 dpavlin 210 my $self = shift;
311     my $args = {@_};
312     my $log = $self->_get_logger;
313    
314 dpavlin 255 $log->debug("estcall: ",Dumper($args));
315 dpavlin 210
316 dpavlin 255 foreach my $p (qw/rest_url validate action/) {
317     $log->die("ectcall needs $p parametar") unless ($args->{$p});
318     }
319 dpavlin 210
320 dpavlin 255 my $url = $args->{rest_url};
321     my $del = '?';
322     $del = '&' if ($url =~ m#\?#);
323 dpavlin 210
324 dpavlin 255 my $url_args;
325 dpavlin 210
326 dpavlin 255 foreach my $arg (@{ $estraier_rest->{ $args->{validate} }->{ $args->{action} } }) {
327     $log->logdie("missing parametar $arg for action $args->{action}") unless ($args->{$arg});
328     $url_args .= $del . $arg . '=' . uri_escape( $args->{$arg} );
329     $del = '&';
330 dpavlin 210 }
331    
332 dpavlin 284 $url .= $url_args if ($url_args);
333 dpavlin 255
334 dpavlin 210 $log->debug("calling $url");
335    
336 dpavlin 238 my $res = $self->est_ua()->get($url);
337 dpavlin 210
338 dpavlin 211 if ($res->is_success) {
339     #$log->debug( $res->content );
340     return split(/\n/, $res->content) if wantarray;
341 dpavlin 255 return $res->content || 0E0;
342 dpavlin 211 } else {
343     $log->warn("unable to call $url: " . $res->status_line);
344 dpavlin 210 return;
345     }
346    
347     }
348    
349 dpavlin 255 =head2 est_ua
350    
351     This is helper function to create C<LWP::UserAgent> object with Super User
352     priviledges
353    
354     my $ua = $self->est_ua( user => 'admin', passwd => 'admin' );
355    
356     =cut
357    
358 dpavlin 256
359    
360 dpavlin 255 sub est_ua {
361     my $self = shift;
362    
363     return $self->{_master_ua} if ($self->{_master_ua});
364    
365 dpavlin 256 {
366     package AdminUserAgent;
367     use base qw/LWP::UserAgent/;
368     sub new {
369     my $self = LWP::UserAgent::new(@_);
370     $self->agent("webpac/$VERSION");
371     $self;
372     }
373     sub get_basic_credentials {
374     my($self, $realm, $uri) = @_;
375     return ($self->{user}, $self->{passwd});
376     }
377     sub set_basic_credentials {
378     my ($self, $user, $passwd) = @_;
379     $self->{user} = $user;
380     $self->{passwd} = $passwd;
381     }
382     };
383    
384     $self->{_master_ua} = AdminUserAgent->new( ) || sub {
385 dpavlin 255 my $log = $self->_get_logger;
386     $log->logdie("can't create LWP::UserAgent: $!");
387     };
388    
389 dpavlin 256 $self->{_master_ua}->set_basic_credentials($self->{user}, $self->{passwd});
390 dpavlin 255
391     return $self->{_master_ua};
392     }
393    
394 dpavlin 1 =head1 AUTHOR
395    
396     Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >>
397    
398     =head1 COPYRIGHT & LICENSE
399    
400     Copyright 2005 Dobrica Pavlinusic, All Rights Reserved.
401    
402     This program is free software; you can redistribute it and/or modify it
403     under the same terms as Perl itself.
404    
405     =cut
406    
407     1; # End of WebPAC::Output::Estraier

  ViewVC Help
Powered by ViewVC 1.1.26