/[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 245 - (hide annotations)
Wed Dec 14 23:04:59 2005 UTC (18 years, 4 months ago) by dpavlin
File size: 6061 byte(s)
 r11698@llin:  dpavlin | 2005-12-15 02:04:58 +0100
 bump version number

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 245 Version 0.04
21 dpavlin 1
22     =cut
23    
24 dpavlin 245 our $VERSION = '0.04';
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     );
44 dpavlin 1
45 dpavlin 74 Options are:
46 dpavlin 1
47 dpavlin 74 =over 4
48 dpavlin 1
49 dpavlin 210 =item masterurl
50 dpavlin 1
51 dpavlin 74 URI to C<estmaster> node
52    
53     =item user
54    
55     C<estmaster> user with sufficient rights
56    
57     =item passwd
58    
59     password for user
60    
61     =item database
62    
63     name of database from which data comes
64    
65     =item encoding
66    
67     character encoding of C<data_structure> if it's differenet than C<ISO-8859-2>
68     (and it probably is). This encoding will be converted to C<UTF-8> for
69     Hyper Estraier.
70    
71     =back
72    
73     Name of database will be used to form URI of documents in index.
74    
75 dpavlin 1 =cut
76    
77 dpavlin 74 sub new {
78     my $class = shift;
79     my $self = {@_};
80     bless($self, $class);
81    
82     my $log = $self->_get_logger;
83    
84 dpavlin 211 #$log->debug("self: ", sub { Dumper($self) });
85 dpavlin 210
86     foreach my $p (qw/masterurl user passwd database/) {
87 dpavlin 74 $log->logdie("need $p") unless ($self->{$p});
88     }
89    
90 dpavlin 210 my $url = $self->{masterurl} . '/node/' . $self->{database};
91     $self->{url} = $url;
92 dpavlin 74
93 dpavlin 210 $log->info("opening Hyper Estraier index $self->{url}");
94 dpavlin 74
95 dpavlin 211 my $nodes = $self->est_master( action => 'nodelist' );
96 dpavlin 210
97 dpavlin 211 $log->debug("nodes found: $nodes");
98    
99     if ($nodes !~ m/^$self->{database}\t/sm) {
100 dpavlin 210 $log->info("creating index $url");
101     $self->est_master(
102     action => 'nodeadd',
103     name => $self->{database},
104     label => "WebPAC $self->{database}",
105     ) || $log->logdie("can't create Hyper Estraier node $self->{database}");
106     }
107    
108     $self->{'db'} = HyperEstraier::Node->new($self->{url});
109     $self->{'db'}->set_auth($self->{'user'}, $self->{passwd});
110    
111 dpavlin 74 my $encoding = $self->{'encoding'} || 'ISO-8859-2';
112     $log->info("using encoding $encoding");
113    
114 dpavlin 75 $self->{'iconv'} = new Text::Iconv($encoding, 'UTF-8') or
115 dpavlin 85 $log->logdie("can't create conversion from $encoding to UTF-8");
116 dpavlin 74
117     $self ? return $self : return undef;
118 dpavlin 1 }
119    
120 dpavlin 75
121 dpavlin 74 =head2 add
122 dpavlin 1
123 dpavlin 74 Adds one entry to database.
124    
125     $est->add(
126     id => 42,
127     ds => $ds,
128     type => 'display',
129     text => 'optional text from which snippet is created',
130     );
131    
132     This function will create entries in index using following URI format:
133    
134 dpavlin 212 C<file:///type/database%20name/000>
135 dpavlin 74
136     Each tag in C<data_structure> with specified C<type> will create one
137     attribute and corresponding hidden text (used for search).
138    
139 dpavlin 1 =cut
140    
141 dpavlin 74 sub add {
142     my $self = shift;
143    
144     my $args = {@_};
145    
146     my $log = $self->_get_logger;
147    
148     my $database = $self->{'database'} || $log->logconfess('no database in $self');
149     $log->logconfess('need db in object') unless ($self->{'db'});
150    
151     foreach my $p (qw/id ds type/) {
152     $log->logdie("need $p") unless ($args->{$p});
153     }
154    
155     my $type = $args->{'type'};
156 dpavlin 213 my $id = $args->{'id'};
157 dpavlin 74
158 dpavlin 213 my $uri = "file:///$type/$database/$id";
159 dpavlin 74 $log->debug("creating $uri");
160    
161     my $doc = HyperEstraier::Document->new;
162 dpavlin 75 $doc->add_attr('@uri', $self->{'iconv'}->convert($uri) );
163 dpavlin 74
164     $log->debug("ds = ", sub { Dumper($args->{'ds'}) } );
165    
166     # filter all tags which have type defined
167     my @tags = grep {
168 dpavlin 113 ref($args->{'ds'}->{$_}) eq 'HASH' && defined( $args->{'ds'}->{$_}->{$type} )
169 dpavlin 74 } keys %{ $args->{'ds'} };
170    
171     $log->debug("tags = ", join(",", @tags));
172    
173     return unless (@tags);
174    
175     foreach my $tag (@tags) {
176    
177     my $vals = join(" ", @{ $args->{'ds'}->{$tag}->{$type} });
178    
179     $log->logconfess("no values for $tag/$type") unless ($vals);
180    
181 dpavlin 75 $vals = $self->{'iconv'}->convert( $vals ) or
182 dpavlin 85 $log->logdie("can't convert '$vals' to UTF-8");
183 dpavlin 75
184     $doc->add_attr( $tag, $vals );
185     $doc->add_hidden_text( $vals );
186 dpavlin 74 }
187    
188     my $text = $args->{'text'};
189 dpavlin 75 if ( $text ) {
190     $text = $self->{'iconv'}->convert( $text ) or
191 dpavlin 85 $log->logdie("can't convert '$text' to UTF-8");
192 dpavlin 75 $doc->add_text( $text );
193     }
194 dpavlin 74
195     $log->debug("adding ", sub { $doc->dump_draft } );
196 dpavlin 212 $self->{'db'}->put_doc($doc) || $log->logdie("can't add document $uri to node " . $self->{url} . " status: " . $self->{db}->status());
197 dpavlin 74
198     return 1;
199 dpavlin 1 }
200    
201 dpavlin 210 =head2 est_master
202    
203     Issue administrative commands to C<estmaster> process and receive response
204     as array of lines
205    
206     my $nodelist = $self->est_master( action => nodelist );
207    
208     =cut
209    
210     my $estmaster_actions = {
211     userdel => [ qw/name/ ],
212     nodelist => [],
213     nodeadd => [ qw/name label/ ],
214     nodedel => [ qw/name/ ],
215     };
216    
217 dpavlin 238 sub est_ua {
218     my $self = shift;
219    
220     return $self->{_master_ua} if ($self->{_master_ua});
221    
222     $self->{_master_ua} = LWP::UserAgent->new( ) || sub {
223     my $log = $self->_get_logger;
224     $log->logdie("can't create LWP::UserAgent: $!");
225     };
226    
227     $self->{_master_ua}->credentials('localhost:1978','Super User', $self->{user} => $self->{passwd});
228    
229     return $self->{_master_ua};
230     }
231    
232 dpavlin 210 sub est_master {
233     my $self = shift;
234     my $args = {@_};
235     my $log = $self->_get_logger;
236    
237     $log->debug(Dumper($args));
238    
239     my $action = $args->{action} || $log->logconfess("no action specified");
240    
241     $log->logdie("action '$action' isn't supported") unless ($estmaster_actions->{$action});
242    
243     my $url = $self->{masterurl} . '/master?action=' . $action;
244    
245     foreach my $arg (@{ $estmaster_actions->{$action} }) {
246     $log->logdie("missing parametar $arg for action $action") unless ($args->{$arg});
247     $url .= '&' . $arg . '=' . uri_escape( $args->{$arg} );
248     }
249    
250     $log->debug("calling $url");
251    
252 dpavlin 238 my $res = $self->est_ua()->get($url);
253 dpavlin 210
254 dpavlin 211 if ($res->is_success) {
255     #$log->debug( $res->content );
256     return split(/\n/, $res->content) if wantarray;
257     return $res->content;
258     } else {
259     $log->warn("unable to call $url: " . $res->status_line);
260 dpavlin 210 return;
261     }
262    
263     }
264    
265 dpavlin 1 =head1 AUTHOR
266    
267     Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >>
268    
269     =head1 COPYRIGHT & LICENSE
270    
271     Copyright 2005 Dobrica Pavlinusic, All Rights Reserved.
272    
273     This program is free software; you can redistribute it and/or modify it
274     under the same terms as Perl itself.
275    
276     =cut
277    
278     1; # End of WebPAC::Output::Estraier

  ViewVC Help
Powered by ViewVC 1.1.26