1 |
package WebPAC::Output::Estraier; |
2 |
|
3 |
use warnings; |
4 |
use strict; |
5 |
|
6 |
use base qw/WebPAC::Common/; |
7 |
|
8 |
use HyperEstraier; |
9 |
use Text::Iconv; |
10 |
use Data::Dumper; |
11 |
|
12 |
=head1 NAME |
13 |
|
14 |
WebPAC::Output::Estraier - Create Hyper Estraier full text index |
15 |
|
16 |
=head1 VERSION |
17 |
|
18 |
Version 0.01 |
19 |
|
20 |
=cut |
21 |
|
22 |
our $VERSION = '0.01'; |
23 |
|
24 |
=head1 SYNOPSIS |
25 |
|
26 |
Create full text index using Hyper Estraier index from data with |
27 |
type C<search>. |
28 |
|
29 |
=head1 FUNCTIONS |
30 |
|
31 |
=head2 new |
32 |
|
33 |
Connect to Hyper Estraier index using HTTP |
34 |
|
35 |
my $est = new WebPAC::Output::Estraier( |
36 |
url => 'http://localhost:1978/node/webpac2', |
37 |
user => 'admin', |
38 |
passwd => 'admin', |
39 |
database => 'demo', |
40 |
encoding => 'iso-8859-2', |
41 |
); |
42 |
|
43 |
Options are: |
44 |
|
45 |
=over 4 |
46 |
|
47 |
=item url |
48 |
|
49 |
URI to C<estmaster> node |
50 |
|
51 |
=item user |
52 |
|
53 |
C<estmaster> user with sufficient rights |
54 |
|
55 |
=item passwd |
56 |
|
57 |
password for user |
58 |
|
59 |
=item database |
60 |
|
61 |
name of database from which data comes |
62 |
|
63 |
=item encoding |
64 |
|
65 |
character encoding of C<data_structure> if it's differenet than C<ISO-8859-2> |
66 |
(and it probably is). This encoding will be converted to C<UTF-8> for |
67 |
Hyper Estraier. |
68 |
|
69 |
=back |
70 |
|
71 |
Name of database will be used to form URI of documents in index. |
72 |
|
73 |
=cut |
74 |
|
75 |
sub new { |
76 |
my $class = shift; |
77 |
my $self = {@_}; |
78 |
bless($self, $class); |
79 |
|
80 |
my $log = $self->_get_logger; |
81 |
|
82 |
foreach my $p (qw/url user passwd/) { |
83 |
$log->logdie("need $p") unless ($self->{$p}); |
84 |
} |
85 |
|
86 |
$log->info("opening Hyper Estraier index $self->{'url'}"); |
87 |
|
88 |
$self->{'db'} = HyperEstraier::Node->new($self->{'url'}); |
89 |
$self->{'db'}->set_auth($self->{'user'}, $self->{'passwd'}); |
90 |
|
91 |
my $encoding = $self->{'encoding'} || 'ISO-8859-2'; |
92 |
$log->info("using encoding $encoding"); |
93 |
|
94 |
$self->{'iconv'} = new Text::Iconv($encoding, 'UTF-8') or |
95 |
$log->logdie("can't create conversion from $encoding to UTF-8"); |
96 |
|
97 |
$self ? return $self : return undef; |
98 |
} |
99 |
|
100 |
|
101 |
=head2 add |
102 |
|
103 |
Adds one entry to database. |
104 |
|
105 |
$est->add( |
106 |
id => 42, |
107 |
ds => $ds, |
108 |
type => 'display', |
109 |
url_prefix => 'database name', |
110 |
text => 'optional text from which snippet is created', |
111 |
); |
112 |
|
113 |
This function will create entries in index using following URI format: |
114 |
|
115 |
C<file:///database%20name/000> |
116 |
|
117 |
Each tag in C<data_structure> with specified C<type> will create one |
118 |
attribute and corresponding hidden text (used for search). |
119 |
|
120 |
=cut |
121 |
|
122 |
sub add { |
123 |
my $self = shift; |
124 |
|
125 |
my $args = {@_}; |
126 |
|
127 |
my $log = $self->_get_logger; |
128 |
|
129 |
my $database = $self->{'database'} || $log->logconfess('no database in $self'); |
130 |
$log->logconfess('need db in object') unless ($self->{'db'}); |
131 |
|
132 |
foreach my $p (qw/id ds type/) { |
133 |
$log->logdie("need $p") unless ($args->{$p}); |
134 |
} |
135 |
|
136 |
my $type = $args->{'type'}; |
137 |
my $mfn = $args->{'id'}; |
138 |
|
139 |
my $uri = "file:///$type/$database/$mfn"; |
140 |
$log->debug("creating $uri"); |
141 |
|
142 |
my $doc = HyperEstraier::Document->new; |
143 |
$doc->add_attr('@uri', $self->{'iconv'}->convert($uri) ); |
144 |
|
145 |
$log->debug("ds = ", sub { Dumper($args->{'ds'}) } ); |
146 |
|
147 |
# filter all tags which have type defined |
148 |
my @tags = grep { |
149 |
defined( $args->{'ds'}->{$_}->{$type} ) |
150 |
} keys %{ $args->{'ds'} }; |
151 |
|
152 |
$log->debug("tags = ", join(",", @tags)); |
153 |
|
154 |
return unless (@tags); |
155 |
|
156 |
foreach my $tag (@tags) { |
157 |
|
158 |
my $vals = join(" ", @{ $args->{'ds'}->{$tag}->{$type} }); |
159 |
|
160 |
$log->logconfess("no values for $tag/$type") unless ($vals); |
161 |
|
162 |
$vals = $self->{'iconv'}->convert( $vals ) or |
163 |
$log->logdie("can't convert '$vals' to UTF-8"); |
164 |
|
165 |
$doc->add_attr( $tag, $vals ); |
166 |
$doc->add_hidden_text( $vals ); |
167 |
} |
168 |
|
169 |
my $text = $args->{'text'}; |
170 |
if ( $text ) { |
171 |
$text = $self->{'iconv'}->convert( $text ) or |
172 |
$log->logdie("can't convert '$text' to UTF-8"); |
173 |
$doc->add_text( $text ); |
174 |
} |
175 |
|
176 |
$log->debug("adding ", sub { $doc->dump_draft } ); |
177 |
$self->{'db'}->put_doc($doc) || $log->logdie("can't add document $uri to index"); |
178 |
|
179 |
return 1; |
180 |
} |
181 |
|
182 |
=head1 AUTHOR |
183 |
|
184 |
Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >> |
185 |
|
186 |
=head1 COPYRIGHT & LICENSE |
187 |
|
188 |
Copyright 2005 Dobrica Pavlinusic, All Rights Reserved. |
189 |
|
190 |
This program is free software; you can redistribute it and/or modify it |
191 |
under the same terms as Perl itself. |
192 |
|
193 |
=cut |
194 |
|
195 |
1; # End of WebPAC::Output::Estraier |