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 |
|
|
|
12 |
dpavlin |
1 |
=head1 NAME |
13 |
|
|
|
14 |
dpavlin |
74 |
WebPAC::Output::Estraier - Create Hyper Estraier full text index |
15 |
dpavlin |
1 |
|
16 |
|
|
=head1 VERSION |
17 |
|
|
|
18 |
|
|
Version 0.01 |
19 |
|
|
|
20 |
|
|
=cut |
21 |
|
|
|
22 |
|
|
our $VERSION = '0.01'; |
23 |
|
|
|
24 |
|
|
=head1 SYNOPSIS |
25 |
|
|
|
26 |
dpavlin |
74 |
Create full text index using Hyper Estraier index from data with |
27 |
|
|
type C<search>. |
28 |
dpavlin |
1 |
|
29 |
dpavlin |
74 |
=head1 FUNCTIONS |
30 |
dpavlin |
1 |
|
31 |
dpavlin |
74 |
=head2 new |
32 |
dpavlin |
1 |
|
33 |
dpavlin |
74 |
Connect to Hyper Estraier index using HTTP |
34 |
dpavlin |
1 |
|
35 |
dpavlin |
74 |
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 |
dpavlin |
1 |
|
43 |
dpavlin |
74 |
Options are: |
44 |
dpavlin |
1 |
|
45 |
dpavlin |
74 |
=over 4 |
46 |
dpavlin |
1 |
|
47 |
dpavlin |
74 |
=item url |
48 |
dpavlin |
1 |
|
49 |
dpavlin |
74 |
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 |
dpavlin |
1 |
=cut |
74 |
|
|
|
75 |
dpavlin |
74 |
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 |
|
|
my $iconv = new Text::Iconv('iso-8859-2', 'utf-8'); |
95 |
|
|
|
96 |
|
|
$self ? return $self : return undef; |
97 |
dpavlin |
1 |
} |
98 |
|
|
|
99 |
dpavlin |
74 |
=head2 add |
100 |
dpavlin |
1 |
|
101 |
dpavlin |
74 |
Adds one entry to database. |
102 |
|
|
|
103 |
|
|
$est->add( |
104 |
|
|
id => 42, |
105 |
|
|
ds => $ds, |
106 |
|
|
type => 'display', |
107 |
|
|
url_prefix => 'database name', |
108 |
|
|
text => 'optional text from which snippet is created', |
109 |
|
|
); |
110 |
|
|
|
111 |
|
|
This function will create entries in index using following URI format: |
112 |
|
|
|
113 |
|
|
C<file:///database%20name/000> |
114 |
|
|
|
115 |
|
|
Each tag in C<data_structure> with specified C<type> will create one |
116 |
|
|
attribute and corresponding hidden text (used for search). |
117 |
|
|
|
118 |
dpavlin |
1 |
=cut |
119 |
|
|
|
120 |
dpavlin |
74 |
sub add { |
121 |
|
|
my $self = shift; |
122 |
|
|
|
123 |
|
|
my $args = {@_}; |
124 |
|
|
|
125 |
|
|
my $log = $self->_get_logger; |
126 |
|
|
|
127 |
|
|
my $database = $self->{'database'} || $log->logconfess('no database in $self'); |
128 |
|
|
$log->logconfess('need db in object') unless ($self->{'db'}); |
129 |
|
|
|
130 |
|
|
foreach my $p (qw/id ds type/) { |
131 |
|
|
$log->logdie("need $p") unless ($args->{$p}); |
132 |
|
|
} |
133 |
|
|
|
134 |
|
|
my $type = $args->{'type'}; |
135 |
|
|
my $mfn = $args->{'id'}; |
136 |
|
|
|
137 |
|
|
my $uri = "file:///$type/$database/$mfn"; |
138 |
|
|
$log->debug("creating $uri"); |
139 |
|
|
|
140 |
|
|
my $doc = HyperEstraier::Document->new; |
141 |
|
|
$doc->add_attr('@uri', $uri); |
142 |
|
|
|
143 |
|
|
$log->debug("ds = ", sub { Dumper($args->{'ds'}) } ); |
144 |
|
|
|
145 |
|
|
# filter all tags which have type defined |
146 |
|
|
my @tags = grep { |
147 |
|
|
defined( $args->{'ds'}->{$_}->{$type} ) |
148 |
|
|
} keys %{ $args->{'ds'} }; |
149 |
|
|
|
150 |
|
|
$log->debug("tags = ", join(",", @tags)); |
151 |
|
|
|
152 |
|
|
return unless (@tags); |
153 |
|
|
|
154 |
|
|
foreach my $tag (@tags) { |
155 |
|
|
|
156 |
|
|
my $vals = join(" ", @{ $args->{'ds'}->{$tag}->{$type} }); |
157 |
|
|
|
158 |
|
|
$log->logconfess("no values for $tag/$type") unless ($vals); |
159 |
|
|
|
160 |
|
|
$doc->add_attr($tag, $vals); |
161 |
|
|
$doc->add_hidden_text($vals); |
162 |
|
|
} |
163 |
|
|
|
164 |
|
|
my $text = $args->{'text'}; |
165 |
|
|
$doc->add_text( $text ) if ( $text ); |
166 |
|
|
|
167 |
|
|
$log->debug("adding ", sub { $doc->dump_draft } ); |
168 |
|
|
$self->{'db'}->put_doc($doc) || $log->die("can't add document $uri to index"); |
169 |
|
|
|
170 |
|
|
return 1; |
171 |
dpavlin |
1 |
} |
172 |
|
|
|
173 |
|
|
=head1 AUTHOR |
174 |
|
|
|
175 |
|
|
Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >> |
176 |
|
|
|
177 |
|
|
=head1 COPYRIGHT & LICENSE |
178 |
|
|
|
179 |
|
|
Copyright 2005 Dobrica Pavlinusic, All Rights Reserved. |
180 |
|
|
|
181 |
|
|
This program is free software; you can redistribute it and/or modify it |
182 |
|
|
under the same terms as Perl itself. |
183 |
|
|
|
184 |
|
|
=cut |
185 |
|
|
|
186 |
|
|
1; # End of WebPAC::Output::Estraier |