| 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 |
use LWP; |
| 12 |
use URI::Escape; |
| 13 |
|
| 14 |
=head1 NAME |
| 15 |
|
| 16 |
WebPAC::Output::Estraier - Create Hyper Estraier full text index |
| 17 |
|
| 18 |
=head1 VERSION |
| 19 |
|
| 20 |
Version 0.03 |
| 21 |
|
| 22 |
=cut |
| 23 |
|
| 24 |
our $VERSION = '0.03'; |
| 25 |
|
| 26 |
=head1 SYNOPSIS |
| 27 |
|
| 28 |
Create full text index using Hyper Estraier index from data with |
| 29 |
type C<search>. |
| 30 |
|
| 31 |
=head1 FUNCTIONS |
| 32 |
|
| 33 |
=head2 new |
| 34 |
|
| 35 |
Connect to Hyper Estraier index using HTTP |
| 36 |
|
| 37 |
my $est = new WebPAC::Output::Estraier( |
| 38 |
masterurl => 'http://localhost:1978/', |
| 39 |
user => 'admin', |
| 40 |
passwd => 'admin', |
| 41 |
database => 'demo', |
| 42 |
encoding => 'iso-8859-2', |
| 43 |
); |
| 44 |
|
| 45 |
Options are: |
| 46 |
|
| 47 |
=over 4 |
| 48 |
|
| 49 |
=item masterurl |
| 50 |
|
| 51 |
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 |
=cut |
| 76 |
|
| 77 |
sub new { |
| 78 |
my $class = shift; |
| 79 |
my $self = {@_}; |
| 80 |
bless($self, $class); |
| 81 |
|
| 82 |
my $log = $self->_get_logger; |
| 83 |
|
| 84 |
#$log->debug("self: ", sub { Dumper($self) }); |
| 85 |
|
| 86 |
foreach my $p (qw/masterurl user passwd database/) { |
| 87 |
$log->logdie("need $p") unless ($self->{$p}); |
| 88 |
} |
| 89 |
|
| 90 |
my $url = $self->{masterurl} . '/node/' . $self->{database}; |
| 91 |
$url =~ s#//#/#g; |
| 92 |
$self->{url} = $url; |
| 93 |
|
| 94 |
$log->info("opening Hyper Estraier index $self->{url}"); |
| 95 |
|
| 96 |
my $nodes = $self->est_master( action => 'nodelist' ); |
| 97 |
|
| 98 |
$log->debug("nodes found: $nodes"); |
| 99 |
|
| 100 |
if ($nodes !~ m/^$self->{database}\t/sm) { |
| 101 |
$log->info("creating index $url"); |
| 102 |
$self->est_master( |
| 103 |
action => 'nodeadd', |
| 104 |
name => $self->{database}, |
| 105 |
label => "WebPAC $self->{database}", |
| 106 |
) || $log->logdie("can't create Hyper Estraier node $self->{database}"); |
| 107 |
} |
| 108 |
|
| 109 |
$self->{'db'} = HyperEstraier::Node->new($self->{url}); |
| 110 |
$self->{'db'}->set_auth($self->{'user'}, $self->{passwd}); |
| 111 |
|
| 112 |
my $encoding = $self->{'encoding'} || 'ISO-8859-2'; |
| 113 |
$log->info("using encoding $encoding"); |
| 114 |
|
| 115 |
$self->{'iconv'} = new Text::Iconv($encoding, 'UTF-8') or |
| 116 |
$log->logdie("can't create conversion from $encoding to UTF-8"); |
| 117 |
|
| 118 |
$self ? return $self : return undef; |
| 119 |
} |
| 120 |
|
| 121 |
|
| 122 |
=head2 add |
| 123 |
|
| 124 |
Adds one entry to database. |
| 125 |
|
| 126 |
$est->add( |
| 127 |
id => 42, |
| 128 |
ds => $ds, |
| 129 |
type => 'display', |
| 130 |
url_prefix => 'database name', |
| 131 |
text => 'optional text from which snippet is created', |
| 132 |
); |
| 133 |
|
| 134 |
This function will create entries in index using following URI format: |
| 135 |
|
| 136 |
C<file:///database%20name/000> |
| 137 |
|
| 138 |
Each tag in C<data_structure> with specified C<type> will create one |
| 139 |
attribute and corresponding hidden text (used for search). |
| 140 |
|
| 141 |
=cut |
| 142 |
|
| 143 |
sub add { |
| 144 |
my $self = shift; |
| 145 |
|
| 146 |
my $args = {@_}; |
| 147 |
|
| 148 |
my $log = $self->_get_logger; |
| 149 |
|
| 150 |
my $database = $self->{'database'} || $log->logconfess('no database in $self'); |
| 151 |
$log->logconfess('need db in object') unless ($self->{'db'}); |
| 152 |
|
| 153 |
foreach my $p (qw/id ds type/) { |
| 154 |
$log->logdie("need $p") unless ($args->{$p}); |
| 155 |
} |
| 156 |
|
| 157 |
my $type = $args->{'type'}; |
| 158 |
my $mfn = $args->{'id'}; |
| 159 |
|
| 160 |
my $uri = "file:///$type/$database/$mfn"; |
| 161 |
$log->debug("creating $uri"); |
| 162 |
|
| 163 |
my $doc = HyperEstraier::Document->new; |
| 164 |
$doc->add_attr('@uri', $self->{'iconv'}->convert($uri) ); |
| 165 |
|
| 166 |
$log->debug("ds = ", sub { Dumper($args->{'ds'}) } ); |
| 167 |
|
| 168 |
# filter all tags which have type defined |
| 169 |
my @tags = grep { |
| 170 |
ref($args->{'ds'}->{$_}) eq 'HASH' && defined( $args->{'ds'}->{$_}->{$type} ) |
| 171 |
} keys %{ $args->{'ds'} }; |
| 172 |
|
| 173 |
$log->debug("tags = ", join(",", @tags)); |
| 174 |
|
| 175 |
return unless (@tags); |
| 176 |
|
| 177 |
foreach my $tag (@tags) { |
| 178 |
|
| 179 |
my $vals = join(" ", @{ $args->{'ds'}->{$tag}->{$type} }); |
| 180 |
|
| 181 |
$log->logconfess("no values for $tag/$type") unless ($vals); |
| 182 |
|
| 183 |
$vals = $self->{'iconv'}->convert( $vals ) or |
| 184 |
$log->logdie("can't convert '$vals' to UTF-8"); |
| 185 |
|
| 186 |
$doc->add_attr( $tag, $vals ); |
| 187 |
$doc->add_hidden_text( $vals ); |
| 188 |
} |
| 189 |
|
| 190 |
my $text = $args->{'text'}; |
| 191 |
if ( $text ) { |
| 192 |
$text = $self->{'iconv'}->convert( $text ) or |
| 193 |
$log->logdie("can't convert '$text' to UTF-8"); |
| 194 |
$doc->add_text( $text ); |
| 195 |
} |
| 196 |
|
| 197 |
$log->debug("adding ", sub { $doc->dump_draft } ); |
| 198 |
$self->{'db'}->put_doc($doc) || $log->logdie("can't add document $uri to index"); |
| 199 |
|
| 200 |
return 1; |
| 201 |
} |
| 202 |
|
| 203 |
=head2 est_master |
| 204 |
|
| 205 |
Issue administrative commands to C<estmaster> process and receive response |
| 206 |
as array of lines |
| 207 |
|
| 208 |
my $nodelist = $self->est_master( action => nodelist ); |
| 209 |
|
| 210 |
=cut |
| 211 |
|
| 212 |
my $estmaster_actions = { |
| 213 |
userdel => [ qw/name/ ], |
| 214 |
nodelist => [], |
| 215 |
nodeadd => [ qw/name label/ ], |
| 216 |
nodedel => [ qw/name/ ], |
| 217 |
}; |
| 218 |
|
| 219 |
sub est_master { |
| 220 |
my $self = shift; |
| 221 |
my $args = {@_}; |
| 222 |
my $log = $self->_get_logger; |
| 223 |
|
| 224 |
$log->debug(Dumper($args)); |
| 225 |
|
| 226 |
my $action = $args->{action} || $log->logconfess("no action specified"); |
| 227 |
|
| 228 |
$log->logdie("action '$action' isn't supported") unless ($estmaster_actions->{$action}); |
| 229 |
|
| 230 |
my $url = $self->{masterurl} . '/master?action=' . $action; |
| 231 |
|
| 232 |
foreach my $arg (@{ $estmaster_actions->{$action} }) { |
| 233 |
$log->logdie("missing parametar $arg for action $action") unless ($args->{$arg}); |
| 234 |
$url .= '&' . $arg . '=' . uri_escape( $args->{$arg} ); |
| 235 |
} |
| 236 |
|
| 237 |
$log->debug("calling $url"); |
| 238 |
|
| 239 |
if (! $self->{_master_ua}) { |
| 240 |
$self->{_master_ua} = LWP::UserAgent->new( ) || $log->logdie("can't create LWP::UserAgent: $!"); |
| 241 |
$self->{_master_ua}->credentials('localhost:1978','Super User', $self->{user} => $self->{passwd}); |
| 242 |
} |
| 243 |
|
| 244 |
my $res = $self->{_master_ua}->get($url); |
| 245 |
|
| 246 |
if ($res->is_success) { |
| 247 |
#$log->debug( $res->content ); |
| 248 |
return split(/\n/, $res->content) if wantarray; |
| 249 |
return $res->content; |
| 250 |
} else { |
| 251 |
$log->warn("unable to call $url: " . $res->status_line); |
| 252 |
#$log->debug(Dumper($res, $self->{'_master_ua'})); |
| 253 |
return; |
| 254 |
} |
| 255 |
|
| 256 |
} |
| 257 |
|
| 258 |
=head1 AUTHOR |
| 259 |
|
| 260 |
Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >> |
| 261 |
|
| 262 |
=head1 COPYRIGHT & LICENSE |
| 263 |
|
| 264 |
Copyright 2005 Dobrica Pavlinusic, All Rights Reserved. |
| 265 |
|
| 266 |
This program is free software; you can redistribute it and/or modify it |
| 267 |
under the same terms as Perl itself. |
| 268 |
|
| 269 |
=cut |
| 270 |
|
| 271 |
1; # End of WebPAC::Output::Estraier |
Parent Directory
| 
Revision Log