| 1 |
package WebPAC::Output::Estraier; |
| 2 |
|
| 3 |
use warnings; |
| 4 |
use strict; |
| 5 |
|
| 6 |
use base qw/WebPAC::Common/; |
| 7 |
|
| 8 |
use Search::Estraier; |
| 9 |
use Encode qw/from_to/; |
| 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 0.10 |
| 21 |
|
| 22 |
=cut |
| 23 |
|
| 24 |
our $VERSION = '0.10'; |
| 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 |
label => 'node label', |
| 43 |
encoding => 'iso-8859-2', |
| 44 |
clean => 1, |
| 45 |
); |
| 46 |
|
| 47 |
Options are: |
| 48 |
|
| 49 |
=over 4 |
| 50 |
|
| 51 |
=item masterurl |
| 52 |
|
| 53 |
URI to C<estmaster> node |
| 54 |
|
| 55 |
=item user |
| 56 |
|
| 57 |
C<estmaster> user with sufficient rights |
| 58 |
|
| 59 |
=item passwd |
| 60 |
|
| 61 |
password for user |
| 62 |
|
| 63 |
=item database |
| 64 |
|
| 65 |
name of database from which data comes |
| 66 |
|
| 67 |
=item label |
| 68 |
|
| 69 |
label for node (optional) |
| 70 |
|
| 71 |
=item encoding |
| 72 |
|
| 73 |
character encoding of C<data_structure> if it's differenet than C<ISO-8859-2> |
| 74 |
(and it probably is). This encoding will be converted to C<UTF-8> for |
| 75 |
Hyper Estraier. |
| 76 |
|
| 77 |
=back |
| 78 |
|
| 79 |
Name of database will be used to form URI of documents in index. |
| 80 |
|
| 81 |
=cut |
| 82 |
|
| 83 |
sub new { |
| 84 |
my $class = shift; |
| 85 |
my $self = {@_}; |
| 86 |
bless($self, $class); |
| 87 |
|
| 88 |
my $log = $self->_get_logger; |
| 89 |
|
| 90 |
#$log->debug("self: ", sub { Dumper($self) }); |
| 91 |
|
| 92 |
foreach my $p (qw/masterurl user passwd database/) { |
| 93 |
$log->logdie("need $p") unless ($self->{$p}); |
| 94 |
} |
| 95 |
|
| 96 |
$self->{encoding} ||= 'ISO-8859-2'; |
| 97 |
|
| 98 |
my $url = $self->{masterurl} . '/node/' . $self->{database}; |
| 99 |
$self->{url} = $url; |
| 100 |
|
| 101 |
if ($self->{clean}) { |
| 102 |
$log->debug("nodedel $self->{database}"); |
| 103 |
$self->master( action => 'nodedel', name => $self->{database} ); |
| 104 |
} else { |
| 105 |
$log->debug("opening index $self->{url}"); |
| 106 |
} |
| 107 |
|
| 108 |
my $nodes = $self->master( action => 'nodelist' ); |
| 109 |
|
| 110 |
$log->debug("nodes found: $nodes"); |
| 111 |
|
| 112 |
if ($nodes !~ m/^$self->{database}\t/sm) { |
| 113 |
my $label = $self->{label} || 'WebPAC ' . $self->{database}; |
| 114 |
$log->warn("creating index $url ($label)"); |
| 115 |
$self->master( |
| 116 |
action => 'nodeadd', |
| 117 |
name => $self->{database}, |
| 118 |
label => $self->convert( $label ), |
| 119 |
) || $log->logdie("can't create Hyper Estraier node $self->{database}"); |
| 120 |
} |
| 121 |
|
| 122 |
$self->{db} = Search::Estraier::Node->new( debug => $self->{debug} ); |
| 123 |
$self->{db}->set_url($self->{url}); |
| 124 |
$self->{db}->set_auth($self->{user}, $self->{passwd}); |
| 125 |
|
| 126 |
$log->info("using index $self->{url} with encoding $self->{encoding}"); |
| 127 |
|
| 128 |
$self ? return $self : return undef; |
| 129 |
} |
| 130 |
|
| 131 |
|
| 132 |
=head2 add |
| 133 |
|
| 134 |
Adds one entry to database. |
| 135 |
|
| 136 |
$est->add( |
| 137 |
id => 42, |
| 138 |
ds => $ds, |
| 139 |
type => 'display', |
| 140 |
text => 'optional text from which snippet is created', |
| 141 |
); |
| 142 |
|
| 143 |
This function will create entries in index using following URI format: |
| 144 |
|
| 145 |
C<file:///type/database%20name/000> |
| 146 |
|
| 147 |
Each tag in C<data_structure> with specified C<type> will create one |
| 148 |
attribute and corresponding hidden text (used for search). |
| 149 |
|
| 150 |
=cut |
| 151 |
|
| 152 |
sub add { |
| 153 |
my $self = shift; |
| 154 |
|
| 155 |
my $args = {@_}; |
| 156 |
|
| 157 |
my $log = $self->_get_logger; |
| 158 |
|
| 159 |
my $database = $self->{'database'} || $log->logconfess('no database in $self'); |
| 160 |
$log->logconfess('need db in object') unless ($self->{'db'}); |
| 161 |
|
| 162 |
foreach my $p (qw/id ds type/) { |
| 163 |
$log->logdie("need $p") unless ($args->{$p}); |
| 164 |
} |
| 165 |
|
| 166 |
my $type = $args->{'type'}; |
| 167 |
my $id = $args->{'id'}; |
| 168 |
|
| 169 |
my $uri = "file:///$type/$database/$id"; |
| 170 |
$log->debug("creating $uri"); |
| 171 |
|
| 172 |
my $doc = Search::Estraier::Document->new; |
| 173 |
$doc->add_attr('@uri', $self->convert($uri) ); |
| 174 |
|
| 175 |
$log->debug("ds = ", sub { Dumper($args->{'ds'}) } ); |
| 176 |
|
| 177 |
# filter all tags which have type defined |
| 178 |
my @tags = grep { |
| 179 |
ref($args->{'ds'}->{$_}) eq 'HASH' && defined( $args->{'ds'}->{$_}->{$type} ) |
| 180 |
} keys %{ $args->{'ds'} }; |
| 181 |
|
| 182 |
$log->debug("tags = ", join(",", @tags)); |
| 183 |
|
| 184 |
return unless (@tags); |
| 185 |
|
| 186 |
foreach my $tag (@tags) { |
| 187 |
|
| 188 |
my $vals = join(" ", @{ $args->{'ds'}->{$tag}->{$type} }); |
| 189 |
|
| 190 |
next if (! $vals); |
| 191 |
|
| 192 |
$vals = $self->convert( $vals ) or |
| 193 |
$log->logdie("can't convert '$vals' to UTF-8"); |
| 194 |
|
| 195 |
$doc->add_attr( $tag, $vals ); |
| 196 |
$doc->add_hidden_text( $vals ); |
| 197 |
} |
| 198 |
|
| 199 |
my $text = $args->{'text'}; |
| 200 |
if ( $text ) { |
| 201 |
$text = $self->convert( $text ) or |
| 202 |
$log->logdie("can't convert '$text' to UTF-8"); |
| 203 |
$doc->add_text( $text ); |
| 204 |
} |
| 205 |
|
| 206 |
$log->debug("adding ", sub { $doc->dump_draft } ); |
| 207 |
$self->{'db'}->put_doc($doc) || $log->warn("can't add document $uri with draft " . $doc->dump_draft . " to node " . $self->{url} . " status: " . $self->{db}->status()); |
| 208 |
|
| 209 |
return 1; |
| 210 |
} |
| 211 |
|
| 212 |
# |
| 213 |
# REST parametars validation data |
| 214 |
# |
| 215 |
|
| 216 |
my $estraier_rest = { |
| 217 |
master => { |
| 218 |
userdel => [ qw/name/ ], |
| 219 |
nodelist => [], |
| 220 |
nodeadd => [ qw/name label/ ], |
| 221 |
nodedel => [ qw/name/ ], |
| 222 |
}, |
| 223 |
node => { |
| 224 |
_set_link => [ qw/url label credit/ ], |
| 225 |
}, |
| 226 |
}; |
| 227 |
|
| 228 |
=head2 master |
| 229 |
|
| 230 |
Issue administrative commands to C<estmaster> process and receive response |
| 231 |
as array of lines |
| 232 |
|
| 233 |
my $nodelist = $est->master( action => 'nodelist' ); |
| 234 |
|
| 235 |
=cut |
| 236 |
|
| 237 |
sub master { |
| 238 |
my $self = shift; |
| 239 |
|
| 240 |
my $args = {@_}; |
| 241 |
my $log = $self->_get_logger; |
| 242 |
|
| 243 |
my $action = $args->{action} || $log->logconfess("no action specified"); |
| 244 |
|
| 245 |
$log->logdie("action '$action' isn't supported") unless ($estraier_rest->{master}->{$action}); |
| 246 |
|
| 247 |
$log->debug("master action: $action"); |
| 248 |
|
| 249 |
return $self->estcall( |
| 250 |
validate => 'master', |
| 251 |
rest_url => $self->{masterurl} . '/master?action=' . $action , |
| 252 |
action => $action, |
| 253 |
%{ $args }, |
| 254 |
); |
| 255 |
} |
| 256 |
|
| 257 |
=head2 add_link |
| 258 |
|
| 259 |
$est->add_link( |
| 260 |
from => 'ps', |
| 261 |
to => 'webpac2', |
| 262 |
credit => 10000, |
| 263 |
); |
| 264 |
|
| 265 |
=cut |
| 266 |
|
| 267 |
sub add_link { |
| 268 |
my $self = shift; |
| 269 |
|
| 270 |
my $args = {@_}; |
| 271 |
my $log = $self->_get_logger; |
| 272 |
|
| 273 |
my @labels = $self->master( action => 'nodelist' ); |
| 274 |
|
| 275 |
$log->debug("got labels: ", join("|", @labels)); |
| 276 |
|
| 277 |
@labels = grep(/^$args->{to}\t/, @labels); |
| 278 |
my $label = shift @labels; |
| 279 |
(undef,$label) = split(/\t/, $label) if ($label); |
| 280 |
|
| 281 |
if (! $label) { |
| 282 |
$log->warn("can't find label for $args->{to}, skipping link creaton"); |
| 283 |
return; |
| 284 |
} |
| 285 |
|
| 286 |
$log->debug("using label $label for $args->{to}"); |
| 287 |
|
| 288 |
return $self->estcall( |
| 289 |
validate => 'node', |
| 290 |
action => '_set_link', |
| 291 |
rest_url => $self->{masterurl} . '/node/' . $args->{from} . '/_set_link' , |
| 292 |
url => $self->{masterurl} . '/node/' . $args->{to}, |
| 293 |
label => $label, |
| 294 |
credit => $args->{credit}, |
| 295 |
); |
| 296 |
} |
| 297 |
|
| 298 |
=head2 estcall |
| 299 |
|
| 300 |
Workhourse which does actual calls to Hyper Estraier |
| 301 |
|
| 302 |
$self->estcall( |
| 303 |
rest_url => '/master?action=' . $action, |
| 304 |
validate => 'master', |
| 305 |
# ... |
| 306 |
); |
| 307 |
|
| 308 |
C<rest_url> is relative URL to C<estmaster> and C<validate> is entry into |
| 309 |
internal hash which will check if all parametars are available before |
| 310 |
calling function. |
| 311 |
|
| 312 |
=cut |
| 313 |
|
| 314 |
sub estcall { |
| 315 |
my $self = shift; |
| 316 |
my $args = {@_}; |
| 317 |
my $log = $self->_get_logger; |
| 318 |
|
| 319 |
$log->debug("estcall: ",Dumper($args)); |
| 320 |
|
| 321 |
foreach my $p (qw/rest_url validate action/) { |
| 322 |
$log->die("ectcall needs $p parametar") unless ($args->{$p}); |
| 323 |
} |
| 324 |
|
| 325 |
my $url = $args->{rest_url}; |
| 326 |
my $del = '?'; |
| 327 |
$del = '&' if ($url =~ m#\?#); |
| 328 |
|
| 329 |
my $url_args; |
| 330 |
|
| 331 |
foreach my $arg (@{ $estraier_rest->{ $args->{validate} }->{ $args->{action} } }) { |
| 332 |
$log->logdie("missing parametar $arg for action $args->{action}") unless ($args->{$arg}); |
| 333 |
$url_args .= $del . $arg . '=' . uri_escape( $args->{$arg} ); |
| 334 |
$del = '&'; |
| 335 |
} |
| 336 |
|
| 337 |
$url .= $url_args if ($url_args); |
| 338 |
|
| 339 |
$log->debug("calling $url"); |
| 340 |
|
| 341 |
my $res = $self->est_ua()->get($url); |
| 342 |
|
| 343 |
if ($res->is_success) { |
| 344 |
#$log->debug( $res->content ); |
| 345 |
return split(/\n/, $res->content) if wantarray; |
| 346 |
return $res->content || 0E0; |
| 347 |
} else { |
| 348 |
$log->warn("unable to call $url: " . $res->status_line); |
| 349 |
return; |
| 350 |
} |
| 351 |
|
| 352 |
} |
| 353 |
|
| 354 |
=head2 est_ua |
| 355 |
|
| 356 |
This is helper function to create C<LWP::UserAgent> object with Super User |
| 357 |
priviledges |
| 358 |
|
| 359 |
my $ua = $self->est_ua( user => 'admin', passwd => 'admin' ); |
| 360 |
|
| 361 |
=cut |
| 362 |
|
| 363 |
|
| 364 |
|
| 365 |
sub est_ua { |
| 366 |
my $self = shift; |
| 367 |
|
| 368 |
return $self->{_master_ua} if ($self->{_master_ua}); |
| 369 |
|
| 370 |
{ |
| 371 |
package AdminUserAgent; |
| 372 |
use base qw/LWP::UserAgent/; |
| 373 |
sub new { |
| 374 |
my $self = LWP::UserAgent::new(@_); |
| 375 |
$self->agent("webpac/$VERSION"); |
| 376 |
$self; |
| 377 |
} |
| 378 |
sub get_basic_credentials { |
| 379 |
my($self, $realm, $uri) = @_; |
| 380 |
return ($self->{user}, $self->{passwd}); |
| 381 |
} |
| 382 |
sub set_basic_credentials { |
| 383 |
my ($self, $user, $passwd) = @_; |
| 384 |
$self->{user} = $user; |
| 385 |
$self->{passwd} = $passwd; |
| 386 |
} |
| 387 |
}; |
| 388 |
|
| 389 |
$self->{_master_ua} = AdminUserAgent->new( ) || sub { |
| 390 |
my $log = $self->_get_logger; |
| 391 |
$log->logdie("can't create LWP::UserAgent: $!"); |
| 392 |
}; |
| 393 |
|
| 394 |
$self->{_master_ua}->set_basic_credentials($self->{user}, $self->{passwd}); |
| 395 |
|
| 396 |
return $self->{_master_ua}; |
| 397 |
} |
| 398 |
|
| 399 |
=head2 convert |
| 400 |
|
| 401 |
my $utf8_string = $self->convert('string in codepage'); |
| 402 |
|
| 403 |
=cut |
| 404 |
|
| 405 |
sub convert { |
| 406 |
my $self = shift; |
| 407 |
|
| 408 |
my $text = shift || return; |
| 409 |
from_to($text, $self->{encoding}, 'UTF-8'); |
| 410 |
return $text; |
| 411 |
} |
| 412 |
|
| 413 |
=head1 AUTHOR |
| 414 |
|
| 415 |
Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >> |
| 416 |
|
| 417 |
=head1 COPYRIGHT & LICENSE |
| 418 |
|
| 419 |
Copyright 2005 Dobrica Pavlinusic, All Rights Reserved. |
| 420 |
|
| 421 |
This program is free software; you can redistribute it and/or modify it |
| 422 |
under the same terms as Perl itself. |
| 423 |
|
| 424 |
=cut |
| 425 |
|
| 426 |
1; # End of WebPAC::Output::Estraier |
Parent Directory
| 
Revision Log