12 |
use Time::HiRes qw/time/; |
use Time::HiRes qw/time/; |
13 |
use Encode qw/encode decode from_to/; |
use Encode qw/encode decode from_to/; |
14 |
use Template; |
use Template; |
|
use Data::Dumper; |
|
15 |
|
|
16 |
=head1 NAME |
=head1 NAME |
17 |
|
|
65 |
|
|
66 |
$est_cfg->{encoding} = $est_cfg->{catalyst_encoding} || $c->config->{catalyst_encoding} or $c->log->fatal("can't find catalyst_encoding"); |
$est_cfg->{encoding} = $est_cfg->{catalyst_encoding} || $c->config->{catalyst_encoding} or $c->log->fatal("can't find catalyst_encoding"); |
67 |
|
|
68 |
$log->debug("using config:" . Dumper($est_cfg) ); |
$log->dumper($est_cfg, 'est_cfg'); |
69 |
|
|
70 |
if (! $est_cfg->{database}) { |
if (! $est_cfg->{database}) { |
71 |
my $defaultnode = $est_cfg->{defaultnode} || $log->logdie("can't find defaultnode in estraier configuration"); |
my $defaultnode = $est_cfg->{defaultnode} || $log->logdie("can't find defaultnode in estraier configuration"); |
86 |
$log->fatal("can't create Search::Estraier::Node $url") unless ($self->{est_node}); |
$log->fatal("can't create Search::Estraier::Node $url") unless ($self->{est_node}); |
87 |
|
|
88 |
# save config parametars in object |
# save config parametars in object |
89 |
foreach my $f (qw/db_path template_path hits_on_page webpac_encoding defaultdepth/) { |
foreach my $f (qw/ |
90 |
|
db_path template_path hits_on_page webpac_encoding defaultdepth |
91 |
|
masterurl defaultnode |
92 |
|
/) { |
93 |
$self->{$f} = $c->config->{hyperestraier}->{$f} || |
$self->{$f} = $c->config->{hyperestraier}->{$f} || |
94 |
$c->config->{webpac}->{$f}; |
$c->config->{webpac}->{$f}; |
95 |
$log->debug("self->{$f} = " . $self->{$f}); |
$log->debug("self->{$f} = " . $self->{$f}); |
140 |
|
|
141 |
} |
} |
142 |
|
|
143 |
|
=head2 setup_site |
144 |
|
|
145 |
|
$self->setup_site('site_name'); |
146 |
|
|
147 |
|
Change node URL and database name according to site name (if available) or fallback |
148 |
|
to C<defaultnode> from configuration. |
149 |
|
|
150 |
|
=cut |
151 |
|
|
152 |
|
sub setup_site { |
153 |
|
my $self = shift; |
154 |
|
|
155 |
|
my $site = shift || $self->{defaultnode}; |
156 |
|
|
157 |
|
$self->{log}->fatal("setup_site can't find site or defaultnode") unless ($site); |
158 |
|
|
159 |
|
my $url = $self->{masterurl} . '/node/' . $site; |
160 |
|
$self->{est_node}->set_url( $url ); |
161 |
|
$self->{log}->debug("setup_site $site using $url"); |
162 |
|
} |
163 |
|
|
164 |
=head2 search |
=head2 search |
165 |
|
|
187 |
|
|
188 |
my $log = $self->{log}; |
my $log = $self->{log}; |
189 |
|
|
190 |
$log->debug("search args: " . Dumper( $args )); |
$log->dumper($args, 'args'); |
191 |
|
|
192 |
my $query = $args->{phrase} || $log->warn("no query phrase") && return; |
my $query = $args->{phrase} || $log->warn("no query phrase") && return; |
193 |
|
|
239 |
$cond->set_max( $page * $max ); |
$cond->set_max( $page * $max ); |
240 |
|
|
241 |
my $result = $self->{est_node}->search($cond, $args->{depth}); |
my $result = $self->{est_node}->search($cond, $args->{depth}); |
242 |
|
if (! $result) { |
243 |
|
$self->{log}->fatal("search didn't return result"); |
244 |
|
return; |
245 |
|
} |
246 |
my $hits = $result->doc_num; |
my $hits = $result->doc_num; |
247 |
|
|
248 |
$times->{est} += time() - $t; |
$times->{est} += time() - $t; |
249 |
|
|
250 |
$log->debug( sprintf("search took %.6fs and returned $hits hits.", $times->{est}) ); |
$log->debug( sprintf("search took %.6fs and returned $hits hits.", $times->{est}) ); |
251 |
|
|
252 |
$log->debug( "hints: " . Dumper($result->{hints}) ); |
$self->{hints} = $result->{hints}; |
253 |
|
#$log->dumper($self->{hints}, 'original hints' ); |
254 |
|
|
255 |
# |
# |
256 |
# fetch results |
# fetch results |
305 |
|
|
306 |
$times->{db} += time() - $t; |
$times->{db} += time() - $t; |
307 |
|
|
|
#$log->debug( "ds = " . Dumper( \@html_results ) ); |
|
|
|
|
308 |
$t = time(); |
$t = time(); |
309 |
|
|
310 |
my $html = $self->apply( |
my $html = $self->apply( |
327 |
|
|
328 |
} |
} |
329 |
|
|
|
#$log->debug( '@results = ' . Dumper( \@results ) ); |
|
|
|
|
330 |
$log->debug( sprintf( |
$log->debug( sprintf( |
331 |
"duration breakdown: estraier %.6fs, hash %.6fs, store %.6fs, apply %.6fs, decode %.06f, total: %.6fs", |
"duration breakdown: estraier %.6fs, hash %.6fs, store %.6fs, apply %.6fs, decode %.06f, total: %.6fs", |
332 |
$times->{est}, $times->{hash}, $times->{db}, $times->{apply}, $times->{decode}, time() - $search_start_t, |
$times->{est}, $times->{hash}, $times->{db}, $times->{apply}, $times->{decode}, time() - $search_start_t, |
335 |
return \@results; |
return \@results; |
336 |
} |
} |
337 |
|
|
338 |
|
=head2 hints |
339 |
|
|
340 |
|
my $hints = $m->hints; |
341 |
|
|
342 |
|
Return various useful hints about result |
343 |
|
|
344 |
|
=cut |
345 |
|
|
346 |
|
sub hints { |
347 |
|
my $self = shift; |
348 |
|
|
349 |
|
unless ($self->{hints}) { |
350 |
|
$self->{log}->fatal("no hints found!"); |
351 |
|
return; |
352 |
|
} |
353 |
|
|
354 |
|
my $hints; |
355 |
|
|
356 |
|
while (my ($key,$val) = each %{ $self->{hints} }) { |
357 |
|
|
358 |
|
#$self->{log}->debug("current hint $key = $val"); |
359 |
|
|
360 |
|
if ($key =~ m/^(?:HITS*|TIME|DOCNUM|WORDNUM)$/) { |
361 |
|
$hints->{ lc($key) } = $val; |
362 |
|
} elsif ($key =~ m/^HINT#/) { |
363 |
|
my ($word,$count) = split(/\t/,$val,2); |
364 |
|
$hints->{words}->{$word} = $count; |
365 |
|
} elsif ($key =~ m/^LINK#/) { |
366 |
|
my ($url,undef,undef,undef,undef,undef,$results) = split(/\t/,$val,7); |
367 |
|
if ($url =~ m#/node/(.+)$#) { |
368 |
|
$hints->{node}->{$1} = $results; |
369 |
|
} else { |
370 |
|
$self->{log}->debug("url $url doesn't have /node/ in it!"); |
371 |
|
} |
372 |
|
} else { |
373 |
|
$self->{log}->debug("unknown hint $key = $val"); |
374 |
|
} |
375 |
|
|
376 |
|
} |
377 |
|
|
378 |
|
$self->{log}->dumper($hints, 'model hints' ); |
379 |
|
|
380 |
|
return $hints; |
381 |
|
} |
382 |
|
|
383 |
|
|
384 |
=head2 record |
=head2 record |
385 |
|
|
386 |
my $html = $m->record( |
my $html = $m->record( |
398 |
|
|
399 |
my $args = {@_}; |
my $args = {@_}; |
400 |
my $log = $self->{log}; |
my $log = $self->{log}; |
401 |
$log->debug("record args: " . Dumper( $args )); |
$log->dumper( $args, 'args' ); |
402 |
|
|
403 |
foreach my $f (qw/record_uri template/) { |
foreach my $f (qw/record_uri template/) { |
404 |
$log->fatal("need $f") unless ($args->{$f}); |
$log->fatal("need $f") unless ($args->{$f}); |
432 |
} |
} |
433 |
|
|
434 |
|
|
435 |
|
=head2 list_nodes |
436 |
|
|
437 |
|
my @nodes = $m->list_nodes( 'site' ); |
438 |
|
|
439 |
|
Return all databases which have records for selected site. Returned array of |
440 |
|
hashes has elements C<name> and C<label>. |
441 |
|
|
442 |
|
=cut |
443 |
|
|
444 |
|
sub list_nodes { |
445 |
|
my $self = shift; |
446 |
|
|
447 |
|
my $site = shift; |
448 |
|
|
449 |
|
$self->{log}->debug("list_nodes use site $site"); |
450 |
|
|
451 |
|
$self->setup_site( $site ); |
452 |
|
|
453 |
|
my @nodes; |
454 |
|
|
455 |
|
if ($self->{est_node}->doc_num > 0) { |
456 |
|
push @nodes, { |
457 |
|
name => $self->{est_node}->name, |
458 |
|
label => $self->{est_node}->label, |
459 |
|
doc_num => $self->{est_node}->doc_num, |
460 |
|
} |
461 |
|
} |
462 |
|
|
463 |
|
# refresh set info |
464 |
|
$self->{est_node}->_set_info; |
465 |
|
|
466 |
|
my $links = $self->{est_node}->links || return @nodes; |
467 |
|
|
468 |
|
$self->{log}->dumper( $links, 'links' ); |
469 |
|
|
470 |
|
foreach my $link (@{ $links }) { |
471 |
|
my ($url, $label, $credit) = split(/\t/, $link, 3); |
472 |
|
if ($url =~ m#/node/(.+)$#) { |
473 |
|
my $node = $1; |
474 |
|
$self->setup_site( $node ); |
475 |
|
$self->{est_node}->_set_info; |
476 |
|
$label = decode('UTF-8', $label); |
477 |
|
push @nodes, { |
478 |
|
name => $node, |
479 |
|
label => $label, |
480 |
|
doc_num => $self->{est_node}->doc_num, |
481 |
|
} |
482 |
|
} else { |
483 |
|
$self->{log}->warn("can't find node name in link $link"); |
484 |
|
} |
485 |
|
} |
486 |
|
|
487 |
|
$self->setup_site( $site ); |
488 |
|
$self->{est_node}->_set_info; |
489 |
|
|
490 |
|
$self->{log}->dumper( \@nodes, 'nodes' ); |
491 |
|
|
492 |
|
return @nodes; |
493 |
|
} |
494 |
|
|
495 |
=head2 save_html |
=head2 save_html |
496 |
|
|
497 |
$m->save_html( '/full/path/to/file', $content ); |
$m->save_html( '/full/path/to/file', $content ); |
563 |
|
|
564 |
=cut |
=cut |
565 |
|
|
566 |
|
# Escape <, >, & and ", and to produce valid XML |
567 |
|
my %escape = ('<'=>'<', '>'=>'>', '&'=>'&', '"'=>'"'); |
568 |
|
my $escape_re = join '|' => keys %escape; |
569 |
|
|
570 |
sub apply { |
sub apply { |
571 |
my $self = shift; |
my $self = shift; |
572 |
|
|
614 |
if (ref($v) eq 'ARRAY') { |
if (ref($v) eq 'ARRAY') { |
615 |
if ($#{$v} == 0) { |
if ($#{$v} == 0) { |
616 |
$v = $v->[0]; |
$v = $v->[0]; |
617 |
|
$v =~ s/($escape_re)/$escape{$1}/g; |
618 |
} else { |
} else { |
619 |
$join = $default_delimiter->{$type} unless defined($join); |
$join = $default_delimiter->{$type} unless defined($join); |
620 |
$v = join($join, @{$v}); |
$v = join($join, map { |
621 |
|
s/($escape_re)/$escape{$1}/g; |
622 |
|
} @{$v}); |
623 |
} |
} |
624 |
} else { |
} else { |
625 |
warn("TT filter $type(): field $name values aren't ARRAY, ignoring"); |
warn("TT filter $type(): field $name values aren't ARRAY, ignoring"); |