7 |
Catalyst::Model |
Catalyst::Model |
8 |
/; |
/; |
9 |
use WebPAC::Store 0.08; |
use WebPAC::Store 0.08; |
10 |
use WebPAC::Search::Estraier 0.05; |
use Search::Estraier 0.04; |
11 |
use File::Slurp; |
use File::Slurp; |
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"); |
73 |
$est_cfg->{database} = $defaultnode; |
$est_cfg->{database} = $defaultnode; |
74 |
} |
} |
75 |
|
|
76 |
$self->{est} = new WebPAC::Search::Estraier( %{ $est_cfg } ); |
my $url = $est_cfg->{masterurl} . '/node/' . $est_cfg->{database}; |
77 |
|
|
78 |
|
$log->info("opening Hyper Estraier index $url as $est_cfg->{'user'}"); |
79 |
|
|
80 |
|
$self->{est_node} = Search::Estraier::Node->new( |
81 |
|
url => $url, |
82 |
|
user => $est_cfg->{user}, |
83 |
|
passwd => $est_cfg->{passwd}, |
84 |
|
); |
85 |
|
|
86 |
|
$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}); |
113 |
"'" |
"'" |
114 |
); |
); |
115 |
|
|
116 |
$self->{databases} = $c->config->{databases} || $log->error("can't find databases in config"); |
$self->{databases} = $c->config->{databases} || $log->fatal("can't find databases in config"); |
117 |
|
|
118 |
# create Template toolkit instance |
# create Template toolkit instance |
119 |
$self->{'tt'} = Template->new( |
$self->{'tt'} = Template->new( |
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 |
|
if (! $site) { |
157 |
|
$self->{log}->warn("not changing site from " . $self->{est_node}->{url}); |
158 |
|
return; |
159 |
|
} |
160 |
|
|
161 |
|
$self->{log}->fatal("setup_site can't find site or defaultnode") unless ($site); |
162 |
|
|
163 |
|
my $url = $self->{masterurl} . '/node/' . $site; |
164 |
|
$self->{est_node}->set_url( $url ); |
165 |
|
$self->{log}->debug("setup_site '$site' using $url"); |
166 |
|
} |
167 |
|
|
168 |
=head2 search |
=head2 search |
169 |
|
|
191 |
|
|
192 |
my $log = $self->{log}; |
my $log = $self->{log}; |
193 |
|
|
194 |
$log->debug("search args: " . Dumper( $args )); |
$log->dumper($args, 'args'); |
195 |
|
|
196 |
my $query = $args->{phrase} || $log->warn("no query phrase") && return; |
my $query = $args->{phrase} || $log->warn("no query phrase") && return; |
197 |
|
|
|
$log->debug("search model query: '$query'"); |
|
|
if ($args->{add_attr}) { |
|
|
$log->debug(" + add_attr: " . |
|
|
join("','", @{ $args->{add_attr} }) |
|
|
); |
|
|
} |
|
|
|
|
198 |
my $template_filename = $args->{template} || $self->{template}; |
my $template_filename = $args->{template} || $self->{template}; |
199 |
|
|
200 |
$args->{max} ||= $self->{'hits_for_pager'}; |
$args->{max} ||= $self->{'hits_for_pager'}; |
213 |
$args->{depth} = $default; |
$args->{depth} = $default; |
214 |
$log->warn("using default search depth $default"); |
$log->warn("using default search depth $default"); |
215 |
} |
} |
216 |
|
$args->{depth} ||= 0; |
217 |
|
|
218 |
my @results = $self->{est}->search( %{ $args } ); |
$log->debug("searching " . $self->{est_node}->{url} . " max: $args->{max} depth: $args->{depth} phrase: " . ($query || '[none]') ); |
219 |
|
|
220 |
$times->{est} += time() - $t; |
# |
221 |
|
# construct condition for Hyper Estraier |
222 |
|
# |
223 |
|
my $cond = Search::Estraier::Condition->new(); |
224 |
|
if ( ref($args->{add_attr}) eq 'ARRAY' ) { |
225 |
|
$log->debug("adding search attributes: " . join(", ", @{ $args->{add_attr} }) ); |
226 |
|
map { |
227 |
|
$cond->add_attr( $_ ); |
228 |
|
$log->debug(" + $_"); |
229 |
|
} @{ $args->{add_attr} }; |
230 |
|
}; |
231 |
|
|
232 |
|
$cond->set_phrase( $query ) if ($query); |
233 |
|
$cond->set_options( $args->{options} ) if ($args->{options}); |
234 |
|
$cond->set_order( $args->{order} ) if ($args->{order}); |
235 |
|
|
236 |
|
my $max = $args->{max} || 7; |
237 |
|
my $page = $args->{page} || 1; |
238 |
|
if ($page < 1) { |
239 |
|
$log->warn("page number $page < 1"); |
240 |
|
$page = 1; |
241 |
|
} |
242 |
|
|
243 |
|
$cond->set_max( $page * $max ); |
244 |
|
|
245 |
|
my $result = $self->{est_node}->search($cond, $args->{depth}); |
246 |
|
if (! $result) { |
247 |
|
$self->{log}->fatal("search didn't return result"); |
248 |
|
return; |
249 |
|
} |
250 |
|
my $hits = $result->doc_num; |
251 |
|
|
252 |
my $hits = $#results + 1; |
$times->{est} += time() - $t; |
253 |
|
|
254 |
$log->debug( sprintf("search took %.6fs and returned $hits hits.", $times->{est}) ); |
$log->debug( sprintf("search took %.6fs and returned $hits hits.", $times->{est}) ); |
255 |
|
|
256 |
# just return results? |
$self->{hints} = $result->{hints}; |
257 |
return @results unless ($args->{'template'}); |
#$log->dumper($self->{hints}, 'original hints' ); |
258 |
|
|
259 |
# |
# |
260 |
# construct HTML results |
# fetch results |
261 |
# |
# |
262 |
|
|
263 |
my @html_results; |
my @results; |
264 |
|
|
265 |
for my $i ( 0 .. $#results ) { |
for my $i ( (($page - 1) * $max) .. ( $hits - 1 ) ) { |
266 |
|
|
267 |
my ($database, $prefix, $id); |
$t = time(); |
268 |
if ( $results[$i]->{'@uri'} =~ m!/([^/]+)/([^/]+)/(\d+)$!) { |
|
269 |
($database, $prefix,$id) = ($1,$2,$3); |
#$log->debug("get_doc($i)"); |
270 |
} else { |
my $doc = $result->get_doc( $i ); |
271 |
$log->warn("can't decode database/prefix/id from " . $results[$i]->{'@uri'}); |
if (! $doc) { |
272 |
|
$log->warn("can't find result $i"); |
273 |
next; |
next; |
274 |
} |
} |
275 |
|
|
276 |
#$log->debug("load_ds( id => $id, prefix => '$prefix' )"); |
my $hash; |
|
|
|
|
$t = time(); |
|
277 |
|
|
278 |
my $ds = $self->{db}->load_ds( database => $database, prefix => $prefix, id => $id ); |
foreach my $attr (@{ $args->{get_attr} }) { |
279 |
if (! $ds) { |
my $val = $doc->attr( $attr ); |
280 |
$log->error("can't load_ds( ${database}/${prefix}/${id} )"); |
#$log->debug("attr $attr = ", $val || 'undef'); |
281 |
next; |
$hash->{$attr} = $val if (defined($val)); |
282 |
} |
} |
283 |
|
|
284 |
$times->{db} += time() - $t; |
$times->{hash} += time() - $t; |
285 |
|
|
286 |
#$log->debug( "ds = " . Dumper( \@html_results ) ); |
next unless ($hash); |
287 |
|
|
288 |
$t = time(); |
if (! $args->{'template'}) { |
289 |
|
push @results, $hash; |
290 |
|
} else { |
291 |
|
my ($database, $prefix, $id); |
292 |
|
|
293 |
my $html = $self->apply( |
if ( $hash->{'@uri'} =~ m!/([^/]+)/([^/]+)/(\d+)$!) { |
294 |
template => $template_filename, |
($database, $prefix,$id) = ($1,$2,$3); |
295 |
data => $ds, |
} else { |
296 |
record_uri => "${database}/${prefix}/${id}", |
$log->warn("can't decode database/prefix/id from " . $hash->{'@uri'}); |
297 |
config => $self->{databases}->{$database}, |
next; |
298 |
); |
} |
299 |
|
|
300 |
$times->{out} += time() - $t; |
#$log->debug("load_ds( id => $id, prefix => '$prefix' )"); |
301 |
|
|
302 |
$t = time(); |
$t = time(); |
303 |
|
|
304 |
$html = decode($self->{webpac_encoding}, $html); |
my $ds = $self->{db}->load_ds( database => $database, prefix => $prefix, id => $id ); |
305 |
|
if (! $ds) { |
306 |
|
$log->error("can't load_ds( ${database}/${prefix}/${id} )"); |
307 |
|
next; |
308 |
|
} |
309 |
|
|
310 |
push @html_results, $html; |
$times->{db} += time() - $t; |
311 |
|
|
312 |
} |
$t = time(); |
313 |
|
|
314 |
|
my $html = $self->apply( |
315 |
|
template => $template_filename, |
316 |
|
data => $ds, |
317 |
|
record_uri => "${database}/${prefix}/${id}", |
318 |
|
config => $self->{databases}->{$database}, |
319 |
|
); |
320 |
|
|
321 |
#$log->debug( '@html_results = ' . Dumper( \@html_results ) ); |
$times->{apply} += time() - $t; |
322 |
|
|
323 |
|
$t = time(); |
324 |
|
|
325 |
|
$html = decode($self->{webpac_encoding}, $html); |
326 |
|
|
327 |
|
$times->{decode} += time() - $t; |
328 |
|
|
329 |
|
push @results, $html; |
330 |
|
} |
331 |
|
|
332 |
|
} |
333 |
|
|
334 |
$log->debug( sprintf( |
$log->debug( sprintf( |
335 |
"duration breakdown: store %.6fs, apply %.6fs, total: %.6fs", |
"duration breakdown: estraier %.6fs, hash %.6fs, store %.6fs, apply %.6fs, decode %.06f, total: %.6fs", |
336 |
$times->{db}, $times->{out}, time() - $search_start_t, |
$times->{est}, $times->{hash}, $times->{db}, $times->{apply}, $times->{decode}, time() - $search_start_t, |
337 |
) ); |
) ); |
338 |
|
|
339 |
return \@html_results; |
return \@results; |
340 |
} |
} |
341 |
|
|
342 |
|
=head2 hints |
343 |
|
|
344 |
|
my $hints = $m->hints; |
345 |
|
|
346 |
|
Return various useful hints about result |
347 |
|
|
348 |
|
=cut |
349 |
|
|
350 |
|
sub hints { |
351 |
|
my $self = shift; |
352 |
|
|
353 |
|
unless ($self->{hints}) { |
354 |
|
$self->{log}->fatal("no hints found!"); |
355 |
|
return; |
356 |
|
} |
357 |
|
|
358 |
|
my $hints; |
359 |
|
|
360 |
|
while (my ($key,$val) = each %{ $self->{hints} }) { |
361 |
|
|
362 |
|
#$self->{log}->debug("current hint $key = $val"); |
363 |
|
|
364 |
|
if ($key =~ m/^(?:HITS*|TIME|DOCNUM|WORDNUM)$/) { |
365 |
|
$hints->{ lc($key) } = $val; |
366 |
|
} elsif ($key =~ m/^HINT#/) { |
367 |
|
my ($word,$count) = split(/\t/,$val,2); |
368 |
|
$hints->{words}->{$word} = $count; |
369 |
|
} elsif ($key =~ m/^LINK#/) { |
370 |
|
my ($url,undef,undef,undef,undef,undef,$results) = split(/\t/,$val,7); |
371 |
|
if ($url =~ m#/node/(.+)$#) { |
372 |
|
$hints->{node}->{$1} = $results; |
373 |
|
} else { |
374 |
|
$self->{log}->debug("url $url doesn't have /node/ in it!"); |
375 |
|
} |
376 |
|
} else { |
377 |
|
$self->{log}->debug("unknown hint $key = $val"); |
378 |
|
} |
379 |
|
|
380 |
|
} |
381 |
|
|
382 |
|
$self->{log}->dumper($hints, 'model hints' ); |
383 |
|
|
384 |
|
return $hints; |
385 |
|
} |
386 |
|
|
387 |
|
|
388 |
=head2 record |
=head2 record |
389 |
|
|
390 |
my $html = $m->record( |
my $html = $m->record( |
402 |
|
|
403 |
my $args = {@_}; |
my $args = {@_}; |
404 |
my $log = $self->{log}; |
my $log = $self->{log}; |
405 |
$log->debug("record args: " . Dumper( $args )); |
$log->dumper( $args, 'args' ); |
406 |
|
|
407 |
foreach my $f (qw/record_uri template/) { |
foreach my $f (qw/record_uri template/) { |
408 |
$log->fatal("need $f") unless ($args->{$f}); |
$log->fatal("need $f") unless ($args->{$f}); |
436 |
} |
} |
437 |
|
|
438 |
|
|
439 |
|
=head2 list_nodes |
440 |
|
|
441 |
|
my @nodes = $m->list_nodes( 'site' ); |
442 |
|
|
443 |
|
Return all databases which have records for selected site. Returned array of |
444 |
|
hashes has elements C<name> and C<label>. |
445 |
|
|
446 |
|
=cut |
447 |
|
|
448 |
|
my $nodes_list; |
449 |
|
|
450 |
|
sub list_nodes { |
451 |
|
my $self = shift; |
452 |
|
|
453 |
|
my $site = shift || $self->{defaultnode}; |
454 |
|
|
455 |
|
# cache? |
456 |
|
if ($nodes_list->{$site} && ref($nodes_list->{$site} eq 'ARRAY')) { |
457 |
|
$self->{log}->debug("list_nodes for site $site and returns from cache"); |
458 |
|
return @{ $nodes_list->{$site} }; |
459 |
|
}; |
460 |
|
|
461 |
|
$self->{log}->debug("list_nodes for site $site"); |
462 |
|
|
463 |
|
$self->setup_site( $site ); |
464 |
|
|
465 |
|
my @nodes; |
466 |
|
|
467 |
|
if ($self->{est_node}->doc_num > 0) { |
468 |
|
push @nodes, { |
469 |
|
name => $self->{est_node}->name, |
470 |
|
label => $self->{est_node}->label, |
471 |
|
doc_num => $self->{est_node}->doc_num, |
472 |
|
} |
473 |
|
} |
474 |
|
|
475 |
|
# refresh set info |
476 |
|
$self->{est_node}->_set_info; |
477 |
|
|
478 |
|
my $links = $self->{est_node}->links || return @nodes; |
479 |
|
|
480 |
|
$self->{log}->dumper( $links, 'links' ); |
481 |
|
|
482 |
|
foreach my $link (@{ $links }) { |
483 |
|
my ($url, $label, $credit) = split(/\t/, $link, 3); |
484 |
|
if ($url =~ m#/node/(.+)$#) { |
485 |
|
my $node = $1; |
486 |
|
$self->setup_site( $node ); |
487 |
|
$self->{est_node}->_set_info; |
488 |
|
$label = decode('UTF-8', $label); |
489 |
|
push @nodes, { |
490 |
|
name => $node, |
491 |
|
label => $label, |
492 |
|
doc_num => $self->{est_node}->doc_num, |
493 |
|
} |
494 |
|
} else { |
495 |
|
$self->{log}->warn("can't find node name in link $link"); |
496 |
|
} |
497 |
|
} |
498 |
|
|
499 |
|
$self->setup_site( $site ); |
500 |
|
$self->{est_node}->_set_info; |
501 |
|
|
502 |
|
$self->{log}->dumper( \@nodes, 'nodes' ); |
503 |
|
|
504 |
|
$nodes_list->{$site} = \@nodes; |
505 |
|
|
506 |
|
return @nodes; |
507 |
|
} |
508 |
|
|
509 |
=head2 save_html |
=head2 save_html |
510 |
|
|
511 |
$m->save_html( '/full/path/to/file', $content ); |
$m->save_html( '/full/path/to/file', $content ); |
577 |
|
|
578 |
=cut |
=cut |
579 |
|
|
580 |
|
# Escape <, >, & and ", and to produce valid XML |
581 |
|
my %escape = ('<'=>'<', '>'=>'>', '&'=>'&', '"'=>'"'); |
582 |
|
my $escape_re = join '|' => keys %escape; |
583 |
|
|
584 |
sub apply { |
sub apply { |
585 |
my $self = shift; |
my $self = shift; |
586 |
|
|
589 |
my $log = $self->{log} || die "no log?"; |
my $log = $self->{log} || die "no log?"; |
590 |
|
|
591 |
foreach my $a (qw/template data/) { |
foreach my $a (qw/template data/) { |
592 |
$log->logconfess("need $a") unless ($args->{$a}); |
$log->fatal("need $a") unless ($args->{$a}); |
593 |
} |
} |
594 |
|
|
595 |
=head3 tt_filter_type |
=head3 tt_filter_type |
628 |
if (ref($v) eq 'ARRAY') { |
if (ref($v) eq 'ARRAY') { |
629 |
if ($#{$v} == 0) { |
if ($#{$v} == 0) { |
630 |
$v = $v->[0]; |
$v = $v->[0]; |
631 |
|
$v =~ s/($escape_re)/$escape{$1}/g; |
632 |
} else { |
} else { |
633 |
$join = $default_delimiter->{$type} unless defined($join); |
$join = $default_delimiter->{$type} unless defined($join); |
634 |
$v = join($join, @{$v}); |
$v = join($join, map { |
635 |
|
s/($escape_re)/$escape{$1}/g; |
636 |
|
} @{$v}); |
637 |
} |
} |
638 |
} else { |
} else { |
639 |
warn("TT filter $type(): field $name values aren't ARRAY, ignoring"); |
warn("TT filter $type(): field $name values aren't ARRAY, ignoring"); |
701 |
|
|
702 |
my $s; |
my $s; |
703 |
if ($s_el > 0) { |
if ($s_el > 0) { |
704 |
$s = $item->{'search'}->[$i] || die "can't find value $i for type search in field $search"; |
$s = $item->{'search'}->[$i] or warn "can't find value $i for type search in field $search"; |
705 |
} else { |
} else { |
706 |
$s = $item->{'search'}->[0]; |
$s = $item->{'search'}->[0]; |
707 |
} |
} |
708 |
#$s =~ s/([^\w.-])/sprintf("%%%02X",ord($1))/eg; |
#$s =~ s/([^\w.-])/sprintf("%%%02X",ord($1))/eg; |
709 |
$s = __quotemeta( $s ); |
$s = __quotemeta( $s ); |
710 |
|
|
711 |
my $d = $item->{'display'}->[$i] || die "can't find value $i for type display in field $display"; |
my $d = $item->{'display'}->[$i] or warn "can't find value $i for type display in field $display"; |
712 |
|
|
713 |
my $template_arg = ''; |
my $template_arg = ''; |
714 |
$template_arg = qq{,'$template'} if ($template); |
$template_arg = qq{,'$template'} if ($template); |