/[webpac2]/Webpacus/lib/Webpacus/Model/WebPAC.pm
This is repository of my old source code which isn't updated any more. Go to git.rot13.org for current projects!
ViewVC logotype

Contents of /Webpacus/lib/Webpacus/Model/WebPAC.pm

Parent Directory Parent Directory | Revision Log Revision Log


Revision 404 - (show annotations)
Sun Feb 19 20:16:11 2006 UTC (18 years, 2 months ago) by dpavlin
File size: 17401 byte(s)
 r472@llin:  dpavlin | 2006-02-19 21:24:56 +0100
 added doc_num to all databases, refactor database selector into databases.tt

1 package Webpacus::Model::WebPAC;
2
3 use strict;
4 use warnings;
5 use lib '/data/webpac2/lib';
6 use base qw/
7 Catalyst::Model
8 /;
9 use WebPAC::Store 0.08;
10 use Search::Estraier 0.04;
11 use File::Slurp;
12 use Time::HiRes qw/time/;
13 use Encode qw/encode decode from_to/;
14 use Template;
15
16 =head1 NAME
17
18 Webpacus::Model::WebPAC - Catalyst Model
19
20 =head1 SYNOPSIS
21
22 See L<Webpacus> and L<WebPAC>.
23
24 =head1 DESCRIPTION
25
26 Catalyst Model for access to WebPAC data.
27
28 =head2 new
29
30 Configuration for hyperestraier in C<config.yaml> like this:
31
32 --- #YAML:1.0
33 # DO NOT USE TABS FOR INDENTATION OR label/value SEPARATION!!!
34
35 # configuration for hyper estraier full text search engine
36 hyperestraier:
37 masterurl: 'http://localhost:1978/node/webpac2'
38 defaultnode: 'webpac2'
39 defaultdepth: 1
40 user: 'admin'
41 passwd: 'admin'
42 hits_on_page: 100
43 hits_for_pager: 1000
44
45 webpac:
46 db_path: '/data/webpac2/db'
47 template_path: '/data/webpac2/conf/output/tt'
48 template: 'html_ffzg_results_short.tt'
49 # encoding comming from webpac
50 webpac_encoding: 'iso-8859-2'
51
52 =cut
53
54 sub new {
55 my ( $self, $c, $config ) = @_;
56
57 $self = $self->NEXT::new($c, $config);
58 $self->config($config);
59
60 my $log = $c->log;
61 $self->{log} = $log;
62
63 my $est_cfg = $c->config->{hyperestraier};
64 $est_cfg->{'log'} = $log;
65
66 $est_cfg->{encoding} = $est_cfg->{catalyst_encoding} || $c->config->{catalyst_encoding} or $c->log->fatal("can't find catalyst_encoding");
67
68 $log->dumper($est_cfg, 'est_cfg');
69
70 if (! $est_cfg->{database}) {
71 my $defaultnode = $est_cfg->{defaultnode} || $log->logdie("can't find defaultnode in estraier configuration");
72 $log->info("using default node $defaultnode");
73 $est_cfg->{database} = $defaultnode;
74 }
75
76 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
89 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} ||
94 $c->config->{webpac}->{$f};
95 $log->debug("self->{$f} = " . $self->{$f});
96 }
97 my $db_path = $self->{db_path};
98 my $template_path = $self->{template_path};
99
100 $log->debug("using db path '$db_path', template path '$template_path'");
101
102 $self->{db} = new WebPAC::Store(
103 path => $db_path,
104 read_only => 1,
105 database => $est_cfg->{database},
106 );
107
108 # default template from config.yaml
109 $self->{template} ||= $c->config->{webpac}->{template};
110
111 $log->debug("converting encoding from webpac_encoding '" .
112 $c->config->{webpac}->{webpac_encoding} .
113 "'"
114 );
115
116 $self->{databases} = $c->config->{databases} || $log->fatal("can't find databases in config");
117
118 # create Template toolkit instance
119 $self->{'tt'} = Template->new(
120 INCLUDE_PATH => $template_path,
121 FILTERS => {
122 dump_html => sub {
123 return unless (@_);
124 my $out;
125 my $i = 1;
126 foreach my $v (@_) {
127 $out .= qq{<div id="dump_$i">} .
128 Data::HTMLDumper->Dump([ $v ],[ "v$i" ]) .
129 qq{</div>};
130 $i++;
131 }
132 $out =~ s!<table[^>/]*>!<table class="dump">!gis if ($out);
133 return $out;
134 }
135 },
136 EVAL_PERL => 1,
137 );
138
139 return $self;
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
165
166 my $m->search(
167 phrase => 'query phrase',
168 add_attr => \@add_attr
169 get_attr => [ '@uri' ],
170 max => 42,
171 template => 'result_template.tt',
172 depth => 1,
173 );
174
175 All fields are standard C<WebPAC::Search::Estraier> parametars except
176 C<template> which will (if specified) return results in HTML using
177 selected template.
178
179 =cut
180
181 sub search {
182 my $self = shift;
183
184 my $search_start_t = time();
185
186 my $args = {@_};
187
188 my $log = $self->{log};
189
190 $log->dumper($args, 'args');
191
192 my $query = $args->{phrase} || $log->warn("no query phrase") && return;
193
194 my $template_filename = $args->{template} || $self->{template};
195
196 $args->{max} ||= $self->{'hits_for_pager'};
197 if (! $args->{max}) {
198 $args->{max} = 100;
199 $log->warn("max not set when calling model. Using default of $args->{max}");
200 }
201
202 my $times; # store some times for benchmarking
203
204 my $t = time();
205
206 # transfer depth of search
207 if (! $args->{depth}) {
208 my $default = $self->{defaultdepth} || $log->logdie("can't find defaultdepth in estraier configuration");
209 $args->{depth} = $default;
210 $log->warn("using default search depth $default");
211 }
212 $args->{depth} ||= 0;
213
214 $log->debug("searching for maximum $args->{max} results using depth $args->{depth} phrase: ", $query || '[none]');
215
216 #
217 # construct condition for Hyper Estraier
218 #
219 my $cond = Search::Estraier::Condition->new();
220 if ( ref($args->{add_attr}) eq 'ARRAY' ) {
221 $log->debug("adding search attributes: " . join(", ", @{ $args->{add_attr} }) );
222 map {
223 $cond->add_attr( $_ );
224 $log->debug(" + $_");
225 } @{ $args->{add_attr} };
226 };
227
228 $cond->set_phrase( $query ) if ($query);
229 $cond->set_options( $args->{options} ) if ($args->{options});
230 $cond->set_order( $args->{order} ) if ($args->{order});
231
232 my $max = $args->{max} || 7;
233 my $page = $args->{page} || 1;
234 if ($page < 1) {
235 $log->warn("page number $page < 1");
236 $page = 1;
237 }
238
239 $cond->set_max( $page * $max );
240
241 my $result = $self->{est_node}->search($cond, $args->{depth});
242 my $hits = $result->doc_num;
243
244 $times->{est} += time() - $t;
245
246 $log->debug( sprintf("search took %.6fs and returned $hits hits.", $times->{est}) );
247
248 $log->dumper($result->{hints}, 'result->hints' );
249
250 #
251 # fetch results
252 #
253
254 my @results;
255
256 for my $i ( (($page - 1) * $max) .. ( $hits - 1 ) ) {
257
258 $t = time();
259
260 #$log->debug("get_doc($i)");
261 my $doc = $result->get_doc( $i );
262 if (! $doc) {
263 $log->warn("can't find result $i");
264 next;
265 }
266
267 my $hash;
268
269 foreach my $attr (@{ $args->{get_attr} }) {
270 my $val = $doc->attr( $attr );
271 #$log->debug("attr $attr = ", $val || 'undef');
272 $hash->{$attr} = $val if (defined($val));
273 }
274
275 $times->{hash} += time() - $t;
276
277 next unless ($hash);
278
279 if (! $args->{'template'}) {
280 push @results, $hash;
281 } else {
282 my ($database, $prefix, $id);
283
284 if ( $hash->{'@uri'} =~ m!/([^/]+)/([^/]+)/(\d+)$!) {
285 ($database, $prefix,$id) = ($1,$2,$3);
286 } else {
287 $log->warn("can't decode database/prefix/id from " . $hash->{'@uri'});
288 next;
289 }
290
291 #$log->debug("load_ds( id => $id, prefix => '$prefix' )");
292
293 $t = time();
294
295 my $ds = $self->{db}->load_ds( database => $database, prefix => $prefix, id => $id );
296 if (! $ds) {
297 $log->error("can't load_ds( ${database}/${prefix}/${id} )");
298 next;
299 }
300
301 $times->{db} += time() - $t;
302
303 $t = time();
304
305 my $html = $self->apply(
306 template => $template_filename,
307 data => $ds,
308 record_uri => "${database}/${prefix}/${id}",
309 config => $self->{databases}->{$database},
310 );
311
312 $times->{apply} += time() - $t;
313
314 $t = time();
315
316 $html = decode($self->{webpac_encoding}, $html);
317
318 $times->{decode} += time() - $t;
319
320 push @results, $html;
321 }
322
323 }
324
325 $log->debug( sprintf(
326 "duration breakdown: estraier %.6fs, hash %.6fs, store %.6fs, apply %.6fs, decode %.06f, total: %.6fs",
327 $times->{est}, $times->{hash}, $times->{db}, $times->{apply}, $times->{decode}, time() - $search_start_t,
328 ) );
329
330 return \@results;
331 }
332
333 =head2 record
334
335 my $html = $m->record(
336 mfn => 42,
337 template => 'foo.tt',
338 );
339
340 This will load one record, convert it to html using C<template> and return
341 it.
342
343 =cut
344
345 sub record {
346 my $self = shift;
347
348 my $args = {@_};
349 my $log = $self->{log};
350 $log->dumper( $args, 'args' );
351
352 foreach my $f (qw/record_uri template/) {
353 $log->fatal("need $f") unless ($args->{$f});
354 }
355
356 my ($database, $prefix, $id);
357
358 if ($args->{record_uri} =~ m#^([^/]+)/([^/]+)/([^/]+)$#) {
359 ($database, $prefix, $id) = ($1,$2,$3);
360 } else {
361 $log->error("can't parse $args->{record_uri} into prefix, database and uri");
362 return;
363 }
364
365 my $ds = $self->{db}->load_ds( id => $id, prefix => $prefix, database => $database );
366 if (! $ds) {
367 $log->error("can't load_ds( $database/$prefix/$id )");
368 return;
369 }
370
371 my $html = $self->apply(
372 template => $args->{template},
373 data => $ds,
374 record_uri => $args->{record_uri},
375 config => $self->{databases}->{$database},
376 );
377
378 $html = decode($self->{webpac_encoding}, $html);
379
380 return $html;
381 }
382
383
384 =head2 list_nodes
385
386 my @nodes = $m->list_nodes( 'site' );
387
388 Return all databases which have records for selected site. Returned array of
389 hashes has elements C<name> and C<label>.
390
391 =cut
392
393 sub list_nodes {
394 my $self = shift;
395
396 my $site = shift;
397
398 $self->{log}->debug("list_nodes use site $site");
399
400 $self->setup_site( $site );
401
402 my @nodes;
403
404 if ($self->{est_node}->doc_num > 0) {
405 push @nodes, {
406 name => $self->{est_node}->name,
407 label => $self->{est_node}->label,
408 doc_num => $self->{est_node}->doc_num,
409 }
410 }
411
412 # refresh set info
413 $self->{est_node}->_set_info;
414
415 my $links = $self->{est_node}->links || return @nodes;
416
417 $self->{log}->dumper( $links, 'links' );
418
419 foreach my $link (@{ $links }) {
420 my ($url, $label, $credit) = split(/\t/, $link, 3);
421 if ($url =~ m#/node/(.+)$#) {
422 my $node = $1;
423 $self->setup_site( $node );
424 $self->{est_node}->_set_info;
425 push @nodes, {
426 name => $node,
427 label => $label,
428 doc_num => $self->{est_node}->doc_num,
429 }
430 } else {
431 $self->{log}->warn("can't find node name in link $link");
432 }
433 }
434
435 $self->setup_site( $site );
436 $self->{est_node}->_set_info;
437
438 $self->{log}->dumper( \@nodes, 'nodes' );
439
440 return @nodes;
441 }
442
443 =cut
444
445
446 =head2 save_html
447
448 $m->save_html( '/full/path/to/file', $content );
449
450 It will use C<Encode> to convert content encoding back to
451 Webpac codepage, recode JavaScript Unicode entities (%u1234),
452 strip extra newlines at beginning and end, and save to
453 C</full/path/to/file.new> and if that succeeds, just rename
454 it over original file which should be atomic on filesystem level.
455
456 =cut
457
458 sub save_html {
459 my ($self, $path, $content) = @_;
460
461 # FIXME Should this be UTF-8 or someting?
462 my $js_encoding = $self->{webpac_encoding};
463 $js_encoding = 'UTF-16';
464
465 sub _conv_js {
466 return '0x' . $_[1];
467 return encode($_[0], chr(hex($_[1])));
468 }
469 #$content =~ s/%u([a-fA-F0-9]{4})/_conv_js($js_encoding,$1)/gex;
470 $content =~ s/^[\n\r]+//s;
471 $content =~ s/[\n\r]+$/\n/s;
472 $content =~ s/\n\r/\n/gs;
473
474 my $disk_encoding = $self->{webpac_encoding} || 'utf-8';
475 $self->{log}->debug("convert encoding to $disk_encoding");
476 from_to($content, 'utf-8', $disk_encoding) || $self->{log}->warn("encoding from utf-8 to $disk_encoding failed for: $content");
477
478 write_file($path . '.new', {binmode => ':raw' }, $content) || die "can't save ${path}.new $!";
479 rename $path . '.new', $path || die "can't rename to $path: $!";
480 }
481
482 =head2 load_html
483
484 my $html = $m->load_html('/full/path/to/file');
485
486 This will convert file from Webpac encoding to Catalyst and
487 convert that data to escaped HTML (for sending into
488 C<< <textarea/> >> tags in html.
489
490 =cut
491
492 sub load_html {
493 my ($self, $path) = @_;
494
495 die "no path?" unless ($path);
496
497 my $content = read_file($path, {binmode => ':raw' }) || die "can't read $path: $!";
498
499 return decode($self->{webpac_encoding}, $content);
500 }
501
502
503 =head2 apply
504
505 Create output from in-memory data structure using Template Toolkit template.
506
507 my $text = $tt->apply(
508 template => 'text.tt',
509 data => $ds,
510 record_uri => 'database/prefix/mfn',
511 );
512
513 It also has follwing template toolikit filter routies defined:
514
515 =cut
516
517 sub apply {
518 my $self = shift;
519
520 my $args = {@_};
521
522 my $log = $self->{log} || die "no log?";
523
524 foreach my $a (qw/template data/) {
525 $log->fatal("need $a") unless ($args->{$a});
526 }
527
528 =head3 tt_filter_type
529
530 filter to return values of specified from $ds, usage from TT template is in form
531 C<d('FieldName','delimiter')>, where C<delimiter> is optional, like this:
532
533 [% d('Title') %]
534 [% d('Author',', ' %]
535
536 =cut
537
538 sub tt_filter_type {
539 my ($data,$type) = @_;
540
541 die "no data?" unless ($data);
542 $type ||= 'display';
543
544 my $default_delimiter = {
545 'display' => '&#182;<br/>',
546 'index' => '\n',
547 };
548
549 return sub {
550
551 my ($name,$join) = @_;
552
553 die "no data hash" unless ($data->{'data'} && ref($data->{'data'}) eq 'HASH');
554 # Hm? Should we die here?
555 return unless ($name);
556
557 my $item = $data->{'data'}->{$name} || return;
558
559 my $v = $item->{$type} || return;
560
561 if (ref($v) eq 'ARRAY') {
562 if ($#{$v} == 0) {
563 $v = $v->[0];
564 } else {
565 $join = $default_delimiter->{$type} unless defined($join);
566 $v = join($join, @{$v});
567 }
568 } else {
569 warn("TT filter $type(): field $name values aren't ARRAY, ignoring");
570 }
571
572 return $v;
573 }
574 }
575
576 $args->{'d'} = tt_filter_type($args, 'display');
577 $args->{'display'} = tt_filter_type($args, 'display');
578
579 =head3 tt_filter_search
580
581 filter to return links to search, usage in TT:
582
583 [% search('FieldToDisplay','FieldToSearch','optional delimiter', 'optional_template.tt') %]
584
585 =cut
586
587 sub tt_filter_search {
588
589 my ($data) = @_;
590
591 die "no data?" unless ($data);
592
593 return sub {
594
595 my ($display,$search,$delimiter,$template) = @_;
596
597 # default delimiter
598 $delimiter ||= '&#182;<br/>',
599
600 die "no data hash" unless ($data->{'data'} && ref($data->{'data'}) eq 'HASH');
601 # Hm? Should we die here?
602 return unless ($display);
603
604 my $item = $data->{'data'}->{$display} || return;
605
606 return unless($item->{'display'});
607 if (! $item->{'search'}) {
608 warn "error in TT template: field $display didn't insert anything into search, use d('$display') and not search('$display'...)";
609 return;
610 }
611
612 my @warn;
613 foreach my $type (qw/display search/) {
614 push @warn, "field $display type $type values aren't ARRAY" unless (ref($item->{$type}) eq 'ARRAY');
615 }
616
617 if (@warn) {
618 warn("TT filter search(): " . join(",", @warn) . ", skipping");
619 return;
620 }
621 my @html;
622
623 my $d_el = $#{ $item->{'display'} };
624 my $s_el = $#{ $item->{'search'} };
625
626 # easy, both fields have same number of elements or there is just
627 # one search and multiple display
628 if ( $d_el == $s_el || $s_el == 0 ) {
629
630 foreach my $i ( 0 .. $d_el ) {
631
632 my $s;
633 if ($s_el > 0) {
634 $s = $item->{'search'}->[$i] or warn "can't find value $i for type search in field $search";
635 } else {
636 $s = $item->{'search'}->[0];
637 }
638 #$s =~ s/([^\w.-])/sprintf("%%%02X",ord($1))/eg;
639 $s = __quotemeta( $s );
640
641 my $d = $item->{'display'}->[$i] or warn "can't find value $i for type display in field $display";
642
643 my $template_arg = '';
644 $template_arg = qq{,'$template'} if ($template);
645
646 push @html, qq{<a href="#" onclick="return search_via_link('$search','$s'${template_arg})">$d</a>};
647 }
648
649 return join($delimiter, @html);
650 } else {
651 my $html = qq{<div class="notice">WARNING: we should really support if there is $d_el display elements and $s_el search elements, but currently there is no nice way to do so, so we will just display values</div>};
652 my $v = $item->{'display'};
653
654 if ($#{$v} == 0) {
655 $html .= $v->[0];
656 } else {
657 $html .= join($delimiter, @{$v});
658 }
659 return $html;
660 }
661 }
662 }
663
664 $args->{'search'} = tt_filter_search($args);
665
666 =head3 load_rec
667
668 Used mostly for onClick events like this:
669
670 <a href="#" onClick="[% load_rec( record_uri, 'template_name.tt') %]>foo</a>
671
672 It will automatically do sanity checking and create correct JavaScript code.
673
674 =cut
675
676 $args->{'load_rec'} = sub {
677 my @errors;
678
679 my $record_uri = shift or push @errors, "record_uri missing";
680 my $template = shift or push @errors, "template missing";
681
682 if ($record_uri !~ m#^[^/]+/[^/]+/[^/]+$#) {
683 push @errors, "invalid format of record_uri: $record_uri";
684 }
685
686 if (@errors) {
687 return "Logger.error('errors in load_rec: " . join(", ", @errors) . "'); return false;";
688 } else {
689 return "load_rec('$record_uri','$template'); return false;";
690 }
691 };
692
693 =head3 load_template
694
695 Used to re-submit search request and load results in different template
696
697 <a href="#" onClick="[% load_template( 'template_name.tt' ) %]">bar</a>
698
699 =cut
700
701 $args->{'load_template'} = sub {
702 my $template = shift or return "Logger.error('load_template missing template name!'); return false;";
703 return "load_template($template); return false;";
704 };
705
706 my $out;
707
708 $self->{'tt'}->process(
709 $args->{'template'},
710 $args,
711 \$out
712 ) || $log->error( "apply can't process template: ", $self->{'tt'}->error() );
713
714 return $out;
715 }
716
717
718 =head2 __quotemeta
719
720 Helper to quote JavaScript-friendly characters
721
722 =cut
723
724 sub __quotemeta {
725 local $_ = shift;
726 $_ = decode('iso-8859-2', $_);
727
728 s<([\x{0080}-\x{fffd}]+)>{sprintf '\u%0*v4X', '\u', $1}ge if ( Encode::is_utf8($_) );
729 {
730 use bytes;
731 s<((?:[^ \x21-\x7E]|(?:\\(?!u)))+)>{sprintf '\x%0*v2X', '\x', $1}ge;
732 }
733
734 s/\\x09/\\t/g;
735 s/\\x0A/\\n/g;
736 s/\\x0D/\\r/g;
737 s/"/\\"/g;
738 s/\\x5C/\\\\/g;
739
740 return $_;
741 }
742
743
744
745 =head1 AUTHOR
746
747 Dobrica Pavlinusic C<< <dpavlin@rot13.org> >>
748
749 =head1 LICENSE
750
751 This library is free software, you can redistribute it and/or modify
752 it under the same terms as Perl itself.
753
754 =cut
755
756 1;

  ViewVC Help
Powered by ViewVC 1.1.26