/[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 385 - (show annotations)
Sun Jan 22 11:32:40 2006 UTC (18 years, 3 months ago) by dpavlin
File size: 15877 byte(s)
debug dump hints

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 use Data::Dumper;
16
17 =head1 NAME
18
19 Webpacus::Model::WebPAC - Catalyst Model
20
21 =head1 SYNOPSIS
22
23 See L<Webpacus> and L<WebPAC>.
24
25 =head1 DESCRIPTION
26
27 Catalyst Model for access to WebPAC data.
28
29 =head2 new
30
31 Configuration for hyperestraier in C<config.yaml> like this:
32
33 --- #YAML:1.0
34 # DO NOT USE TABS FOR INDENTATION OR label/value SEPARATION!!!
35
36 # configuration for hyper estraier full text search engine
37 hyperestraier:
38 masterurl: 'http://localhost:1978/node/webpac2'
39 defaultnode: 'webpac2'
40 defaultdepth: 1
41 user: 'admin'
42 passwd: 'admin'
43 hits_on_page: 100
44 hits_for_pager: 1000
45
46 webpac:
47 db_path: '/data/webpac2/db'
48 template_path: '/data/webpac2/conf/output/tt'
49 template: 'html_ffzg_results_short.tt'
50 # encoding comming from webpac
51 webpac_encoding: 'iso-8859-2'
52
53 =cut
54
55 sub new {
56 my ( $self, $c, $config ) = @_;
57
58 $self = $self->NEXT::new($c, $config);
59 $self->config($config);
60
61 my $log = $c->log;
62 $self->{log} = $log;
63
64 my $est_cfg = $c->config->{hyperestraier};
65 $est_cfg->{'log'} = $log;
66
67 $est_cfg->{encoding} = $est_cfg->{catalyst_encoding} || $c->config->{catalyst_encoding} or $c->log->fatal("can't find catalyst_encoding");
68
69 $log->debug("using config:" . Dumper($est_cfg) );
70
71 if (! $est_cfg->{database}) {
72 my $defaultnode = $est_cfg->{defaultnode} || $log->logdie("can't find defaultnode in estraier configuration");
73 $log->info("using default node $defaultnode");
74 $est_cfg->{database} = $defaultnode;
75 }
76
77 my $url = $est_cfg->{masterurl} . '/node/' . $est_cfg->{database};
78
79 $log->info("opening Hyper Estraier index $url as $est_cfg->{'user'}");
80
81 $self->{est_node} = Search::Estraier::Node->new(
82 url => $url,
83 user => $est_cfg->{user},
84 passwd => $est_cfg->{passwd},
85 );
86
87 $log->fatal("can't create Search::Estraier::Node $url") unless ($self->{est_node});
88
89 # save config parametars in object
90 foreach my $f (qw/db_path template_path hits_on_page webpac_encoding defaultdepth/) {
91 $self->{$f} = $c->config->{hyperestraier}->{$f} ||
92 $c->config->{webpac}->{$f};
93 $log->debug("self->{$f} = " . $self->{$f});
94 }
95 my $db_path = $self->{db_path};
96 my $template_path = $self->{template_path};
97
98 $log->debug("using db path '$db_path', template path '$template_path'");
99
100 $self->{db} = new WebPAC::Store(
101 path => $db_path,
102 read_only => 1,
103 database => $est_cfg->{database},
104 );
105
106 # default template from config.yaml
107 $self->{template} ||= $c->config->{webpac}->{template};
108
109 $log->debug("converting encoding from webpac_encoding '" .
110 $c->config->{webpac}->{webpac_encoding} .
111 "'"
112 );
113
114 $self->{databases} = $c->config->{databases} || $log->fatal("can't find databases in config");
115
116 # create Template toolkit instance
117 $self->{'tt'} = Template->new(
118 INCLUDE_PATH => $template_path,
119 FILTERS => {
120 dump_html => sub {
121 return unless (@_);
122 my $out;
123 my $i = 1;
124 foreach my $v (@_) {
125 $out .= qq{<div id="dump_$i">} .
126 Data::HTMLDumper->Dump([ $v ],[ "v$i" ]) .
127 qq{</div>};
128 $i++;
129 }
130 $out =~ s!<table[^>/]*>!<table class="dump">!gis if ($out);
131 return $out;
132 }
133 },
134 EVAL_PERL => 1,
135 );
136
137 return $self;
138
139 }
140
141
142 =head2 search
143
144 my $m->search(
145 phrase => 'query phrase',
146 add_attr => \@add_attr
147 get_attr => [ '@uri' ],
148 max => 42,
149 template => 'result_template.tt',
150 depth => 1,
151 );
152
153 All fields are standard C<WebPAC::Search::Estraier> parametars except
154 C<template> which will (if specified) return results in HTML using
155 selected template.
156
157 =cut
158
159 sub search {
160 my $self = shift;
161
162 my $search_start_t = time();
163
164 my $args = {@_};
165
166 my $log = $self->{log};
167
168 $log->debug("search args: " . Dumper( $args ));
169
170 my $query = $args->{phrase} || $log->warn("no query phrase") && return;
171
172 my $template_filename = $args->{template} || $self->{template};
173
174 $args->{max} ||= $self->{'hits_for_pager'};
175 if (! $args->{max}) {
176 $args->{max} = 100;
177 $log->warn("max not set when calling model. Using default of $args->{max}");
178 }
179
180 my $times; # store some times for benchmarking
181
182 my $t = time();
183
184 # transfer depth of search
185 if (! $args->{depth}) {
186 my $default = $self->{defaultdepth} || $log->logdie("can't find defaultdepth in estraier configuration");
187 $args->{depth} = $default;
188 $log->warn("using default search depth $default");
189 }
190 $args->{depth} ||= 0;
191
192 $log->debug("searching for maximum $args->{max} results using depth $args->{depth} phrase: ", $query || '[none]');
193
194 #
195 # construct condition for Hyper Estraier
196 #
197 my $cond = Search::Estraier::Condition->new();
198 if ( ref($args->{add_attr}) eq 'ARRAY' ) {
199 $log->debug("adding search attributes: " . join(", ", @{ $args->{add_attr} }) );
200 map {
201 $cond->add_attr( $_ );
202 $log->debug(" + $_");
203 } @{ $args->{add_attr} };
204 };
205
206 $cond->set_phrase( $query ) if ($query);
207 $cond->set_options( $args->{options} ) if ($args->{options});
208 $cond->set_order( $args->{order} ) if ($args->{order});
209
210 my $max = $args->{max} || 7;
211 my $page = $args->{page} || 1;
212 if ($page < 1) {
213 $log->warn("page number $page < 1");
214 $page = 1;
215 }
216
217 $cond->set_max( $page * $max );
218
219 my $result = $self->{est_node}->search($cond, $args->{depth});
220 my $hits = $result->doc_num;
221
222 $times->{est} += time() - $t;
223
224 $log->debug( sprintf("search took %.6fs and returned $hits hits.", $times->{est}) );
225
226 $log->debug( "hints: " . Dumper($result->{hints}) );
227
228 #
229 # fetch results
230 #
231
232 my @results;
233
234 for my $i ( (($page - 1) * $max) .. ( $hits - 1 ) ) {
235
236 $t = time();
237
238 #$log->debug("get_doc($i)");
239 my $doc = $result->get_doc( $i );
240 if (! $doc) {
241 $log->warn("can't find result $i");
242 next;
243 }
244
245 my $hash;
246
247 foreach my $attr (@{ $args->{get_attr} }) {
248 my $val = $doc->attr( $attr );
249 #$log->debug("attr $attr = ", $val || 'undef');
250 $hash->{$attr} = $val if (defined($val));
251 }
252
253 $times->{hash} += time() - $t;
254
255 next unless ($hash);
256
257 if (! $args->{'template'}) {
258 push @results, $hash;
259 } else {
260 my ($database, $prefix, $id);
261
262 if ( $hash->{'@uri'} =~ m!/([^/]+)/([^/]+)/(\d+)$!) {
263 ($database, $prefix,$id) = ($1,$2,$3);
264 } else {
265 $log->warn("can't decode database/prefix/id from " . $hash->{'@uri'});
266 next;
267 }
268
269 #$log->debug("load_ds( id => $id, prefix => '$prefix' )");
270
271 $t = time();
272
273 my $ds = $self->{db}->load_ds( database => $database, prefix => $prefix, id => $id );
274 if (! $ds) {
275 $log->error("can't load_ds( ${database}/${prefix}/${id} )");
276 next;
277 }
278
279 $times->{db} += time() - $t;
280
281 #$log->debug( "ds = " . Dumper( \@html_results ) );
282
283 $t = time();
284
285 my $html = $self->apply(
286 template => $template_filename,
287 data => $ds,
288 record_uri => "${database}/${prefix}/${id}",
289 config => $self->{databases}->{$database},
290 );
291
292 $times->{apply} += time() - $t;
293
294 $t = time();
295
296 $html = decode($self->{webpac_encoding}, $html);
297
298 $times->{decode} += time() - $t;
299
300 push @results, $html;
301 }
302
303 }
304
305 #$log->debug( '@results = ' . Dumper( \@results ) );
306
307 $log->debug( sprintf(
308 "duration breakdown: estraier %.6fs, hash %.6fs, store %.6fs, apply %.6fs, decode %.06f, total: %.6fs",
309 $times->{est}, $times->{hash}, $times->{db}, $times->{apply}, $times->{decode}, time() - $search_start_t,
310 ) );
311
312 return \@results;
313 }
314
315 =head2 record
316
317 my $html = $m->record(
318 mfn => 42,
319 template => 'foo.tt',
320 );
321
322 This will load one record, convert it to html using C<template> and return
323 it.
324
325 =cut
326
327 sub record {
328 my $self = shift;
329
330 my $args = {@_};
331 my $log = $self->{log};
332 $log->debug("record args: " . Dumper( $args ));
333
334 foreach my $f (qw/record_uri template/) {
335 $log->fatal("need $f") unless ($args->{$f});
336 }
337
338 my ($database, $prefix, $id);
339
340 if ($args->{record_uri} =~ m#^([^/]+)/([^/]+)/([^/]+)$#) {
341 ($database, $prefix, $id) = ($1,$2,$3);
342 } else {
343 $log->error("can't parse $args->{record_uri} into prefix, database and uri");
344 return;
345 }
346
347 my $ds = $self->{db}->load_ds( id => $id, prefix => $prefix, database => $database );
348 if (! $ds) {
349 $log->error("can't load_ds( $database/$prefix/$id )");
350 return;
351 }
352
353 my $html = $self->apply(
354 template => $args->{template},
355 data => $ds,
356 record_uri => $args->{record_uri},
357 config => $self->{databases}->{$database},
358 );
359
360 $html = decode($self->{webpac_encoding}, $html);
361
362 return $html;
363 }
364
365
366 =head2 save_html
367
368 $m->save_html( '/full/path/to/file', $content );
369
370 It will use C<Encode> to convert content encoding back to
371 Webpac codepage, recode JavaScript Unicode entities (%u1234),
372 strip extra newlines at beginning and end, and save to
373 C</full/path/to/file.new> and if that succeeds, just rename
374 it over original file which should be atomic on filesystem level.
375
376 =cut
377
378 sub save_html {
379 my ($self, $path, $content) = @_;
380
381 # FIXME Should this be UTF-8 or someting?
382 my $js_encoding = $self->{webpac_encoding};
383 $js_encoding = 'UTF-16';
384
385 sub _conv_js {
386 return '0x' . $_[1];
387 return encode($_[0], chr(hex($_[1])));
388 }
389 #$content =~ s/%u([a-fA-F0-9]{4})/_conv_js($js_encoding,$1)/gex;
390 $content =~ s/^[\n\r]+//s;
391 $content =~ s/[\n\r]+$/\n/s;
392 $content =~ s/\n\r/\n/gs;
393
394 my $disk_encoding = $self->{webpac_encoding} || 'utf-8';
395 $self->{log}->debug("convert encoding to $disk_encoding");
396 from_to($content, 'utf-8', $disk_encoding) || $self->{log}->warn("encoding from utf-8 to $disk_encoding failed for: $content");
397
398 write_file($path . '.new', {binmode => ':raw' }, $content) || die "can't save ${path}.new $!";
399 rename $path . '.new', $path || die "can't rename to $path: $!";
400 }
401
402 =head2 load_html
403
404 my $html = $m->load_html('/full/path/to/file');
405
406 This will convert file from Webpac encoding to Catalyst and
407 convert that data to escaped HTML (for sending into
408 C<< <textarea/> >> tags in html.
409
410 =cut
411
412 sub load_html {
413 my ($self, $path) = @_;
414
415 die "no path?" unless ($path);
416
417 my $content = read_file($path, {binmode => ':raw' }) || die "can't read $path: $!";
418
419 return decode($self->{webpac_encoding}, $content);
420 }
421
422
423 =head2 apply
424
425 Create output from in-memory data structure using Template Toolkit template.
426
427 my $text = $tt->apply(
428 template => 'text.tt',
429 data => $ds,
430 record_uri => 'database/prefix/mfn',
431 );
432
433 It also has follwing template toolikit filter routies defined:
434
435 =cut
436
437 sub apply {
438 my $self = shift;
439
440 my $args = {@_};
441
442 my $log = $self->{log} || die "no log?";
443
444 foreach my $a (qw/template data/) {
445 $log->fatal("need $a") unless ($args->{$a});
446 }
447
448 =head3 tt_filter_type
449
450 filter to return values of specified from $ds, usage from TT template is in form
451 C<d('FieldName','delimiter')>, where C<delimiter> is optional, like this:
452
453 [% d('Title') %]
454 [% d('Author',', ' %]
455
456 =cut
457
458 sub tt_filter_type {
459 my ($data,$type) = @_;
460
461 die "no data?" unless ($data);
462 $type ||= 'display';
463
464 my $default_delimiter = {
465 'display' => '&#182;<br/>',
466 'index' => '\n',
467 };
468
469 return sub {
470
471 my ($name,$join) = @_;
472
473 die "no data hash" unless ($data->{'data'} && ref($data->{'data'}) eq 'HASH');
474 # Hm? Should we die here?
475 return unless ($name);
476
477 my $item = $data->{'data'}->{$name} || return;
478
479 my $v = $item->{$type} || return;
480
481 if (ref($v) eq 'ARRAY') {
482 if ($#{$v} == 0) {
483 $v = $v->[0];
484 } else {
485 $join = $default_delimiter->{$type} unless defined($join);
486 $v = join($join, @{$v});
487 }
488 } else {
489 warn("TT filter $type(): field $name values aren't ARRAY, ignoring");
490 }
491
492 return $v;
493 }
494 }
495
496 $args->{'d'} = tt_filter_type($args, 'display');
497 $args->{'display'} = tt_filter_type($args, 'display');
498
499 =head3 tt_filter_search
500
501 filter to return links to search, usage in TT:
502
503 [% search('FieldToDisplay','FieldToSearch','optional delimiter', 'optional_template.tt') %]
504
505 =cut
506
507 sub tt_filter_search {
508
509 my ($data) = @_;
510
511 die "no data?" unless ($data);
512
513 return sub {
514
515 my ($display,$search,$delimiter,$template) = @_;
516
517 # default delimiter
518 $delimiter ||= '&#182;<br/>',
519
520 die "no data hash" unless ($data->{'data'} && ref($data->{'data'}) eq 'HASH');
521 # Hm? Should we die here?
522 return unless ($display);
523
524 my $item = $data->{'data'}->{$display} || return;
525
526 return unless($item->{'display'});
527 if (! $item->{'search'}) {
528 warn "error in TT template: field $display didn't insert anything into search, use d('$display') and not search('$display'...)";
529 return;
530 }
531
532 my @warn;
533 foreach my $type (qw/display search/) {
534 push @warn, "field $display type $type values aren't ARRAY" unless (ref($item->{$type}) eq 'ARRAY');
535 }
536
537 if (@warn) {
538 warn("TT filter search(): " . join(",", @warn) . ", skipping");
539 return;
540 }
541 my @html;
542
543 my $d_el = $#{ $item->{'display'} };
544 my $s_el = $#{ $item->{'search'} };
545
546 # easy, both fields have same number of elements or there is just
547 # one search and multiple display
548 if ( $d_el == $s_el || $s_el == 0 ) {
549
550 foreach my $i ( 0 .. $d_el ) {
551
552 my $s;
553 if ($s_el > 0) {
554 $s = $item->{'search'}->[$i] or warn "can't find value $i for type search in field $search";
555 } else {
556 $s = $item->{'search'}->[0];
557 }
558 #$s =~ s/([^\w.-])/sprintf("%%%02X",ord($1))/eg;
559 $s = __quotemeta( $s );
560
561 my $d = $item->{'display'}->[$i] or warn "can't find value $i for type display in field $display";
562
563 my $template_arg = '';
564 $template_arg = qq{,'$template'} if ($template);
565
566 push @html, qq{<a href="#" onclick="return search_via_link('$search','$s'${template_arg})">$d</a>};
567 }
568
569 return join($delimiter, @html);
570 } else {
571 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>};
572 my $v = $item->{'display'};
573
574 if ($#{$v} == 0) {
575 $html .= $v->[0];
576 } else {
577 $html .= join($delimiter, @{$v});
578 }
579 return $html;
580 }
581 }
582 }
583
584 $args->{'search'} = tt_filter_search($args);
585
586 =head3 load_rec
587
588 Used mostly for onClick events like this:
589
590 <a href="#" onClick="[% load_rec( record_uri, 'template_name.tt') %]>foo</a>
591
592 It will automatically do sanity checking and create correct JavaScript code.
593
594 =cut
595
596 $args->{'load_rec'} = sub {
597 my @errors;
598
599 my $record_uri = shift or push @errors, "record_uri missing";
600 my $template = shift or push @errors, "template missing";
601
602 if ($record_uri !~ m#^[^/]+/[^/]+/[^/]+$#) {
603 push @errors, "invalid format of record_uri: $record_uri";
604 }
605
606 if (@errors) {
607 return "Logger.error('errors in load_rec: " . join(", ", @errors) . "'); return false;";
608 } else {
609 return "load_rec('$record_uri','$template'); return false;";
610 }
611 };
612
613 =head3 load_template
614
615 Used to re-submit search request and load results in different template
616
617 <a href="#" onClick="[% load_template( 'template_name.tt' ) %]">bar</a>
618
619 =cut
620
621 $args->{'load_template'} = sub {
622 my $template = shift or return "Logger.error('load_template missing template name!'); return false;";
623 return "load_template($template); return false;";
624 };
625
626 my $out;
627
628 $self->{'tt'}->process(
629 $args->{'template'},
630 $args,
631 \$out
632 ) || $log->error( "apply can't process template: ", $self->{'tt'}->error() );
633
634 return $out;
635 }
636
637
638 =head2 __quotemeta
639
640 Helper to quote JavaScript-friendly characters
641
642 =cut
643
644 sub __quotemeta {
645 local $_ = shift;
646 $_ = decode('iso-8859-2', $_);
647
648 s<([\x{0080}-\x{fffd}]+)>{sprintf '\u%0*v4X', '\u', $1}ge if ( Encode::is_utf8($_) );
649 {
650 use bytes;
651 s<((?:[^ \x21-\x7E]|(?:\\(?!u)))+)>{sprintf '\x%0*v2X', '\x', $1}ge;
652 }
653
654 s/\\x09/\\t/g;
655 s/\\x0A/\\n/g;
656 s/\\x0D/\\r/g;
657 s/"/\\"/g;
658 s/\\x5C/\\\\/g;
659
660 return $_;
661 }
662
663
664
665 =head1 AUTHOR
666
667 Dobrica Pavlinusic C<< <dpavlin@rot13.org> >>
668
669 =head1 LICENSE
670
671 This library is free software, you can redistribute it and/or modify
672 it under the same terms as Perl itself.
673
674 =cut
675
676 1;

  ViewVC Help
Powered by ViewVC 1.1.26