/[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 384 - (show annotations)
Sun Jan 22 10:58:58 2006 UTC (18 years, 3 months ago) by dpavlin
File size: 15822 byte(s)
fix estraier time reporting

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 #
227 # fetch results
228 #
229
230 my @results;
231
232 for my $i ( (($page - 1) * $max) .. ( $hits - 1 ) ) {
233
234 $t = time();
235
236 #$log->debug("get_doc($i)");
237 my $doc = $result->get_doc( $i );
238 if (! $doc) {
239 $log->warn("can't find result $i");
240 next;
241 }
242
243 my $hash;
244
245 foreach my $attr (@{ $args->{get_attr} }) {
246 my $val = $doc->attr( $attr );
247 #$log->debug("attr $attr = ", $val || 'undef');
248 $hash->{$attr} = $val if (defined($val));
249 }
250
251 $times->{hash} += time() - $t;
252
253 next unless ($hash);
254
255 if (! $args->{'template'}) {
256 push @results, $hash;
257 } else {
258 my ($database, $prefix, $id);
259
260 if ( $hash->{'@uri'} =~ m!/([^/]+)/([^/]+)/(\d+)$!) {
261 ($database, $prefix,$id) = ($1,$2,$3);
262 } else {
263 $log->warn("can't decode database/prefix/id from " . $hash->{'@uri'});
264 next;
265 }
266
267 #$log->debug("load_ds( id => $id, prefix => '$prefix' )");
268
269 $t = time();
270
271 my $ds = $self->{db}->load_ds( database => $database, prefix => $prefix, id => $id );
272 if (! $ds) {
273 $log->error("can't load_ds( ${database}/${prefix}/${id} )");
274 next;
275 }
276
277 $times->{db} += time() - $t;
278
279 #$log->debug( "ds = " . Dumper( \@html_results ) );
280
281 $t = time();
282
283 my $html = $self->apply(
284 template => $template_filename,
285 data => $ds,
286 record_uri => "${database}/${prefix}/${id}",
287 config => $self->{databases}->{$database},
288 );
289
290 $times->{apply} += time() - $t;
291
292 $t = time();
293
294 $html = decode($self->{webpac_encoding}, $html);
295
296 $times->{decode} += time() - $t;
297
298 push @results, $html;
299 }
300
301 }
302
303 #$log->debug( '@results = ' . Dumper( \@results ) );
304
305 $log->debug( sprintf(
306 "duration breakdown: estraier %.6fs, hash %.6fs, store %.6fs, apply %.6fs, decode %.06f, total: %.6fs",
307 $times->{est}, $times->{hash}, $times->{db}, $times->{apply}, $times->{decode}, time() - $search_start_t,
308 ) );
309
310 return \@results;
311 }
312
313 =head2 record
314
315 my $html = $m->record(
316 mfn => 42,
317 template => 'foo.tt',
318 );
319
320 This will load one record, convert it to html using C<template> and return
321 it.
322
323 =cut
324
325 sub record {
326 my $self = shift;
327
328 my $args = {@_};
329 my $log = $self->{log};
330 $log->debug("record args: " . Dumper( $args ));
331
332 foreach my $f (qw/record_uri template/) {
333 $log->fatal("need $f") unless ($args->{$f});
334 }
335
336 my ($database, $prefix, $id);
337
338 if ($args->{record_uri} =~ m#^([^/]+)/([^/]+)/([^/]+)$#) {
339 ($database, $prefix, $id) = ($1,$2,$3);
340 } else {
341 $log->error("can't parse $args->{record_uri} into prefix, database and uri");
342 return;
343 }
344
345 my $ds = $self->{db}->load_ds( id => $id, prefix => $prefix, database => $database );
346 if (! $ds) {
347 $log->error("can't load_ds( $database/$prefix/$id )");
348 return;
349 }
350
351 my $html = $self->apply(
352 template => $args->{template},
353 data => $ds,
354 record_uri => $args->{record_uri},
355 config => $self->{databases}->{$database},
356 );
357
358 $html = decode($self->{webpac_encoding}, $html);
359
360 return $html;
361 }
362
363
364 =head2 save_html
365
366 $m->save_html( '/full/path/to/file', $content );
367
368 It will use C<Encode> to convert content encoding back to
369 Webpac codepage, recode JavaScript Unicode entities (%u1234),
370 strip extra newlines at beginning and end, and save to
371 C</full/path/to/file.new> and if that succeeds, just rename
372 it over original file which should be atomic on filesystem level.
373
374 =cut
375
376 sub save_html {
377 my ($self, $path, $content) = @_;
378
379 # FIXME Should this be UTF-8 or someting?
380 my $js_encoding = $self->{webpac_encoding};
381 $js_encoding = 'UTF-16';
382
383 sub _conv_js {
384 return '0x' . $_[1];
385 return encode($_[0], chr(hex($_[1])));
386 }
387 #$content =~ s/%u([a-fA-F0-9]{4})/_conv_js($js_encoding,$1)/gex;
388 $content =~ s/^[\n\r]+//s;
389 $content =~ s/[\n\r]+$/\n/s;
390 $content =~ s/\n\r/\n/gs;
391
392 my $disk_encoding = $self->{webpac_encoding} || 'utf-8';
393 $self->{log}->debug("convert encoding to $disk_encoding");
394 from_to($content, 'utf-8', $disk_encoding) || $self->{log}->warn("encoding from utf-8 to $disk_encoding failed for: $content");
395
396 write_file($path . '.new', {binmode => ':raw' }, $content) || die "can't save ${path}.new $!";
397 rename $path . '.new', $path || die "can't rename to $path: $!";
398 }
399
400 =head2 load_html
401
402 my $html = $m->load_html('/full/path/to/file');
403
404 This will convert file from Webpac encoding to Catalyst and
405 convert that data to escaped HTML (for sending into
406 C<< <textarea/> >> tags in html.
407
408 =cut
409
410 sub load_html {
411 my ($self, $path) = @_;
412
413 die "no path?" unless ($path);
414
415 my $content = read_file($path, {binmode => ':raw' }) || die "can't read $path: $!";
416
417 return decode($self->{webpac_encoding}, $content);
418 }
419
420
421 =head2 apply
422
423 Create output from in-memory data structure using Template Toolkit template.
424
425 my $text = $tt->apply(
426 template => 'text.tt',
427 data => $ds,
428 record_uri => 'database/prefix/mfn',
429 );
430
431 It also has follwing template toolikit filter routies defined:
432
433 =cut
434
435 sub apply {
436 my $self = shift;
437
438 my $args = {@_};
439
440 my $log = $self->{log} || die "no log?";
441
442 foreach my $a (qw/template data/) {
443 $log->fatal("need $a") unless ($args->{$a});
444 }
445
446 =head3 tt_filter_type
447
448 filter to return values of specified from $ds, usage from TT template is in form
449 C<d('FieldName','delimiter')>, where C<delimiter> is optional, like this:
450
451 [% d('Title') %]
452 [% d('Author',', ' %]
453
454 =cut
455
456 sub tt_filter_type {
457 my ($data,$type) = @_;
458
459 die "no data?" unless ($data);
460 $type ||= 'display';
461
462 my $default_delimiter = {
463 'display' => '&#182;<br/>',
464 'index' => '\n',
465 };
466
467 return sub {
468
469 my ($name,$join) = @_;
470
471 die "no data hash" unless ($data->{'data'} && ref($data->{'data'}) eq 'HASH');
472 # Hm? Should we die here?
473 return unless ($name);
474
475 my $item = $data->{'data'}->{$name} || return;
476
477 my $v = $item->{$type} || return;
478
479 if (ref($v) eq 'ARRAY') {
480 if ($#{$v} == 0) {
481 $v = $v->[0];
482 } else {
483 $join = $default_delimiter->{$type} unless defined($join);
484 $v = join($join, @{$v});
485 }
486 } else {
487 warn("TT filter $type(): field $name values aren't ARRAY, ignoring");
488 }
489
490 return $v;
491 }
492 }
493
494 $args->{'d'} = tt_filter_type($args, 'display');
495 $args->{'display'} = tt_filter_type($args, 'display');
496
497 =head3 tt_filter_search
498
499 filter to return links to search, usage in TT:
500
501 [% search('FieldToDisplay','FieldToSearch','optional delimiter', 'optional_template.tt') %]
502
503 =cut
504
505 sub tt_filter_search {
506
507 my ($data) = @_;
508
509 die "no data?" unless ($data);
510
511 return sub {
512
513 my ($display,$search,$delimiter,$template) = @_;
514
515 # default delimiter
516 $delimiter ||= '&#182;<br/>',
517
518 die "no data hash" unless ($data->{'data'} && ref($data->{'data'}) eq 'HASH');
519 # Hm? Should we die here?
520 return unless ($display);
521
522 my $item = $data->{'data'}->{$display} || return;
523
524 return unless($item->{'display'});
525 if (! $item->{'search'}) {
526 warn "error in TT template: field $display didn't insert anything into search, use d('$display') and not search('$display'...)";
527 return;
528 }
529
530 my @warn;
531 foreach my $type (qw/display search/) {
532 push @warn, "field $display type $type values aren't ARRAY" unless (ref($item->{$type}) eq 'ARRAY');
533 }
534
535 if (@warn) {
536 warn("TT filter search(): " . join(",", @warn) . ", skipping");
537 return;
538 }
539 my @html;
540
541 my $d_el = $#{ $item->{'display'} };
542 my $s_el = $#{ $item->{'search'} };
543
544 # easy, both fields have same number of elements or there is just
545 # one search and multiple display
546 if ( $d_el == $s_el || $s_el == 0 ) {
547
548 foreach my $i ( 0 .. $d_el ) {
549
550 my $s;
551 if ($s_el > 0) {
552 $s = $item->{'search'}->[$i] or warn "can't find value $i for type search in field $search";
553 } else {
554 $s = $item->{'search'}->[0];
555 }
556 #$s =~ s/([^\w.-])/sprintf("%%%02X",ord($1))/eg;
557 $s = __quotemeta( $s );
558
559 my $d = $item->{'display'}->[$i] or warn "can't find value $i for type display in field $display";
560
561 my $template_arg = '';
562 $template_arg = qq{,'$template'} if ($template);
563
564 push @html, qq{<a href="#" onclick="return search_via_link('$search','$s'${template_arg})">$d</a>};
565 }
566
567 return join($delimiter, @html);
568 } else {
569 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>};
570 my $v = $item->{'display'};
571
572 if ($#{$v} == 0) {
573 $html .= $v->[0];
574 } else {
575 $html .= join($delimiter, @{$v});
576 }
577 return $html;
578 }
579 }
580 }
581
582 $args->{'search'} = tt_filter_search($args);
583
584 =head3 load_rec
585
586 Used mostly for onClick events like this:
587
588 <a href="#" onClick="[% load_rec( record_uri, 'template_name.tt') %]>foo</a>
589
590 It will automatically do sanity checking and create correct JavaScript code.
591
592 =cut
593
594 $args->{'load_rec'} = sub {
595 my @errors;
596
597 my $record_uri = shift or push @errors, "record_uri missing";
598 my $template = shift or push @errors, "template missing";
599
600 if ($record_uri !~ m#^[^/]+/[^/]+/[^/]+$#) {
601 push @errors, "invalid format of record_uri: $record_uri";
602 }
603
604 if (@errors) {
605 return "Logger.error('errors in load_rec: " . join(", ", @errors) . "'); return false;";
606 } else {
607 return "load_rec('$record_uri','$template'); return false;";
608 }
609 };
610
611 =head3 load_template
612
613 Used to re-submit search request and load results in different template
614
615 <a href="#" onClick="[% load_template( 'template_name.tt' ) %]">bar</a>
616
617 =cut
618
619 $args->{'load_template'} = sub {
620 my $template = shift or return "Logger.error('load_template missing template name!'); return false;";
621 return "load_template($template); return false;";
622 };
623
624 my $out;
625
626 $self->{'tt'}->process(
627 $args->{'template'},
628 $args,
629 \$out
630 ) || $log->error( "apply can't process template: ", $self->{'tt'}->error() );
631
632 return $out;
633 }
634
635
636 =head2 __quotemeta
637
638 Helper to quote JavaScript-friendly characters
639
640 =cut
641
642 sub __quotemeta {
643 local $_ = shift;
644 $_ = decode('iso-8859-2', $_);
645
646 s<([\x{0080}-\x{fffd}]+)>{sprintf '\u%0*v4X', '\u', $1}ge if ( Encode::is_utf8($_) );
647 {
648 use bytes;
649 s<((?:[^ \x21-\x7E]|(?:\\(?!u)))+)>{sprintf '\x%0*v2X', '\x', $1}ge;
650 }
651
652 s/\\x09/\\t/g;
653 s/\\x0A/\\n/g;
654 s/\\x0D/\\r/g;
655 s/"/\\"/g;
656 s/\\x5C/\\\\/g;
657
658 return $_;
659 }
660
661
662
663 =head1 AUTHOR
664
665 Dobrica Pavlinusic C<< <dpavlin@rot13.org> >>
666
667 =head1 LICENSE
668
669 This library is free software, you can redistribute it and/or modify
670 it under the same terms as Perl itself.
671
672 =cut
673
674 1;

  ViewVC Help
Powered by ViewVC 1.1.26