/[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 382 - (show annotations)
Sun Jan 22 02:52:24 2006 UTC (18 years, 3 months ago) by dpavlin
File size: 15940 byte(s)
 r432@llin:  dpavlin | 2006-01-22 03:43:55 +0100
 remove WebPAC::Search::Estraier usage and replace it with calls to Search::Estraier
 directly, simplifing code in process

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 $log->debug("search model query: '$query'");
173 if ($args->{add_attr}) {
174 $log->debug(" + add_attr: " .
175 join("','", @{ $args->{add_attr} })
176 );
177 }
178
179 my $template_filename = $args->{template} || $self->{template};
180
181 $args->{max} ||= $self->{'hits_for_pager'};
182 if (! $args->{max}) {
183 $args->{max} = 100;
184 $log->warn("max not set when calling model. Using default of $args->{max}");
185 }
186
187 my $times; # store some times for benchmarking
188
189 my $t = time();
190
191 # transfer depth of search
192 if (! $args->{depth}) {
193 my $default = $self->{defaultdepth} || $log->logdie("can't find defaultdepth in estraier configuration");
194 $args->{depth} = $default;
195 $log->warn("using default search depth $default");
196 }
197
198 $log->debug("searching for maximum $args->{max} results using depth $args->{depth}");
199
200 #
201 # construct condition for Hyper Estraier
202 #
203 my $cond = Search::Estraier::Condition->new();
204 if ( ref($args->{add_attr}) eq 'ARRAY' ) {
205 $log->debug("adding search attributes: " . join(", ", @{ $args->{add_attr} }) );
206 map {
207 $cond->add_attr( _convert( $_ ) );
208 $log->debug(" + $_");
209 } @{ $args->{add_attr} };
210 };
211
212 $cond->set_phrase( $query ) if ($query);
213 $cond->set_options( $args->{options} ) if ($args->{options});
214 $cond->set_order( $args->{order} ) if ($args->{order});
215
216 my $max = $args->{max} || 7;
217 my $page = $args->{page} || 1;
218 if ($page < 1) {
219 $log->warn("page number $page < 1");
220 $page = 1;
221 }
222
223 $times->{est} += time() - $t;
224
225 $cond->set_max( $page * $max );
226
227 my $result = $self->{est_node}->search($cond, ( $args->{depth} || 0 ));
228 my $hits = $result->doc_num;
229
230 $log->debug( sprintf("search took %.6fs and returned $hits hits.", $times->{est}) );
231
232 #
233 # fetch results
234 #
235
236 my @results;
237
238 for my $i ( (($page - 1) * $max) .. ( $hits - 1 ) ) {
239
240 $t = time();
241
242 #$log->debug("get_doc($i)");
243 my $doc = $result->get_doc( $i );
244 if (! $doc) {
245 $log->warn("can't find result $i");
246 next;
247 }
248
249 my $hash;
250
251 foreach my $attr (@{ $args->{get_attr} }) {
252 my $val = $doc->attr( $attr );
253 #$log->debug("attr $attr = ", $val || 'undef');
254 $hash->{$attr} = $val if (defined($val));
255 }
256
257 $times->{hash} += time() - $t;
258
259 next unless ($hash);
260
261 if (! $args->{'template'}) {
262 push @results, $hash;
263 } else {
264 my ($database, $prefix, $id);
265
266 if ( $hash->{'@uri'} =~ m!/([^/]+)/([^/]+)/(\d+)$!) {
267 ($database, $prefix,$id) = ($1,$2,$3);
268 } else {
269 $log->warn("can't decode database/prefix/id from " . $hash->{'@uri'});
270 next;
271 }
272
273 #$log->debug("load_ds( id => $id, prefix => '$prefix' )");
274
275 $t = time();
276
277 my $ds = $self->{db}->load_ds( database => $database, prefix => $prefix, id => $id );
278 if (! $ds) {
279 $log->error("can't load_ds( ${database}/${prefix}/${id} )");
280 next;
281 }
282
283 $times->{db} += time() - $t;
284
285 #$log->debug( "ds = " . Dumper( \@html_results ) );
286
287 $t = time();
288
289 my $html = $self->apply(
290 template => $template_filename,
291 data => $ds,
292 record_uri => "${database}/${prefix}/${id}",
293 config => $self->{databases}->{$database},
294 );
295
296 $times->{apply} += time() - $t;
297
298 $t = time();
299
300 $html = decode($self->{webpac_encoding}, $html);
301
302 $times->{decode} += time() - $t;
303
304 push @results, $html;
305 }
306
307 }
308
309 $log->debug( '@results = ' . Dumper( \@results ) );
310
311 $log->debug( sprintf(
312 "duration breakdown: estraier %.6fs, hash %.6fs, store %.6fs, apply %.6fs, decode %.06f, total: %.6fs",
313 $times->{est}, $times->{hash}, $times->{db}, $times->{apply}, $times->{decode}, time() - $search_start_t,
314 ) );
315
316 return \@results;
317 }
318
319 =head2 record
320
321 my $html = $m->record(
322 mfn => 42,
323 template => 'foo.tt',
324 );
325
326 This will load one record, convert it to html using C<template> and return
327 it.
328
329 =cut
330
331 sub record {
332 my $self = shift;
333
334 my $args = {@_};
335 my $log = $self->{log};
336 $log->debug("record args: " . Dumper( $args ));
337
338 foreach my $f (qw/record_uri template/) {
339 $log->fatal("need $f") unless ($args->{$f});
340 }
341
342 my ($database, $prefix, $id);
343
344 if ($args->{record_uri} =~ m#^([^/]+)/([^/]+)/([^/]+)$#) {
345 ($database, $prefix, $id) = ($1,$2,$3);
346 } else {
347 $log->error("can't parse $args->{record_uri} into prefix, database and uri");
348 return;
349 }
350
351 my $ds = $self->{db}->load_ds( id => $id, prefix => $prefix, database => $database );
352 if (! $ds) {
353 $log->error("can't load_ds( $database/$prefix/$id )");
354 return;
355 }
356
357 my $html = $self->apply(
358 template => $args->{template},
359 data => $ds,
360 record_uri => $args->{record_uri},
361 config => $self->{databases}->{$database},
362 );
363
364 $html = decode($self->{webpac_encoding}, $html);
365
366 return $html;
367 }
368
369
370 =head2 save_html
371
372 $m->save_html( '/full/path/to/file', $content );
373
374 It will use C<Encode> to convert content encoding back to
375 Webpac codepage, recode JavaScript Unicode entities (%u1234),
376 strip extra newlines at beginning and end, and save to
377 C</full/path/to/file.new> and if that succeeds, just rename
378 it over original file which should be atomic on filesystem level.
379
380 =cut
381
382 sub save_html {
383 my ($self, $path, $content) = @_;
384
385 # FIXME Should this be UTF-8 or someting?
386 my $js_encoding = $self->{webpac_encoding};
387 $js_encoding = 'UTF-16';
388
389 sub _conv_js {
390 return '0x' . $_[1];
391 return encode($_[0], chr(hex($_[1])));
392 }
393 #$content =~ s/%u([a-fA-F0-9]{4})/_conv_js($js_encoding,$1)/gex;
394 $content =~ s/^[\n\r]+//s;
395 $content =~ s/[\n\r]+$/\n/s;
396 $content =~ s/\n\r/\n/gs;
397
398 my $disk_encoding = $self->{webpac_encoding} || 'utf-8';
399 $self->{log}->debug("convert encoding to $disk_encoding");
400 from_to($content, 'utf-8', $disk_encoding) || $self->{log}->warn("encoding from utf-8 to $disk_encoding failed for: $content");
401
402 write_file($path . '.new', {binmode => ':raw' }, $content) || die "can't save ${path}.new $!";
403 rename $path . '.new', $path || die "can't rename to $path: $!";
404 }
405
406 =head2 load_html
407
408 my $html = $m->load_html('/full/path/to/file');
409
410 This will convert file from Webpac encoding to Catalyst and
411 convert that data to escaped HTML (for sending into
412 C<< <textarea/> >> tags in html.
413
414 =cut
415
416 sub load_html {
417 my ($self, $path) = @_;
418
419 die "no path?" unless ($path);
420
421 my $content = read_file($path, {binmode => ':raw' }) || die "can't read $path: $!";
422
423 return decode($self->{webpac_encoding}, $content);
424 }
425
426
427 =head2 apply
428
429 Create output from in-memory data structure using Template Toolkit template.
430
431 my $text = $tt->apply(
432 template => 'text.tt',
433 data => $ds,
434 record_uri => 'database/prefix/mfn',
435 );
436
437 It also has follwing template toolikit filter routies defined:
438
439 =cut
440
441 sub apply {
442 my $self = shift;
443
444 my $args = {@_};
445
446 my $log = $self->{log} || die "no log?";
447
448 foreach my $a (qw/template data/) {
449 $log->fatal("need $a") unless ($args->{$a});
450 }
451
452 =head3 tt_filter_type
453
454 filter to return values of specified from $ds, usage from TT template is in form
455 C<d('FieldName','delimiter')>, where C<delimiter> is optional, like this:
456
457 [% d('Title') %]
458 [% d('Author',', ' %]
459
460 =cut
461
462 sub tt_filter_type {
463 my ($data,$type) = @_;
464
465 die "no data?" unless ($data);
466 $type ||= 'display';
467
468 my $default_delimiter = {
469 'display' => '&#182;<br/>',
470 'index' => '\n',
471 };
472
473 return sub {
474
475 my ($name,$join) = @_;
476
477 die "no data hash" unless ($data->{'data'} && ref($data->{'data'}) eq 'HASH');
478 # Hm? Should we die here?
479 return unless ($name);
480
481 my $item = $data->{'data'}->{$name} || return;
482
483 my $v = $item->{$type} || return;
484
485 if (ref($v) eq 'ARRAY') {
486 if ($#{$v} == 0) {
487 $v = $v->[0];
488 } else {
489 $join = $default_delimiter->{$type} unless defined($join);
490 $v = join($join, @{$v});
491 }
492 } else {
493 warn("TT filter $type(): field $name values aren't ARRAY, ignoring");
494 }
495
496 return $v;
497 }
498 }
499
500 $args->{'d'} = tt_filter_type($args, 'display');
501 $args->{'display'} = tt_filter_type($args, 'display');
502
503 =head3 tt_filter_search
504
505 filter to return links to search, usage in TT:
506
507 [% search('FieldToDisplay','FieldToSearch','optional delimiter', 'optional_template.tt') %]
508
509 =cut
510
511 sub tt_filter_search {
512
513 my ($data) = @_;
514
515 die "no data?" unless ($data);
516
517 return sub {
518
519 my ($display,$search,$delimiter,$template) = @_;
520
521 # default delimiter
522 $delimiter ||= '&#182;<br/>',
523
524 die "no data hash" unless ($data->{'data'} && ref($data->{'data'}) eq 'HASH');
525 # Hm? Should we die here?
526 return unless ($display);
527
528 my $item = $data->{'data'}->{$display} || return;
529
530 return unless($item->{'display'});
531 if (! $item->{'search'}) {
532 warn "error in TT template: field $display didn't insert anything into search, use d('$display') and not search('$display'...)";
533 return;
534 }
535
536 my @warn;
537 foreach my $type (qw/display search/) {
538 push @warn, "field $display type $type values aren't ARRAY" unless (ref($item->{$type}) eq 'ARRAY');
539 }
540
541 if (@warn) {
542 warn("TT filter search(): " . join(",", @warn) . ", skipping");
543 return;
544 }
545 my @html;
546
547 my $d_el = $#{ $item->{'display'} };
548 my $s_el = $#{ $item->{'search'} };
549
550 # easy, both fields have same number of elements or there is just
551 # one search and multiple display
552 if ( $d_el == $s_el || $s_el == 0 ) {
553
554 foreach my $i ( 0 .. $d_el ) {
555
556 my $s;
557 if ($s_el > 0) {
558 $s = $item->{'search'}->[$i] || die "can't find value $i for type search in field $search";
559 } else {
560 $s = $item->{'search'}->[0];
561 }
562 #$s =~ s/([^\w.-])/sprintf("%%%02X",ord($1))/eg;
563 $s = __quotemeta( $s );
564
565 my $d = $item->{'display'}->[$i] || die "can't find value $i for type display in field $display";
566
567 my $template_arg = '';
568 $template_arg = qq{,'$template'} if ($template);
569
570 push @html, qq{<a href="#" onclick="return search_via_link('$search','$s'${template_arg})">$d</a>};
571 }
572
573 return join($delimiter, @html);
574 } else {
575 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>};
576 my $v = $item->{'display'};
577
578 if ($#{$v} == 0) {
579 $html .= $v->[0];
580 } else {
581 $html .= join($delimiter, @{$v});
582 }
583 return $html;
584 }
585 }
586 }
587
588 $args->{'search'} = tt_filter_search($args);
589
590 =head3 load_rec
591
592 Used mostly for onClick events like this:
593
594 <a href="#" onClick="[% load_rec( record_uri, 'template_name.tt') %]>foo</a>
595
596 It will automatically do sanity checking and create correct JavaScript code.
597
598 =cut
599
600 $args->{'load_rec'} = sub {
601 my @errors;
602
603 my $record_uri = shift or push @errors, "record_uri missing";
604 my $template = shift or push @errors, "template missing";
605
606 if ($record_uri !~ m#^[^/]+/[^/]+/[^/]+$#) {
607 push @errors, "invalid format of record_uri: $record_uri";
608 }
609
610 if (@errors) {
611 return "Logger.error('errors in load_rec: " . join(", ", @errors) . "'); return false;";
612 } else {
613 return "load_rec('$record_uri','$template'); return false;";
614 }
615 };
616
617 =head3 load_template
618
619 Used to re-submit search request and load results in different template
620
621 <a href="#" onClick="[% load_template( 'template_name.tt' ) %]">bar</a>
622
623 =cut
624
625 $args->{'load_template'} = sub {
626 my $template = shift or return "Logger.error('load_template missing template name!'); return false;";
627 return "load_template($template); return false;";
628 };
629
630 my $out;
631
632 $self->{'tt'}->process(
633 $args->{'template'},
634 $args,
635 \$out
636 ) || $log->error( "apply can't process template: ", $self->{'tt'}->error() );
637
638 return $out;
639 }
640
641
642 =head2 __quotemeta
643
644 Helper to quote JavaScript-friendly characters
645
646 =cut
647
648 sub __quotemeta {
649 local $_ = shift;
650 $_ = decode('iso-8859-2', $_);
651
652 s<([\x{0080}-\x{fffd}]+)>{sprintf '\u%0*v4X', '\u', $1}ge if ( Encode::is_utf8($_) );
653 {
654 use bytes;
655 s<((?:[^ \x21-\x7E]|(?:\\(?!u)))+)>{sprintf '\x%0*v2X', '\x', $1}ge;
656 }
657
658 s/\\x09/\\t/g;
659 s/\\x0A/\\n/g;
660 s/\\x0D/\\r/g;
661 s/"/\\"/g;
662 s/\\x5C/\\\\/g;
663
664 return $_;
665 }
666
667
668
669 =head1 AUTHOR
670
671 Dobrica Pavlinusic C<< <dpavlin@rot13.org> >>
672
673 =head1 LICENSE
674
675 This library is free software, you can redistribute it and/or modify
676 it under the same terms as Perl itself.
677
678 =cut
679
680 1;

  ViewVC Help
Powered by ViewVC 1.1.26