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 WebPAC::Search::Estraier 0.05; |
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 |
$self->{est} = new WebPAC::Search::Estraier( %{ $est_cfg } ); |
78 |
|
79 |
# save config parametars in object |
80 |
foreach my $f (qw/db_path template_path hits_on_page webpac_encoding defaultdepth/) { |
81 |
$self->{$f} = $c->config->{hyperestraier}->{$f} || |
82 |
$c->config->{webpac}->{$f}; |
83 |
$log->debug("self->{$f} = " . $self->{$f}); |
84 |
} |
85 |
my $db_path = $self->{db_path}; |
86 |
my $template_path = $self->{template_path}; |
87 |
|
88 |
$log->debug("using db path '$db_path', template path '$template_path'"); |
89 |
|
90 |
$self->{db} = new WebPAC::Store( |
91 |
path => $db_path, |
92 |
read_only => 1, |
93 |
database => $est_cfg->{database}, |
94 |
); |
95 |
|
96 |
# default template from config.yaml |
97 |
$self->{template} ||= $c->config->{webpac}->{template}; |
98 |
|
99 |
$log->debug("converting encoding from webpac_encoding '" . |
100 |
$c->config->{webpac}->{webpac_encoding} . |
101 |
"'" |
102 |
); |
103 |
|
104 |
$self->{databases} = $c->config->{databases} || $log->error("can't find databases in config"); |
105 |
|
106 |
# create Template toolkit instance |
107 |
$self->{'tt'} = Template->new( |
108 |
INCLUDE_PATH => $template_path, |
109 |
FILTERS => { |
110 |
dump_html => sub { |
111 |
return unless (@_); |
112 |
my $out; |
113 |
my $i = 1; |
114 |
foreach my $v (@_) { |
115 |
$out .= qq{<div id="dump_$i">} . |
116 |
Data::HTMLDumper->Dump([ $v ],[ "v$i" ]) . |
117 |
qq{</div>}; |
118 |
$i++; |
119 |
} |
120 |
$out =~ s!<table[^>/]*>!<table class="dump">!gis if ($out); |
121 |
return $out; |
122 |
} |
123 |
}, |
124 |
EVAL_PERL => 1, |
125 |
); |
126 |
|
127 |
return $self; |
128 |
|
129 |
} |
130 |
|
131 |
|
132 |
=head2 search |
133 |
|
134 |
my $m->search( |
135 |
phrase => 'query phrase', |
136 |
add_attr => \@add_attr |
137 |
get_attr => [ '@uri' ], |
138 |
max => 42, |
139 |
template => 'result_template.tt', |
140 |
depth => 1, |
141 |
); |
142 |
|
143 |
All fields are standard C<WebPAC::Search::Estraier> parametars except |
144 |
C<template> which will (if specified) return results in HTML using |
145 |
selected template. |
146 |
|
147 |
=cut |
148 |
|
149 |
sub search { |
150 |
my $self = shift; |
151 |
|
152 |
my $search_start_t = time(); |
153 |
|
154 |
my $args = {@_}; |
155 |
|
156 |
my $log = $self->{log}; |
157 |
|
158 |
$log->debug("search args: " . Dumper( $args )); |
159 |
|
160 |
my $query = $args->{phrase} || $log->warn("no query phrase") && return; |
161 |
|
162 |
$log->debug("search model query: '$query'"); |
163 |
if ($args->{add_attr}) { |
164 |
$log->debug(" + add_attr: " . |
165 |
join("','", @{ $args->{add_attr} }) |
166 |
); |
167 |
} |
168 |
|
169 |
my $template_filename = $args->{template} || $self->{template}; |
170 |
|
171 |
$args->{max} ||= $self->{'hits_for_pager'}; |
172 |
if (! $args->{max}) { |
173 |
$args->{max} = 100; |
174 |
$log->warn("max not set when calling model. Using default of $args->{max}"); |
175 |
} |
176 |
|
177 |
my $times; # store some times for benchmarking |
178 |
|
179 |
my $t = time(); |
180 |
|
181 |
# transfer depth of search |
182 |
if (! $args->{depth}) { |
183 |
my $default = $self->{defaultdepth} || $log->logdie("can't find defaultdepth in estraier configuration"); |
184 |
$args->{depth} = $default; |
185 |
$log->warn("using default search depth $default"); |
186 |
} |
187 |
|
188 |
my @results = $self->{est}->search( %{ $args } ); |
189 |
|
190 |
$times->{est} += time() - $t; |
191 |
|
192 |
my $hits = $#results + 1; |
193 |
|
194 |
$log->debug( sprintf("search took %.6fs and returned $hits hits.", $times->{est}) ); |
195 |
|
196 |
# just return results? |
197 |
return @results unless ($args->{'template'}); |
198 |
|
199 |
# |
200 |
# construct HTML results |
201 |
# |
202 |
|
203 |
my @html_results; |
204 |
|
205 |
for my $i ( 0 .. $#results ) { |
206 |
|
207 |
my ($database, $prefix, $id); |
208 |
if ( $results[$i]->{'@uri'} =~ m!/([^/]+)/([^/]+)/(\d+)$!) { |
209 |
($database, $prefix,$id) = ($1,$2,$3); |
210 |
} else { |
211 |
$log->warn("can't decode database/prefix/id from " . $results[$i]->{'@uri'}); |
212 |
next; |
213 |
} |
214 |
|
215 |
#$log->debug("load_ds( id => $id, prefix => '$prefix' )"); |
216 |
|
217 |
$t = time(); |
218 |
|
219 |
my $ds = $self->{db}->load_ds( database => $database, prefix => $prefix, id => $id ); |
220 |
if (! $ds) { |
221 |
$log->error("can't load_ds( ${database}/${prefix}/${id} )"); |
222 |
next; |
223 |
} |
224 |
|
225 |
$times->{db} += time() - $t; |
226 |
|
227 |
#$log->debug( "ds = " . Dumper( \@html_results ) ); |
228 |
|
229 |
$t = time(); |
230 |
|
231 |
my $html = $self->apply( |
232 |
template => $template_filename, |
233 |
data => $ds, |
234 |
record_uri => "${database}/${prefix}/${id}", |
235 |
config => $self->{databases}->{$database}, |
236 |
); |
237 |
|
238 |
$times->{out} += time() - $t; |
239 |
|
240 |
$t = time(); |
241 |
|
242 |
$html = decode($self->{webpac_encoding}, $html); |
243 |
|
244 |
push @html_results, $html; |
245 |
|
246 |
} |
247 |
|
248 |
#$log->debug( '@html_results = ' . Dumper( \@html_results ) ); |
249 |
|
250 |
$log->debug( sprintf( |
251 |
"duration breakdown: store %.6fs, apply %.6fs, total: %.6fs", |
252 |
$times->{db}, $times->{out}, time() - $search_start_t, |
253 |
) ); |
254 |
|
255 |
return \@html_results; |
256 |
} |
257 |
|
258 |
=head2 record |
259 |
|
260 |
my $html = $m->record( |
261 |
mfn => 42, |
262 |
template => 'foo.tt', |
263 |
); |
264 |
|
265 |
This will load one record, convert it to html using C<template> and return |
266 |
it. |
267 |
|
268 |
=cut |
269 |
|
270 |
sub record { |
271 |
my $self = shift; |
272 |
|
273 |
my $args = {@_}; |
274 |
my $log = $self->{log}; |
275 |
$log->debug("record args: " . Dumper( $args )); |
276 |
|
277 |
foreach my $f (qw/record_uri template/) { |
278 |
$log->fatal("need $f") unless ($args->{$f}); |
279 |
} |
280 |
|
281 |
my ($database, $prefix, $id); |
282 |
|
283 |
if ($args->{record_uri} =~ m#^([^/]+)/([^/]+)/([^/]+)$#) { |
284 |
($database, $prefix, $id) = ($1,$2,$3); |
285 |
} else { |
286 |
$log->error("can't parse $args->{record_uri} into prefix, database and uri"); |
287 |
return; |
288 |
} |
289 |
|
290 |
my $ds = $self->{db}->load_ds( id => $id, prefix => $prefix, database => $database ); |
291 |
if (! $ds) { |
292 |
$log->error("can't load_ds( $database/$prefix/$id )"); |
293 |
return; |
294 |
} |
295 |
|
296 |
my $html = $self->apply( |
297 |
template => $args->{template}, |
298 |
data => $ds, |
299 |
record_uri => $args->{record_uri}, |
300 |
config => $self->{databases}->{$database}, |
301 |
); |
302 |
|
303 |
$html = decode($self->{webpac_encoding}, $html); |
304 |
|
305 |
return $html; |
306 |
} |
307 |
|
308 |
|
309 |
=head2 save_html |
310 |
|
311 |
$m->save_html( '/full/path/to/file', $content ); |
312 |
|
313 |
It will use C<Encode> to convert content encoding back to |
314 |
Webpac codepage, recode JavaScript Unicode entities (%u1234), |
315 |
strip extra newlines at beginning and end, and save to |
316 |
C</full/path/to/file.new> and if that succeeds, just rename |
317 |
it over original file which should be atomic on filesystem level. |
318 |
|
319 |
=cut |
320 |
|
321 |
sub save_html { |
322 |
my ($self, $path, $content) = @_; |
323 |
|
324 |
# FIXME Should this be UTF-8 or someting? |
325 |
my $js_encoding = $self->{webpac_encoding}; |
326 |
$js_encoding = 'UTF-16'; |
327 |
|
328 |
sub _conv_js { |
329 |
return '0x' . $_[1]; |
330 |
return encode($_[0], chr(hex($_[1]))); |
331 |
} |
332 |
#$content =~ s/%u([a-fA-F0-9]{4})/_conv_js($js_encoding,$1)/gex; |
333 |
$content =~ s/^[\n\r]+//s; |
334 |
$content =~ s/[\n\r]+$/\n/s; |
335 |
$content =~ s/\n\r/\n/gs; |
336 |
|
337 |
my $disk_encoding = $self->{webpac_encoding} || 'utf-8'; |
338 |
$self->{log}->debug("convert encoding to $disk_encoding"); |
339 |
from_to($content, 'utf-8', $disk_encoding) || $self->{log}->warn("encoding from utf-8 to $disk_encoding failed for: $content"); |
340 |
|
341 |
write_file($path . '.new', {binmode => ':raw' }, $content) || die "can't save ${path}.new $!"; |
342 |
rename $path . '.new', $path || die "can't rename to $path: $!"; |
343 |
} |
344 |
|
345 |
=head2 load_html |
346 |
|
347 |
my $html = $m->load_html('/full/path/to/file'); |
348 |
|
349 |
This will convert file from Webpac encoding to Catalyst and |
350 |
convert that data to escaped HTML (for sending into |
351 |
C<< <textarea/> >> tags in html. |
352 |
|
353 |
=cut |
354 |
|
355 |
sub load_html { |
356 |
my ($self, $path) = @_; |
357 |
|
358 |
die "no path?" unless ($path); |
359 |
|
360 |
my $content = read_file($path, {binmode => ':raw' }) || die "can't read $path: $!"; |
361 |
|
362 |
return decode($self->{webpac_encoding}, $content); |
363 |
} |
364 |
|
365 |
|
366 |
=head2 apply |
367 |
|
368 |
Create output from in-memory data structure using Template Toolkit template. |
369 |
|
370 |
my $text = $tt->apply( |
371 |
template => 'text.tt', |
372 |
data => $ds, |
373 |
record_uri => 'database/prefix/mfn', |
374 |
); |
375 |
|
376 |
It also has follwing template toolikit filter routies defined: |
377 |
|
378 |
=cut |
379 |
|
380 |
sub apply { |
381 |
my $self = shift; |
382 |
|
383 |
my $args = {@_}; |
384 |
|
385 |
my $log = $self->{log} || die "no log?"; |
386 |
|
387 |
foreach my $a (qw/template data/) { |
388 |
$log->logconfess("need $a") unless ($args->{$a}); |
389 |
} |
390 |
|
391 |
=head3 tt_filter_type |
392 |
|
393 |
filter to return values of specified from $ds, usage from TT template is in form |
394 |
C<d('FieldName','delimiter')>, where C<delimiter> is optional, like this: |
395 |
|
396 |
[% d('Title') %] |
397 |
[% d('Author',', ' %] |
398 |
|
399 |
=cut |
400 |
|
401 |
sub tt_filter_type { |
402 |
my ($data,$type) = @_; |
403 |
|
404 |
die "no data?" unless ($data); |
405 |
$type ||= 'display'; |
406 |
|
407 |
my $default_delimiter = { |
408 |
'display' => '¶<br/>', |
409 |
'index' => '\n', |
410 |
}; |
411 |
|
412 |
return sub { |
413 |
|
414 |
my ($name,$join) = @_; |
415 |
|
416 |
die "no data hash" unless ($data->{'data'} && ref($data->{'data'}) eq 'HASH'); |
417 |
# Hm? Should we die here? |
418 |
return unless ($name); |
419 |
|
420 |
my $item = $data->{'data'}->{$name} || return; |
421 |
|
422 |
my $v = $item->{$type} || return; |
423 |
|
424 |
if (ref($v) eq 'ARRAY') { |
425 |
if ($#{$v} == 0) { |
426 |
$v = $v->[0]; |
427 |
} else { |
428 |
$join = $default_delimiter->{$type} unless defined($join); |
429 |
$v = join($join, @{$v}); |
430 |
} |
431 |
} else { |
432 |
warn("TT filter $type(): field $name values aren't ARRAY, ignoring"); |
433 |
} |
434 |
|
435 |
return $v; |
436 |
} |
437 |
} |
438 |
|
439 |
$args->{'d'} = tt_filter_type($args, 'display'); |
440 |
$args->{'display'} = tt_filter_type($args, 'display'); |
441 |
|
442 |
=head3 tt_filter_search |
443 |
|
444 |
filter to return links to search, usage in TT: |
445 |
|
446 |
[% search('FieldToDisplay','FieldToSearch','optional delimiter', 'optional_template.tt') %] |
447 |
|
448 |
=cut |
449 |
|
450 |
sub tt_filter_search { |
451 |
|
452 |
my ($data) = @_; |
453 |
|
454 |
die "no data?" unless ($data); |
455 |
|
456 |
return sub { |
457 |
|
458 |
my ($display,$search,$delimiter,$template) = @_; |
459 |
|
460 |
# default delimiter |
461 |
$delimiter ||= '¶<br/>', |
462 |
|
463 |
die "no data hash" unless ($data->{'data'} && ref($data->{'data'}) eq 'HASH'); |
464 |
# Hm? Should we die here? |
465 |
return unless ($display); |
466 |
|
467 |
my $item = $data->{'data'}->{$display} || return; |
468 |
|
469 |
return unless($item->{'display'}); |
470 |
if (! $item->{'search'}) { |
471 |
warn "error in TT template: field $display didn't insert anything into search, use d('$display') and not search('$display'...)"; |
472 |
return; |
473 |
} |
474 |
|
475 |
my @warn; |
476 |
foreach my $type (qw/display search/) { |
477 |
push @warn, "field $display type $type values aren't ARRAY" unless (ref($item->{$type}) eq 'ARRAY'); |
478 |
} |
479 |
|
480 |
if (@warn) { |
481 |
warn("TT filter search(): " . join(",", @warn) . ", skipping"); |
482 |
return; |
483 |
} |
484 |
my @html; |
485 |
|
486 |
my $d_el = $#{ $item->{'display'} }; |
487 |
my $s_el = $#{ $item->{'search'} }; |
488 |
|
489 |
# easy, both fields have same number of elements or there is just |
490 |
# one search and multiple display |
491 |
if ( $d_el == $s_el || $s_el == 0 ) { |
492 |
|
493 |
foreach my $i ( 0 .. $d_el ) { |
494 |
|
495 |
my $s; |
496 |
if ($s_el > 0) { |
497 |
$s = $item->{'search'}->[$i] || die "can't find value $i for type search in field $search"; |
498 |
} else { |
499 |
$s = $item->{'search'}->[0]; |
500 |
} |
501 |
#$s =~ s/([^\w.-])/sprintf("%%%02X",ord($1))/eg; |
502 |
$s = __quotemeta( $s ); |
503 |
|
504 |
my $d = $item->{'display'}->[$i] || die "can't find value $i for type display in field $display"; |
505 |
|
506 |
my $template_arg = ''; |
507 |
$template_arg = qq{,'$template'} if ($template); |
508 |
|
509 |
push @html, qq{<a href="#" onclick="return search_via_link('$search','$s'${template_arg})">$d</a>}; |
510 |
} |
511 |
|
512 |
return join($delimiter, @html); |
513 |
} else { |
514 |
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>}; |
515 |
my $v = $item->{'display'}; |
516 |
|
517 |
if ($#{$v} == 0) { |
518 |
$html .= $v->[0]; |
519 |
} else { |
520 |
$html .= join($delimiter, @{$v}); |
521 |
} |
522 |
return $html; |
523 |
} |
524 |
} |
525 |
} |
526 |
|
527 |
$args->{'search'} = tt_filter_search($args); |
528 |
|
529 |
=head3 load_rec |
530 |
|
531 |
Used mostly for onClick events like this: |
532 |
|
533 |
<a href="#" onClick="[% load_rec( record_uri, 'template_name.tt') %]>foo</a> |
534 |
|
535 |
It will automatically do sanity checking and create correct JavaScript code. |
536 |
|
537 |
=cut |
538 |
|
539 |
$args->{'load_rec'} = sub { |
540 |
my @errors; |
541 |
|
542 |
my $record_uri = shift or push @errors, "record_uri missing"; |
543 |
my $template = shift or push @errors, "template missing"; |
544 |
|
545 |
if ($record_uri !~ m#^[^/]+/[^/]+/[^/]+$#) { |
546 |
push @errors, "invalid format of record_uri: $record_uri"; |
547 |
} |
548 |
|
549 |
if (@errors) { |
550 |
return "Logger.error('errors in load_rec: " . join(", ", @errors) . "'); return false;"; |
551 |
} else { |
552 |
return "load_rec('$record_uri','$template'); return false;"; |
553 |
} |
554 |
}; |
555 |
|
556 |
=head3 load_template |
557 |
|
558 |
Used to re-submit search request and load results in different template |
559 |
|
560 |
<a href="#" onClick="[% load_template( 'template_name.tt' ) %]">bar</a> |
561 |
|
562 |
=cut |
563 |
|
564 |
$args->{'load_template'} = sub { |
565 |
my $template = shift or return "Logger.error('load_template missing template name!'); return false;"; |
566 |
return "load_template($template); return false;"; |
567 |
}; |
568 |
|
569 |
my $out; |
570 |
|
571 |
$self->{'tt'}->process( |
572 |
$args->{'template'}, |
573 |
$args, |
574 |
\$out |
575 |
) || $log->error( "apply can't process template: ", $self->{'tt'}->error() ); |
576 |
|
577 |
return $out; |
578 |
} |
579 |
|
580 |
|
581 |
=head2 __quotemeta |
582 |
|
583 |
Helper to quote JavaScript-friendly characters |
584 |
|
585 |
=cut |
586 |
|
587 |
sub __quotemeta { |
588 |
local $_ = shift; |
589 |
$_ = decode('iso-8859-2', $_); |
590 |
|
591 |
s<([\x{0080}-\x{fffd}]+)>{sprintf '\u%0*v4X', '\u', $1}ge if ( Encode::is_utf8($_) ); |
592 |
{ |
593 |
use bytes; |
594 |
s<((?:[^ \x21-\x7E]|(?:\\(?!u)))+)>{sprintf '\x%0*v2X', '\x', $1}ge; |
595 |
} |
596 |
|
597 |
s/\\x09/\\t/g; |
598 |
s/\\x0A/\\n/g; |
599 |
s/\\x0D/\\r/g; |
600 |
s/"/\\"/g; |
601 |
s/\\x5C/\\\\/g; |
602 |
|
603 |
return $_; |
604 |
} |
605 |
|
606 |
|
607 |
|
608 |
=head1 AUTHOR |
609 |
|
610 |
Dobrica Pavlinusic C<< <dpavlin@rot13.org> >> |
611 |
|
612 |
=head1 LICENSE |
613 |
|
614 |
This library is free software, you can redistribute it and/or modify |
615 |
it under the same terms as Perl itself. |
616 |
|
617 |
=cut |
618 |
|
619 |
1; |