1 |
dpavlin |
92 |
package Webpacus::Model::WebPAC; |
2 |
|
|
|
3 |
|
|
use strict; |
4 |
|
|
use warnings; |
5 |
|
|
use lib '/data/webpac2/lib'; |
6 |
dpavlin |
93 |
use base qw/ |
7 |
|
|
Catalyst::Model |
8 |
|
|
/; |
9 |
dpavlin |
237 |
use WebPAC::Store 0.08; |
10 |
dpavlin |
222 |
use WebPAC::Search::Estraier 0.05; |
11 |
dpavlin |
135 |
use File::Slurp; |
12 |
dpavlin |
378 |
use Time::HiRes qw/time/; |
13 |
dpavlin |
348 |
use Encode qw/encode decode from_to/; |
14 |
dpavlin |
378 |
use Data::Dumper; |
15 |
dpavlin |
92 |
|
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 |
dpavlin |
222 |
masterurl: 'http://localhost:1978/node/webpac2' |
38 |
|
|
defaultnode: 'webpac2' |
39 |
|
|
defaultdepth: 1 |
40 |
dpavlin |
96 |
user: 'admin' |
41 |
|
|
passwd: 'admin' |
42 |
dpavlin |
143 |
hits_on_page: 100 |
43 |
dpavlin |
305 |
hits_for_pager: 1000 |
44 |
dpavlin |
92 |
|
45 |
dpavlin |
96 |
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 |
dpavlin |
92 |
=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 |
dpavlin |
94 |
$self->{log} = $log; |
62 |
dpavlin |
92 |
|
63 |
dpavlin |
93 |
my $est_cfg = $c->config->{hyperestraier}; |
64 |
|
|
$est_cfg->{'log'} = $log; |
65 |
dpavlin |
92 |
|
66 |
dpavlin |
271 |
$est_cfg->{encoding} = $est_cfg->{catalyst_encoding} || $c->config->{catalyst_encoding} or $c->log->fatal("can't find catalyst_encoding"); |
67 |
dpavlin |
142 |
|
68 |
dpavlin |
93 |
$log->debug("using config:" . Dumper($est_cfg) ); |
69 |
dpavlin |
92 |
|
70 |
dpavlin |
222 |
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 |
dpavlin |
93 |
$self->{est} = new WebPAC::Search::Estraier( %{ $est_cfg } ); |
77 |
dpavlin |
92 |
|
78 |
dpavlin |
167 |
# save config parametars in object |
79 |
dpavlin |
348 |
foreach my $f (qw/db_path template_path hits_on_page webpac_encoding defaultdepth/) { |
80 |
dpavlin |
167 |
$self->{$f} = $c->config->{hyperestraier}->{$f} || |
81 |
|
|
$c->config->{webpac}->{$f}; |
82 |
|
|
$log->debug("self->{$f} = " . $self->{$f}); |
83 |
|
|
} |
84 |
|
|
my $db_path = $self->{db_path}; |
85 |
|
|
my $template_path = $self->{template_path}; |
86 |
dpavlin |
95 |
|
87 |
|
|
$log->debug("using db path '$db_path', template path '$template_path'"); |
88 |
|
|
|
89 |
dpavlin |
222 |
$self->{db} = new WebPAC::Store( |
90 |
dpavlin |
95 |
path => $db_path, |
91 |
|
|
read_only => 1, |
92 |
dpavlin |
224 |
database => $est_cfg->{database}, |
93 |
dpavlin |
95 |
); |
94 |
|
|
|
95 |
dpavlin |
101 |
# default template from config.yaml |
96 |
dpavlin |
95 |
$self->{template} ||= $c->config->{webpac}->{template}; |
97 |
|
|
|
98 |
dpavlin |
100 |
$log->debug("converting encoding from webpac_encoding '" . |
99 |
|
|
$c->config->{webpac}->{webpac_encoding} . |
100 |
dpavlin |
99 |
"'" |
101 |
|
|
); |
102 |
dpavlin |
96 |
|
103 |
dpavlin |
305 |
$self->{databases} = $c->config->{databases} || $log->error("can't find databases in config"); |
104 |
dpavlin |
155 |
|
105 |
dpavlin |
378 |
$self->{model_record} = $c->comp('Model::Record') or $log->error("can't find Model::Record"); |
106 |
|
|
|
107 |
dpavlin |
92 |
return $self; |
108 |
|
|
|
109 |
|
|
} |
110 |
|
|
|
111 |
dpavlin |
135 |
|
112 |
|
|
=head2 search |
113 |
|
|
|
114 |
dpavlin |
150 |
my $m->search( |
115 |
|
|
phrase => 'query phrase', |
116 |
dpavlin |
155 |
add_attr => \@add_attr |
117 |
|
|
get_attr => [ '@uri' ], |
118 |
|
|
max => 42, |
119 |
dpavlin |
150 |
template => 'result_template.tt', |
120 |
dpavlin |
222 |
depth => 1, |
121 |
dpavlin |
150 |
); |
122 |
dpavlin |
135 |
|
123 |
dpavlin |
155 |
All fields are standard C<WebPAC::Search::Estraier> parametars except |
124 |
|
|
C<template> which will (if specified) return results in HTML using |
125 |
|
|
selected template. |
126 |
|
|
|
127 |
dpavlin |
135 |
=cut |
128 |
|
|
|
129 |
dpavlin |
93 |
sub search { |
130 |
dpavlin |
150 |
my $self = shift; |
131 |
dpavlin |
93 |
|
132 |
dpavlin |
378 |
my $search_start_t = time(); |
133 |
|
|
|
134 |
dpavlin |
150 |
my $args = {@_}; |
135 |
|
|
|
136 |
dpavlin |
95 |
my $log = $self->{log}; |
137 |
dpavlin |
94 |
|
138 |
dpavlin |
351 |
$log->debug("search args: " . Dumper( $args )); |
139 |
dpavlin |
95 |
|
140 |
dpavlin |
150 |
my $query = $args->{phrase} || $log->warn("no query phrase") && return; |
141 |
dpavlin |
95 |
|
142 |
dpavlin |
155 |
$log->debug("search model query: '$query'"); |
143 |
|
|
if ($args->{add_attr}) { |
144 |
|
|
$log->debug(" + add_attr: " . |
145 |
|
|
join("','", @{ $args->{add_attr} }) |
146 |
|
|
); |
147 |
|
|
} |
148 |
dpavlin |
150 |
|
149 |
|
|
my $template_filename = $args->{template} || $self->{template}; |
150 |
|
|
|
151 |
dpavlin |
305 |
$args->{max} ||= $self->{'hits_for_pager'}; |
152 |
dpavlin |
155 |
if (! $args->{max}) { |
153 |
dpavlin |
305 |
$args->{max} = 100; |
154 |
|
|
$log->warn("max not set when calling model. Using default of $args->{max}"); |
155 |
dpavlin |
155 |
} |
156 |
dpavlin |
93 |
|
157 |
dpavlin |
155 |
my $times; # store some times for benchmarking |
158 |
|
|
|
159 |
|
|
my $t = time(); |
160 |
|
|
|
161 |
dpavlin |
222 |
# transfer depth of search |
162 |
|
|
if (! $args->{depth}) { |
163 |
|
|
my $default = $self->{defaultdepth} || $log->logdie("can't find defaultdepth in estraier configuration"); |
164 |
|
|
$args->{depth} = $default; |
165 |
|
|
$log->warn("using default search depth $default"); |
166 |
|
|
} |
167 |
|
|
|
168 |
dpavlin |
155 |
my @results = $self->{est}->search( %{ $args } ); |
169 |
|
|
|
170 |
|
|
$times->{est} += time() - $t; |
171 |
|
|
|
172 |
dpavlin |
150 |
my $hits = $#results + 1; |
173 |
dpavlin |
99 |
|
174 |
dpavlin |
378 |
$log->debug( sprintf("search took %.6fs and returned $hits hits.", $times->{est}) ); |
175 |
dpavlin |
150 |
|
176 |
dpavlin |
155 |
# just return results? |
177 |
|
|
return @results unless ($args->{'template'}); |
178 |
|
|
|
179 |
|
|
# |
180 |
|
|
# construct HTML results |
181 |
|
|
# |
182 |
|
|
|
183 |
dpavlin |
100 |
my @html_results; |
184 |
|
|
|
185 |
dpavlin |
95 |
for my $i ( 0 .. $#results ) { |
186 |
|
|
|
187 |
dpavlin |
237 |
my ($database, $prefix, $id); |
188 |
dpavlin |
293 |
if ( $results[$i]->{'@uri'} =~ m!/([^/]+)/([^/]+)/(\d+)$!) { |
189 |
dpavlin |
237 |
($database, $prefix,$id) = ($1,$2,$3); |
190 |
dpavlin |
224 |
} else { |
191 |
dpavlin |
293 |
$log->warn("can't decode database/prefix/id from " . $results[$i]->{'@uri'}); |
192 |
dpavlin |
224 |
next; |
193 |
|
|
} |
194 |
dpavlin |
95 |
|
195 |
dpavlin |
224 |
#$log->debug("load_ds( id => $id, prefix => '$prefix' )"); |
196 |
dpavlin |
95 |
|
197 |
dpavlin |
155 |
$t = time(); |
198 |
|
|
|
199 |
dpavlin |
242 |
my $ds = $self->{db}->load_ds( database => $database, prefix => $prefix, id => $id ); |
200 |
|
|
if (! $ds) { |
201 |
dpavlin |
293 |
$log->error("can't load_ds( ${database}/${prefix}/${id} )"); |
202 |
dpavlin |
242 |
next; |
203 |
|
|
} |
204 |
dpavlin |
155 |
|
205 |
|
|
$times->{db} += time() - $t; |
206 |
|
|
|
207 |
dpavlin |
115 |
#$log->debug( "ds = " . Dumper( \@html_results ) ); |
208 |
|
|
|
209 |
dpavlin |
155 |
$t = time(); |
210 |
|
|
|
211 |
dpavlin |
378 |
my $html = '[no output]'; |
212 |
dpavlin |
100 |
|
213 |
dpavlin |
378 |
if ($self->{model_record}) { |
214 |
|
|
$html = $self->{model_record}->apply( |
215 |
|
|
template => $template_filename, |
216 |
|
|
data => $ds, |
217 |
|
|
record_uri => "${database}/${prefix}/${id}", |
218 |
|
|
config => $self->{databases}->{$database}, |
219 |
|
|
); |
220 |
|
|
} else { |
221 |
|
|
$log->warn("skipped apply"); |
222 |
|
|
} |
223 |
|
|
|
224 |
dpavlin |
155 |
$times->{out} += time() - $t; |
225 |
|
|
|
226 |
|
|
$t = time(); |
227 |
|
|
|
228 |
dpavlin |
348 |
$html = decode($self->{webpac_encoding}, $html); |
229 |
dpavlin |
100 |
|
230 |
|
|
push @html_results, $html; |
231 |
|
|
|
232 |
dpavlin |
95 |
} |
233 |
|
|
|
234 |
dpavlin |
115 |
#$log->debug( '@html_results = ' . Dumper( \@html_results ) ); |
235 |
|
|
|
236 |
dpavlin |
155 |
$log->debug( sprintf( |
237 |
dpavlin |
378 |
"duration breakdown: store %.6fs, apply %.6fs, total: %.6fs", |
238 |
|
|
$times->{db}, $times->{out}, time() - $search_start_t, |
239 |
dpavlin |
155 |
) ); |
240 |
|
|
|
241 |
dpavlin |
100 |
return \@html_results; |
242 |
dpavlin |
93 |
} |
243 |
|
|
|
244 |
dpavlin |
165 |
=head2 record |
245 |
|
|
|
246 |
|
|
my $html = $m->record( |
247 |
|
|
mfn => 42, |
248 |
|
|
template => 'foo.tt', |
249 |
|
|
); |
250 |
|
|
|
251 |
|
|
This will load one record, convert it to html using C<template> and return |
252 |
|
|
it. |
253 |
|
|
|
254 |
|
|
=cut |
255 |
|
|
|
256 |
|
|
sub record { |
257 |
|
|
my $self = shift; |
258 |
|
|
|
259 |
|
|
my $args = {@_}; |
260 |
|
|
my $log = $self->{log}; |
261 |
dpavlin |
351 |
$log->debug("record args: " . Dumper( $args )); |
262 |
dpavlin |
165 |
|
263 |
dpavlin |
242 |
foreach my $f (qw/record_uri template/) { |
264 |
dpavlin |
228 |
$log->fatal("need $f") unless ($args->{$f}); |
265 |
dpavlin |
165 |
} |
266 |
|
|
|
267 |
dpavlin |
242 |
my ($database, $prefix, $id); |
268 |
dpavlin |
165 |
|
269 |
dpavlin |
242 |
if ($args->{record_uri} =~ m#^([^/]+)/([^/]+)/([^/]+)$#) { |
270 |
|
|
($database, $prefix, $id) = ($1,$2,$3); |
271 |
|
|
} else { |
272 |
|
|
$log->error("can't parse $args->{record_uri} into prefix, database and uri"); |
273 |
|
|
return; |
274 |
|
|
} |
275 |
dpavlin |
165 |
|
276 |
dpavlin |
242 |
my $ds = $self->{db}->load_ds( id => $id, prefix => $prefix, database => $database ); |
277 |
|
|
if (! $ds) { |
278 |
|
|
$log->error("can't load_ds( $database/$prefix/$id )"); |
279 |
|
|
return; |
280 |
|
|
} |
281 |
|
|
|
282 |
dpavlin |
378 |
my $html = $self->{model_record}->apply( |
283 |
dpavlin |
165 |
template => $args->{template}, |
284 |
|
|
data => $ds, |
285 |
dpavlin |
242 |
record_uri => $args->{record_uri}, |
286 |
dpavlin |
305 |
config => $self->{databases}->{$database}, |
287 |
dpavlin |
165 |
); |
288 |
|
|
|
289 |
dpavlin |
348 |
$html = decode($self->{webpac_encoding}, $html); |
290 |
dpavlin |
165 |
|
291 |
|
|
return $html; |
292 |
|
|
} |
293 |
|
|
|
294 |
dpavlin |
305 |
|
295 |
dpavlin |
135 |
=head2 save_html |
296 |
dpavlin |
93 |
|
297 |
dpavlin |
135 |
$m->save_html( '/full/path/to/file', $content ); |
298 |
dpavlin |
93 |
|
299 |
dpavlin |
348 |
It will use C<Encode> to convert content encoding back to |
300 |
dpavlin |
135 |
Webpac codepage, recode JavaScript Unicode entities (%u1234), |
301 |
|
|
strip extra newlines at beginning and end, and save to |
302 |
|
|
C</full/path/to/file.new> and if that succeeds, just rename |
303 |
|
|
it over original file which should be atomic on filesystem level. |
304 |
|
|
|
305 |
|
|
=cut |
306 |
|
|
|
307 |
|
|
sub save_html { |
308 |
|
|
my ($self, $path, $content) = @_; |
309 |
|
|
|
310 |
dpavlin |
348 |
# FIXME Should this be UTF-8 or someting? |
311 |
|
|
my $js_encoding = $self->{webpac_encoding}; |
312 |
|
|
$js_encoding = 'UTF-16'; |
313 |
|
|
|
314 |
dpavlin |
135 |
sub _conv_js { |
315 |
dpavlin |
348 |
return '0x' . $_[1]; |
316 |
|
|
return encode($_[0], chr(hex($_[1]))); |
317 |
dpavlin |
135 |
} |
318 |
dpavlin |
348 |
#$content =~ s/%u([a-fA-F0-9]{4})/_conv_js($js_encoding,$1)/gex; |
319 |
dpavlin |
135 |
$content =~ s/^[\n\r]+//s; |
320 |
|
|
$content =~ s/[\n\r]+$/\n/s; |
321 |
dpavlin |
282 |
$content =~ s/\n\r/\n/gs; |
322 |
dpavlin |
135 |
|
323 |
dpavlin |
348 |
my $disk_encoding = $self->{webpac_encoding} || 'utf-8'; |
324 |
|
|
$self->{log}->debug("convert encoding to $disk_encoding"); |
325 |
|
|
from_to($content, 'utf-8', $disk_encoding) || $self->{log}->warn("encoding from utf-8 to $disk_encoding failed for: $content"); |
326 |
dpavlin |
179 |
|
327 |
|
|
write_file($path . '.new', {binmode => ':raw' }, $content) || die "can't save ${path}.new $!"; |
328 |
dpavlin |
135 |
rename $path . '.new', $path || die "can't rename to $path: $!"; |
329 |
|
|
} |
330 |
|
|
|
331 |
|
|
=head2 load_html |
332 |
|
|
|
333 |
|
|
my $html = $m->load_html('/full/path/to/file'); |
334 |
|
|
|
335 |
|
|
This will convert file from Webpac encoding to Catalyst and |
336 |
|
|
convert that data to escaped HTML (for sending into |
337 |
|
|
C<< <textarea/> >> tags in html. |
338 |
|
|
|
339 |
|
|
=cut |
340 |
|
|
|
341 |
|
|
sub load_html { |
342 |
|
|
my ($self, $path) = @_; |
343 |
|
|
|
344 |
|
|
die "no path?" unless ($path); |
345 |
|
|
|
346 |
dpavlin |
179 |
my $content = read_file($path, {binmode => ':raw' }) || die "can't read $path: $!"; |
347 |
dpavlin |
135 |
|
348 |
dpavlin |
348 |
return decode($self->{webpac_encoding}, $content); |
349 |
dpavlin |
135 |
} |
350 |
|
|
|
351 |
dpavlin |
92 |
=head1 AUTHOR |
352 |
|
|
|
353 |
dpavlin |
348 |
Dobrica Pavlinusic C<< <dpavlin@rot13.org> >> |
354 |
dpavlin |
92 |
|
355 |
|
|
=head1 LICENSE |
356 |
|
|
|
357 |
|
|
This library is free software, you can redistribute it and/or modify |
358 |
|
|
it under the same terms as Perl itself. |
359 |
|
|
|
360 |
|
|
=cut |
361 |
|
|
|
362 |
|
|
1; |