5 |
use lib '/data/webpac2/lib'; |
use lib '/data/webpac2/lib'; |
6 |
use base qw/ |
use base qw/ |
7 |
Catalyst::Model |
Catalyst::Model |
|
WebPAC::Search::Estraier |
|
8 |
/; |
/; |
9 |
use Data::Dumper; |
use Data::Dumper; |
10 |
|
use WebPAC::Store 0.08; |
11 |
|
use WebPAC::Output::TT 0.06; |
12 |
|
use WebPAC::Search::Estraier 0.05; |
13 |
|
use File::Slurp; |
14 |
|
use Time::HiRes; |
15 |
|
use Encode qw/encode decode from_to/; |
16 |
|
|
17 |
=head1 NAME |
=head1 NAME |
18 |
|
|
35 |
|
|
36 |
# configuration for hyper estraier full text search engine |
# configuration for hyper estraier full text search engine |
37 |
hyperestraier: |
hyperestraier: |
38 |
url: 'http://localhost:1978/node/webpac2' |
masterurl: 'http://localhost:1978/node/webpac2' |
39 |
user: 'admin' |
defaultnode: 'webpac2' |
40 |
passwd: 'admin' |
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 |
=cut |
54 |
|
|
59 |
$self->config($config); |
$self->config($config); |
60 |
|
|
61 |
my $log = $c->log; |
my $log = $c->log; |
62 |
|
$self->{log} = $log; |
63 |
|
|
64 |
my $est_cfg = $c->config->{hyperestraier}; |
my $est_cfg = $c->config->{hyperestraier}; |
65 |
$est_cfg->{'log'} = $log; |
$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) ); |
$log->debug("using config:" . Dumper($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 } ); |
$self->{est} = new WebPAC::Search::Estraier( %{ $est_cfg } ); |
78 |
|
|
79 |
# $c->stash->{est}->search( |
# save config parametars in object |
80 |
# query => $c->req->params->{Title}, |
foreach my $f (qw/db_path template_path hits_on_page webpac_encoding defaultdepth/) { |
81 |
# max => 100, |
$self->{$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 |
|
$self->{out} = new WebPAC::Output::TT( |
97 |
|
include_path => $template_path, |
98 |
|
filters => { foo => sub { shift } }, |
99 |
|
); |
100 |
|
|
101 |
|
# default template from config.yaml |
102 |
|
$self->{template} ||= $c->config->{webpac}->{template}; |
103 |
|
|
104 |
|
$log->debug("converting encoding from webpac_encoding '" . |
105 |
|
$c->config->{webpac}->{webpac_encoding} . |
106 |
|
"'" |
107 |
|
); |
108 |
|
|
109 |
|
$self->{databases} = $c->config->{databases} || $log->error("can't find databases in config"); |
110 |
|
|
111 |
return $self; |
return $self; |
112 |
|
|
113 |
} |
} |
114 |
|
|
115 |
|
|
116 |
|
=head2 search |
117 |
|
|
118 |
|
my $m->search( |
119 |
|
phrase => 'query phrase', |
120 |
|
add_attr => \@add_attr |
121 |
|
get_attr => [ '@uri' ], |
122 |
|
max => 42, |
123 |
|
template => 'result_template.tt', |
124 |
|
depth => 1, |
125 |
|
); |
126 |
|
|
127 |
|
All fields are standard C<WebPAC::Search::Estraier> parametars except |
128 |
|
C<template> which will (if specified) return results in HTML using |
129 |
|
selected template. |
130 |
|
|
131 |
|
=cut |
132 |
|
|
133 |
sub search { |
sub search { |
134 |
my ( $self, $query ) = @_; |
my $self = shift; |
135 |
|
|
136 |
|
my $args = {@_}; |
137 |
|
|
138 |
|
my $log = $self->{log}; |
139 |
|
|
140 |
|
$log->debug("args: " . Dumper( $args )); |
141 |
|
|
142 |
|
my $query = $args->{phrase} || $log->warn("no query phrase") && return; |
143 |
|
|
144 |
|
$log->debug("search model query: '$query'"); |
145 |
|
if ($args->{add_attr}) { |
146 |
|
$log->debug(" + add_attr: " . |
147 |
|
join("','", @{ $args->{add_attr} }) |
148 |
|
); |
149 |
|
} |
150 |
|
|
151 |
|
my $template_filename = $args->{template} || $self->{template}; |
152 |
|
|
153 |
|
$args->{max} ||= $self->{'hits_for_pager'}; |
154 |
|
if (! $args->{max}) { |
155 |
|
$args->{max} = 100; |
156 |
|
$log->warn("max not set when calling model. Using default of $args->{max}"); |
157 |
|
} |
158 |
|
|
159 |
|
my $times; # store some times for benchmarking |
160 |
|
|
161 |
|
my $t = time(); |
162 |
|
|
163 |
|
# transfer depth of search |
164 |
|
if (! $args->{depth}) { |
165 |
|
my $default = $self->{defaultdepth} || $log->logdie("can't find defaultdepth in estraier configuration"); |
166 |
|
$args->{depth} = $default; |
167 |
|
$log->warn("using default search depth $default"); |
168 |
|
} |
169 |
|
|
170 |
|
my @results = $self->{est}->search( %{ $args } ); |
171 |
|
|
172 |
|
$times->{est} += time() - $t; |
173 |
|
|
174 |
|
my $hits = $#results + 1; |
175 |
|
|
176 |
|
$log->debug( sprintf("search took %.2fs and returned $hits hits.", $times->{est}) ); |
177 |
|
|
178 |
|
# just return results? |
179 |
|
return @results unless ($args->{'template'}); |
180 |
|
|
181 |
|
# |
182 |
|
# construct HTML results |
183 |
|
# |
184 |
|
|
185 |
|
my @html_results; |
186 |
|
|
187 |
|
for my $i ( 0 .. $#results ) { |
188 |
|
|
189 |
|
my ($database, $prefix, $id); |
190 |
|
if ( $results[$i]->{'@uri'} =~ m!/([^/]+)/([^/]+)/(\d+)$!) { |
191 |
|
($database, $prefix,$id) = ($1,$2,$3); |
192 |
|
} else { |
193 |
|
$log->warn("can't decode database/prefix/id from " . $results[$i]->{'@uri'}); |
194 |
|
next; |
195 |
|
} |
196 |
|
|
197 |
|
#$log->debug("load_ds( id => $id, prefix => '$prefix' )"); |
198 |
|
|
199 |
|
$t = time(); |
200 |
|
|
201 |
|
my $ds = $self->{db}->load_ds( database => $database, prefix => $prefix, id => $id ); |
202 |
|
if (! $ds) { |
203 |
|
$log->error("can't load_ds( ${database}/${prefix}/${id} )"); |
204 |
|
next; |
205 |
|
} |
206 |
|
|
207 |
|
$times->{db} += time() - $t; |
208 |
|
|
209 |
my @results = $self->{est}->search( |
#$log->debug( "ds = " . Dumper( \@html_results ) ); |
210 |
query => $query, |
|
211 |
attr => [ '@uri' ], |
$t = time(); |
212 |
max => 100, |
|
213 |
|
my $html = $self->{out}->apply( |
214 |
|
template => $template_filename, |
215 |
|
data => $ds, |
216 |
|
record_uri => "${database}/${prefix}/${id}", |
217 |
|
config => $self->{databases}->{$database}, |
218 |
|
); |
219 |
|
|
220 |
|
$times->{out} += time() - $t; |
221 |
|
|
222 |
|
$t = time(); |
223 |
|
|
224 |
|
$html = decode($self->{webpac_encoding}, $html); |
225 |
|
|
226 |
|
push @html_results, $html; |
227 |
|
|
228 |
|
} |
229 |
|
|
230 |
|
#$log->debug( '@html_results = ' . Dumper( \@html_results ) ); |
231 |
|
|
232 |
|
$log->debug( sprintf( |
233 |
|
"time spent: db = %.2f, out = %.2f", |
234 |
|
$times->{db}, $times->{out}, |
235 |
|
) ); |
236 |
|
|
237 |
|
return \@html_results; |
238 |
|
} |
239 |
|
|
240 |
|
=head2 record |
241 |
|
|
242 |
|
my $html = $m->record( |
243 |
|
mfn => 42, |
244 |
|
template => 'foo.tt', |
245 |
|
); |
246 |
|
|
247 |
|
This will load one record, convert it to html using C<template> and return |
248 |
|
it. |
249 |
|
|
250 |
|
=cut |
251 |
|
|
252 |
|
sub record { |
253 |
|
my $self = shift; |
254 |
|
|
255 |
|
my $args = {@_}; |
256 |
|
my $log = $self->{log}; |
257 |
|
$log->debug("args: " . Dumper( $args )); |
258 |
|
|
259 |
|
foreach my $f (qw/record_uri template/) { |
260 |
|
$log->fatal("need $f") unless ($args->{$f}); |
261 |
|
} |
262 |
|
|
263 |
|
my ($database, $prefix, $id); |
264 |
|
|
265 |
|
if ($args->{record_uri} =~ m#^([^/]+)/([^/]+)/([^/]+)$#) { |
266 |
|
($database, $prefix, $id) = ($1,$2,$3); |
267 |
|
} else { |
268 |
|
$log->error("can't parse $args->{record_uri} into prefix, database and uri"); |
269 |
|
return; |
270 |
|
} |
271 |
|
|
272 |
|
my $ds = $self->{db}->load_ds( id => $id, prefix => $prefix, database => $database ); |
273 |
|
if (! $ds) { |
274 |
|
$log->error("can't load_ds( $database/$prefix/$id )"); |
275 |
|
return; |
276 |
|
} |
277 |
|
|
278 |
|
my $html = $self->{out}->apply( |
279 |
|
template => $args->{template}, |
280 |
|
data => $ds, |
281 |
|
record_uri => $args->{record_uri}, |
282 |
|
config => $self->{databases}->{$database}, |
283 |
); |
); |
284 |
|
|
285 |
return \@results; |
$html = decode($self->{webpac_encoding}, $html); |
286 |
|
|
287 |
|
return $html; |
288 |
} |
} |
289 |
|
|
290 |
|
|
291 |
|
=head2 save_html |
292 |
|
|
293 |
|
$m->save_html( '/full/path/to/file', $content ); |
294 |
|
|
295 |
|
It will use C<Encode> to convert content encoding back to |
296 |
|
Webpac codepage, recode JavaScript Unicode entities (%u1234), |
297 |
|
strip extra newlines at beginning and end, and save to |
298 |
|
C</full/path/to/file.new> and if that succeeds, just rename |
299 |
|
it over original file which should be atomic on filesystem level. |
300 |
|
|
301 |
|
=cut |
302 |
|
|
303 |
|
sub save_html { |
304 |
|
my ($self, $path, $content) = @_; |
305 |
|
|
306 |
|
# FIXME Should this be UTF-8 or someting? |
307 |
|
my $js_encoding = $self->{webpac_encoding}; |
308 |
|
$js_encoding = 'UTF-16'; |
309 |
|
|
310 |
|
sub _conv_js { |
311 |
|
return '0x' . $_[1]; |
312 |
|
return encode($_[0], chr(hex($_[1]))); |
313 |
|
} |
314 |
|
#$content =~ s/%u([a-fA-F0-9]{4})/_conv_js($js_encoding,$1)/gex; |
315 |
|
$content =~ s/^[\n\r]+//s; |
316 |
|
$content =~ s/[\n\r]+$/\n/s; |
317 |
|
$content =~ s/\n\r/\n/gs; |
318 |
|
|
319 |
|
my $disk_encoding = $self->{webpac_encoding} || 'utf-8'; |
320 |
|
$self->{log}->debug("convert encoding to $disk_encoding"); |
321 |
|
from_to($content, 'utf-8', $disk_encoding) || $self->{log}->warn("encoding from utf-8 to $disk_encoding failed for: $content"); |
322 |
|
|
323 |
|
write_file($path . '.new', {binmode => ':raw' }, $content) || die "can't save ${path}.new $!"; |
324 |
|
rename $path . '.new', $path || die "can't rename to $path: $!"; |
325 |
|
} |
326 |
|
|
327 |
|
=head2 load_html |
328 |
|
|
329 |
|
my $html = $m->load_html('/full/path/to/file'); |
330 |
|
|
331 |
|
This will convert file from Webpac encoding to Catalyst and |
332 |
|
convert that data to escaped HTML (for sending into |
333 |
|
C<< <textarea/> >> tags in html. |
334 |
|
|
335 |
|
=cut |
336 |
|
|
337 |
|
sub load_html { |
338 |
|
my ($self, $path) = @_; |
339 |
|
|
340 |
|
die "no path?" unless ($path); |
341 |
|
|
342 |
|
my $content = read_file($path, {binmode => ':raw' }) || die "can't read $path: $!"; |
343 |
|
|
344 |
|
return decode($self->{webpac_encoding}, $content); |
345 |
|
} |
346 |
|
|
347 |
=head1 AUTHOR |
=head1 AUTHOR |
348 |
|
|
349 |
Dobrica Pavlinusic |
Dobrica Pavlinusic C<< <dpavlin@rot13.org> >> |
350 |
|
|
351 |
=head1 LICENSE |
=head1 LICENSE |
352 |
|
|