7 |
Catalyst::Model |
Catalyst::Model |
8 |
/; |
/; |
9 |
use Data::Dumper; |
use Data::Dumper; |
10 |
use WebPAC::DB; |
use WebPAC::Store 0.08; |
11 |
use WebPAC::Output::TT; |
use WebPAC::Output::TT 0.06; |
12 |
use WebPAC::Search::Estraier 0.02; |
use WebPAC::Search::Estraier 0.05; |
13 |
use File::Slurp; |
use File::Slurp; |
14 |
use Time::HiRes; |
use Time::HiRes; |
15 |
|
|
34 |
|
|
35 |
# configuration for hyper estraier full text search engine |
# configuration for hyper estraier full text search engine |
36 |
hyperestraier: |
hyperestraier: |
37 |
url: 'http://localhost:1978/node/webpac2' |
masterurl: 'http://localhost:1978/node/webpac2' |
38 |
|
defaultnode: 'webpac2' |
39 |
|
defaultdepth: 1 |
40 |
user: 'admin' |
user: 'admin' |
41 |
passwd: 'admin' |
passwd: 'admin' |
42 |
hits_on_page: 100 |
hits_on_page: 100 |
43 |
|
hits_for_pager: 1000 |
44 |
|
|
45 |
webpac: |
webpac: |
46 |
db_path: '/data/webpac2/db' |
db_path: '/data/webpac2/db' |
65 |
my $est_cfg = $c->config->{hyperestraier}; |
my $est_cfg = $c->config->{hyperestraier}; |
66 |
$est_cfg->{'log'} = $log; |
$est_cfg->{'log'} = $log; |
67 |
|
|
68 |
$est_cfg->{encoding} = $est_cfg->{catalyst_encoding}; |
$est_cfg->{encoding} = $est_cfg->{catalyst_encoding} || $c->config->{catalyst_encoding} or $c->log->fatal("can't find catalyst_encoding"); |
69 |
|
|
70 |
$log->debug("using config:" . Dumper($est_cfg) ); |
$log->debug("using config:" . Dumper($est_cfg) ); |
71 |
|
|
72 |
|
if (! $est_cfg->{database}) { |
73 |
|
my $defaultnode = $est_cfg->{defaultnode} || $log->logdie("can't find defaultnode in estraier configuration"); |
74 |
|
$log->info("using default node $defaultnode"); |
75 |
|
$est_cfg->{database} = $defaultnode; |
76 |
|
} |
77 |
|
|
78 |
$self->{est} = new WebPAC::Search::Estraier( %{ $est_cfg } ); |
$self->{est} = new WebPAC::Search::Estraier( %{ $est_cfg } ); |
79 |
|
|
80 |
# save config parametars in object |
# save config parametars in object |
81 |
foreach my $f (qw/db_path template_path hits_on_page webpac_encoding out_encoding/) { |
foreach my $f (qw/db_path template_path hits_on_page webpac_encoding out_encoding defaultdepth/) { |
82 |
$self->{$f} = $c->config->{hyperestraier}->{$f} || |
$self->{$f} = $c->config->{hyperestraier}->{$f} || |
83 |
$c->config->{webpac}->{$f}; |
$c->config->{webpac}->{$f}; |
84 |
$log->debug("self->{$f} = " . $self->{$f}); |
$log->debug("self->{$f} = " . $self->{$f}); |
88 |
|
|
89 |
$log->debug("using db path '$db_path', template path '$template_path'"); |
$log->debug("using db path '$db_path', template path '$template_path'"); |
90 |
|
|
91 |
$self->{db} = new WebPAC::DB( |
$self->{db} = new WebPAC::Store( |
92 |
path => $db_path, |
path => $db_path, |
93 |
read_only => 1, |
read_only => 1, |
94 |
|
database => $est_cfg->{database}, |
95 |
); |
); |
96 |
|
|
97 |
$self->{out} = new WebPAC::Output::TT( |
$self->{out} = new WebPAC::Output::TT( |
114 |
"'" |
"'" |
115 |
); |
); |
116 |
|
|
117 |
|
$self->{databases} = $c->config->{databases} || $log->error("can't find databases in config"); |
118 |
|
|
119 |
return $self; |
return $self; |
120 |
|
|
129 |
get_attr => [ '@uri' ], |
get_attr => [ '@uri' ], |
130 |
max => 42, |
max => 42, |
131 |
template => 'result_template.tt', |
template => 'result_template.tt', |
132 |
|
depth => 1, |
133 |
); |
); |
134 |
|
|
135 |
All fields are standard C<WebPAC::Search::Estraier> parametars except |
All fields are standard C<WebPAC::Search::Estraier> parametars except |
158 |
|
|
159 |
my $template_filename = $args->{template} || $self->{template}; |
my $template_filename = $args->{template} || $self->{template}; |
160 |
|
|
161 |
$args->{max} ||= $self->{'hits_on_page'}; |
$args->{max} ||= $self->{'hits_for_pager'}; |
162 |
if (! $args->{max}) { |
if (! $args->{max}) { |
163 |
$args->{max} = 10; |
$args->{max} = 100; |
164 |
$log->warn("max not set when calling model. Using default of 10"); |
$log->warn("max not set when calling model. Using default of $args->{max}"); |
165 |
} |
} |
166 |
|
|
167 |
my $times; # store some times for benchmarking |
my $times; # store some times for benchmarking |
168 |
|
|
169 |
my $t = time(); |
my $t = time(); |
170 |
|
|
171 |
|
# transfer depth of search |
172 |
|
if (! $args->{depth}) { |
173 |
|
my $default = $self->{defaultdepth} || $log->logdie("can't find defaultdepth in estraier configuration"); |
174 |
|
$args->{depth} = $default; |
175 |
|
$log->warn("using default search depth $default"); |
176 |
|
} |
177 |
|
|
178 |
my @results = $self->{est}->search( %{ $args } ); |
my @results = $self->{est}->search( %{ $args } ); |
179 |
|
|
180 |
$times->{est} += time() - $t; |
$times->{est} += time() - $t; |
194 |
|
|
195 |
for my $i ( 0 .. $#results ) { |
for my $i ( 0 .. $#results ) { |
196 |
|
|
197 |
my $mfn = $1 if ( $results[$i]->{'@uri'} =~ m#/(\d+)$#); |
my ($database, $prefix, $id); |
198 |
|
if ( $results[$i]->{'@uri'} =~ m!/([^/]+)/([^/]+)/(\d+)$!) { |
199 |
|
($database, $prefix,$id) = ($1,$2,$3); |
200 |
|
} else { |
201 |
|
$log->warn("can't decode database/prefix/id from " . $results[$i]->{'@uri'}); |
202 |
|
next; |
203 |
|
} |
204 |
|
|
205 |
#$log->debug("load_ds( $mfn )"); |
#$log->debug("load_ds( id => $id, prefix => '$prefix' )"); |
206 |
|
|
207 |
$t = time(); |
$t = time(); |
208 |
|
|
209 |
my $ds = $self->{db}->load_ds( $mfn ) || $log->error("can't load_ds( $mfn )") && next; |
my $ds = $self->{db}->load_ds( database => $database, prefix => $prefix, id => $id ); |
210 |
|
if (! $ds) { |
211 |
|
$log->error("can't load_ds( ${database}/${prefix}/${id} )"); |
212 |
|
next; |
213 |
|
} |
214 |
|
|
215 |
$times->{db} += time() - $t; |
$times->{db} += time() - $t; |
216 |
|
|
221 |
my $html = $self->{out}->apply( |
my $html = $self->{out}->apply( |
222 |
template => $template_filename, |
template => $template_filename, |
223 |
data => $ds, |
data => $ds, |
224 |
|
record_uri => "${database}/${prefix}/${id}", |
225 |
|
config => $self->{databases}->{$database}, |
226 |
); |
); |
227 |
|
|
228 |
$times->{out} += time() - $t; |
$times->{out} += time() - $t; |
266 |
my $log = $self->{log}; |
my $log = $self->{log}; |
267 |
$log->debug("args: " . Dumper( $args )); |
$log->debug("args: " . Dumper( $args )); |
268 |
|
|
269 |
foreach my $f (qw/mfn template/) { |
foreach my $f (qw/record_uri template/) { |
270 |
$log->die("need $f") unless ($args->{$f}); |
$log->fatal("need $f") unless ($args->{$f}); |
271 |
} |
} |
272 |
|
|
273 |
my $mfn = $args->{mfn}; |
my ($database, $prefix, $id); |
274 |
|
|
275 |
my $ds = $self->{db}->load_ds( $mfn ) || $log->error("can't load_ds( $mfn )") && next; |
if ($args->{record_uri} =~ m#^([^/]+)/([^/]+)/([^/]+)$#) { |
276 |
|
($database, $prefix, $id) = ($1,$2,$3); |
277 |
|
} else { |
278 |
|
$log->error("can't parse $args->{record_uri} into prefix, database and uri"); |
279 |
|
return; |
280 |
|
} |
281 |
|
|
282 |
|
my $ds = $self->{db}->load_ds( id => $id, prefix => $prefix, database => $database ); |
283 |
|
if (! $ds) { |
284 |
|
$log->error("can't load_ds( $database/$prefix/$id )"); |
285 |
|
return; |
286 |
|
} |
287 |
|
|
288 |
my $html = $self->{out}->apply( |
my $html = $self->{out}->apply( |
289 |
template => $args->{template}, |
template => $args->{template}, |
290 |
data => $ds, |
data => $ds, |
291 |
|
record_uri => $args->{record_uri}, |
292 |
|
config => $self->{databases}->{$database}, |
293 |
); |
); |
294 |
|
|
295 |
$html = $self->{iconv}->convert( $html ) || $log->error("can't convert: $html"); |
$html = $self->{iconv}->convert( $html ) || $log->error("can't convert: $html"); |
297 |
return $html; |
return $html; |
298 |
} |
} |
299 |
|
|
300 |
|
|
301 |
=head2 save_html |
=head2 save_html |
302 |
|
|
303 |
$m->save_html( '/full/path/to/file', $content ); |
$m->save_html( '/full/path/to/file', $content ); |
320 |
$content =~ s/%u([a-fA-F0-9]{4})/_conv_js($1)/gex; |
$content =~ s/%u([a-fA-F0-9]{4})/_conv_js($1)/gex; |
321 |
$content =~ s/^[\n\r]+//s; |
$content =~ s/^[\n\r]+//s; |
322 |
$content =~ s/[\n\r]+$/\n/s; |
$content =~ s/[\n\r]+$/\n/s; |
323 |
|
$content =~ s/\n\r/\n/gs; |
324 |
|
|
325 |
my ($from, $to) = ( |
my ($from, $to) = ( |
326 |
$self->{out_encoding}, |
$self->{out_encoding}, |