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 0.02; |
12 |
|
use WebPAC::Search::Estraier 0.05; |
13 |
|
use File::Slurp; |
14 |
|
use Time::HiRes; |
15 |
|
|
16 |
=head1 NAME |
=head1 NAME |
17 |
|
|
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 |
user: 'admin' |
defaultnode: 'webpac2' |
39 |
passwd: 'admin' |
defaultdepth: 1 |
40 |
|
user: 'admin' |
41 |
|
passwd: 'admin' |
42 |
|
hits_on_page: 100 |
43 |
|
|
44 |
|
webpac: |
45 |
|
db_path: '/data/webpac2/db' |
46 |
|
template_path: '/data/webpac2/conf/output/tt' |
47 |
|
template: 'html_ffzg_results_short.tt' |
48 |
|
# encoding comming from webpac |
49 |
|
webpac_encoding: 'iso-8859-2' |
50 |
|
# encoding expected by Catalyst |
51 |
|
out_encoding: 'UTF-8' |
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}; |
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 out_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 |
|
$self->{iconv} = new Text::Iconv( |
105 |
|
$c->config->{webpac}->{webpac_encoding}, |
106 |
|
$c->config->{webpac}->{out_encoding} |
107 |
|
); |
108 |
|
|
109 |
|
$log->debug("converting encoding from webpac_encoding '" . |
110 |
|
$c->config->{webpac}->{webpac_encoding} . |
111 |
|
"' to '" . |
112 |
|
$c->config->{webpac}->{out_encoding} . |
113 |
|
"'" |
114 |
|
); |
115 |
|
|
116 |
|
|
117 |
return $self; |
return $self; |
118 |
|
|
119 |
} |
} |
120 |
|
|
121 |
|
|
122 |
|
=head2 search |
123 |
|
|
124 |
|
my $m->search( |
125 |
|
phrase => 'query phrase', |
126 |
|
add_attr => \@add_attr |
127 |
|
get_attr => [ '@uri' ], |
128 |
|
max => 42, |
129 |
|
template => 'result_template.tt', |
130 |
|
depth => 1, |
131 |
|
); |
132 |
|
|
133 |
|
All fields are standard C<WebPAC::Search::Estraier> parametars except |
134 |
|
C<template> which will (if specified) return results in HTML using |
135 |
|
selected template. |
136 |
|
|
137 |
|
=cut |
138 |
|
|
139 |
sub search { |
sub search { |
140 |
my ( $self, $query ) = @_; |
my $self = shift; |
141 |
|
|
142 |
|
my $args = {@_}; |
143 |
|
|
144 |
|
my $log = $self->{log}; |
145 |
|
|
146 |
|
$log->debug("args: " . Dumper( $args )); |
147 |
|
|
148 |
|
my $query = $args->{phrase} || $log->warn("no query phrase") && return; |
149 |
|
|
150 |
|
$log->debug("search model query: '$query'"); |
151 |
|
if ($args->{add_attr}) { |
152 |
|
$log->debug(" + add_attr: " . |
153 |
|
join("','", @{ $args->{add_attr} }) |
154 |
|
); |
155 |
|
} |
156 |
|
|
157 |
|
my $template_filename = $args->{template} || $self->{template}; |
158 |
|
|
159 |
|
$args->{max} ||= $self->{'hits_on_page'}; |
160 |
|
if (! $args->{max}) { |
161 |
|
$args->{max} = 10; |
162 |
|
$log->warn("max not set when calling model. Using default of 10"); |
163 |
|
} |
164 |
|
|
165 |
|
my $times; # store some times for benchmarking |
166 |
|
|
167 |
|
my $t = time(); |
168 |
|
|
169 |
|
# transfer depth of search |
170 |
|
if (! $args->{depth}) { |
171 |
|
my $default = $self->{defaultdepth} || $log->logdie("can't find defaultdepth in estraier configuration"); |
172 |
|
$args->{depth} = $default; |
173 |
|
$log->warn("using default search depth $default"); |
174 |
|
} |
175 |
|
|
176 |
|
my @results = $self->{est}->search( %{ $args } ); |
177 |
|
|
178 |
|
$times->{est} += time() - $t; |
179 |
|
|
180 |
|
my $hits = $#results + 1; |
181 |
|
|
182 |
|
$log->debug( sprintf("search took %.2fs and returned $hits hits.", $times->{est}) ); |
183 |
|
|
184 |
|
# just return results? |
185 |
|
return @results unless ($args->{'template'}); |
186 |
|
|
187 |
|
# |
188 |
|
# construct HTML results |
189 |
|
# |
190 |
|
|
191 |
|
my @html_results; |
192 |
|
|
193 |
|
for my $i ( 0 .. $#results ) { |
194 |
|
|
195 |
|
my ($database, $prefix, $id); |
196 |
|
if ( $results[$i]->{'@uri'} =~ m!/([^/]+)/([^/]+)#(\d+)$!) { |
197 |
|
($database, $prefix,$id) = ($1,$2,$3); |
198 |
|
} else { |
199 |
|
$log->warn("can't decode prefix#id from " . $results[$i]->{'@uri'}); |
200 |
|
next; |
201 |
|
} |
202 |
|
|
203 |
|
#$log->debug("load_ds( id => $id, prefix => '$prefix' )"); |
204 |
|
|
205 |
|
$t = time(); |
206 |
|
|
207 |
|
my $ds = $self->{db}->load_ds( database => $database, prefix => $prefix, id => $id ) || $log->error("can't load_ds( $database / $prefix # $id )") && next; |
208 |
|
|
209 |
|
$times->{db} += time() - $t; |
210 |
|
|
211 |
|
#$log->debug( "ds = " . Dumper( \@html_results ) ); |
212 |
|
|
213 |
|
$t = time(); |
214 |
|
|
215 |
|
my $html = $self->{out}->apply( |
216 |
|
template => $template_filename, |
217 |
|
data => $ds, |
218 |
|
record_uri => "${database}/${prefix}#${id}", |
219 |
|
); |
220 |
|
|
221 |
my @results = $self->{est}->search( |
$times->{out} += time() - $t; |
222 |
query => $query, |
|
223 |
attr => [ '@uri' ], |
$t = time(); |
224 |
max => 100, |
|
225 |
|
$html = $self->{iconv}->convert( $html ) || $log->error("can't convert: $html"); |
226 |
|
|
227 |
|
$times->{iconv} += time() - $t; |
228 |
|
|
229 |
|
push @html_results, $html; |
230 |
|
|
231 |
|
} |
232 |
|
|
233 |
|
#$log->debug( '@html_results = ' . Dumper( \@html_results ) ); |
234 |
|
|
235 |
|
$log->debug( sprintf( |
236 |
|
"time spent: db = %.2f, out = %.2f, iconv = %.2f", |
237 |
|
$times->{db}, $times->{out}, $times->{iconv}, |
238 |
|
) ); |
239 |
|
|
240 |
|
return \@html_results; |
241 |
|
} |
242 |
|
|
243 |
|
=head2 record |
244 |
|
|
245 |
|
my $html = $m->record( |
246 |
|
mfn => 42, |
247 |
|
template => 'foo.tt', |
248 |
|
); |
249 |
|
|
250 |
|
This will load one record, convert it to html using C<template> and return |
251 |
|
it. |
252 |
|
|
253 |
|
=cut |
254 |
|
|
255 |
|
sub record { |
256 |
|
my $self = shift; |
257 |
|
|
258 |
|
my $args = {@_}; |
259 |
|
my $log = $self->{log}; |
260 |
|
$log->debug("args: " . Dumper( $args )); |
261 |
|
|
262 |
|
foreach my $f (qw/mfn template prefix/) { |
263 |
|
$log->fatal("need $f") unless ($args->{$f}); |
264 |
|
} |
265 |
|
|
266 |
|
my $mfn = $args->{mfn}; |
267 |
|
|
268 |
|
my $ds = $self->{db}->load_ds( id => $mfn, prefix => 'FIXME' ) || $log->error("can't load_ds( $mfn )") && return; |
269 |
|
|
270 |
|
my $html = $self->{out}->apply( |
271 |
|
template => $args->{template}, |
272 |
|
data => $ds, |
273 |
|
); |
274 |
|
|
275 |
|
$html = $self->{iconv}->convert( $html ) || $log->error("can't convert: $html"); |
276 |
|
|
277 |
|
return $html; |
278 |
|
} |
279 |
|
|
280 |
|
=head2 save_html |
281 |
|
|
282 |
|
$m->save_html( '/full/path/to/file', $content ); |
283 |
|
|
284 |
|
It will use C<Text::Iconv> to convert content encoding back to |
285 |
|
Webpac codepage, recode JavaScript Unicode entities (%u1234), |
286 |
|
strip extra newlines at beginning and end, and save to |
287 |
|
C</full/path/to/file.new> and if that succeeds, just rename |
288 |
|
it over original file which should be atomic on filesystem level. |
289 |
|
|
290 |
|
=cut |
291 |
|
|
292 |
|
sub save_html { |
293 |
|
my ($self, $path, $content) = @_; |
294 |
|
|
295 |
|
sub _conv_js { |
296 |
|
my $t = shift || return; |
297 |
|
return $self->{iconv}->convert(chr(hex($t))); |
298 |
|
} |
299 |
|
$content =~ s/%u([a-fA-F0-9]{4})/_conv_js($1)/gex; |
300 |
|
$content =~ s/^[\n\r]+//s; |
301 |
|
$content =~ s/[\n\r]+$/\n/s; |
302 |
|
|
303 |
|
my ($from, $to) = ( |
304 |
|
$self->{out_encoding}, |
305 |
|
$self->{webpac_encoding}, |
306 |
); |
); |
307 |
|
|
308 |
return \@results; |
$self->{log}->debug("using iconv to convert from $from to $to encoding"); |
309 |
|
|
310 |
|
my $iconv_on_save = new Text::Iconv($from, $to) |
311 |
|
|| $self->{log}->fatal("can't create iconv for saving"); |
312 |
|
|
313 |
|
$content = $iconv_on_save->convert( $content ) || die "no content?"; |
314 |
|
|
315 |
|
write_file($path . '.new', {binmode => ':raw' }, $content) || die "can't save ${path}.new $!"; |
316 |
|
rename $path . '.new', $path || die "can't rename to $path: $!"; |
317 |
} |
} |
318 |
|
|
319 |
|
=head2 load_html |
320 |
|
|
321 |
|
my $html = $m->load_html('/full/path/to/file'); |
322 |
|
|
323 |
|
This will convert file from Webpac encoding to Catalyst and |
324 |
|
convert that data to escaped HTML (for sending into |
325 |
|
C<< <textarea/> >> tags in html. |
326 |
|
|
327 |
|
=cut |
328 |
|
|
329 |
|
sub load_html { |
330 |
|
my ($self, $path) = @_; |
331 |
|
|
332 |
|
die "no path?" unless ($path); |
333 |
|
|
334 |
|
my $content = read_file($path, {binmode => ':raw' }) || die "can't read $path: $!"; |
335 |
|
#$content = $q->escapeHTML($iconv_utf8->convert($content)); |
336 |
|
$content = $self->{iconv}->convert($content); |
337 |
|
|
338 |
|
return $content; |
339 |
|
} |
340 |
|
|
341 |
=head1 AUTHOR |
=head1 AUTHOR |
342 |
|
|