| 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 Data::Dumper; |
| 10 |
use WebPAC::DB; |
| 11 |
use WebPAC::Output::TT; |
| 12 |
use WebPAC::Search::Estraier 0.02; |
| 13 |
use File::Slurp; |
| 14 |
use Time::HiRes; |
| 15 |
|
| 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 |
url: 'http://localhost:1978/node/webpac2' |
| 38 |
user: 'admin' |
| 39 |
passwd: 'admin' |
| 40 |
hits_on_page: 100 |
| 41 |
|
| 42 |
webpac: |
| 43 |
db_path: '/data/webpac2/db' |
| 44 |
template_path: '/data/webpac2/conf/output/tt' |
| 45 |
template: 'html_ffzg_results_short.tt' |
| 46 |
# encoding comming from webpac |
| 47 |
webpac_encoding: 'iso-8859-2' |
| 48 |
# encoding expected by Catalyst |
| 49 |
out_encoding: 'UTF-8' |
| 50 |
|
| 51 |
=cut |
| 52 |
|
| 53 |
sub new { |
| 54 |
my ( $self, $c, $config ) = @_; |
| 55 |
|
| 56 |
$self = $self->NEXT::new($c, $config); |
| 57 |
$self->config($config); |
| 58 |
|
| 59 |
my $log = $c->log; |
| 60 |
$self->{log} = $log; |
| 61 |
|
| 62 |
my $est_cfg = $c->config->{hyperestraier}; |
| 63 |
$est_cfg->{'log'} = $log; |
| 64 |
|
| 65 |
$est_cfg->{encoding} = $est_cfg->{catalyst_encoding}; |
| 66 |
|
| 67 |
$log->debug("using config:" . Dumper($est_cfg) ); |
| 68 |
|
| 69 |
$self->{est} = new WebPAC::Search::Estraier( %{ $est_cfg } ); |
| 70 |
|
| 71 |
my $db_path = $c->config->{webpac}->{db_path}; |
| 72 |
my $template_path = $c->config->{webpac}->{template_path}; |
| 73 |
$self->{template_path} = $template_path; |
| 74 |
|
| 75 |
$log->debug("using db path '$db_path', template path '$template_path'"); |
| 76 |
|
| 77 |
$self->{db} = new WebPAC::DB( |
| 78 |
path => $db_path, |
| 79 |
read_only => 1, |
| 80 |
); |
| 81 |
|
| 82 |
$self->{out} = new WebPAC::Output::TT( |
| 83 |
include_path => $template_path, |
| 84 |
filters => { foo => sub { shift } }, |
| 85 |
); |
| 86 |
|
| 87 |
# default template from config.yaml |
| 88 |
$self->{template} ||= $c->config->{webpac}->{template}; |
| 89 |
|
| 90 |
$self->{iconv} = new Text::Iconv( |
| 91 |
$c->config->{webpac}->{webpac_encoding}, |
| 92 |
$c->config->{webpac}->{out_encoding} |
| 93 |
); |
| 94 |
|
| 95 |
$log->debug("converting encoding from webpac_encoding '" . |
| 96 |
$c->config->{webpac}->{webpac_encoding} . |
| 97 |
"' to '" . |
| 98 |
$c->config->{webpac}->{out_encoding} . |
| 99 |
"'" |
| 100 |
); |
| 101 |
|
| 102 |
# save config parametars in object |
| 103 |
foreach my $f (qw/hits_on_page/) { |
| 104 |
$self->{$_} = $c->config->{$_}; |
| 105 |
$log->debug("self->{$_} = " . $c->config->{$_}); |
| 106 |
} |
| 107 |
|
| 108 |
return $self; |
| 109 |
|
| 110 |
} |
| 111 |
|
| 112 |
=head2 iconv_on_save |
| 113 |
|
| 114 |
my $out = $m->iconv_on_save( $content ); |
| 115 |
|
| 116 |
Convert data saved to disk in Webpac encoding. |
| 117 |
|
| 118 |
=cut |
| 119 |
|
| 120 |
sub iconv_on_save { |
| 121 |
my $self = shift; |
| 122 |
|
| 123 |
$self->{iconv_save} ||= new Text::Iconv( |
| 124 |
$self->config->{webpac}->{out_encoding}, |
| 125 |
$self->config->{webpac}->{webpac_encoding}, |
| 126 |
); |
| 127 |
|
| 128 |
$self->{iconv_save}->convert( @_ ); |
| 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 |
); |
| 141 |
|
| 142 |
All fields are standard C<WebPAC::Search::Estraier> parametars except |
| 143 |
C<template> which will (if specified) return results in HTML using |
| 144 |
selected template. |
| 145 |
|
| 146 |
=cut |
| 147 |
|
| 148 |
sub search { |
| 149 |
my $self = shift; |
| 150 |
|
| 151 |
my $args = {@_}; |
| 152 |
|
| 153 |
my $log = $self->{log}; |
| 154 |
|
| 155 |
$log->debug("args: " . Dumper( $args )); |
| 156 |
|
| 157 |
my $query = $args->{phrase} || $log->warn("no query phrase") && return; |
| 158 |
|
| 159 |
$log->debug("search model query: '$query'"); |
| 160 |
if ($args->{add_attr}) { |
| 161 |
$log->debug(" + add_attr: " . |
| 162 |
join("','", @{ $args->{add_attr} }) |
| 163 |
); |
| 164 |
} |
| 165 |
|
| 166 |
my $template_filename = $args->{template} || $self->{template}; |
| 167 |
|
| 168 |
$args->{max} ||= $self->{'hits_on_page'}; |
| 169 |
if (! $args->{max}) { |
| 170 |
$args->{max} = 10; |
| 171 |
$log->warn("max not set when calling model. Using default of 10"); |
| 172 |
} |
| 173 |
|
| 174 |
my $times; # store some times for benchmarking |
| 175 |
|
| 176 |
my $t = time(); |
| 177 |
|
| 178 |
my @results = $self->{est}->search( %{ $args } ); |
| 179 |
|
| 180 |
$times->{est} += time() - $t; |
| 181 |
|
| 182 |
my $hits = $#results + 1; |
| 183 |
|
| 184 |
$log->debug( sprintf("search took %.2fs and returned $hits hits.", $times->{est}) ); |
| 185 |
|
| 186 |
# just return results? |
| 187 |
return @results unless ($args->{'template'}); |
| 188 |
|
| 189 |
# |
| 190 |
# construct HTML results |
| 191 |
# |
| 192 |
|
| 193 |
my @html_results; |
| 194 |
|
| 195 |
for my $i ( 0 .. $#results ) { |
| 196 |
|
| 197 |
my $mfn = $1 if ( $results[$i]->{'@uri'} =~ m#/(\d+)$#); |
| 198 |
|
| 199 |
#$log->debug("load_ds( $mfn )"); |
| 200 |
|
| 201 |
$t = time(); |
| 202 |
|
| 203 |
my $ds = $self->{db}->load_ds( $mfn ) || $log->error("can't load_ds( $mfn )") && next; |
| 204 |
|
| 205 |
$times->{db} += time() - $t; |
| 206 |
|
| 207 |
#$log->debug( "ds = " . Dumper( \@html_results ) ); |
| 208 |
|
| 209 |
$t = time(); |
| 210 |
|
| 211 |
my $html = $self->{out}->apply( |
| 212 |
template => $template_filename, |
| 213 |
data => $ds, |
| 214 |
); |
| 215 |
|
| 216 |
$times->{out} += time() - $t; |
| 217 |
|
| 218 |
$t = time(); |
| 219 |
|
| 220 |
$html = $self->{iconv}->convert( $html ) || $log->error("can't convert: $html"); |
| 221 |
|
| 222 |
$times->{iconv} += time() - $t; |
| 223 |
|
| 224 |
push @html_results, $html; |
| 225 |
|
| 226 |
} |
| 227 |
|
| 228 |
#$log->debug( '@html_results = ' . Dumper( \@html_results ) ); |
| 229 |
|
| 230 |
$log->debug( sprintf( |
| 231 |
"time spent: db = %.2f, out = %.2f, iconv = %.2f", |
| 232 |
$times->{db}, $times->{out}, $times->{iconv}, |
| 233 |
) ); |
| 234 |
|
| 235 |
return \@html_results; |
| 236 |
} |
| 237 |
|
| 238 |
=head2 save_html |
| 239 |
|
| 240 |
$m->save_html( '/full/path/to/file', $content ); |
| 241 |
|
| 242 |
It will use C<iconv_on_save> to convert content encoding back to |
| 243 |
Webpac codepage, recode JavaScript Unicode entities (%u1234), |
| 244 |
strip extra newlines at beginning and end, and save to |
| 245 |
C</full/path/to/file.new> and if that succeeds, just rename |
| 246 |
it over original file which should be atomic on filesystem level. |
| 247 |
|
| 248 |
=cut |
| 249 |
|
| 250 |
sub save_html { |
| 251 |
my ($self, $path, $content) = @_; |
| 252 |
|
| 253 |
$content = $self->iconv_on_save( $content ) || die "no content?"; |
| 254 |
|
| 255 |
sub _conv_js { |
| 256 |
my $t = shift || return; |
| 257 |
return $self->{iconv}->convert(chr(hex($t))); |
| 258 |
} |
| 259 |
$content =~ s/%u([a-fA-F0-9]{4})/_conv_js($1)/gex; |
| 260 |
$content =~ s/^[\n\r]+//s; |
| 261 |
$content =~ s/[\n\r]+$/\n/s; |
| 262 |
|
| 263 |
write_file($path . '.new', $content) || die "can't save ${path}.new $!"; |
| 264 |
rename $path . '.new', $path || die "can't rename to $path: $!"; |
| 265 |
} |
| 266 |
|
| 267 |
=head2 load_html |
| 268 |
|
| 269 |
my $html = $m->load_html('/full/path/to/file'); |
| 270 |
|
| 271 |
This will convert file from Webpac encoding to Catalyst and |
| 272 |
convert that data to escaped HTML (for sending into |
| 273 |
C<< <textarea/> >> tags in html. |
| 274 |
|
| 275 |
=cut |
| 276 |
|
| 277 |
sub load_html { |
| 278 |
my ($self, $path) = @_; |
| 279 |
|
| 280 |
die "no path?" unless ($path); |
| 281 |
|
| 282 |
my $content = read_file($path) || die "can't read $path: $!"; |
| 283 |
#$content = $q->escapeHTML($iconv_utf8->convert($content)); |
| 284 |
$content = $self->{iconv}->convert($content); |
| 285 |
|
| 286 |
return $content; |
| 287 |
} |
| 288 |
|
| 289 |
=head1 AUTHOR |
| 290 |
|
| 291 |
Dobrica Pavlinusic |
| 292 |
|
| 293 |
=head1 LICENSE |
| 294 |
|
| 295 |
This library is free software, you can redistribute it and/or modify |
| 296 |
it under the same terms as Perl itself. |
| 297 |
|
| 298 |
=cut |
| 299 |
|
| 300 |
1; |
Parent Directory
| 
Revision Log