3 |
use strict; |
use strict; |
4 |
use warnings; |
use warnings; |
5 |
use lib '/data/webpac2/lib'; |
use lib '/data/webpac2/lib'; |
6 |
use base qw/Catalyst::Model/; |
use base qw/ |
7 |
use WebPAC::Search::Estraier; |
Catalyst::Model |
8 |
|
/; |
9 |
use Data::Dumper; |
use Data::Dumper; |
10 |
|
use WebPAC::DB; |
11 |
|
use WebPAC::Output::TT; |
12 |
|
use WebPAC::Search::Estraier 0.02; |
13 |
|
use File::Slurp; |
14 |
|
|
15 |
=head1 NAME |
=head1 NAME |
16 |
|
|
33 |
|
|
34 |
# configuration for hyper estraier full text search engine |
# configuration for hyper estraier full text search engine |
35 |
hyperestraier: |
hyperestraier: |
36 |
url: 'http://localhost:1978/node/webpac2' |
url: 'http://localhost:1978/node/webpac2' |
37 |
user: 'admin' |
user: 'admin' |
38 |
passwd: 'admin' |
passwd: 'admin' |
39 |
|
hits_on_page: 100 |
40 |
|
|
41 |
|
webpac: |
42 |
|
db_path: '/data/webpac2/db' |
43 |
|
template_path: '/data/webpac2/conf/output/tt' |
44 |
|
template: 'html_ffzg_results_short.tt' |
45 |
|
# encoding comming from webpac |
46 |
|
webpac_encoding: 'iso-8859-2' |
47 |
|
# encoding expected by Catalyst |
48 |
|
out_encoding: 'UTF-8' |
49 |
|
|
50 |
=cut |
=cut |
51 |
|
|
56 |
$self->config($config); |
$self->config($config); |
57 |
|
|
58 |
my $log = $c->log; |
my $log = $c->log; |
59 |
|
$self->{log} = $log; |
60 |
|
|
61 |
# if (! $c->stash->{est}) { |
my $est_cfg = $c->config->{hyperestraier}; |
62 |
|
$est_cfg->{'log'} = $log; |
63 |
|
|
64 |
my $est_cfg = $c->config->{hyperestraier}; |
$est_cfg->{encoding} = $est_cfg->{catalyst_encoding}; |
|
$est_cfg->{'log'} = $log; |
|
65 |
|
|
66 |
$log->debug("using config:" . Dumper($est_cfg) ); |
$log->debug("using config:" . Dumper($est_cfg) ); |
67 |
|
|
68 |
# $c->stash->{est} = new WebPAC::Search::Estraier( %{ $est_cfg } ); |
$self->{est} = new WebPAC::Search::Estraier( %{ $est_cfg } ); |
|
# } |
|
69 |
|
|
70 |
# $log->debug("param: " . Dumper($c->req->params)); |
my $db_path = $c->config->{webpac}->{db_path}; |
71 |
|
my $template_path = $c->config->{webpac}->{template_path}; |
72 |
|
$self->{template_path} = $template_path; |
73 |
|
|
74 |
# $c->stash->{est}->search( |
$log->debug("using db path '$db_path', template path '$template_path'"); |
75 |
# query => $c->req->params->{Title}, |
|
76 |
# max => 100, |
$self->{db} = new WebPAC::DB( |
77 |
# ); |
path => $db_path, |
78 |
|
read_only => 1, |
79 |
|
); |
80 |
|
|
81 |
|
$self->{out} = new WebPAC::Output::TT( |
82 |
|
include_path => $template_path, |
83 |
|
filters => { foo => sub { shift } }, |
84 |
|
); |
85 |
|
|
86 |
|
# default template from config.yaml |
87 |
|
$self->{template} ||= $c->config->{webpac}->{template}; |
88 |
|
|
89 |
|
$self->{iconv} = new Text::Iconv( |
90 |
|
$c->config->{webpac}->{webpac_encoding}, |
91 |
|
$c->config->{webpac}->{out_encoding} |
92 |
|
); |
93 |
|
|
94 |
|
$log->debug("converting encoding from webpac_encoding '" . |
95 |
|
$c->config->{webpac}->{webpac_encoding} . |
96 |
|
"' to '" . |
97 |
|
$c->config->{webpac}->{out_encoding} . |
98 |
|
"'" |
99 |
|
); |
100 |
|
|
101 |
return $self; |
return $self; |
102 |
|
|
103 |
} |
} |
104 |
|
|
105 |
|
=head2 iconv_on_save |
106 |
|
|
107 |
|
my $out = $m->iconv_on_save( $content ); |
108 |
|
|
109 |
|
Convert data saved to disk in Webpac encoding. |
110 |
|
|
111 |
|
=cut |
112 |
|
|
113 |
|
sub iconv_on_save { |
114 |
|
my $self = shift; |
115 |
|
|
116 |
|
$self->{iconv_save} ||= new Text::Iconv( |
117 |
|
$self->config->{webpac}->{out_encoding}, |
118 |
|
$self->config->{webpac}->{webpac_encoding}, |
119 |
|
); |
120 |
|
|
121 |
|
$self->{iconv_save}->convert( @_ ); |
122 |
|
} |
123 |
|
|
124 |
|
|
125 |
|
=head2 search |
126 |
|
|
127 |
|
my $m->search( 'query phrase', 'result_template.tt', \@add_attr ); |
128 |
|
|
129 |
|
=cut |
130 |
|
|
131 |
|
sub search { |
132 |
|
my ( $self, $query, $template, $add_attr ) = @_; |
133 |
|
|
134 |
|
my $log = $self->{log}; |
135 |
|
|
136 |
|
$log->debug("search model query: '$query', add_attr: '" . join("','", @{$add_attr}) . "'"); |
137 |
|
|
138 |
|
my $template_filename = $template || $self->{template}; |
139 |
|
|
140 |
|
my @results = $self->{est}->search( |
141 |
|
phrase => $query, |
142 |
|
get_attr => [ '@uri' ], |
143 |
|
max => $self->{est}->{hits_on_page} || 30, |
144 |
|
add_attr => $add_attr, |
145 |
|
); |
146 |
|
|
147 |
|
$log->debug("loading " . ($#results + 1) . " results"); |
148 |
|
|
149 |
|
my @html_results; |
150 |
|
|
151 |
|
for my $i ( 0 .. $#results ) { |
152 |
|
|
153 |
|
my $mfn = $1 if ( $results[$i]->{'@uri'} =~ m#/(\d+)$#); |
154 |
|
|
155 |
|
#$log->debug("load_ds( $mfn )"); |
156 |
|
|
157 |
|
my $ds = $self->{db}->load_ds( $mfn ) || $log->error("can't load_ds( $mfn )") && next; |
158 |
|
|
159 |
|
#$log->debug( "ds = " . Dumper( \@html_results ) ); |
160 |
|
|
161 |
|
my $html = $self->{out}->apply( |
162 |
|
template => $template_filename, |
163 |
|
data => $ds, |
164 |
|
); |
165 |
|
|
166 |
|
$html = $self->{iconv}->convert( $html ) || $log->error("can't convert: $html"); |
167 |
|
|
168 |
|
push @html_results, $html; |
169 |
|
|
170 |
|
} |
171 |
|
|
172 |
|
#$log->debug( '@html_results = ' . Dumper( \@html_results ) ); |
173 |
|
|
174 |
|
return \@html_results; |
175 |
|
} |
176 |
|
|
177 |
|
=head2 save_html |
178 |
|
|
179 |
|
$m->save_html( '/full/path/to/file', $content ); |
180 |
|
|
181 |
|
It will use C<iconv_on_save> to convert content encoding back to |
182 |
|
Webpac codepage, recode JavaScript Unicode entities (%u1234), |
183 |
|
strip extra newlines at beginning and end, and save to |
184 |
|
C</full/path/to/file.new> and if that succeeds, just rename |
185 |
|
it over original file which should be atomic on filesystem level. |
186 |
|
|
187 |
|
=cut |
188 |
|
|
189 |
|
sub save_html { |
190 |
|
my ($self, $path, $content) = @_; |
191 |
|
|
192 |
|
$content = $self->iconv_on_save( $content ) || die "no content?"; |
193 |
|
|
194 |
|
sub _conv_js { |
195 |
|
my $t = shift || return; |
196 |
|
return $self->{iconv}->convert(chr(hex($t))); |
197 |
|
} |
198 |
|
$content =~ s/%u([a-fA-F0-9]{4})/_conv_js($1)/gex; |
199 |
|
$content =~ s/^[\n\r]+//s; |
200 |
|
$content =~ s/[\n\r]+$/\n/s; |
201 |
|
|
202 |
|
write_file($path . '.new', $content) || die "can't save ${path}.new $!"; |
203 |
|
rename $path . '.new', $path || die "can't rename to $path: $!"; |
204 |
|
} |
205 |
|
|
206 |
|
=head2 load_html |
207 |
|
|
208 |
|
my $html = $m->load_html('/full/path/to/file'); |
209 |
|
|
210 |
|
This will convert file from Webpac encoding to Catalyst and |
211 |
|
convert that data to escaped HTML (for sending into |
212 |
|
C<< <textarea/> >> tags in html. |
213 |
|
|
214 |
|
=cut |
215 |
|
|
216 |
|
sub load_html { |
217 |
|
my ($self, $path) = @_; |
218 |
|
|
219 |
|
die "no path?" unless ($path); |
220 |
|
|
221 |
|
my $content = read_file($path) || die "can't read $path: $!"; |
222 |
|
#$content = $q->escapeHTML($iconv_utf8->convert($content)); |
223 |
|
$content = $self->{iconv}->convert($content); |
224 |
|
|
225 |
|
return $content; |
226 |
|
} |
227 |
|
|
228 |
=head1 AUTHOR |
=head1 AUTHOR |
229 |
|
|
230 |
Dobrica Pavlinusic |
Dobrica Pavlinusic |