/[webpac2]/Webpacus/lib/Webpacus/Model/WebPAC.pm
This is repository of my old source code which isn't updated any more. Go to git.rot13.org for current projects!
ViewVC logotype

Annotation of /Webpacus/lib/Webpacus/Model/WebPAC.pm

Parent Directory Parent Directory | Revision Log Revision Log


Revision 200 - (hide annotations)
Wed Nov 30 23:21:30 2005 UTC (18 years, 5 months ago) by dpavlin
File size: 7063 byte(s)
TEMPORARY BROKEN VERSION [0.11]
minor tweaks, usage of WebPAC::Output::TT 0.02 (to support search() filter). It segfaults
on 8-bit characters in URL.

1 dpavlin 92 package Webpacus::Model::WebPAC;
2    
3     use strict;
4     use warnings;
5     use lib '/data/webpac2/lib';
6 dpavlin 93 use base qw/
7     Catalyst::Model
8     /;
9 dpavlin 92 use Data::Dumper;
10 dpavlin 95 use WebPAC::DB;
11 dpavlin 200 use WebPAC::Output::TT 0.02;
12 dpavlin 116 use WebPAC::Search::Estraier 0.02;
13 dpavlin 135 use File::Slurp;
14 dpavlin 155 use Time::HiRes;
15 dpavlin 92
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 dpavlin 96 url: 'http://localhost:1978/node/webpac2'
38     user: 'admin'
39     passwd: 'admin'
40 dpavlin 143 hits_on_page: 100
41 dpavlin 92
42 dpavlin 96 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 dpavlin 92 =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 dpavlin 94 $self->{log} = $log;
61 dpavlin 92
62 dpavlin 93 my $est_cfg = $c->config->{hyperestraier};
63     $est_cfg->{'log'} = $log;
64 dpavlin 92
65 dpavlin 142 $est_cfg->{encoding} = $est_cfg->{catalyst_encoding};
66    
67 dpavlin 93 $log->debug("using config:" . Dumper($est_cfg) );
68 dpavlin 92
69 dpavlin 93 $self->{est} = new WebPAC::Search::Estraier( %{ $est_cfg } );
70 dpavlin 92
71 dpavlin 167 # save config parametars in object
72 dpavlin 179 foreach my $f (qw/db_path template_path hits_on_page webpac_encoding out_encoding/) {
73 dpavlin 167 $self->{$f} = $c->config->{hyperestraier}->{$f} ||
74     $c->config->{webpac}->{$f};
75     $log->debug("self->{$f} = " . $self->{$f});
76     }
77     my $db_path = $self->{db_path};
78     my $template_path = $self->{template_path};
79 dpavlin 95
80     $log->debug("using db path '$db_path', template path '$template_path'");
81    
82     $self->{db} = new WebPAC::DB(
83     path => $db_path,
84     read_only => 1,
85     );
86    
87     $self->{out} = new WebPAC::Output::TT(
88     include_path => $template_path,
89     filters => { foo => sub { shift } },
90     );
91    
92 dpavlin 101 # default template from config.yaml
93 dpavlin 95 $self->{template} ||= $c->config->{webpac}->{template};
94    
95 dpavlin 96 $self->{iconv} = new Text::Iconv(
96     $c->config->{webpac}->{webpac_encoding},
97     $c->config->{webpac}->{out_encoding}
98     );
99    
100 dpavlin 100 $log->debug("converting encoding from webpac_encoding '" .
101     $c->config->{webpac}->{webpac_encoding} .
102     "' to '" .
103     $c->config->{webpac}->{out_encoding} .
104 dpavlin 99 "'"
105     );
106 dpavlin 96
107 dpavlin 155
108 dpavlin 92 return $self;
109    
110     }
111    
112 dpavlin 135
113     =head2 search
114    
115 dpavlin 150 my $m->search(
116     phrase => 'query phrase',
117 dpavlin 155 add_attr => \@add_attr
118     get_attr => [ '@uri' ],
119     max => 42,
120 dpavlin 150 template => 'result_template.tt',
121     );
122 dpavlin 135
123 dpavlin 155 All fields are standard C<WebPAC::Search::Estraier> parametars except
124     C<template> which will (if specified) return results in HTML using
125     selected template.
126    
127 dpavlin 135 =cut
128    
129 dpavlin 93 sub search {
130 dpavlin 150 my $self = shift;
131 dpavlin 93
132 dpavlin 150 my $args = {@_};
133    
134 dpavlin 95 my $log = $self->{log};
135 dpavlin 94
136 dpavlin 150 $log->debug("args: " . Dumper( $args ));
137 dpavlin 95
138 dpavlin 150 my $query = $args->{phrase} || $log->warn("no query phrase") && return;
139 dpavlin 95
140 dpavlin 155 $log->debug("search model query: '$query'");
141     if ($args->{add_attr}) {
142     $log->debug(" + add_attr: " .
143     join("','", @{ $args->{add_attr} })
144     );
145     }
146 dpavlin 150
147     my $template_filename = $args->{template} || $self->{template};
148    
149 dpavlin 155 $args->{max} ||= $self->{'hits_on_page'};
150     if (! $args->{max}) {
151     $args->{max} = 10;
152     $log->warn("max not set when calling model. Using default of 10");
153     }
154 dpavlin 93
155 dpavlin 155 my $times; # store some times for benchmarking
156    
157     my $t = time();
158    
159     my @results = $self->{est}->search( %{ $args } );
160    
161     $times->{est} += time() - $t;
162    
163 dpavlin 150 my $hits = $#results + 1;
164 dpavlin 99
165 dpavlin 155 $log->debug( sprintf("search took %.2fs and returned $hits hits.", $times->{est}) );
166 dpavlin 150
167 dpavlin 155 # just return results?
168     return @results unless ($args->{'template'});
169    
170     #
171     # construct HTML results
172     #
173    
174 dpavlin 100 my @html_results;
175    
176 dpavlin 95 for my $i ( 0 .. $#results ) {
177    
178     my $mfn = $1 if ( $results[$i]->{'@uri'} =~ m#/(\d+)$#);
179    
180 dpavlin 115 #$log->debug("load_ds( $mfn )");
181 dpavlin 95
182 dpavlin 155 $t = time();
183    
184 dpavlin 115 my $ds = $self->{db}->load_ds( $mfn ) || $log->error("can't load_ds( $mfn )") && next;
185 dpavlin 155
186     $times->{db} += time() - $t;
187    
188 dpavlin 115 #$log->debug( "ds = " . Dumper( \@html_results ) );
189    
190 dpavlin 155 $t = time();
191    
192 dpavlin 100 my $html = $self->{out}->apply(
193 dpavlin 95 template => $template_filename,
194     data => $ds,
195 dpavlin 100 );
196    
197 dpavlin 155 $times->{out} += time() - $t;
198    
199     $t = time();
200    
201 dpavlin 100 $html = $self->{iconv}->convert( $html ) || $log->error("can't convert: $html");
202    
203 dpavlin 155 $times->{iconv} += time() - $t;
204    
205 dpavlin 100 push @html_results, $html;
206    
207 dpavlin 95 }
208    
209 dpavlin 115 #$log->debug( '@html_results = ' . Dumper( \@html_results ) );
210    
211 dpavlin 155 $log->debug( sprintf(
212     "time spent: db = %.2f, out = %.2f, iconv = %.2f",
213     $times->{db}, $times->{out}, $times->{iconv},
214     ) );
215    
216 dpavlin 100 return \@html_results;
217 dpavlin 93 }
218    
219 dpavlin 165 =head2 record
220    
221     my $html = $m->record(
222     mfn => 42,
223     template => 'foo.tt',
224     );
225    
226     This will load one record, convert it to html using C<template> and return
227     it.
228    
229     =cut
230    
231     sub record {
232     my $self = shift;
233    
234     my $args = {@_};
235     my $log = $self->{log};
236     $log->debug("args: " . Dumper( $args ));
237    
238     foreach my $f (qw/mfn template/) {
239     $log->die("need $f") unless ($args->{$f});
240     }
241    
242     my $mfn = $args->{mfn};
243    
244     my $ds = $self->{db}->load_ds( $mfn ) || $log->error("can't load_ds( $mfn )") && next;
245    
246     my $html = $self->{out}->apply(
247     template => $args->{template},
248     data => $ds,
249     );
250    
251     $html = $self->{iconv}->convert( $html ) || $log->error("can't convert: $html");
252    
253     return $html;
254     }
255    
256 dpavlin 135 =head2 save_html
257 dpavlin 93
258 dpavlin 135 $m->save_html( '/full/path/to/file', $content );
259 dpavlin 93
260 dpavlin 179 It will use C<Text::Iconv> to convert content encoding back to
261 dpavlin 135 Webpac codepage, recode JavaScript Unicode entities (%u1234),
262     strip extra newlines at beginning and end, and save to
263     C</full/path/to/file.new> and if that succeeds, just rename
264     it over original file which should be atomic on filesystem level.
265    
266     =cut
267    
268     sub save_html {
269     my ($self, $path, $content) = @_;
270    
271     sub _conv_js {
272     my $t = shift || return;
273     return $self->{iconv}->convert(chr(hex($t)));
274     }
275     $content =~ s/%u([a-fA-F0-9]{4})/_conv_js($1)/gex;
276     $content =~ s/^[\n\r]+//s;
277     $content =~ s/[\n\r]+$/\n/s;
278    
279 dpavlin 179 my ($from, $to) = (
280     $self->{out_encoding},
281     $self->{webpac_encoding},
282 dpavlin 178 );
283 dpavlin 179
284     $self->{log}->debug("using iconv to convert from $from to $to encoding");
285    
286     my $iconv_on_save = new Text::Iconv($from, $to)
287     || $self->{log}->fatal("can't create iconv for saving");
288    
289 dpavlin 178 $content = $iconv_on_save->convert( $content ) || die "no content?";
290    
291 dpavlin 179 write_file($path . '.new', {binmode => ':raw' }, $content) || die "can't save ${path}.new $!";
292 dpavlin 135 rename $path . '.new', $path || die "can't rename to $path: $!";
293     }
294    
295     =head2 load_html
296    
297     my $html = $m->load_html('/full/path/to/file');
298    
299     This will convert file from Webpac encoding to Catalyst and
300     convert that data to escaped HTML (for sending into
301     C<< <textarea/> >> tags in html.
302    
303     =cut
304    
305     sub load_html {
306     my ($self, $path) = @_;
307    
308     die "no path?" unless ($path);
309    
310 dpavlin 179 my $content = read_file($path, {binmode => ':raw' }) || die "can't read $path: $!";
311 dpavlin 135 #$content = $q->escapeHTML($iconv_utf8->convert($content));
312     $content = $self->{iconv}->convert($content);
313    
314     return $content;
315     }
316    
317 dpavlin 92 =head1 AUTHOR
318    
319     Dobrica Pavlinusic
320    
321     =head1 LICENSE
322    
323     This library is free software, you can redistribute it and/or modify
324     it under the same terms as Perl itself.
325    
326     =cut
327    
328     1;

  ViewVC Help
Powered by ViewVC 1.1.26