/[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 305 - (hide annotations)
Tue Dec 20 00:00:57 2005 UTC (18 years, 5 months ago) by dpavlin
File size: 8694 byte(s)
 r330@athlon:  dpavlin | 2005-12-20 01:01:58 +0100
 make database config available to each record as [% config.something %], for
 example [% config.name %] is actually name field for that record

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 237 use WebPAC::Store 0.08;
11 dpavlin 281 use WebPAC::Output::TT 0.05;
12 dpavlin 222 use WebPAC::Search::Estraier 0.05;
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 222 masterurl: 'http://localhost:1978/node/webpac2'
38     defaultnode: 'webpac2'
39     defaultdepth: 1
40 dpavlin 96 user: 'admin'
41     passwd: 'admin'
42 dpavlin 143 hits_on_page: 100
43 dpavlin 305 hits_for_pager: 1000
44 dpavlin 92
45 dpavlin 96 webpac:
46     db_path: '/data/webpac2/db'
47     template_path: '/data/webpac2/conf/output/tt'
48     template: 'html_ffzg_results_short.tt'
49     # encoding comming from webpac
50     webpac_encoding: 'iso-8859-2'
51     # encoding expected by Catalyst
52     out_encoding: 'UTF-8'
53    
54 dpavlin 92 =cut
55    
56     sub new {
57     my ( $self, $c, $config ) = @_;
58    
59     $self = $self->NEXT::new($c, $config);
60     $self->config($config);
61    
62     my $log = $c->log;
63 dpavlin 94 $self->{log} = $log;
64 dpavlin 92
65 dpavlin 93 my $est_cfg = $c->config->{hyperestraier};
66     $est_cfg->{'log'} = $log;
67 dpavlin 92
68 dpavlin 271 $est_cfg->{encoding} = $est_cfg->{catalyst_encoding} || $c->config->{catalyst_encoding} or $c->log->fatal("can't find catalyst_encoding");
69 dpavlin 142
70 dpavlin 93 $log->debug("using config:" . Dumper($est_cfg) );
71 dpavlin 92
72 dpavlin 222 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 dpavlin 93 $self->{est} = new WebPAC::Search::Estraier( %{ $est_cfg } );
79 dpavlin 92
80 dpavlin 167 # save config parametars in object
81 dpavlin 222 foreach my $f (qw/db_path template_path hits_on_page webpac_encoding out_encoding defaultdepth/) {
82 dpavlin 167 $self->{$f} = $c->config->{hyperestraier}->{$f} ||
83     $c->config->{webpac}->{$f};
84     $log->debug("self->{$f} = " . $self->{$f});
85     }
86     my $db_path = $self->{db_path};
87     my $template_path = $self->{template_path};
88 dpavlin 95
89     $log->debug("using db path '$db_path', template path '$template_path'");
90    
91 dpavlin 222 $self->{db} = new WebPAC::Store(
92 dpavlin 95 path => $db_path,
93     read_only => 1,
94 dpavlin 224 database => $est_cfg->{database},
95 dpavlin 95 );
96    
97     $self->{out} = new WebPAC::Output::TT(
98     include_path => $template_path,
99     filters => { foo => sub { shift } },
100     );
101    
102 dpavlin 101 # default template from config.yaml
103 dpavlin 95 $self->{template} ||= $c->config->{webpac}->{template};
104    
105 dpavlin 96 $self->{iconv} = new Text::Iconv(
106     $c->config->{webpac}->{webpac_encoding},
107     $c->config->{webpac}->{out_encoding}
108     );
109    
110 dpavlin 100 $log->debug("converting encoding from webpac_encoding '" .
111     $c->config->{webpac}->{webpac_encoding} .
112     "' to '" .
113     $c->config->{webpac}->{out_encoding} .
114 dpavlin 99 "'"
115     );
116 dpavlin 96
117 dpavlin 305 $self->{databases} = $c->config->{databases} || $log->error("can't find databases in config");
118 dpavlin 155
119 dpavlin 92 return $self;
120    
121     }
122    
123 dpavlin 135
124     =head2 search
125    
126 dpavlin 150 my $m->search(
127     phrase => 'query phrase',
128 dpavlin 155 add_attr => \@add_attr
129     get_attr => [ '@uri' ],
130     max => 42,
131 dpavlin 150 template => 'result_template.tt',
132 dpavlin 222 depth => 1,
133 dpavlin 150 );
134 dpavlin 135
135 dpavlin 155 All fields are standard C<WebPAC::Search::Estraier> parametars except
136     C<template> which will (if specified) return results in HTML using
137     selected template.
138    
139 dpavlin 135 =cut
140    
141 dpavlin 93 sub search {
142 dpavlin 150 my $self = shift;
143 dpavlin 93
144 dpavlin 150 my $args = {@_};
145    
146 dpavlin 95 my $log = $self->{log};
147 dpavlin 94
148 dpavlin 150 $log->debug("args: " . Dumper( $args ));
149 dpavlin 95
150 dpavlin 150 my $query = $args->{phrase} || $log->warn("no query phrase") && return;
151 dpavlin 95
152 dpavlin 155 $log->debug("search model query: '$query'");
153     if ($args->{add_attr}) {
154     $log->debug(" + add_attr: " .
155     join("','", @{ $args->{add_attr} })
156     );
157     }
158 dpavlin 150
159     my $template_filename = $args->{template} || $self->{template};
160    
161 dpavlin 305 $args->{max} ||= $self->{'hits_for_pager'};
162 dpavlin 155 if (! $args->{max}) {
163 dpavlin 305 $args->{max} = 100;
164     $log->warn("max not set when calling model. Using default of $args->{max}");
165 dpavlin 155 }
166 dpavlin 93
167 dpavlin 155 my $times; # store some times for benchmarking
168    
169     my $t = time();
170    
171 dpavlin 222 # 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 dpavlin 155 my @results = $self->{est}->search( %{ $args } );
179    
180     $times->{est} += time() - $t;
181    
182 dpavlin 150 my $hits = $#results + 1;
183 dpavlin 99
184 dpavlin 155 $log->debug( sprintf("search took %.2fs and returned $hits hits.", $times->{est}) );
185 dpavlin 150
186 dpavlin 155 # just return results?
187     return @results unless ($args->{'template'});
188    
189     #
190     # construct HTML results
191     #
192    
193 dpavlin 100 my @html_results;
194    
195 dpavlin 95 for my $i ( 0 .. $#results ) {
196    
197 dpavlin 237 my ($database, $prefix, $id);
198 dpavlin 293 if ( $results[$i]->{'@uri'} =~ m!/([^/]+)/([^/]+)/(\d+)$!) {
199 dpavlin 237 ($database, $prefix,$id) = ($1,$2,$3);
200 dpavlin 224 } else {
201 dpavlin 293 $log->warn("can't decode database/prefix/id from " . $results[$i]->{'@uri'});
202 dpavlin 224 next;
203     }
204 dpavlin 95
205 dpavlin 224 #$log->debug("load_ds( id => $id, prefix => '$prefix' )");
206 dpavlin 95
207 dpavlin 155 $t = time();
208    
209 dpavlin 242 my $ds = $self->{db}->load_ds( database => $database, prefix => $prefix, id => $id );
210     if (! $ds) {
211 dpavlin 293 $log->error("can't load_ds( ${database}/${prefix}/${id} )");
212 dpavlin 242 next;
213     }
214 dpavlin 155
215     $times->{db} += time() - $t;
216    
217 dpavlin 115 #$log->debug( "ds = " . Dumper( \@html_results ) );
218    
219 dpavlin 155 $t = time();
220    
221 dpavlin 100 my $html = $self->{out}->apply(
222 dpavlin 95 template => $template_filename,
223     data => $ds,
224 dpavlin 242 record_uri => "${database}/${prefix}/${id}",
225 dpavlin 305 config => $self->{databases}->{$database},
226 dpavlin 100 );
227    
228 dpavlin 155 $times->{out} += time() - $t;
229    
230     $t = time();
231    
232 dpavlin 100 $html = $self->{iconv}->convert( $html ) || $log->error("can't convert: $html");
233    
234 dpavlin 155 $times->{iconv} += time() - $t;
235    
236 dpavlin 100 push @html_results, $html;
237    
238 dpavlin 95 }
239    
240 dpavlin 115 #$log->debug( '@html_results = ' . Dumper( \@html_results ) );
241    
242 dpavlin 155 $log->debug( sprintf(
243     "time spent: db = %.2f, out = %.2f, iconv = %.2f",
244     $times->{db}, $times->{out}, $times->{iconv},
245     ) );
246    
247 dpavlin 100 return \@html_results;
248 dpavlin 93 }
249    
250 dpavlin 165 =head2 record
251    
252     my $html = $m->record(
253     mfn => 42,
254     template => 'foo.tt',
255     );
256    
257     This will load one record, convert it to html using C<template> and return
258     it.
259    
260     =cut
261    
262     sub record {
263     my $self = shift;
264    
265     my $args = {@_};
266     my $log = $self->{log};
267     $log->debug("args: " . Dumper( $args ));
268    
269 dpavlin 242 foreach my $f (qw/record_uri template/) {
270 dpavlin 228 $log->fatal("need $f") unless ($args->{$f});
271 dpavlin 165 }
272    
273 dpavlin 242 my ($database, $prefix, $id);
274 dpavlin 165
275 dpavlin 242 if ($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 dpavlin 165
282 dpavlin 242 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 dpavlin 165 my $html = $self->{out}->apply(
289     template => $args->{template},
290     data => $ds,
291 dpavlin 242 record_uri => $args->{record_uri},
292 dpavlin 305 config => $self->{databases}->{$database},
293 dpavlin 165 );
294    
295     $html = $self->{iconv}->convert( $html ) || $log->error("can't convert: $html");
296    
297     return $html;
298     }
299    
300 dpavlin 305
301 dpavlin 135 =head2 save_html
302 dpavlin 93
303 dpavlin 135 $m->save_html( '/full/path/to/file', $content );
304 dpavlin 93
305 dpavlin 179 It will use C<Text::Iconv> to convert content encoding back to
306 dpavlin 135 Webpac codepage, recode JavaScript Unicode entities (%u1234),
307     strip extra newlines at beginning and end, and save to
308     C</full/path/to/file.new> and if that succeeds, just rename
309     it over original file which should be atomic on filesystem level.
310    
311     =cut
312    
313     sub save_html {
314     my ($self, $path, $content) = @_;
315    
316     sub _conv_js {
317     my $t = shift || return;
318     return $self->{iconv}->convert(chr(hex($t)));
319     }
320     $content =~ s/%u([a-fA-F0-9]{4})/_conv_js($1)/gex;
321     $content =~ s/^[\n\r]+//s;
322     $content =~ s/[\n\r]+$/\n/s;
323 dpavlin 282 $content =~ s/\n\r/\n/gs;
324 dpavlin 135
325 dpavlin 179 my ($from, $to) = (
326     $self->{out_encoding},
327     $self->{webpac_encoding},
328 dpavlin 178 );
329 dpavlin 179
330     $self->{log}->debug("using iconv to convert from $from to $to encoding");
331    
332     my $iconv_on_save = new Text::Iconv($from, $to)
333     || $self->{log}->fatal("can't create iconv for saving");
334    
335 dpavlin 178 $content = $iconv_on_save->convert( $content ) || die "no content?";
336    
337 dpavlin 179 write_file($path . '.new', {binmode => ':raw' }, $content) || die "can't save ${path}.new $!";
338 dpavlin 135 rename $path . '.new', $path || die "can't rename to $path: $!";
339     }
340    
341     =head2 load_html
342    
343     my $html = $m->load_html('/full/path/to/file');
344    
345     This will convert file from Webpac encoding to Catalyst and
346     convert that data to escaped HTML (for sending into
347     C<< <textarea/> >> tags in html.
348    
349     =cut
350    
351     sub load_html {
352     my ($self, $path) = @_;
353    
354     die "no path?" unless ($path);
355    
356 dpavlin 179 my $content = read_file($path, {binmode => ':raw' }) || die "can't read $path: $!";
357 dpavlin 135 #$content = $q->escapeHTML($iconv_utf8->convert($content));
358     $content = $self->{iconv}->convert($content);
359    
360     return $content;
361     }
362    
363 dpavlin 92 =head1 AUTHOR
364    
365     Dobrica Pavlinusic
366    
367     =head1 LICENSE
368    
369     This library is free software, you can redistribute it and/or modify
370     it under the same terms as Perl itself.
371    
372     =cut
373    
374     1;

  ViewVC Help
Powered by ViewVC 1.1.26