/[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 281 - (hide annotations)
Sat Dec 17 22:35:00 2005 UTC (18 years, 5 months ago) by dpavlin
File size: 8434 byte(s)
 r11769@llin:  dpavlin | 2005-12-18 04:35:31 +0100
 use WebPAC::Output::TT 0.05 (needed for 0.22 to work)

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 92
44 dpavlin 96 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 dpavlin 92 =cut
54    
55     sub new {
56     my ( $self, $c, $config ) = @_;
57    
58     $self = $self->NEXT::new($c, $config);
59     $self->config($config);
60    
61     my $log = $c->log;
62 dpavlin 94 $self->{log} = $log;
63 dpavlin 92
64 dpavlin 93 my $est_cfg = $c->config->{hyperestraier};
65     $est_cfg->{'log'} = $log;
66 dpavlin 92
67 dpavlin 271 $est_cfg->{encoding} = $est_cfg->{catalyst_encoding} || $c->config->{catalyst_encoding} or $c->log->fatal("can't find catalyst_encoding");
68 dpavlin 142
69 dpavlin 93 $log->debug("using config:" . Dumper($est_cfg) );
70 dpavlin 92
71 dpavlin 222 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 dpavlin 93 $self->{est} = new WebPAC::Search::Estraier( %{ $est_cfg } );
78 dpavlin 92
79 dpavlin 167 # save config parametars in object
80 dpavlin 222 foreach my $f (qw/db_path template_path hits_on_page webpac_encoding out_encoding defaultdepth/) {
81 dpavlin 167 $self->{$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 dpavlin 95
88     $log->debug("using db path '$db_path', template path '$template_path'");
89    
90 dpavlin 222 $self->{db} = new WebPAC::Store(
91 dpavlin 95 path => $db_path,
92     read_only => 1,
93 dpavlin 224 database => $est_cfg->{database},
94 dpavlin 95 );
95    
96     $self->{out} = new WebPAC::Output::TT(
97     include_path => $template_path,
98     filters => { foo => sub { shift } },
99     );
100    
101 dpavlin 101 # default template from config.yaml
102 dpavlin 95 $self->{template} ||= $c->config->{webpac}->{template};
103    
104 dpavlin 96 $self->{iconv} = new Text::Iconv(
105     $c->config->{webpac}->{webpac_encoding},
106     $c->config->{webpac}->{out_encoding}
107     );
108    
109 dpavlin 100 $log->debug("converting encoding from webpac_encoding '" .
110     $c->config->{webpac}->{webpac_encoding} .
111     "' to '" .
112     $c->config->{webpac}->{out_encoding} .
113 dpavlin 99 "'"
114     );
115 dpavlin 96
116 dpavlin 155
117 dpavlin 92 return $self;
118    
119     }
120    
121 dpavlin 135
122     =head2 search
123    
124 dpavlin 150 my $m->search(
125     phrase => 'query phrase',
126 dpavlin 155 add_attr => \@add_attr
127     get_attr => [ '@uri' ],
128     max => 42,
129 dpavlin 150 template => 'result_template.tt',
130 dpavlin 222 depth => 1,
131 dpavlin 150 );
132 dpavlin 135
133 dpavlin 155 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 dpavlin 135 =cut
138    
139 dpavlin 93 sub search {
140 dpavlin 150 my $self = shift;
141 dpavlin 93
142 dpavlin 150 my $args = {@_};
143    
144 dpavlin 95 my $log = $self->{log};
145 dpavlin 94
146 dpavlin 150 $log->debug("args: " . Dumper( $args ));
147 dpavlin 95
148 dpavlin 150 my $query = $args->{phrase} || $log->warn("no query phrase") && return;
149 dpavlin 95
150 dpavlin 155 $log->debug("search model query: '$query'");
151     if ($args->{add_attr}) {
152     $log->debug(" + add_attr: " .
153     join("','", @{ $args->{add_attr} })
154     );
155     }
156 dpavlin 150
157     my $template_filename = $args->{template} || $self->{template};
158    
159 dpavlin 155 $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 dpavlin 93
165 dpavlin 155 my $times; # store some times for benchmarking
166    
167     my $t = time();
168    
169 dpavlin 222 # 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 dpavlin 155 my @results = $self->{est}->search( %{ $args } );
177    
178     $times->{est} += time() - $t;
179    
180 dpavlin 150 my $hits = $#results + 1;
181 dpavlin 99
182 dpavlin 155 $log->debug( sprintf("search took %.2fs and returned $hits hits.", $times->{est}) );
183 dpavlin 150
184 dpavlin 155 # just return results?
185     return @results unless ($args->{'template'});
186    
187     #
188     # construct HTML results
189     #
190    
191 dpavlin 100 my @html_results;
192    
193 dpavlin 95 for my $i ( 0 .. $#results ) {
194    
195 dpavlin 237 my ($database, $prefix, $id);
196     if ( $results[$i]->{'@uri'} =~ m!/([^/]+)/([^/]+)#(\d+)$!) {
197     ($database, $prefix,$id) = ($1,$2,$3);
198 dpavlin 224 } else {
199     $log->warn("can't decode prefix#id from " . $results[$i]->{'@uri'});
200     next;
201     }
202 dpavlin 95
203 dpavlin 224 #$log->debug("load_ds( id => $id, prefix => '$prefix' )");
204 dpavlin 95
205 dpavlin 155 $t = time();
206    
207 dpavlin 242 my $ds = $self->{db}->load_ds( database => $database, prefix => $prefix, id => $id );
208     if (! $ds) {
209     $log->error("can't load_ds( ${database}/${prefix}#${id} )");
210     next;
211     }
212 dpavlin 155
213     $times->{db} += time() - $t;
214    
215 dpavlin 115 #$log->debug( "ds = " . Dumper( \@html_results ) );
216    
217 dpavlin 155 $t = time();
218    
219 dpavlin 100 my $html = $self->{out}->apply(
220 dpavlin 95 template => $template_filename,
221     data => $ds,
222 dpavlin 242 record_uri => "${database}/${prefix}/${id}",
223 dpavlin 100 );
224    
225 dpavlin 155 $times->{out} += time() - $t;
226    
227     $t = time();
228    
229 dpavlin 100 $html = $self->{iconv}->convert( $html ) || $log->error("can't convert: $html");
230    
231 dpavlin 155 $times->{iconv} += time() - $t;
232    
233 dpavlin 100 push @html_results, $html;
234    
235 dpavlin 95 }
236    
237 dpavlin 115 #$log->debug( '@html_results = ' . Dumper( \@html_results ) );
238    
239 dpavlin 155 $log->debug( sprintf(
240     "time spent: db = %.2f, out = %.2f, iconv = %.2f",
241     $times->{db}, $times->{out}, $times->{iconv},
242     ) );
243    
244 dpavlin 100 return \@html_results;
245 dpavlin 93 }
246    
247 dpavlin 165 =head2 record
248    
249     my $html = $m->record(
250     mfn => 42,
251     template => 'foo.tt',
252     );
253    
254     This will load one record, convert it to html using C<template> and return
255     it.
256    
257     =cut
258    
259     sub record {
260     my $self = shift;
261    
262     my $args = {@_};
263     my $log = $self->{log};
264     $log->debug("args: " . Dumper( $args ));
265    
266 dpavlin 242 foreach my $f (qw/record_uri template/) {
267 dpavlin 228 $log->fatal("need $f") unless ($args->{$f});
268 dpavlin 165 }
269    
270 dpavlin 242 my ($database, $prefix, $id);
271 dpavlin 165
272 dpavlin 242 if ($args->{record_uri} =~ m#^([^/]+)/([^/]+)/([^/]+)$#) {
273     ($database, $prefix, $id) = ($1,$2,$3);
274     } else {
275     $log->error("can't parse $args->{record_uri} into prefix, database and uri");
276     return;
277     }
278 dpavlin 165
279 dpavlin 242 my $ds = $self->{db}->load_ds( id => $id, prefix => $prefix, database => $database );
280     if (! $ds) {
281     $log->error("can't load_ds( $database/$prefix/$id )");
282     return;
283     }
284    
285 dpavlin 165 my $html = $self->{out}->apply(
286     template => $args->{template},
287     data => $ds,
288 dpavlin 242 record_uri => $args->{record_uri},
289 dpavlin 165 );
290    
291     $html = $self->{iconv}->convert( $html ) || $log->error("can't convert: $html");
292    
293     return $html;
294     }
295    
296 dpavlin 135 =head2 save_html
297 dpavlin 93
298 dpavlin 135 $m->save_html( '/full/path/to/file', $content );
299 dpavlin 93
300 dpavlin 179 It will use C<Text::Iconv> to convert content encoding back to
301 dpavlin 135 Webpac codepage, recode JavaScript Unicode entities (%u1234),
302     strip extra newlines at beginning and end, and save to
303     C</full/path/to/file.new> and if that succeeds, just rename
304     it over original file which should be atomic on filesystem level.
305    
306     =cut
307    
308     sub save_html {
309     my ($self, $path, $content) = @_;
310    
311     sub _conv_js {
312     my $t = shift || return;
313     return $self->{iconv}->convert(chr(hex($t)));
314     }
315     $content =~ s/%u([a-fA-F0-9]{4})/_conv_js($1)/gex;
316     $content =~ s/^[\n\r]+//s;
317     $content =~ s/[\n\r]+$/\n/s;
318    
319 dpavlin 179 my ($from, $to) = (
320     $self->{out_encoding},
321     $self->{webpac_encoding},
322 dpavlin 178 );
323 dpavlin 179
324     $self->{log}->debug("using iconv to convert from $from to $to encoding");
325    
326     my $iconv_on_save = new Text::Iconv($from, $to)
327     || $self->{log}->fatal("can't create iconv for saving");
328    
329 dpavlin 178 $content = $iconv_on_save->convert( $content ) || die "no content?";
330    
331 dpavlin 179 write_file($path . '.new', {binmode => ':raw' }, $content) || die "can't save ${path}.new $!";
332 dpavlin 135 rename $path . '.new', $path || die "can't rename to $path: $!";
333     }
334    
335     =head2 load_html
336    
337     my $html = $m->load_html('/full/path/to/file');
338    
339     This will convert file from Webpac encoding to Catalyst and
340     convert that data to escaped HTML (for sending into
341     C<< <textarea/> >> tags in html.
342    
343     =cut
344    
345     sub load_html {
346     my ($self, $path) = @_;
347    
348     die "no path?" unless ($path);
349    
350 dpavlin 179 my $content = read_file($path, {binmode => ':raw' }) || die "can't read $path: $!";
351 dpavlin 135 #$content = $q->escapeHTML($iconv_utf8->convert($content));
352     $content = $self->{iconv}->convert($content);
353    
354     return $content;
355     }
356    
357 dpavlin 92 =head1 AUTHOR
358    
359     Dobrica Pavlinusic
360    
361     =head1 LICENSE
362    
363     This library is free software, you can redistribute it and/or modify
364     it under the same terms as Perl itself.
365    
366     =cut
367    
368     1;

  ViewVC Help
Powered by ViewVC 1.1.26