/[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 348 - (hide annotations)
Sat Jan 7 17:34:16 2006 UTC (18 years, 4 months ago) by dpavlin
File size: 8334 byte(s)
 r360@llin:  dpavlin | 2006-01-07 17:38:48 +0100
 use Encode instead of Text::Iconv

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 322 use WebPAC::Output::TT 0.06;
12 dpavlin 222 use WebPAC::Search::Estraier 0.05;
13 dpavlin 135 use File::Slurp;
14 dpavlin 155 use Time::HiRes;
15 dpavlin 348 use Encode qw/encode decode from_to/;
16 dpavlin 92
17     =head1 NAME
18    
19     Webpacus::Model::WebPAC - Catalyst Model
20    
21     =head1 SYNOPSIS
22    
23     See L<Webpacus> and L<WebPAC>.
24    
25     =head1 DESCRIPTION
26    
27     Catalyst Model for access to WebPAC data.
28    
29     =head2 new
30    
31     Configuration for hyperestraier in C<config.yaml> like this:
32    
33     --- #YAML:1.0
34     # DO NOT USE TABS FOR INDENTATION OR label/value SEPARATION!!!
35    
36     # configuration for hyper estraier full text search engine
37     hyperestraier:
38 dpavlin 222 masterurl: 'http://localhost:1978/node/webpac2'
39     defaultnode: 'webpac2'
40     defaultdepth: 1
41 dpavlin 96 user: 'admin'
42     passwd: 'admin'
43 dpavlin 143 hits_on_page: 100
44 dpavlin 305 hits_for_pager: 1000
45 dpavlin 92
46 dpavlin 96 webpac:
47     db_path: '/data/webpac2/db'
48     template_path: '/data/webpac2/conf/output/tt'
49     template: 'html_ffzg_results_short.tt'
50     # encoding comming from webpac
51     webpac_encoding: 'iso-8859-2'
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 348 foreach my $f (qw/db_path template_path hits_on_page webpac_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 100 $log->debug("converting encoding from webpac_encoding '" .
105     $c->config->{webpac}->{webpac_encoding} .
106 dpavlin 99 "'"
107     );
108 dpavlin 96
109 dpavlin 305 $self->{databases} = $c->config->{databases} || $log->error("can't find databases in config");
110 dpavlin 155
111 dpavlin 92 return $self;
112    
113     }
114    
115 dpavlin 135
116     =head2 search
117    
118 dpavlin 150 my $m->search(
119     phrase => 'query phrase',
120 dpavlin 155 add_attr => \@add_attr
121     get_attr => [ '@uri' ],
122     max => 42,
123 dpavlin 150 template => 'result_template.tt',
124 dpavlin 222 depth => 1,
125 dpavlin 150 );
126 dpavlin 135
127 dpavlin 155 All fields are standard C<WebPAC::Search::Estraier> parametars except
128     C<template> which will (if specified) return results in HTML using
129     selected template.
130    
131 dpavlin 135 =cut
132    
133 dpavlin 93 sub search {
134 dpavlin 150 my $self = shift;
135 dpavlin 93
136 dpavlin 150 my $args = {@_};
137    
138 dpavlin 95 my $log = $self->{log};
139 dpavlin 94
140 dpavlin 150 $log->debug("args: " . Dumper( $args ));
141 dpavlin 95
142 dpavlin 150 my $query = $args->{phrase} || $log->warn("no query phrase") && return;
143 dpavlin 95
144 dpavlin 155 $log->debug("search model query: '$query'");
145     if ($args->{add_attr}) {
146     $log->debug(" + add_attr: " .
147     join("','", @{ $args->{add_attr} })
148     );
149     }
150 dpavlin 150
151     my $template_filename = $args->{template} || $self->{template};
152    
153 dpavlin 305 $args->{max} ||= $self->{'hits_for_pager'};
154 dpavlin 155 if (! $args->{max}) {
155 dpavlin 305 $args->{max} = 100;
156     $log->warn("max not set when calling model. Using default of $args->{max}");
157 dpavlin 155 }
158 dpavlin 93
159 dpavlin 155 my $times; # store some times for benchmarking
160    
161     my $t = time();
162    
163 dpavlin 222 # transfer depth of search
164     if (! $args->{depth}) {
165     my $default = $self->{defaultdepth} || $log->logdie("can't find defaultdepth in estraier configuration");
166     $args->{depth} = $default;
167     $log->warn("using default search depth $default");
168     }
169    
170 dpavlin 155 my @results = $self->{est}->search( %{ $args } );
171    
172     $times->{est} += time() - $t;
173    
174 dpavlin 150 my $hits = $#results + 1;
175 dpavlin 99
176 dpavlin 155 $log->debug( sprintf("search took %.2fs and returned $hits hits.", $times->{est}) );
177 dpavlin 150
178 dpavlin 155 # just return results?
179     return @results unless ($args->{'template'});
180    
181     #
182     # construct HTML results
183     #
184    
185 dpavlin 100 my @html_results;
186    
187 dpavlin 95 for my $i ( 0 .. $#results ) {
188    
189 dpavlin 237 my ($database, $prefix, $id);
190 dpavlin 293 if ( $results[$i]->{'@uri'} =~ m!/([^/]+)/([^/]+)/(\d+)$!) {
191 dpavlin 237 ($database, $prefix,$id) = ($1,$2,$3);
192 dpavlin 224 } else {
193 dpavlin 293 $log->warn("can't decode database/prefix/id from " . $results[$i]->{'@uri'});
194 dpavlin 224 next;
195     }
196 dpavlin 95
197 dpavlin 224 #$log->debug("load_ds( id => $id, prefix => '$prefix' )");
198 dpavlin 95
199 dpavlin 155 $t = time();
200    
201 dpavlin 242 my $ds = $self->{db}->load_ds( database => $database, prefix => $prefix, id => $id );
202     if (! $ds) {
203 dpavlin 293 $log->error("can't load_ds( ${database}/${prefix}/${id} )");
204 dpavlin 242 next;
205     }
206 dpavlin 155
207     $times->{db} += time() - $t;
208    
209 dpavlin 115 #$log->debug( "ds = " . Dumper( \@html_results ) );
210    
211 dpavlin 155 $t = time();
212    
213 dpavlin 100 my $html = $self->{out}->apply(
214 dpavlin 95 template => $template_filename,
215     data => $ds,
216 dpavlin 242 record_uri => "${database}/${prefix}/${id}",
217 dpavlin 305 config => $self->{databases}->{$database},
218 dpavlin 100 );
219    
220 dpavlin 155 $times->{out} += time() - $t;
221    
222     $t = time();
223    
224 dpavlin 348 $html = decode($self->{webpac_encoding}, $html);
225 dpavlin 100
226     push @html_results, $html;
227    
228 dpavlin 95 }
229    
230 dpavlin 115 #$log->debug( '@html_results = ' . Dumper( \@html_results ) );
231    
232 dpavlin 155 $log->debug( sprintf(
233 dpavlin 348 "time spent: db = %.2f, out = %.2f",
234     $times->{db}, $times->{out},
235 dpavlin 155 ) );
236    
237 dpavlin 100 return \@html_results;
238 dpavlin 93 }
239    
240 dpavlin 165 =head2 record
241    
242     my $html = $m->record(
243     mfn => 42,
244     template => 'foo.tt',
245     );
246    
247     This will load one record, convert it to html using C<template> and return
248     it.
249    
250     =cut
251    
252     sub record {
253     my $self = shift;
254    
255     my $args = {@_};
256     my $log = $self->{log};
257     $log->debug("args: " . Dumper( $args ));
258    
259 dpavlin 242 foreach my $f (qw/record_uri template/) {
260 dpavlin 228 $log->fatal("need $f") unless ($args->{$f});
261 dpavlin 165 }
262    
263 dpavlin 242 my ($database, $prefix, $id);
264 dpavlin 165
265 dpavlin 242 if ($args->{record_uri} =~ m#^([^/]+)/([^/]+)/([^/]+)$#) {
266     ($database, $prefix, $id) = ($1,$2,$3);
267     } else {
268     $log->error("can't parse $args->{record_uri} into prefix, database and uri");
269     return;
270     }
271 dpavlin 165
272 dpavlin 242 my $ds = $self->{db}->load_ds( id => $id, prefix => $prefix, database => $database );
273     if (! $ds) {
274     $log->error("can't load_ds( $database/$prefix/$id )");
275     return;
276     }
277    
278 dpavlin 165 my $html = $self->{out}->apply(
279     template => $args->{template},
280     data => $ds,
281 dpavlin 242 record_uri => $args->{record_uri},
282 dpavlin 305 config => $self->{databases}->{$database},
283 dpavlin 165 );
284    
285 dpavlin 348 $html = decode($self->{webpac_encoding}, $html);
286 dpavlin 165
287     return $html;
288     }
289    
290 dpavlin 305
291 dpavlin 135 =head2 save_html
292 dpavlin 93
293 dpavlin 135 $m->save_html( '/full/path/to/file', $content );
294 dpavlin 93
295 dpavlin 348 It will use C<Encode> to convert content encoding back to
296 dpavlin 135 Webpac codepage, recode JavaScript Unicode entities (%u1234),
297     strip extra newlines at beginning and end, and save to
298     C</full/path/to/file.new> and if that succeeds, just rename
299     it over original file which should be atomic on filesystem level.
300    
301     =cut
302    
303     sub save_html {
304     my ($self, $path, $content) = @_;
305    
306 dpavlin 348 # FIXME Should this be UTF-8 or someting?
307     my $js_encoding = $self->{webpac_encoding};
308     $js_encoding = 'UTF-16';
309    
310 dpavlin 135 sub _conv_js {
311 dpavlin 348 return '0x' . $_[1];
312     return encode($_[0], chr(hex($_[1])));
313 dpavlin 135 }
314 dpavlin 348 #$content =~ s/%u([a-fA-F0-9]{4})/_conv_js($js_encoding,$1)/gex;
315 dpavlin 135 $content =~ s/^[\n\r]+//s;
316     $content =~ s/[\n\r]+$/\n/s;
317 dpavlin 282 $content =~ s/\n\r/\n/gs;
318 dpavlin 135
319 dpavlin 348 my $disk_encoding = $self->{webpac_encoding} || 'utf-8';
320     $self->{log}->debug("convert encoding to $disk_encoding");
321     from_to($content, 'utf-8', $disk_encoding) || $self->{log}->warn("encoding from utf-8 to $disk_encoding failed for: $content");
322 dpavlin 179
323     write_file($path . '.new', {binmode => ':raw' }, $content) || die "can't save ${path}.new $!";
324 dpavlin 135 rename $path . '.new', $path || die "can't rename to $path: $!";
325     }
326    
327     =head2 load_html
328    
329     my $html = $m->load_html('/full/path/to/file');
330    
331     This will convert file from Webpac encoding to Catalyst and
332     convert that data to escaped HTML (for sending into
333     C<< <textarea/> >> tags in html.
334    
335     =cut
336    
337     sub load_html {
338     my ($self, $path) = @_;
339    
340     die "no path?" unless ($path);
341    
342 dpavlin 179 my $content = read_file($path, {binmode => ':raw' }) || die "can't read $path: $!";
343 dpavlin 135
344 dpavlin 348 return decode($self->{webpac_encoding}, $content);
345 dpavlin 135 }
346    
347 dpavlin 92 =head1 AUTHOR
348    
349 dpavlin 348 Dobrica Pavlinusic C<< <dpavlin@rot13.org> >>
350 dpavlin 92
351     =head1 LICENSE
352    
353     This library is free software, you can redistribute it and/or modify
354     it under the same terms as Perl itself.
355    
356     =cut
357    
358     1;

  ViewVC Help
Powered by ViewVC 1.1.26