Webpacus/Model/WebPAC.pm

package Webpacus::Model::WebPAC;

use strict;
use warnings;
use lib '/data/webpac2/lib';
use base qw/
        Catalyst::Model
/;
use Data::Dumper;
use WebPAC::DB;
use WebPAC::Output::TT;
use WebPAC::Search::Estraier 0.02;
use File::Slurp;
use Time::HiRes;

=head1 NAME

Webpacus::Model::WebPAC - Catalyst Model

=head1 SYNOPSIS

See L<Webpacus> and L<WebPAC>.

=head1 DESCRIPTION

Catalyst Model for access to WebPAC data.

=head2 new

Configuration for hyperestraier in C<config.yaml> like this:

 --- #YAML:1.0
 # DO NOT USE TABS FOR INDENTATION OR label/value SEPARATION!!!

 # configuration for hyper estraier full text search engine
 hyperestraier:
  url: 'http://localhost:1978/node/webpac2'
  user: 'admin'
  passwd: 'admin'
  hits_on_page: 100

 webpac:
  db_path: '/data/webpac2/db'
  template_path: '/data/webpac2/conf/output/tt'
  template: 'html_ffzg_results_short.tt'
  # encoding comming from webpac
  webpac_encoding: 'iso-8859-2'
  # encoding expected by Catalyst
  out_encoding: 'UTF-8'

=cut

sub new {
        my ( $self, $c, $config ) = @_;

        $self = $self->NEXT::new($c, $config);
        $self->config($config);

        my $log = $c->log;
        $self->{log} = $log;

        my $est_cfg = $c->config->{hyperestraier};
        $est_cfg->{'log'} = $log;

        $est_cfg->{encoding} = $est_cfg->{catalyst_encoding};

        $log->debug("using config:" . Dumper($est_cfg) );

        $self->{est} = new WebPAC::Search::Estraier( %{ $est_cfg } );

        # save config parametars in object
        foreach my $f (qw/db_path template_path hits_on_page/) {
                $self->{$f} = $c->config->{hyperestraier}->{$f} ||
                        $c->config->{webpac}->{$f};
                $log->debug("self->{$f} = " . $self->{$f});
        }
        my $db_path = $self->{db_path};
        my $template_path = $self->{template_path};

        $log->debug("using db path '$db_path', template path '$template_path'");

        $self->{db} = new WebPAC::DB(
                path => $db_path,
                read_only => 1,
        );

        $self->{out} = new WebPAC::Output::TT(
                include_path => $template_path,
                filters => { foo => sub { shift } },
        );

        # default template from config.yaml
        $self->{template} ||= $c->config->{webpac}->{template};

        $self->{iconv} = new Text::Iconv(
                $c->config->{webpac}->{webpac_encoding},
                $c->config->{webpac}->{out_encoding}
        );

        $log->debug("converting encoding from webpac_encoding '" .
                $c->config->{webpac}->{webpac_encoding} .
                "' to '" .
                $c->config->{webpac}->{out_encoding} .
                "'"
        );


        return $self;

}

=head2 iconv_on_save

  my $out = $m->iconv_on_save( $content );

Convert data saved to disk in Webpac encoding.

=cut

sub iconv_on_save {
        my $self = shift;

        $self->{iconv_save} ||= new Text::Iconv(
                $self->config->{webpac}->{out_encoding},
                $self->config->{webpac}->{webpac_encoding},
        );

        $self->{iconv_save}->convert( @_ );
}


=head2 search

  my $m->search(
        phrase => 'query phrase',
        add_attr => \@add_attr
        get_attr => [ '@uri' ],
        max => 42,
        template => 'result_template.tt',
  );

All fields are standard C<WebPAC::Search::Estraier> parametars except
C<template> which will (if specified) return results in HTML using
selected template.

=cut

sub search {
        my $self = shift;

        my $args = {@_};

        my $log = $self->{log};

        $log->debug("args: " . Dumper( $args ));

        my $query = $args->{phrase} || $log->warn("no query phrase") && return;

        $log->debug("search model query: '$query'");
        if ($args->{add_attr}) {
                $log->debug(" + add_attr: " .
                        join("','", @{ $args->{add_attr} })
                );
        }

        my $template_filename = $args->{template} || $self->{template};

        $args->{max} ||= $self->{'hits_on_page'};
        if (! $args->{max}) {
                $args->{max} = 10;
                $log->warn("max not set when calling model. Using default of 10");
        }

        my $times;      # store some times for benchmarking

        my $t = time();

        my @results = $self->{est}->search( %{ $args } );

        $times->{est} += time() - $t;

        my $hits = $#results + 1;

        $log->debug( sprintf("search took %.2fs and returned $hits hits.", $times->{est}) );

        # just return results?
        return @results unless ($args->{'template'});

        #
        # construct HTML results
        #

        my @html_results;

        for my $i ( 0 .. $#results ) {

                my $mfn = $1 if ( $results[$i]->{'@uri'} =~ m#/(\d+)$#);

                #$log->debug("load_ds( $mfn )");

                $t = time();

                my $ds = $self->{db}->load_ds( $mfn ) || $log->error("can't load_ds( $mfn )") && next;

                $times->{db} += time() - $t;

                #$log->debug( "ds = " . Dumper( \@html_results ) );

                $t = time();

                my $html = $self->{out}->apply(
                        template => $template_filename,
                        data => $ds,
                );

                $times->{out} += time() - $t;

                $t = time();

                $html = $self->{iconv}->convert( $html ) || $log->error("can't convert: $html");

                $times->{iconv} += time() - $t;

                push @html_results, $html;

        }

        #$log->debug( '@html_results = ' . Dumper( \@html_results ) );

        $log->debug( sprintf(
                "time spent: db = %.2f, out = %.2f, iconv = %.2f",
                $times->{db}, $times->{out}, $times->{iconv},
        ) );

        return \@html_results;
}

=head2 record

  my $html = $m->record(
        mfn => 42,
        template => 'foo.tt',
  );

This will load one record, convert it to html using C<template> and return
it.

=cut

sub record {
        my $self = shift;

        my $args = {@_};
        my $log = $self->{log};
        $log->debug("args: " . Dumper( $args ));

        foreach my $f (qw/mfn template/) {
                $log->die("need $f") unless ($args->{$f});
        }

        my $mfn = $args->{mfn};

        my $ds = $self->{db}->load_ds( $mfn ) || $log->error("can't load_ds( $mfn )") && next;

        my $html = $self->{out}->apply(
                template => $args->{template},
                data => $ds,
        );

        $html = $self->{iconv}->convert( $html ) || $log->error("can't convert: $html");

        return $html;
}

=head2 save_html

  $m->save_html( '/full/path/to/file', $content );

It will use C<iconv_on_save> to convert content encoding back to
Webpac codepage, recode JavaScript Unicode entities (%u1234),
strip extra newlines at beginning and end, and save to 
C</full/path/to/file.new> and if that succeeds, just rename
it over original file which should be atomic on filesystem level.

=cut

sub save_html {
        my ($self, $path, $content) = @_;

        $content = $self->iconv_on_save( $content ) || die "no content?";

        sub _conv_js {
                my $t = shift || return;
                return $self->{iconv}->convert(chr(hex($t)));
        }
        $content =~ s/%u([a-fA-F0-9]{4})/_conv_js($1)/gex;
        $content =~ s/^[\n\r]+//s;
        $content =~ s/[\n\r]+$/\n/s;

        write_file($path . '.new', $content) || die "can't save ${path}.new $!";
        rename $path . '.new', $path || die "can't rename to $path: $!";
}

=head2 load_html

  my $html = $m->load_html('/full/path/to/file');

This will convert file from Webpac encoding to Catalyst and
convert that data to escaped HTML (for sending into
C<< <textarea/> >> tags in html.

=cut

sub load_html {
        my ($self, $path) = @_;

        die "no path?" unless ($path);

        my $content = read_file($path) || die "can't read $path: $!";
        #$content = $q->escapeHTML($iconv_utf8->convert($content));
        $content = $self->{iconv}->convert($content);

        return $content;
}

=head1 AUTHOR

Dobrica Pavlinusic

=head1 LICENSE

This library is free software, you can redistribute it and/or modify
it under the same terms as Perl itself.

=cut

1;
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			use WebPAC::Output::TT;
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			foreach my $f (qw/db_path template_path hits_on_page/) {
73			$self->{$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	=head2 iconv_on_save
113
114			my $out = $m->iconv_on_save( $content );
115
116			Convert data saved to disk in Webpac encoding.
117
118			=cut
119
120			sub iconv_on_save {
121			my $self = shift;
122
123			$self->{iconv_save} \|\|= new Text::Iconv(
124			$self->config->{webpac}->{out_encoding},
125			$self->config->{webpac}->{webpac_encoding},
126			);
127
128			$self->{iconv_save}->convert( @_ );
129			}
130
131
132			=head2 search
133
134	dpavlin	150	my $m->search(
135			phrase => 'query phrase',
136	dpavlin	155	add_attr => \@add_attr
137			get_attr => [ '@uri' ],
138			max => 42,
139	dpavlin	150	template => 'result_template.tt',
140			);
141	dpavlin	135
142	dpavlin	155	All fields are standard C<WebPAC::Search::Estraier> parametars except
143			C<template> which will (if specified) return results in HTML using
144			selected template.
145
146	dpavlin	135	=cut
147
148	dpavlin	93	sub search {
149	dpavlin	150	my $self = shift;
150	dpavlin	93
151	dpavlin	150	my $args = {@_};
152
153	dpavlin	95	my $log = $self->{log};
154	dpavlin	94
155	dpavlin	150	$log->debug("args: " . Dumper( $args ));
156	dpavlin	95
157	dpavlin	150	my $query = $args->{phrase} \|\| $log->warn("no query phrase") && return;
158	dpavlin	95
159	dpavlin	155	$log->debug("search model query: '$query'");
160			if ($args->{add_attr}) {
161			$log->debug(" + add_attr: " .
162			join("','", @{ $args->{add_attr} })
163			);
164			}
165	dpavlin	150
166			my $template_filename = $args->{template} \|\| $self->{template};
167
168	dpavlin	155	$args->{max} \|\|= $self->{'hits_on_page'};
169			if (! $args->{max}) {
170			$args->{max} = 10;
171			$log->warn("max not set when calling model. Using default of 10");
172			}
173	dpavlin	93
174	dpavlin	155	my $times; # store some times for benchmarking
175
176			my $t = time();
177
178			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			my $mfn = $1 if ( $results[$i]->{'@uri'} =~ m#/(\d+)$#);
198
199	dpavlin	115	#$log->debug("load_ds( $mfn )");
200	dpavlin	95
201	dpavlin	155	$t = time();
202
203	dpavlin	115	my $ds = $self->{db}->load_ds( $mfn ) \|\| $log->error("can't load_ds( $mfn )") && next;
204	dpavlin	155
205			$times->{db} += time() - $t;
206
207	dpavlin	115	#$log->debug( "ds = " . Dumper( \@html_results ) );
208
209	dpavlin	155	$t = time();
210
211	dpavlin	100	my $html = $self->{out}->apply(
212	dpavlin	95	template => $template_filename,
213			data => $ds,
214	dpavlin	100	);
215
216	dpavlin	155	$times->{out} += time() - $t;
217
218			$t = time();
219
220	dpavlin	100	$html = $self->{iconv}->convert( $html ) \|\| $log->error("can't convert: $html");
221
222	dpavlin	155	$times->{iconv} += time() - $t;
223
224	dpavlin	100	push @html_results, $html;
225
226	dpavlin	95	}
227
228	dpavlin	115	#$log->debug( '@html_results = ' . Dumper( \@html_results ) );
229
230	dpavlin	155	$log->debug( sprintf(
231			"time spent: db = %.2f, out = %.2f, iconv = %.2f",
232			$times->{db}, $times->{out}, $times->{iconv},
233			) );
234
235	dpavlin	100	return \@html_results;
236	dpavlin	93	}
237
238	dpavlin	165	=head2 record
239
240			my $html = $m->record(
241			mfn => 42,
242			template => 'foo.tt',
243			);
244
245			This will load one record, convert it to html using C<template> and return
246			it.
247
248			=cut
249
250			sub record {
251			my $self = shift;
252
253			my $args = {@_};
254			my $log = $self->{log};
255			$log->debug("args: " . Dumper( $args ));
256
257			foreach my $f (qw/mfn template/) {
258			$log->die("need $f") unless ($args->{$f});
259			}
260
261			my $mfn = $args->{mfn};
262
263			my $ds = $self->{db}->load_ds( $mfn ) \|\| $log->error("can't load_ds( $mfn )") && next;
264
265			my $html = $self->{out}->apply(
266			template => $args->{template},
267			data => $ds,
268			);
269
270			$html = $self->{iconv}->convert( $html ) \|\| $log->error("can't convert: $html");
271
272			return $html;
273			}
274
275	dpavlin	135	=head2 save_html
276	dpavlin	93
277	dpavlin	135	$m->save_html( '/full/path/to/file', $content );
278	dpavlin	93
279	dpavlin	135	It will use C<iconv_on_save> to convert content encoding back to
280			Webpac codepage, recode JavaScript Unicode entities (%u1234),
281			strip extra newlines at beginning and end, and save to
282			C</full/path/to/file.new> and if that succeeds, just rename
283			it over original file which should be atomic on filesystem level.
284
285			=cut
286
287			sub save_html {
288			my ($self, $path, $content) = @_;
289
290			$content = $self->iconv_on_save( $content ) \|\| die "no content?";
291
292			sub _conv_js {
293			my $t = shift \|\| return;
294			return $self->{iconv}->convert(chr(hex($t)));
295			}
296			$content =~ s/%u([a-fA-F0-9]{4})/_conv_js($1)/gex;
297			$content =~ s/^[\n\r]+//s;
298			$content =~ s/[\n\r]+$/\n/s;
299
300			write_file($path . '.new', $content) \|\| die "can't save ${path}.new $!";
301			rename $path . '.new', $path \|\| die "can't rename to $path: $!";
302			}
303
304			=head2 load_html
305
306			my $html = $m->load_html('/full/path/to/file');
307
308			This will convert file from Webpac encoding to Catalyst and
309			convert that data to escaped HTML (for sending into
310			C<< <textarea/> >> tags in html.
311
312			=cut
313
314			sub load_html {
315			my ($self, $path) = @_;
316
317			die "no path?" unless ($path);
318
319			my $content = read_file($path) \|\| die "can't read $path: $!";
320			#$content = $q->escapeHTML($iconv_utf8->convert($content));
321			$content = $self->{iconv}->convert($content);
322
323			return $content;
324			}
325
326	dpavlin	92	=head1 AUTHOR
327
328			Dobrica Pavlinusic
329
330			=head1 LICENSE
331
332			This library is free software, you can redistribute it and/or modify
333			it under the same terms as Perl itself.
334
335			=cut
336
337			1;