5 |
|
|
6 |
use base qw/WebPAC::Common/; |
use base qw/WebPAC::Common/; |
7 |
|
|
8 |
use HyperEstraier; |
use Search::Estraier; |
9 |
use Text::Iconv; |
use Encode qw/from_to/; |
10 |
use Data::Dumper; |
use Data::Dumper; |
11 |
use LWP; |
use LWP; |
12 |
use URI::Escape; |
use URI::Escape; |
17 |
|
|
18 |
=head1 VERSION |
=head1 VERSION |
19 |
|
|
20 |
Version 0.05 |
Version 0.10 |
21 |
|
|
22 |
=cut |
=cut |
23 |
|
|
24 |
our $VERSION = '0.05'; |
our $VERSION = '0.10'; |
25 |
|
|
26 |
=head1 SYNOPSIS |
=head1 SYNOPSIS |
27 |
|
|
39 |
user => 'admin', |
user => 'admin', |
40 |
passwd => 'admin', |
passwd => 'admin', |
41 |
database => 'demo', |
database => 'demo', |
42 |
|
label => 'node label', |
43 |
encoding => 'iso-8859-2', |
encoding => 'iso-8859-2', |
44 |
|
clean => 1, |
45 |
); |
); |
46 |
|
|
47 |
Options are: |
Options are: |
64 |
|
|
65 |
name of database from which data comes |
name of database from which data comes |
66 |
|
|
67 |
|
=item label |
68 |
|
|
69 |
|
label for node (optional) |
70 |
|
|
71 |
=item encoding |
=item encoding |
72 |
|
|
73 |
character encoding of C<data_structure> if it's differenet than C<ISO-8859-2> |
character encoding of C<data_structure> if it's differenet than C<ISO-8859-2> |
82 |
|
|
83 |
sub new { |
sub new { |
84 |
my $class = shift; |
my $class = shift; |
85 |
my $self = {@_}; |
my $self = {@_}; |
86 |
bless($self, $class); |
bless($self, $class); |
87 |
|
|
88 |
my $log = $self->_get_logger; |
my $log = $self->_get_logger; |
89 |
|
|
93 |
$log->logdie("need $p") unless ($self->{$p}); |
$log->logdie("need $p") unless ($self->{$p}); |
94 |
} |
} |
95 |
|
|
96 |
|
$self->{encoding} ||= 'ISO-8859-2'; |
97 |
|
|
98 |
my $url = $self->{masterurl} . '/node/' . $self->{database}; |
my $url = $self->{masterurl} . '/node/' . $self->{database}; |
99 |
$self->{url} = $url; |
$self->{url} = $url; |
100 |
|
|
101 |
$log->info("opening Hyper Estraier index $self->{url}"); |
if ($self->{clean}) { |
102 |
|
$log->debug("nodedel $self->{database}"); |
103 |
|
$self->master( action => 'nodedel', name => $self->{database} ); |
104 |
|
} else { |
105 |
|
$log->debug("opening index $self->{url}"); |
106 |
|
} |
107 |
|
|
108 |
my $nodes = $self->master( action => 'nodelist' ); |
my $nodes = $self->master( action => 'nodelist' ); |
109 |
|
|
110 |
$log->debug("nodes found: $nodes"); |
$log->debug("nodes found: $nodes"); |
111 |
|
|
112 |
if ($nodes !~ m/^$self->{database}\t/sm) { |
if ($nodes !~ m/^$self->{database}\t/sm) { |
113 |
$log->info("creating index $url"); |
my $label = $self->{label} || 'WebPAC ' . $self->{database}; |
114 |
|
$log->warn("creating index $url ($label)"); |
115 |
$self->master( |
$self->master( |
116 |
action => 'nodeadd', |
action => 'nodeadd', |
117 |
name => $self->{database}, |
name => $self->{database}, |
118 |
label => "WebPAC $self->{database}", |
label => $self->convert( $label ), |
119 |
) || $log->logdie("can't create Hyper Estraier node $self->{database}"); |
) || $log->logdie("can't create Hyper Estraier node $self->{database}"); |
120 |
} |
} |
121 |
|
|
122 |
$self->{'db'} = HyperEstraier::Node->new($self->{url}); |
$self->{db} = Search::Estraier::Node->new( debug => $self->{debug} ); |
123 |
$self->{'db'}->set_auth($self->{'user'}, $self->{passwd}); |
$self->{db}->set_url($self->{url}); |
124 |
|
$self->{db}->set_auth($self->{user}, $self->{passwd}); |
|
my $encoding = $self->{'encoding'} || 'ISO-8859-2'; |
|
|
$log->info("using encoding $encoding"); |
|
125 |
|
|
126 |
$self->{'iconv'} = new Text::Iconv($encoding, 'UTF-8') or |
$log->info("using index $self->{url} with encoding $self->{encoding}"); |
|
$log->logdie("can't create conversion from $encoding to UTF-8"); |
|
127 |
|
|
128 |
$self ? return $self : return undef; |
$self ? return $self : return undef; |
129 |
} |
} |
169 |
my $uri = "file:///$type/$database/$id"; |
my $uri = "file:///$type/$database/$id"; |
170 |
$log->debug("creating $uri"); |
$log->debug("creating $uri"); |
171 |
|
|
172 |
my $doc = HyperEstraier::Document->new; |
my $doc = Search::Estraier::Document->new; |
173 |
$doc->add_attr('@uri', $self->{'iconv'}->convert($uri) ); |
$doc->add_attr('@uri', $self->convert($uri) ); |
174 |
|
|
175 |
$log->debug("ds = ", sub { Dumper($args->{'ds'}) } ); |
$log->debug("ds = ", sub { Dumper($args->{'ds'}) } ); |
176 |
|
|
187 |
|
|
188 |
my $vals = join(" ", @{ $args->{'ds'}->{$tag}->{$type} }); |
my $vals = join(" ", @{ $args->{'ds'}->{$tag}->{$type} }); |
189 |
|
|
190 |
$log->logconfess("no values for $tag/$type") unless ($vals); |
next if (! $vals); |
191 |
|
|
192 |
$vals = $self->{'iconv'}->convert( $vals ) or |
$vals = $self->convert( $vals ) or |
193 |
$log->logdie("can't convert '$vals' to UTF-8"); |
$log->logdie("can't convert '$vals' to UTF-8"); |
194 |
|
|
195 |
$doc->add_attr( $tag, $vals ); |
$doc->add_attr( $tag, $vals ); |
198 |
|
|
199 |
my $text = $args->{'text'}; |
my $text = $args->{'text'}; |
200 |
if ( $text ) { |
if ( $text ) { |
201 |
$text = $self->{'iconv'}->convert( $text ) or |
$text = $self->convert( $text ) or |
202 |
$log->logdie("can't convert '$text' to UTF-8"); |
$log->logdie("can't convert '$text' to UTF-8"); |
203 |
$doc->add_text( $text ); |
$doc->add_text( $text ); |
204 |
} |
} |
205 |
|
|
206 |
$log->debug("adding ", sub { $doc->dump_draft } ); |
$log->debug("adding ", sub { $doc->dump_draft } ); |
207 |
$self->{'db'}->put_doc($doc) || $log->logdie("can't add document $uri to node " . $self->{url} . " status: " . $self->{db}->status()); |
$self->{'db'}->put_doc($doc) || $log->warn("can't add document $uri with draft " . $doc->dump_draft . " to node " . $self->{url} . " status: " . $self->{db}->status()); |
208 |
|
|
209 |
return 1; |
return 1; |
210 |
} |
} |
250 |
validate => 'master', |
validate => 'master', |
251 |
rest_url => $self->{masterurl} . '/master?action=' . $action , |
rest_url => $self->{masterurl} . '/master?action=' . $action , |
252 |
action => $action, |
action => $action, |
253 |
|
%{ $args }, |
254 |
); |
); |
255 |
} |
} |
256 |
|
|
274 |
|
|
275 |
$log->debug("got labels: ", join("|", @labels)); |
$log->debug("got labels: ", join("|", @labels)); |
276 |
|
|
277 |
@labels = grep(/^$args->{to}/, @labels); |
@labels = grep(/^$args->{to}\t/, @labels); |
278 |
|
my $label = shift @labels; |
279 |
my (undef,$label) = split(/\t/, shift @labels); |
(undef,$label) = split(/\t/, $label) if ($label); |
280 |
|
|
281 |
if (! $label) { |
if (! $label) { |
282 |
$log->warn("can't find label for $args->{to}, skipping link creaton"); |
$log->warn("can't find label for $args->{to}, skipping link creaton"); |
334 |
$del = '&'; |
$del = '&'; |
335 |
} |
} |
336 |
|
|
337 |
if ($url_args) { |
$url .= $url_args if ($url_args); |
|
$url_args =~ s#^\&#?# if ($url =~ m#\?#); |
|
|
$url .= $url_args; |
|
|
} |
|
338 |
|
|
339 |
$log->debug("calling $url"); |
$log->debug("calling $url"); |
340 |
|
|
396 |
return $self->{_master_ua}; |
return $self->{_master_ua}; |
397 |
} |
} |
398 |
|
|
399 |
|
=head2 convert |
400 |
|
|
401 |
|
my $utf8_string = $self->convert('string in codepage'); |
402 |
|
|
403 |
|
=cut |
404 |
|
|
405 |
|
sub convert { |
406 |
|
my $self = shift; |
407 |
|
|
408 |
|
my $text = shift || return; |
409 |
|
from_to($text, $self->{encoding}, 'UTF-8'); |
410 |
|
return $text; |
411 |
|
} |
412 |
|
|
413 |
=head1 AUTHOR |
=head1 AUTHOR |
414 |
|
|
415 |
Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >> |
Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >> |