5 |
|
|
6 |
use base qw/WebPAC::Common/; |
use base qw/WebPAC::Common/; |
7 |
|
|
8 |
use HyperEstraier; |
use Search::Estraier 0.06; |
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; |
13 |
|
use List::Util qw/first/; |
14 |
|
|
15 |
=head1 NAME |
=head1 NAME |
16 |
|
|
18 |
|
|
19 |
=head1 VERSION |
=head1 VERSION |
20 |
|
|
21 |
Version 0.06 |
Version 0.11 |
22 |
|
|
23 |
=cut |
=cut |
24 |
|
|
25 |
our $VERSION = '0.06'; |
our $VERSION = '0.11'; |
26 |
|
|
27 |
=head1 SYNOPSIS |
=head1 SYNOPSIS |
28 |
|
|
40 |
user => 'admin', |
user => 'admin', |
41 |
passwd => 'admin', |
passwd => 'admin', |
42 |
database => 'demo', |
database => 'demo', |
43 |
|
label => 'node label', |
44 |
encoding => 'iso-8859-2', |
encoding => 'iso-8859-2', |
45 |
|
clean => 1, |
46 |
); |
); |
47 |
|
|
48 |
Options are: |
Options are: |
65 |
|
|
66 |
name of database from which data comes |
name of database from which data comes |
67 |
|
|
68 |
|
=item label |
69 |
|
|
70 |
|
label for node (optional) |
71 |
|
|
72 |
=item encoding |
=item encoding |
73 |
|
|
74 |
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> |
83 |
|
|
84 |
sub new { |
sub new { |
85 |
my $class = shift; |
my $class = shift; |
86 |
my $self = {@_}; |
my $self = {@_}; |
87 |
bless($self, $class); |
bless($self, $class); |
88 |
|
|
89 |
my $log = $self->_get_logger; |
my $log = $self->_get_logger; |
90 |
|
|
94 |
$log->logdie("need $p") unless ($self->{$p}); |
$log->logdie("need $p") unless ($self->{$p}); |
95 |
} |
} |
96 |
|
|
97 |
|
$self->{encoding} ||= 'ISO-8859-2'; |
98 |
|
|
99 |
my $url = $self->{masterurl} . '/node/' . $self->{database}; |
my $url = $self->{masterurl} . '/node/' . $self->{database}; |
100 |
$self->{url} = $url; |
$self->{url} = $url; |
101 |
|
|
102 |
$log->info("opening Hyper Estraier index $self->{url}"); |
$self->{db} = Search::Estraier::Node->new( |
103 |
|
url => $url, |
104 |
my $nodes = $self->master( action => 'nodelist' ); |
user => $self->{user}, |
105 |
|
passwd => $self->{passwd}, |
106 |
|
debug => $self->{debug}, |
107 |
|
create => 1, |
108 |
|
label => "WebPAC $self->{database}", |
109 |
|
); |
110 |
|
|
111 |
$log->debug("nodes found: $nodes"); |
$log->info("using index $self->{url} with encoding $self->{encoding}"); |
112 |
|
|
113 |
if ($nodes !~ m/^$self->{database}\t/sm) { |
if ($self->{clean}) { |
114 |
$log->info("creating index $url"); |
$log->debug("clean $self->{database}"); |
115 |
$self->master( |
$self->master( action => 'nodeclr', name => $self->{database} ); |
116 |
action => 'nodeadd', |
} else { |
117 |
name => $self->{database}, |
$log->debug("opening index $self->{url}"); |
|
label => "WebPAC $self->{database}", |
|
|
) || $log->logdie("can't create Hyper Estraier node $self->{database}"); |
|
118 |
} |
} |
119 |
|
|
|
$self->{'db'} = HyperEstraier::Node->new($self->{url}); |
|
|
$self->{'db'}->set_auth($self->{'user'}, $self->{passwd}); |
|
|
|
|
|
my $encoding = $self->{'encoding'} || 'ISO-8859-2'; |
|
|
$log->info("using encoding $encoding"); |
|
|
|
|
|
$self->{'iconv'} = new Text::Iconv($encoding, 'UTF-8') or |
|
|
$log->logdie("can't create conversion from $encoding to UTF-8"); |
|
|
|
|
120 |
$self ? return $self : return undef; |
$self ? return $self : return undef; |
121 |
} |
} |
122 |
|
|
144 |
sub add { |
sub add { |
145 |
my $self = shift; |
my $self = shift; |
146 |
|
|
147 |
my $args = {@_}; |
my $args = {@_}; |
148 |
|
|
149 |
my $log = $self->_get_logger; |
my $log = $self->_get_logger; |
150 |
|
|
161 |
my $uri = "file:///$type/$database/$id"; |
my $uri = "file:///$type/$database/$id"; |
162 |
$log->debug("creating $uri"); |
$log->debug("creating $uri"); |
163 |
|
|
164 |
my $doc = HyperEstraier::Document->new; |
my $doc = Search::Estraier::Document->new; |
165 |
$doc->add_attr('@uri', $self->{'iconv'}->convert($uri) ); |
$doc->add_attr('@uri', $self->convert($uri) ); |
166 |
|
|
167 |
$log->debug("ds = ", sub { Dumper($args->{'ds'}) } ); |
$log->debug("ds = ", sub { Dumper($args->{'ds'}) } ); |
168 |
|
|
179 |
|
|
180 |
my $vals = join(" ", @{ $args->{'ds'}->{$tag}->{$type} }); |
my $vals = join(" ", @{ $args->{'ds'}->{$tag}->{$type} }); |
181 |
|
|
182 |
$log->logconfess("no values for $tag/$type") unless ($vals); |
next if (! $vals); |
183 |
|
|
184 |
$vals = $self->{'iconv'}->convert( $vals ) or |
$vals = $self->convert( $vals ) or |
185 |
$log->logdie("can't convert '$vals' to UTF-8"); |
$log->logdie("can't convert '$vals' to UTF-8"); |
186 |
|
|
187 |
$doc->add_attr( $tag, $vals ); |
$doc->add_attr( $tag, $vals ); |
190 |
|
|
191 |
my $text = $args->{'text'}; |
my $text = $args->{'text'}; |
192 |
if ( $text ) { |
if ( $text ) { |
193 |
$text = $self->{'iconv'}->convert( $text ) or |
$text = $self->convert( $text ) or |
194 |
$log->logdie("can't convert '$text' to UTF-8"); |
$log->logdie("can't convert '$text' to UTF-8"); |
195 |
$doc->add_text( $text ); |
$doc->add_text( $text ); |
196 |
} |
} |
197 |
|
|
198 |
$log->debug("adding ", sub { $doc->dump_draft } ); |
$log->debug("adding ", sub { $doc->dump_draft } ); |
199 |
$self->{'db'}->put_doc($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()); |
200 |
|
|
201 |
return 1; |
return 1; |
202 |
} |
} |
203 |
|
|
|
# |
|
|
# REST parametars validation data |
|
|
# |
|
|
|
|
|
my $estraier_rest = { |
|
|
master => { |
|
|
userdel => [ qw/name/ ], |
|
|
nodelist => [], |
|
|
nodeadd => [ qw/name label/ ], |
|
|
nodedel => [ qw/name/ ], |
|
|
}, |
|
|
node => { |
|
|
_set_link => [ qw/url label credit/ ], |
|
|
}, |
|
|
}; |
|
|
|
|
|
=head2 master |
|
|
|
|
|
Issue administrative commands to C<estmaster> process and receive response |
|
|
as array of lines |
|
|
|
|
|
my $nodelist = $est->master( action => 'nodelist' ); |
|
|
|
|
|
=cut |
|
|
|
|
|
sub master { |
|
|
my $self = shift; |
|
|
|
|
|
my $args = {@_}; |
|
|
my $log = $self->_get_logger; |
|
|
|
|
|
my $action = $args->{action} || $log->logconfess("no action specified"); |
|
|
|
|
|
$log->logdie("action '$action' isn't supported") unless ($estraier_rest->{master}->{$action}); |
|
|
|
|
|
$log->debug("master action: $action"); |
|
|
|
|
|
return $self->estcall( |
|
|
validate => 'master', |
|
|
rest_url => $self->{masterurl} . '/master?action=' . $action , |
|
|
action => $action, |
|
|
%{ $args }, |
|
|
); |
|
|
} |
|
|
|
|
204 |
=head2 add_link |
=head2 add_link |
205 |
|
|
206 |
$est->add_link( |
$est->add_link( |
217 |
my $args = {@_}; |
my $args = {@_}; |
218 |
my $log = $self->_get_logger; |
my $log = $self->_get_logger; |
219 |
|
|
220 |
my @labels = $self->master( action => 'nodelist' ); |
my $node = first { $_->{name} eq $args->{to} } $self->master( action => 'nodelist' ); |
|
|
|
|
$log->debug("got labels: ", join("|", @labels)); |
|
221 |
|
|
222 |
@labels = grep(/^$args->{to}/, @labels); |
if (! $node) { |
223 |
|
$log->warn("can't find node $args->{to}, skipping link creaton"); |
224 |
|
return; |
225 |
|
} |
226 |
|
|
227 |
my (undef,$label) = split(/\t/, shift @labels); |
my $label = $node->{label}; |
228 |
|
|
229 |
if (! $label) { |
if (! $label) { |
230 |
$log->warn("can't find label for $args->{to}, skipping link creaton"); |
$log->warn("can't find label for $args->{to}, skipping link creaton"); |
233 |
|
|
234 |
$log->debug("using label $label for $args->{to}"); |
$log->debug("using label $label for $args->{to}"); |
235 |
|
|
236 |
return $self->estcall( |
return $self->{db}->set_link( |
237 |
validate => 'node', |
$self->{masterurl} . '/node/' . $args->{to}, |
238 |
action => '_set_link', |
$label, |
239 |
rest_url => $self->{masterurl} . '/node/' . $args->{from} . '/_set_link' , |
$args->{credit}, |
|
url => $self->{masterurl} . '/node/' . $args->{to}, |
|
|
label => $label, |
|
|
credit => $args->{credit}, |
|
240 |
); |
); |
241 |
} |
} |
242 |
|
|
|
=head2 estcall |
|
243 |
|
|
244 |
Workhourse which does actual calls to Hyper Estraier |
=head2 master |
245 |
|
|
246 |
$self->estcall( |
Issue administrative commands to C<estmaster> process. See documentation for |
247 |
rest_url => '/master?action=' . $action, |
C<master> in L<Search::Estraier>::Node. |
|
validate => 'master', |
|
|
# ... |
|
|
); |
|
248 |
|
|
249 |
C<rest_url> is relative URL to C<estmaster> and C<validate> is entry into |
$self->master( |
250 |
internal hash which will check if all parametars are available before |
action => 'nodeclr', |
251 |
calling function. |
name => 'foobar', |
252 |
|
); |
253 |
|
|
254 |
=cut |
=cut |
255 |
|
|
256 |
sub estcall { |
sub master { |
257 |
my $self = shift; |
my $self = shift; |
258 |
my $args = {@_}; |
$self->{db}->master( @_ ); |
|
my $log = $self->_get_logger; |
|
|
|
|
|
$log->debug("estcall: ",Dumper($args)); |
|
|
|
|
|
foreach my $p (qw/rest_url validate action/) { |
|
|
$log->die("ectcall needs $p parametar") unless ($args->{$p}); |
|
|
} |
|
|
|
|
|
my $url = $args->{rest_url}; |
|
|
my $del = '?'; |
|
|
$del = '&' if ($url =~ m#\?#); |
|
|
|
|
|
my $url_args; |
|
|
|
|
|
foreach my $arg (@{ $estraier_rest->{ $args->{validate} }->{ $args->{action} } }) { |
|
|
$log->logdie("missing parametar $arg for action $args->{action}") unless ($args->{$arg}); |
|
|
$url_args .= $del . $arg . '=' . uri_escape( $args->{$arg} ); |
|
|
$del = '&'; |
|
|
} |
|
|
|
|
|
$url .= $url_args if ($url_args); |
|
|
|
|
|
$log->debug("calling $url"); |
|
|
|
|
|
my $res = $self->est_ua()->get($url); |
|
|
|
|
|
if ($res->is_success) { |
|
|
#$log->debug( $res->content ); |
|
|
return split(/\n/, $res->content) if wantarray; |
|
|
return $res->content || 0E0; |
|
|
} else { |
|
|
$log->warn("unable to call $url: " . $res->status_line); |
|
|
return; |
|
|
} |
|
|
|
|
259 |
} |
} |
260 |
|
|
|
=head2 est_ua |
|
261 |
|
|
262 |
This is helper function to create C<LWP::UserAgent> object with Super User |
=head2 convert |
|
priviledges |
|
263 |
|
|
264 |
my $ua = $self->est_ua( user => 'admin', passwd => 'admin' ); |
my $utf8_string = $self->convert('string in codepage'); |
265 |
|
|
266 |
=cut |
=cut |
267 |
|
|
268 |
|
sub convert { |
|
|
|
|
sub est_ua { |
|
269 |
my $self = shift; |
my $self = shift; |
270 |
|
|
271 |
return $self->{_master_ua} if ($self->{_master_ua}); |
my $text = shift || return; |
272 |
|
from_to($text, $self->{encoding}, 'UTF-8'); |
273 |
{ |
return $text; |
|
package AdminUserAgent; |
|
|
use base qw/LWP::UserAgent/; |
|
|
sub new { |
|
|
my $self = LWP::UserAgent::new(@_); |
|
|
$self->agent("webpac/$VERSION"); |
|
|
$self; |
|
|
} |
|
|
sub get_basic_credentials { |
|
|
my($self, $realm, $uri) = @_; |
|
|
return ($self->{user}, $self->{passwd}); |
|
|
} |
|
|
sub set_basic_credentials { |
|
|
my ($self, $user, $passwd) = @_; |
|
|
$self->{user} = $user; |
|
|
$self->{passwd} = $passwd; |
|
|
} |
|
|
}; |
|
|
|
|
|
$self->{_master_ua} = AdminUserAgent->new( ) || sub { |
|
|
my $log = $self->_get_logger; |
|
|
$log->logdie("can't create LWP::UserAgent: $!"); |
|
|
}; |
|
|
|
|
|
$self->{_master_ua}->set_basic_credentials($self->{user}, $self->{passwd}); |
|
|
|
|
|
return $self->{_master_ua}; |
|
274 |
} |
} |
275 |
|
|
276 |
=head1 AUTHOR |
=head1 AUTHOR |