27 |
|
|
28 |
=cut |
=cut |
29 |
|
|
30 |
|
=head1 Inheritable common methods |
31 |
|
|
32 |
|
This methods should really move somewhere else. |
33 |
|
|
34 |
=head2 _s |
=head2 _s |
35 |
|
|
36 |
Remove multiple whitespaces from string, as well as whitespaces at beginning or end |
Remove multiple whitespaces from string, as well as whitespaces at beginning or end |
60 |
This class implements Document which is collection of attributes |
This class implements Document which is collection of attributes |
61 |
(key=value), vectors (also key value) display text and hidden text. |
(key=value), vectors (also key value) display text and hidden text. |
62 |
|
|
63 |
|
|
64 |
=head2 new |
=head2 new |
65 |
|
|
66 |
Create new document, empty or from draft. |
Create new document, empty or from draft. |
180 |
push @{ $self->{htexts} }, $self->_s($text); |
push @{ $self->{htexts} }, $self->_s($text); |
181 |
} |
} |
182 |
|
|
183 |
|
|
184 |
=head2 id |
=head2 id |
185 |
|
|
186 |
Get the ID number of document. If the object has never been registred, C<-1> is returned. |
Get the ID number of document. If the object has never been registred, C<-1> is returned. |
194 |
return $self->{id}; |
return $self->{id}; |
195 |
} |
} |
196 |
|
|
197 |
|
|
198 |
=head2 attr_names |
=head2 attr_names |
199 |
|
|
200 |
Returns array with attribute names from document object. |
Returns array with attribute names from document object. |
240 |
return @{ $self->{dtexts} }; |
return @{ $self->{dtexts} }; |
241 |
} |
} |
242 |
|
|
243 |
|
|
244 |
=head2 cat_texts |
=head2 cat_texts |
245 |
|
|
246 |
Return whole text as single scalar. |
Return whole text as single scalar. |
254 |
return join(' ',@{ $self->{dtexts} }); |
return join(' ',@{ $self->{dtexts} }); |
255 |
} |
} |
256 |
|
|
257 |
|
|
258 |
=head2 dump_draft |
=head2 dump_draft |
259 |
|
|
260 |
Dump draft data from document object. |
Dump draft data from document object. |
281 |
|
|
282 |
$draft .= "\n"; |
$draft .= "\n"; |
283 |
|
|
284 |
$draft .= join("\n", @{ $self->{dtexts} }) . "\n"; |
$draft .= join("\n", @{ $self->{dtexts} }) . "\n" if ($self->{dtexts}); |
285 |
$draft .= "\t" . join("\n\t", @{ $self->{htexts} }) . "\n"; |
$draft .= "\t" . join("\n\t", @{ $self->{htexts} }) . "\n" if ($self->{htexts}); |
286 |
|
|
287 |
return $draft; |
return $draft; |
288 |
} |
} |
289 |
|
|
290 |
|
|
291 |
=head2 delete |
=head2 delete |
292 |
|
|
293 |
Empty document object |
Empty document object |
340 |
$self ? return $self : return undef; |
$self ? return $self : return undef; |
341 |
} |
} |
342 |
|
|
343 |
|
|
344 |
=head2 set_phrase |
=head2 set_phrase |
345 |
|
|
346 |
$cond->set_phrase('search phrase'); |
$cond->set_phrase('search phrase'); |
352 |
$self->{phrase} = $self->_s( shift ); |
$self->{phrase} = $self->_s( shift ); |
353 |
} |
} |
354 |
|
|
355 |
|
|
356 |
=head2 add_attr |
=head2 add_attr |
357 |
|
|
358 |
$cond->add_attr('@URI STRINC /~dpavlin/'); |
$cond->add_attr('@URI STRINC /~dpavlin/'); |
365 |
push @{ $self->{attrs} }, $self->_s( $attr ); |
push @{ $self->{attrs} }, $self->_s( $attr ); |
366 |
} |
} |
367 |
|
|
368 |
|
|
369 |
=head2 set_order |
=head2 set_order |
370 |
|
|
371 |
$cond->set_order('@mdate NUMD'); |
$cond->set_order('@mdate NUMD'); |
377 |
$self->{order} = shift; |
$self->{order} = shift; |
378 |
} |
} |
379 |
|
|
380 |
|
|
381 |
=head2 set_max |
=head2 set_max |
382 |
|
|
383 |
$cond->set_max(42); |
$cond->set_max(42); |
391 |
$self->{max} = $max; |
$self->{max} = $max; |
392 |
} |
} |
393 |
|
|
394 |
|
|
395 |
=head2 set_options |
=head2 set_options |
396 |
|
|
397 |
$cond->set_options( SURE => 1 ); |
$cond->set_options( SURE => 1 ); |
420 |
$self->{options} ||= $options->{$option}; |
$self->{options} ||= $options->{$option}; |
421 |
} |
} |
422 |
|
|
423 |
|
|
424 |
=head2 phrase |
=head2 phrase |
425 |
|
|
426 |
Return search phrase. |
Return search phrase. |
434 |
return $self->{phrase}; |
return $self->{phrase}; |
435 |
} |
} |
436 |
|
|
437 |
|
|
438 |
=head2 order |
=head2 order |
439 |
|
|
440 |
Return search result order. |
Return search result order. |
448 |
return $self->{order}; |
return $self->{order}; |
449 |
} |
} |
450 |
|
|
451 |
|
|
452 |
=head2 attrs |
=head2 attrs |
453 |
|
|
454 |
Return search result attrs. |
Return search result attrs. |
463 |
return @{ $self->{attrs} }; |
return @{ $self->{attrs} }; |
464 |
} |
} |
465 |
|
|
466 |
|
|
467 |
=head2 max |
=head2 max |
468 |
|
|
469 |
Return maximum number of results. |
Return maximum number of results. |
479 |
return $self->{max}; |
return $self->{max}; |
480 |
} |
} |
481 |
|
|
482 |
|
|
483 |
=head2 options |
=head2 options |
484 |
|
|
485 |
Return options for this condition. |
Return options for this condition. |
531 |
$self ? return $self : return undef; |
$self ? return $self : return undef; |
532 |
} |
} |
533 |
|
|
534 |
|
|
535 |
=head2 uri |
=head2 uri |
536 |
|
|
537 |
Return URI of result document |
Return URI of result document |
560 |
return sort keys %{ $self->{attrs} }; |
return sort keys %{ $self->{attrs} }; |
561 |
} |
} |
562 |
|
|
563 |
|
|
564 |
=head2 attr |
=head2 attr |
565 |
|
|
566 |
Returns value of an attribute. |
Returns value of an attribute. |
575 |
return $self->{attrs}->{ $name }; |
return $self->{attrs}->{ $name }; |
576 |
} |
} |
577 |
|
|
578 |
|
|
579 |
=head2 snippet |
=head2 snippet |
580 |
|
|
581 |
Return snippet from result document |
Return snippet from result document |
589 |
return $self->{snippet}; |
return $self->{snippet}; |
590 |
} |
} |
591 |
|
|
592 |
|
|
593 |
=head2 keywords |
=head2 keywords |
594 |
|
|
595 |
Return keywords from result document |
Return keywords from result document |
634 |
$self ? return $self : return undef; |
$self ? return $self : return undef; |
635 |
} |
} |
636 |
|
|
637 |
|
|
638 |
=head2 doc_num |
=head2 doc_num |
639 |
|
|
640 |
Return number of documents |
Return number of documents |
648 |
return $#{$self->{docs}}; |
return $#{$self->{docs}}; |
649 |
} |
} |
650 |
|
|
651 |
|
|
652 |
=head2 get_doc |
=head2 get_doc |
653 |
|
|
654 |
Return single document |
Return single document |
667 |
return $self->{docs}->[$num]; |
return $self->{docs}->[$num]; |
668 |
} |
} |
669 |
|
|
670 |
|
|
671 |
=head2 hint |
=head2 hint |
672 |
|
|
673 |
Return specific hint from results. |
Return specific hint from results. |
688 |
|
|
689 |
package Search::Estraier::Node; |
package Search::Estraier::Node; |
690 |
|
|
691 |
use Carp qw/croak/; |
use Carp qw/carp croak/; |
692 |
|
use URI; |
693 |
|
use MIME::Base64; |
694 |
|
use IO::Socket::INET; |
695 |
|
|
696 |
=head1 Search::Estraier::Node |
=head1 Search::Estraier::Node |
697 |
|
|
705 |
my $class = shift; |
my $class = shift; |
706 |
my $self = { |
my $self = { |
707 |
pxport => -1, |
pxport => -1, |
708 |
timeout => -1, |
timeout => 0, # this used to be -1 |
709 |
dnum => -1, |
dnum => -1, |
710 |
wnum => -1, |
wnum => -1, |
711 |
size => -1.0, |
size => -1.0, |
716 |
}; |
}; |
717 |
bless($self, $class); |
bless($self, $class); |
718 |
|
|
719 |
|
if (@_) { |
720 |
|
$self->{debug} = shift; |
721 |
|
warn "## Node debug on\n"; |
722 |
|
} |
723 |
|
|
724 |
$self ? return $self : return undef; |
$self ? return $self : return undef; |
725 |
} |
} |
726 |
|
|
727 |
|
|
728 |
=head2 set_url |
=head2 set_url |
729 |
|
|
730 |
Specify URL to node server |
Specify URL to node server |
738 |
$self->{url} = shift; |
$self->{url} = shift; |
739 |
} |
} |
740 |
|
|
741 |
|
|
742 |
=head2 set_proxy |
=head2 set_proxy |
743 |
|
|
744 |
Specify proxy server to connect to node server |
Specify proxy server to connect to node server |
755 |
$self->{pxport} = $port; |
$self->{pxport} = $port; |
756 |
} |
} |
757 |
|
|
|
package Search::Estraier::Master; |
|
758 |
|
|
759 |
use Carp; |
=head2 set_timeout |
760 |
|
|
761 |
=head1 Search::Estraier::Master |
Specify timeout of connection in seconds |
762 |
|
|
763 |
Controll node master. This requires user with administration priviledges. |
$node->set_timeout( 15 ); |
764 |
|
|
765 |
=cut |
=cut |
766 |
|
|
767 |
{ |
sub set_timeout { |
768 |
package RequestAgent; |
my $self = shift; |
769 |
our @ISA = qw(LWP::UserAgent); |
my $sec = shift; |
770 |
|
croak "timeout must be number" unless ($sec =~ m/^\d+$/); |
771 |
|
$self->{timeout} = $sec; |
772 |
|
} |
773 |
|
|
|
sub new { |
|
|
my $self = LWP::UserAgent::new(@_); |
|
|
$self->agent("Search-Estraier/$Search::Estraer::VERSION"); |
|
|
$self; |
|
|
} |
|
774 |
|
|
775 |
sub get_basic_credentials { |
=head2 set_auth |
776 |
my($self, $realm, $uri) = @_; |
|
777 |
# return ($user, $password); |
Specify name and password for authentication to node server. |
778 |
} |
|
779 |
|
$node->set_auth('clint','eastwood'); |
780 |
|
|
781 |
|
=cut |
782 |
|
|
783 |
|
sub set_auth { |
784 |
|
my $self = shift; |
785 |
|
my ($login,$passwd) = @_; |
786 |
|
my $basic_auth = encode_base64( "$login:$passwd" ); |
787 |
|
chomp($basic_auth); |
788 |
|
$self->{auth} = $basic_auth; |
789 |
} |
} |
790 |
|
|
791 |
|
|
792 |
|
=head2 status |
793 |
|
|
794 |
=head2 new |
Return status code of last request. |
795 |
|
|
796 |
Create new connection to node master. |
print $node->status; |
797 |
|
|
798 |
my $master = new Search::Estraier::Master( |
C<-1> means connection failure. |
|
url => 'http://localhost:1978', |
|
|
user => 'admin', |
|
|
passwd => 'admin', |
|
|
); |
|
799 |
|
|
800 |
=cut |
=cut |
801 |
|
|
802 |
sub new { |
sub status { |
803 |
my $class = shift; |
my $self = shift; |
804 |
my $self = {@_}; |
return $self->{status}; |
805 |
bless($self, $class); |
} |
806 |
|
|
|
foreach my $p (qw/url user passwd/) { |
|
|
croak "need $p" unless ($self->{$p}); |
|
|
} |
|
807 |
|
|
808 |
$self ? return $self : return undef; |
=head2 put_doc |
809 |
|
|
810 |
|
Add a document |
811 |
|
|
812 |
|
$node->put_doc( $document_draft ) or die "can't add document"; |
813 |
|
|
814 |
|
Return true on success or false on failture. |
815 |
|
|
816 |
|
=cut |
817 |
|
|
818 |
|
sub put_doc { |
819 |
|
my $self = shift; |
820 |
|
my $doc = shift || return; |
821 |
|
return unless ($self->{url}); |
822 |
|
$self->shuttle_url( $self->{url} . '/put_doc', |
823 |
|
'text/x-estraier-draft', |
824 |
|
$doc->dump_draft, |
825 |
|
undef |
826 |
|
) == 200; |
827 |
|
} |
828 |
|
|
829 |
|
|
830 |
|
=head2 out_doc |
831 |
|
|
832 |
|
Remove a document |
833 |
|
|
834 |
|
$node->out_doc( document_id ) or "can't remove document"; |
835 |
|
|
836 |
|
Return true on success or false on failture. |
837 |
|
|
838 |
|
=cut |
839 |
|
|
840 |
|
sub out_doc { |
841 |
|
my $self = shift; |
842 |
|
my $id = shift || return; |
843 |
|
return unless ($self->{url}); |
844 |
|
croak "id must be number" unless ($id =~ m/^\d+$/); |
845 |
|
$self->shuttle_url( $self->{url} . '/out_doc', |
846 |
|
'application/x-www-form-urlencoded', |
847 |
|
"id=$id", |
848 |
|
undef |
849 |
|
) == 200; |
850 |
|
} |
851 |
|
|
852 |
|
|
853 |
|
=head2 out_doc_by_uri |
854 |
|
|
855 |
|
Remove a registrated document using it's uri |
856 |
|
|
857 |
|
$node->out_doc_by_uri( 'file:///document_url' ) or "can't remove document"; |
858 |
|
|
859 |
|
Return true on success or false on failture. |
860 |
|
|
861 |
|
=cut |
862 |
|
|
863 |
|
sub out_doc_by_uri { |
864 |
|
my $self = shift; |
865 |
|
my $uri = shift || return; |
866 |
|
return unless ($self->{url}); |
867 |
|
$self->shuttle_url( $self->{url} . '/out_doc', |
868 |
|
'application/x-www-form-urlencoded', |
869 |
|
"uri=$uri", |
870 |
|
undef |
871 |
|
) == 200; |
872 |
} |
} |
873 |
|
|
874 |
|
|
875 |
|
=head2 edit_doc |
876 |
|
|
877 |
|
Edit attributes of a document |
878 |
|
|
879 |
|
$node->edit_doc( $document_draft ) or die "can't edit document"; |
880 |
|
|
881 |
|
Return true on success or false on failture. |
882 |
|
|
883 |
|
=cut |
884 |
|
|
885 |
|
sub edit_doc { |
886 |
|
my $self = shift; |
887 |
|
my $doc = shift || return; |
888 |
|
return unless ($self->{url}); |
889 |
|
$self->shuttle_url( $self->{url} . '/edit_doc', |
890 |
|
'text/x-estraier-draft', |
891 |
|
$doc->dump_draft, |
892 |
|
undef |
893 |
|
) == 200; |
894 |
|
} |
895 |
|
|
896 |
|
|
897 |
|
=head2 shuttle_url |
898 |
|
|
899 |
|
This is method which uses C<IO::Socket::INET> to communicate with Hyper Estraier node |
900 |
|
master. |
901 |
|
|
902 |
|
my $rv = shuttle_url( $url, $content_type, \$req_body, \$resbody ); |
903 |
|
|
904 |
|
C<$resheads> and C<$resbody> booleans controll if response headers and/or response |
905 |
|
body will be saved within object. |
906 |
|
|
907 |
|
=cut |
908 |
|
|
909 |
|
sub shuttle_url { |
910 |
|
my $self = shift; |
911 |
|
|
912 |
|
my ($url, $content_type, $reqbody, $resbody) = @_; |
913 |
|
|
914 |
|
$self->{status} = -1; |
915 |
|
|
916 |
|
warn "## $url\n" if ($self->{debug}); |
917 |
|
|
918 |
|
$url = new URI($url); |
919 |
|
if ( |
920 |
|
!$url || !$url->scheme || !$url->scheme eq 'http' || |
921 |
|
!$url->host || !$url->port || $url->port < 1 |
922 |
|
) { |
923 |
|
carp "can't parse $url\n"; |
924 |
|
return -1; |
925 |
|
} |
926 |
|
|
927 |
|
my ($host,$port,$query) = ($url->host, $url->port, $url->path); |
928 |
|
|
929 |
|
if ($self->{pxhost}) { |
930 |
|
($host,$port) = ($self->{pxhost}, $self->{pxport}); |
931 |
|
$query = "http://$host:$port/$query"; |
932 |
|
} |
933 |
|
|
934 |
|
$query .= '?' . $url->query if ($url->query && ! $reqbody); |
935 |
|
|
936 |
|
my $headers; |
937 |
|
|
938 |
|
if ($reqbody) { |
939 |
|
$headers .= "POST $query HTTP/1.0\r\n"; |
940 |
|
} else { |
941 |
|
$headers .= "GET $query HTTP/1.0\r\n"; |
942 |
|
} |
943 |
|
|
944 |
|
$headers .= "Host: " . $url->host . ":" . $url->port . "\r\n"; |
945 |
|
$headers .= "Connection: close\r\n"; |
946 |
|
$headers .= "User-Agent: Search-Estraier/$Search::Estraier::VERSION\r\n"; |
947 |
|
$headers .= "Content-Type: $content_type\r\n"; |
948 |
|
$headers .= "Authorization: Basic $self->{auth}\r\n"; |
949 |
|
my $len = 0; |
950 |
|
{ |
951 |
|
use bytes; |
952 |
|
$len = length($reqbody) if ($reqbody); |
953 |
|
} |
954 |
|
$headers .= "Content-Length: $len\r\n"; |
955 |
|
$headers .= "\r\n"; |
956 |
|
|
957 |
|
my $sock = IO::Socket::INET->new( |
958 |
|
PeerAddr => $host, |
959 |
|
PeerPort => $port, |
960 |
|
Proto => 'tcp', |
961 |
|
Timeout => $self->{timeout} || 90, |
962 |
|
); |
963 |
|
|
964 |
|
if (! $sock) { |
965 |
|
carp "can't open socket to $host:$port"; |
966 |
|
return -1; |
967 |
|
} |
968 |
|
|
969 |
|
warn $headers if ($self->{debug}); |
970 |
|
|
971 |
|
print $sock $headers or |
972 |
|
carp "can't send headers to network:\n$headers\n" and return -1; |
973 |
|
|
974 |
|
if ($reqbody) { |
975 |
|
warn "$reqbody\n" if ($self->{debug}); |
976 |
|
print $sock $reqbody or |
977 |
|
carp "can't send request body to network:\n$$reqbody\n" and return -1; |
978 |
|
} |
979 |
|
|
980 |
|
my $line = <$sock>; |
981 |
|
chomp($line); |
982 |
|
my ($schema, $res_status, undef) = split(/ */, $line, 3); |
983 |
|
return if ($schema !~ /^HTTP/ || ! $res_status); |
984 |
|
|
985 |
|
$self->{status} = $res_status; |
986 |
|
warn "## response status: $res_status\n" if ($self->{debug}); |
987 |
|
|
988 |
|
# skip rest of headers |
989 |
|
$line = <$sock>; |
990 |
|
while ($line) { |
991 |
|
$line = <$sock>; |
992 |
|
$line =~ s/[\r\n]+$//; |
993 |
|
warn "## ", $line || 'NULL', " ##\n" if ($self->{debug}); |
994 |
|
}; |
995 |
|
|
996 |
|
# read body |
997 |
|
$len = 0; |
998 |
|
do { |
999 |
|
$len = read($sock, my $buf, 8192); |
1000 |
|
$$resbody .= $buf if ($resbody); |
1001 |
|
} while ($len); |
1002 |
|
|
1003 |
|
warn "## response body:\n$$resbody\n" if ($resbody && $self->{debug}); |
1004 |
|
|
1005 |
|
return $self->{status}; |
1006 |
|
} |
1007 |
|
|
1008 |
### |
### |
1009 |
|
|