387 |
sub set_max { |
sub set_max { |
388 |
my $self = shift; |
my $self = shift; |
389 |
my $max = shift; |
my $max = shift; |
390 |
croak "set_max needs number" unless ($max =~ m/^\d+$/); |
croak "set_max needs number, not '$max'" unless ($max =~ m/^\d+$/); |
391 |
$self->{max} = $max; |
$self->{max} = $max; |
392 |
} |
} |
393 |
|
|
662 |
sub get_doc { |
sub get_doc { |
663 |
my $self = shift; |
my $self = shift; |
664 |
my $num = shift; |
my $num = shift; |
665 |
croak "expect number as argument" unless ($num =~ m/^\d+$/); |
croak "expect number as argument, not '$num'" unless ($num =~ m/^\d+$/); |
666 |
return undef if ($num < 0 || $num > $self->{docs}); |
return undef if ($num < 0 || $num > $self->{docs}); |
667 |
return $self->{docs}->[$num]; |
return $self->{docs}->[$num]; |
668 |
} |
} |
688 |
|
|
689 |
package Search::Estraier::Node; |
package Search::Estraier::Node; |
690 |
|
|
691 |
use Carp qw/carp croak/; |
use Carp qw/carp croak confess/; |
692 |
use URI; |
use URI; |
693 |
use MIME::Base64; |
use MIME::Base64; |
694 |
use IO::Socket::INET; |
use IO::Socket::INET; |
750 |
sub set_proxy { |
sub set_proxy { |
751 |
my $self = shift; |
my $self = shift; |
752 |
my ($host,$port) = @_; |
my ($host,$port) = @_; |
753 |
croak "proxy port must be number" unless ($port =~ m/^\d+$/); |
croak "proxy port must be number, not '$port'" unless ($port =~ m/^\d+$/); |
754 |
$self->{pxhost} = $host; |
$self->{pxhost} = $host; |
755 |
$self->{pxport} = $port; |
$self->{pxport} = $port; |
756 |
} |
} |
767 |
sub set_timeout { |
sub set_timeout { |
768 |
my $self = shift; |
my $self = shift; |
769 |
my $sec = shift; |
my $sec = shift; |
770 |
croak "timeout must be number" unless ($sec =~ m/^\d+$/); |
croak "timeout must be number, not '$sec'" unless ($sec =~ m/^\d+$/); |
771 |
$self->{timeout} = $sec; |
$self->{timeout} = $sec; |
772 |
} |
} |
773 |
|
|
818 |
sub put_doc { |
sub put_doc { |
819 |
my $self = shift; |
my $self = shift; |
820 |
my $doc = shift || return; |
my $doc = shift || return; |
821 |
return unless ($self->{url}); |
return unless ($self->{url} && $doc->isa('Search::Estraier::Document')); |
822 |
$self->shuttle_url( $self->{url} . '/put_doc', |
$self->shuttle_url( $self->{url} . '/put_doc', |
823 |
'text/x-estraier-draft', |
'text/x-estraier-draft', |
824 |
$doc->dump_draft, |
$doc->dump_draft, |
841 |
my $self = shift; |
my $self = shift; |
842 |
my $id = shift || return; |
my $id = shift || return; |
843 |
return unless ($self->{url}); |
return unless ($self->{url}); |
844 |
croak "id must be number" unless ($id =~ m/^\d+$/); |
croak "id must be number, not '$id'" unless ($id =~ m/^\d+$/); |
845 |
$self->shuttle_url( $self->{url} . '/out_doc', |
$self->shuttle_url( $self->{url} . '/out_doc', |
846 |
'application/x-www-form-urlencoded', |
'application/x-www-form-urlencoded', |
847 |
"id=$id", |
"id=$id", |
854 |
|
|
855 |
Remove a registrated document using it's uri |
Remove a registrated document using it's uri |
856 |
|
|
857 |
$node->out_doc_by_uri( 'file:///document_url' ) or "can't remove document"; |
$node->out_doc_by_uri( 'file:///document/uri/42' ) or "can't remove document"; |
858 |
|
|
859 |
Return true on success or false on failture. |
Return true on success or false on failture. |
860 |
|
|
885 |
sub edit_doc { |
sub edit_doc { |
886 |
my $self = shift; |
my $self = shift; |
887 |
my $doc = shift || return; |
my $doc = shift || return; |
888 |
return unless ($self->{url}); |
return unless ($self->{url} && $doc->isa('Search::Estraier::Document')); |
889 |
$self->shuttle_url( $self->{url} . '/edit_doc', |
$self->shuttle_url( $self->{url} . '/edit_doc', |
890 |
'text/x-estraier-draft', |
'text/x-estraier-draft', |
891 |
$doc->dump_draft, |
$doc->dump_draft, |
894 |
} |
} |
895 |
|
|
896 |
|
|
897 |
|
=head2 get_doc |
898 |
|
|
899 |
|
Retreive document |
900 |
|
|
901 |
|
my $doc = $node->get_doc( document_id ) or die "can't get document"; |
902 |
|
|
903 |
|
Return true on success or false on failture. |
904 |
|
|
905 |
|
=cut |
906 |
|
|
907 |
|
sub get_doc { |
908 |
|
my $self = shift; |
909 |
|
my $id = shift || return; |
910 |
|
return $self->_fetch_doc( id => $id ); |
911 |
|
} |
912 |
|
|
913 |
|
|
914 |
|
=head2 get_doc_by_uri |
915 |
|
|
916 |
|
Retreive document |
917 |
|
|
918 |
|
my $doc = $node->get_doc_by_uri( 'file:///document/uri/42' ) or die "can't get document"; |
919 |
|
|
920 |
|
Return true on success or false on failture. |
921 |
|
|
922 |
|
=cut |
923 |
|
|
924 |
|
sub get_doc_by_uri { |
925 |
|
my $self = shift; |
926 |
|
my $uri = shift || return; |
927 |
|
return $self->_fetch_doc( uri => $uri ); |
928 |
|
} |
929 |
|
|
930 |
|
|
931 |
|
=head2 etch_doc |
932 |
|
|
933 |
|
Exctract document keywords |
934 |
|
|
935 |
|
my $keywords = $node->etch_doc( document_id ) or die "can't etch document"; |
936 |
|
|
937 |
|
=cut |
938 |
|
|
939 |
|
sub erch_doc { |
940 |
|
my $self = shift; |
941 |
|
my $id = shift || return; |
942 |
|
return $self->_fetch_doc( id => $id, etch => 1 ); |
943 |
|
} |
944 |
|
|
945 |
|
=head2 etch_doc_by_uri |
946 |
|
|
947 |
|
Retreive document |
948 |
|
|
949 |
|
my $keywords = $node->etch_doc_by_uri( 'file:///document/uri/42' ) or die "can't etch document"; |
950 |
|
|
951 |
|
Return true on success or false on failture. |
952 |
|
|
953 |
|
=cut |
954 |
|
|
955 |
|
sub etch_doc_by_uri { |
956 |
|
my $self = shift; |
957 |
|
my $uri = shift || return; |
958 |
|
return $self->_fetch_doc( uri => $uri, etch => 1 ); |
959 |
|
} |
960 |
|
|
961 |
|
|
962 |
|
=head2 uri_to_id |
963 |
|
|
964 |
|
Get ID of document specified by URI |
965 |
|
|
966 |
|
my $id = $node->uri_to_id( 'file:///document/uri/42' ); |
967 |
|
|
968 |
|
=cut |
969 |
|
|
970 |
|
sub uri_to_id { |
971 |
|
my $self = shift; |
972 |
|
my $uri = shift || return; |
973 |
|
return $self->_fetch_doc( uri => $uri, path => '/uri_to_id', chomp_resbody => 1 ); |
974 |
|
} |
975 |
|
|
976 |
|
|
977 |
|
=head2 _fetch_doc |
978 |
|
|
979 |
|
Private function used for implementing of C<get_doc>, C<get_doc_by_uri>, |
980 |
|
C<etch_doc>, C<etch_doc_by_uri>. |
981 |
|
|
982 |
|
# this will decode received draft into Search::Estraier::Document object |
983 |
|
my $doc = $node->_fetch_doc( id => 42 ); |
984 |
|
my $doc = $node->_fetch_doc( uri => 'file:///document/uri/42' ); |
985 |
|
|
986 |
|
# to extract keywords, add etch |
987 |
|
my $doc = $node->_fetch_doc( id => 42, etch => 1 ); |
988 |
|
my $doc = $node->_fetch_doc( uri => 'file:///document/uri/42', etch => 1 ); |
989 |
|
|
990 |
|
# more general form which allows implementation of |
991 |
|
# uri_to_id |
992 |
|
my $id = $node->_fetch_doc( |
993 |
|
uri => 'file:///document/uri/42', |
994 |
|
path => '/uri_to_id', |
995 |
|
chomp_resbody => 1 |
996 |
|
); |
997 |
|
|
998 |
|
=cut |
999 |
|
|
1000 |
|
sub _fetch_doc { |
1001 |
|
my $self = shift; |
1002 |
|
my $a = {@_}; |
1003 |
|
return unless ( ($a->{id} || $a->{uri}) && $self->{url} ); |
1004 |
|
|
1005 |
|
my ($arg, $resbody); |
1006 |
|
|
1007 |
|
my $path = $a->{path} || '/get_doc'; |
1008 |
|
$path = '/etch_doc' if ($a->{etch}); |
1009 |
|
|
1010 |
|
if ($a->{id}) { |
1011 |
|
croak "id must be numberm not '$a->{id}'" unless ($a->{id} =~ m/^\d+$/); |
1012 |
|
$arg = 'id=' . $a->{id}; |
1013 |
|
} elsif ($a->{uri}) { |
1014 |
|
$arg = 'uri=' . $a->{uri}; |
1015 |
|
} else { |
1016 |
|
confess "unhandled argument. Need id or uri."; |
1017 |
|
} |
1018 |
|
|
1019 |
|
my $rv = $self->shuttle_url( $self->{url} . $path, |
1020 |
|
'application/x-www-form-urlencoded', |
1021 |
|
$arg, |
1022 |
|
\$resbody, |
1023 |
|
); |
1024 |
|
|
1025 |
|
return if ($rv != 200); |
1026 |
|
|
1027 |
|
if ($a->{etch}) { |
1028 |
|
$self->{kwords} = {}; |
1029 |
|
return +{} unless ($resbody); |
1030 |
|
foreach my $l (split(/\n/, $resbody)) { |
1031 |
|
my ($k,$v) = split(/\t/, $l, 2); |
1032 |
|
$self->{kwords}->{$k} = $v if ($v); |
1033 |
|
} |
1034 |
|
return $self->{kwords}; |
1035 |
|
} elsif ($a->{chomp_resbody}) { |
1036 |
|
return unless (defined($resbody)); |
1037 |
|
chomp($resbody); |
1038 |
|
return $resbody; |
1039 |
|
} else { |
1040 |
|
return new Search::Estraier::Document($resbody); |
1041 |
|
} |
1042 |
|
} |
1043 |
|
|
1044 |
|
|
1045 |
|
|
1046 |
|
|
1047 |
=head2 shuttle_url |
=head2 shuttle_url |
1048 |
|
|
1049 |
This is method which uses C<IO::Socket::INET> to communicate with Hyper Estraier node |
This is method which uses C<IO::Socket::INET> to communicate with Hyper Estraier node |