| 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; |
| 910 |
return $self->_fetch_doc( id => $id ); |
return $self->_fetch_doc( id => $id ); |
| 911 |
} |
} |
| 912 |
|
|
| 913 |
|
|
| 914 |
=head2 get_doc_by_uri |
=head2 get_doc_by_uri |
| 915 |
|
|
| 916 |
Retreive document |
Retreive document |
| 927 |
return $self->_fetch_doc( uri => $uri ); |
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' ) 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 _fetch_doc |
=head2 _fetch_doc |
| 963 |
|
|
| 964 |
Private function used for implementation of C<get_doc> and C<get_doc_by_uri>. |
Private function used for implementing of C<get_doc>, C<get_doc_by_uri>, |
| 965 |
|
C<etch_doc>, C<etch_doc_by_uri>. |
| 966 |
|
|
| 967 |
my $doc = $node->fetch_doc( id => 42 ); |
my $doc = $node->fetch_doc( id => 42, etch => 1 ); |
| 968 |
my $doc = $node->fetch_doc( uri => 'file://uri/42' ); |
my $doc = $node->fetch_doc( uri => 'file://uri/42' ); |
| 969 |
|
|
| 970 |
=cut |
=cut |
| 971 |
|
|
| 972 |
sub _fetch_doc { |
sub _fetch_doc { |
| 973 |
my $self = shift; |
my $self = shift; |
| 974 |
my ($name,$val) = @_; |
my $a = {@_}; |
| 975 |
return unless ($name && defined($val) && $self->{url}); |
return unless ( ($a->{id} || $a->{uri}) && $self->{url} ); |
| 976 |
if ($name eq 'id') { |
|
| 977 |
croak "id must be numberm not '$val'" unless ($val =~ m/^\d+$/); |
my ($arg, $resbody); |
| 978 |
|
|
| 979 |
|
my $path = '/get_doc'; |
| 980 |
|
$path = '/etch_doc' if ($a->{etch}); |
| 981 |
|
|
| 982 |
|
if ($a->{id}) { |
| 983 |
|
croak "id must be numberm not '$a->{id}'" unless ($a->{id} =~ m/^\d+$/); |
| 984 |
|
$arg = 'id=' . $a->{id}; |
| 985 |
|
} elsif ($a->{uri}) { |
| 986 |
|
$arg = 'uri=' . $a->{uri}; |
| 987 |
|
} else { |
| 988 |
|
confess "unhandled argument. Need id or uri."; |
| 989 |
} |
} |
| 990 |
my $rv = $self->shuttle_url( $self->{url} . '/get_doc', |
|
| 991 |
|
my $rv = $self->shuttle_url( $self->{url} . $path, |
| 992 |
'application/x-www-form-urlencoded', |
'application/x-www-form-urlencoded', |
| 993 |
"$name=$val", |
$arg, |
| 994 |
my $draft, |
$resbody, |
| 995 |
); |
); |
| 996 |
|
|
| 997 |
return if ($rv != 200); |
return if ($rv != 200); |
| 998 |
return new Search::Estraier::Document($draft); |
|
| 999 |
|
if ($a->{etch}) { |
| 1000 |
|
$self->{kwords} = {}; |
| 1001 |
|
return +{} unless ($resbody); |
| 1002 |
|
foreach my $l (split(/\n/, $resbody)) { |
| 1003 |
|
my ($k,$v) = split(/\t/, $l, 2); |
| 1004 |
|
$self->{kwords}->{$k} = $v if ($v); |
| 1005 |
|
} |
| 1006 |
|
return $self->{kwords}; |
| 1007 |
|
} else { |
| 1008 |
|
return new Search::Estraier::Document($resbody); |
| 1009 |
|
} |
| 1010 |
} |
} |
| 1011 |
|
|
| 1012 |
|
|
Parent Directory
| 
Revision Log
| 
Patch