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; |
695 |
|
use URI::Escape qw/uri_escape/; |
696 |
|
|
697 |
=head1 Search::Estraier::Node |
=head1 Search::Estraier::Node |
698 |
|
|
751 |
sub set_proxy { |
sub set_proxy { |
752 |
my $self = shift; |
my $self = shift; |
753 |
my ($host,$port) = @_; |
my ($host,$port) = @_; |
754 |
croak "proxy port must be number" unless ($port =~ m/^\d+$/); |
croak "proxy port must be number, not '$port'" unless ($port =~ m/^\d+$/); |
755 |
$self->{pxhost} = $host; |
$self->{pxhost} = $host; |
756 |
$self->{pxport} = $port; |
$self->{pxport} = $port; |
757 |
} |
} |
768 |
sub set_timeout { |
sub set_timeout { |
769 |
my $self = shift; |
my $self = shift; |
770 |
my $sec = shift; |
my $sec = shift; |
771 |
croak "timeout must be number" unless ($sec =~ m/^\d+$/); |
croak "timeout must be number, not '$sec'" unless ($sec =~ m/^\d+$/); |
772 |
$self->{timeout} = $sec; |
$self->{timeout} = $sec; |
773 |
} |
} |
774 |
|
|
819 |
sub put_doc { |
sub put_doc { |
820 |
my $self = shift; |
my $self = shift; |
821 |
my $doc = shift || return; |
my $doc = shift || return; |
822 |
return unless ($self->{url}); |
return unless ($self->{url} && $doc->isa('Search::Estraier::Document')); |
823 |
$self->shuttle_url( $self->{url} . '/put_doc', |
$self->shuttle_url( $self->{url} . '/put_doc', |
824 |
'text/x-estraier-draft', |
'text/x-estraier-draft', |
825 |
$doc->dump_draft, |
$doc->dump_draft, |
842 |
my $self = shift; |
my $self = shift; |
843 |
my $id = shift || return; |
my $id = shift || return; |
844 |
return unless ($self->{url}); |
return unless ($self->{url}); |
845 |
croak "id must be number" unless ($id =~ m/^\d+$/); |
croak "id must be number, not '$id'" unless ($id =~ m/^\d+$/); |
846 |
$self->shuttle_url( $self->{url} . '/out_doc', |
$self->shuttle_url( $self->{url} . '/out_doc', |
847 |
'application/x-www-form-urlencoded', |
'application/x-www-form-urlencoded', |
848 |
"id=$id", |
"id=$id", |
855 |
|
|
856 |
Remove a registrated document using it's uri |
Remove a registrated document using it's uri |
857 |
|
|
858 |
$node->out_doc_by_uri( 'file:///document_url' ) or "can't remove document"; |
$node->out_doc_by_uri( 'file:///document/uri/42' ) or "can't remove document"; |
859 |
|
|
860 |
Return true on success or false on failture. |
Return true on success or false on failture. |
861 |
|
|
867 |
return unless ($self->{url}); |
return unless ($self->{url}); |
868 |
$self->shuttle_url( $self->{url} . '/out_doc', |
$self->shuttle_url( $self->{url} . '/out_doc', |
869 |
'application/x-www-form-urlencoded', |
'application/x-www-form-urlencoded', |
870 |
"uri=$uri", |
"uri=" . uri_escape($uri), |
871 |
undef |
undef |
872 |
) == 200; |
) == 200; |
873 |
} |
} |
886 |
sub edit_doc { |
sub edit_doc { |
887 |
my $self = shift; |
my $self = shift; |
888 |
my $doc = shift || return; |
my $doc = shift || return; |
889 |
return unless ($self->{url}); |
return unless ($self->{url} && $doc->isa('Search::Estraier::Document')); |
890 |
$self->shuttle_url( $self->{url} . '/edit_doc', |
$self->shuttle_url( $self->{url} . '/edit_doc', |
891 |
'text/x-estraier-draft', |
'text/x-estraier-draft', |
892 |
$doc->dump_draft, |
$doc->dump_draft, |
895 |
} |
} |
896 |
|
|
897 |
|
|
898 |
|
=head2 get_doc |
899 |
|
|
900 |
|
Retreive document |
901 |
|
|
902 |
|
my $doc = $node->get_doc( document_id ) or die "can't get document"; |
903 |
|
|
904 |
|
Return true on success or false on failture. |
905 |
|
|
906 |
|
=cut |
907 |
|
|
908 |
|
sub get_doc { |
909 |
|
my $self = shift; |
910 |
|
my $id = shift || return; |
911 |
|
return $self->_fetch_doc( id => $id ); |
912 |
|
} |
913 |
|
|
914 |
|
|
915 |
|
=head2 get_doc_by_uri |
916 |
|
|
917 |
|
Retreive document |
918 |
|
|
919 |
|
my $doc = $node->get_doc_by_uri( 'file:///document/uri/42' ) or die "can't get document"; |
920 |
|
|
921 |
|
Return true on success or false on failture. |
922 |
|
|
923 |
|
=cut |
924 |
|
|
925 |
|
sub get_doc_by_uri { |
926 |
|
my $self = shift; |
927 |
|
my $uri = shift || return; |
928 |
|
return $self->_fetch_doc( uri => $uri ); |
929 |
|
} |
930 |
|
|
931 |
|
|
932 |
|
=head2 get_doc_attr |
933 |
|
|
934 |
|
Retrieve the value of an atribute from object |
935 |
|
|
936 |
|
my $val = $node->get_doc_attr( document_id, 'attribute_name' ) or |
937 |
|
die "can't get document attribute"; |
938 |
|
|
939 |
|
=cut |
940 |
|
|
941 |
|
sub get_doc_attr { |
942 |
|
my $self = shift; |
943 |
|
my ($id,$name) = @_; |
944 |
|
return unless ($id && $name); |
945 |
|
return $self->_fetch_doc( id => $id, attr => $name ); |
946 |
|
} |
947 |
|
|
948 |
|
|
949 |
|
=head2 get_doc_attr_by_uri |
950 |
|
|
951 |
|
Retrieve the value of an atribute from object |
952 |
|
|
953 |
|
my $val = $node->get_doc_attr_by_uri( document_id, 'attribute_name' ) or |
954 |
|
die "can't get document attribute"; |
955 |
|
|
956 |
|
=cut |
957 |
|
|
958 |
|
sub get_doc_attr_by_uri { |
959 |
|
my $self = shift; |
960 |
|
my ($uri,$name) = @_; |
961 |
|
return unless ($uri && $name); |
962 |
|
return $self->_fetch_doc( uri => $uri, attr => $name ); |
963 |
|
} |
964 |
|
|
965 |
|
|
966 |
|
=head2 etch_doc |
967 |
|
|
968 |
|
Exctract document keywords |
969 |
|
|
970 |
|
my $keywords = $node->etch_doc( document_id ) or die "can't etch document"; |
971 |
|
|
972 |
|
=cut |
973 |
|
|
974 |
|
sub etch_doc { |
975 |
|
my $self = shift; |
976 |
|
my $id = shift || return; |
977 |
|
return $self->_fetch_doc( id => $id, etch => 1 ); |
978 |
|
} |
979 |
|
|
980 |
|
=head2 etch_doc_by_uri |
981 |
|
|
982 |
|
Retreive document |
983 |
|
|
984 |
|
my $keywords = $node->etch_doc_by_uri( 'file:///document/uri/42' ) or die "can't etch document"; |
985 |
|
|
986 |
|
Return true on success or false on failture. |
987 |
|
|
988 |
|
=cut |
989 |
|
|
990 |
|
sub etch_doc_by_uri { |
991 |
|
my $self = shift; |
992 |
|
my $uri = shift || return; |
993 |
|
return $self->_fetch_doc( uri => $uri, etch => 1 ); |
994 |
|
} |
995 |
|
|
996 |
|
|
997 |
|
=head2 uri_to_id |
998 |
|
|
999 |
|
Get ID of document specified by URI |
1000 |
|
|
1001 |
|
my $id = $node->uri_to_id( 'file:///document/uri/42' ); |
1002 |
|
|
1003 |
|
=cut |
1004 |
|
|
1005 |
|
sub uri_to_id { |
1006 |
|
my $self = shift; |
1007 |
|
my $uri = shift || return; |
1008 |
|
return $self->_fetch_doc( uri => $uri, path => '/uri_to_id', chomp_resbody => 1 ); |
1009 |
|
} |
1010 |
|
|
1011 |
|
|
1012 |
|
=head2 _fetch_doc |
1013 |
|
|
1014 |
|
Private function used for implementing of C<get_doc>, C<get_doc_by_uri>, |
1015 |
|
C<etch_doc>, C<etch_doc_by_uri>. |
1016 |
|
|
1017 |
|
# this will decode received draft into Search::Estraier::Document object |
1018 |
|
my $doc = $node->_fetch_doc( id => 42 ); |
1019 |
|
my $doc = $node->_fetch_doc( uri => 'file:///document/uri/42' ); |
1020 |
|
|
1021 |
|
# to extract keywords, add etch |
1022 |
|
my $doc = $node->_fetch_doc( id => 42, etch => 1 ); |
1023 |
|
my $doc = $node->_fetch_doc( uri => 'file:///document/uri/42', etch => 1 ); |
1024 |
|
|
1025 |
|
# to get document attrubute add attr |
1026 |
|
my $doc = $node->_fetch_doc( id => 42, attr => '@mdate' ); |
1027 |
|
my $doc = $node->_fetch_doc( uri => 'file:///document/uri/42', attr => '@mdate' ); |
1028 |
|
|
1029 |
|
# more general form which allows implementation of |
1030 |
|
# uri_to_id |
1031 |
|
my $id = $node->_fetch_doc( |
1032 |
|
uri => 'file:///document/uri/42', |
1033 |
|
path => '/uri_to_id', |
1034 |
|
chomp_resbody => 1 |
1035 |
|
); |
1036 |
|
|
1037 |
|
=cut |
1038 |
|
|
1039 |
|
sub _fetch_doc { |
1040 |
|
my $self = shift; |
1041 |
|
my $a = {@_}; |
1042 |
|
return unless ( ($a->{id} || $a->{uri}) && $self->{url} ); |
1043 |
|
|
1044 |
|
my ($arg, $resbody); |
1045 |
|
|
1046 |
|
my $path = $a->{path} || '/get_doc'; |
1047 |
|
$path = '/etch_doc' if ($a->{etch}); |
1048 |
|
|
1049 |
|
if ($a->{id}) { |
1050 |
|
croak "id must be numberm not '$a->{id}'" unless ($a->{id} =~ m/^\d+$/); |
1051 |
|
$arg = 'id=' . $a->{id}; |
1052 |
|
} elsif ($a->{uri}) { |
1053 |
|
$arg = 'uri=' . uri_escape($a->{uri}); |
1054 |
|
} else { |
1055 |
|
confess "unhandled argument. Need id or uri."; |
1056 |
|
} |
1057 |
|
|
1058 |
|
if ($a->{attr}) { |
1059 |
|
$path = '/get_doc_attr'; |
1060 |
|
$arg .= '&attr=' . uri_escape($a->{attr}); |
1061 |
|
$a->{chomp_resbody} = 1; |
1062 |
|
} |
1063 |
|
|
1064 |
|
my $rv = $self->shuttle_url( $self->{url} . $path, |
1065 |
|
'application/x-www-form-urlencoded', |
1066 |
|
$arg, |
1067 |
|
\$resbody, |
1068 |
|
); |
1069 |
|
|
1070 |
|
return if ($rv != 200); |
1071 |
|
|
1072 |
|
if ($a->{etch}) { |
1073 |
|
$self->{kwords} = {}; |
1074 |
|
return +{} unless ($resbody); |
1075 |
|
foreach my $l (split(/\n/, $resbody)) { |
1076 |
|
my ($k,$v) = split(/\t/, $l, 2); |
1077 |
|
$self->{kwords}->{$k} = $v if ($v); |
1078 |
|
} |
1079 |
|
return $self->{kwords}; |
1080 |
|
} elsif ($a->{chomp_resbody}) { |
1081 |
|
return unless (defined($resbody)); |
1082 |
|
chomp($resbody); |
1083 |
|
return $resbody; |
1084 |
|
} else { |
1085 |
|
return new Search::Estraier::Document($resbody); |
1086 |
|
} |
1087 |
|
} |
1088 |
|
|
1089 |
|
|
1090 |
|
=head2 name |
1091 |
|
|
1092 |
|
my $node_name = $node->name; |
1093 |
|
|
1094 |
|
=cut |
1095 |
|
|
1096 |
|
sub name { |
1097 |
|
my $self = shift; |
1098 |
|
$self->set_info unless ($self->{name}); |
1099 |
|
return $self->{name}; |
1100 |
|
} |
1101 |
|
|
1102 |
|
|
1103 |
|
=head2 label |
1104 |
|
|
1105 |
|
my $node_label = $node->label; |
1106 |
|
|
1107 |
|
=cut |
1108 |
|
|
1109 |
|
sub label { |
1110 |
|
my $self = shift; |
1111 |
|
$self->set_info unless ($self->{label}); |
1112 |
|
return $self->{label}; |
1113 |
|
} |
1114 |
|
|
1115 |
|
|
1116 |
|
=head2 doc_num |
1117 |
|
|
1118 |
|
my $documents_in_node = $node->doc_num; |
1119 |
|
|
1120 |
|
=cut |
1121 |
|
|
1122 |
|
sub doc_num { |
1123 |
|
my $self = shift; |
1124 |
|
$self->set_info if ($self->{dnum} < 0); |
1125 |
|
return $self->{dnum}; |
1126 |
|
} |
1127 |
|
|
1128 |
|
|
1129 |
|
=head2 word_num |
1130 |
|
|
1131 |
|
my $words_in_node = $node->word_num; |
1132 |
|
|
1133 |
|
=cut |
1134 |
|
|
1135 |
|
sub word_num { |
1136 |
|
my $self = shift; |
1137 |
|
$self->set_info if ($self->{wnum} < 0); |
1138 |
|
return $self->{wnum}; |
1139 |
|
} |
1140 |
|
|
1141 |
|
|
1142 |
|
=head2 size |
1143 |
|
|
1144 |
|
my $node_size = $node->size; |
1145 |
|
|
1146 |
|
=cut |
1147 |
|
|
1148 |
|
sub size { |
1149 |
|
my $self = shift; |
1150 |
|
$self->set_info if ($self->{size} < 0); |
1151 |
|
return $self->{size}; |
1152 |
|
} |
1153 |
|
|
1154 |
|
|
1155 |
|
=head2 search |
1156 |
|
|
1157 |
|
Search documents which match condition |
1158 |
|
|
1159 |
|
my $nres = $node->search( $cond, $depth ); |
1160 |
|
|
1161 |
|
C<$cond> is C<Search::Estraier::Condition> object, while <$depth> specifies |
1162 |
|
depth for meta search. |
1163 |
|
|
1164 |
|
Function results C<Search::Estraier::NodeResult> object. |
1165 |
|
|
1166 |
|
=cut |
1167 |
|
|
1168 |
|
sub search { |
1169 |
|
my $self = shift; |
1170 |
|
my ($cond, $depth) = @_; |
1171 |
|
return unless ($cond && defined($depth) && $self->{url}); |
1172 |
|
croak "cond mush be Search::Estraier::Condition, not '$cond->isa'" unless ($cond->isa('Search::Estraier::Condition')); |
1173 |
|
croak "depth needs number, not '$depth'" unless ($depth =~ m/^\d+$/); |
1174 |
|
|
1175 |
|
|
1176 |
|
} |
1177 |
|
|
1178 |
|
|
1179 |
|
=head2 cond_to_query |
1180 |
|
|
1181 |
|
my $args = $node->cond_to_query( $cond ); |
1182 |
|
|
1183 |
|
=cut |
1184 |
|
|
1185 |
|
sub cond_to_query { |
1186 |
|
my $self = shift; |
1187 |
|
|
1188 |
|
my $cond = shift || return; |
1189 |
|
croak "condition must be Search::Estraier::Condition, not '$cond->isa'" unless ($cond->isa('Search::Estraier::Condition')); |
1190 |
|
|
1191 |
|
my @args; |
1192 |
|
|
1193 |
|
if (my $phrase = $cond->phrase) { |
1194 |
|
push @args, 'phrase=' . uri_escape($phrase); |
1195 |
|
} |
1196 |
|
|
1197 |
|
if (my @attrs = $cond->attrs) { |
1198 |
|
for my $i ( 0 .. $#attrs ) { |
1199 |
|
push @args,'attr' . ($i+1) . '=' . uri_escape( $attrs[$i] ); |
1200 |
|
} |
1201 |
|
} |
1202 |
|
|
1203 |
|
if (my $order = $cond->order) { |
1204 |
|
push @args, 'order=' . uri_escape($order); |
1205 |
|
} |
1206 |
|
|
1207 |
|
if (my $max = $cond->max) { |
1208 |
|
push @args, 'max=' . $max; |
1209 |
|
} else { |
1210 |
|
push @args, 'max=' . (1 << 30); |
1211 |
|
} |
1212 |
|
|
1213 |
|
if (my $options = $cond->options) { |
1214 |
|
push @args, 'options=' . $options; |
1215 |
|
} |
1216 |
|
|
1217 |
|
push @args, 'depth=' . $self->{depth} if ($self->{depth}); |
1218 |
|
push @args, 'wwidth=' . $self->{wwidth}; |
1219 |
|
push @args, 'hwidth=' . $self->{hwidth}; |
1220 |
|
push @args, 'awidth=' . $self->{awidth}; |
1221 |
|
|
1222 |
|
return join('&', @args); |
1223 |
|
} |
1224 |
|
|
1225 |
|
|
1226 |
=head2 shuttle_url |
=head2 shuttle_url |
1227 |
|
|
1228 |
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 |
1334 |
return $self->{status}; |
return $self->{status}; |
1335 |
} |
} |
1336 |
|
|
1337 |
|
|
1338 |
|
=head2 set_info |
1339 |
|
|
1340 |
|
Set information for node |
1341 |
|
|
1342 |
|
$node->set_info; |
1343 |
|
|
1344 |
|
=cut |
1345 |
|
|
1346 |
|
sub set_info { |
1347 |
|
my $self = shift; |
1348 |
|
|
1349 |
|
$self->{status} = -1; |
1350 |
|
return unless ($self->{url}); |
1351 |
|
|
1352 |
|
my $resbody; |
1353 |
|
my $rv = $self->shuttle_url( $self->{url} . '/inform', |
1354 |
|
'text/plain', |
1355 |
|
undef, |
1356 |
|
\$resbody, |
1357 |
|
); |
1358 |
|
|
1359 |
|
return if ($rv != 200 || !$resbody); |
1360 |
|
|
1361 |
|
chomp($resbody); |
1362 |
|
|
1363 |
|
( $self->{name}, $self->{label}, $self->{dnum}, $self->{wnum}, $self->{size} ) = |
1364 |
|
split(/\t/, $resbody, 5); |
1365 |
|
|
1366 |
|
} |
1367 |
|
|
1368 |
### |
### |
1369 |
|
|
1370 |
=head1 EXPORT |
=head1 EXPORT |