645 |
|
|
646 |
sub doc_num { |
sub doc_num { |
647 |
my $self = shift; |
my $self = shift; |
648 |
return $#{$self->{docs}}; |
return $#{$self->{docs}} + 1; |
649 |
} |
} |
650 |
|
|
651 |
|
|
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($uri), |
871 |
undef |
undef |
872 |
) == 200; |
) == 200; |
873 |
} |
} |
1050 |
croak "id must be numberm not '$a->{id}'" unless ($a->{id} =~ m/^\d+$/); |
croak "id must be numberm not '$a->{id}'" unless ($a->{id} =~ m/^\d+$/); |
1051 |
$arg = 'id=' . $a->{id}; |
$arg = 'id=' . $a->{id}; |
1052 |
} elsif ($a->{uri}) { |
} elsif ($a->{uri}) { |
1053 |
$arg = 'uri=' . $a->{uri}; |
$arg = 'uri=' . uri_escape($a->{uri}); |
1054 |
} else { |
} else { |
1055 |
confess "unhandled argument. Need id or uri."; |
confess "unhandled argument. Need id or uri."; |
1056 |
} |
} |
1095 |
|
|
1096 |
sub name { |
sub name { |
1097 |
my $self = shift; |
my $self = shift; |
1098 |
$self->set_info unless ($self->{name}); |
$self->_set_info unless ($self->{name}); |
1099 |
return $self->{name}; |
return $self->{name}; |
1100 |
} |
} |
1101 |
|
|
1108 |
|
|
1109 |
sub label { |
sub label { |
1110 |
my $self = shift; |
my $self = shift; |
1111 |
$self->set_info unless ($self->{label}); |
$self->_set_info unless ($self->{label}); |
1112 |
return $self->{label}; |
return $self->{label}; |
1113 |
} |
} |
1114 |
|
|
1121 |
|
|
1122 |
sub doc_num { |
sub doc_num { |
1123 |
my $self = shift; |
my $self = shift; |
1124 |
$self->set_info if ($self->{dnum} < 0); |
$self->_set_info if ($self->{dnum} < 0); |
1125 |
return $self->{dnum}; |
return $self->{dnum}; |
1126 |
} |
} |
1127 |
|
|
1134 |
|
|
1135 |
sub word_num { |
sub word_num { |
1136 |
my $self = shift; |
my $self = shift; |
1137 |
$self->set_info if ($self->{wnum} < 0); |
$self->_set_info if ($self->{wnum} < 0); |
1138 |
return $self->{wnum}; |
return $self->{wnum}; |
1139 |
} |
} |
1140 |
|
|
1147 |
|
|
1148 |
sub size { |
sub size { |
1149 |
my $self = shift; |
my $self = shift; |
1150 |
$self->set_info if ($self->{size} < 0); |
$self->_set_info if ($self->{size} < 0); |
1151 |
return $self->{size}; |
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 |
|
my $resbody; |
1176 |
|
|
1177 |
|
my $rv = $self->shuttle_url( $self->{url} . '/search', |
1178 |
|
'application/x-www-form-urlencoded', |
1179 |
|
$self->cond_to_query( $cond ), |
1180 |
|
\$resbody, |
1181 |
|
); |
1182 |
|
return if ($rv != 200); |
1183 |
|
|
1184 |
|
my (@docs, $hints); |
1185 |
|
|
1186 |
|
my @lines = split(/\n/, $resbody); |
1187 |
|
return unless (@lines); |
1188 |
|
|
1189 |
|
my $border = $lines[0]; |
1190 |
|
my $isend = 0; |
1191 |
|
my $lnum = 1; |
1192 |
|
|
1193 |
|
while ( $lnum <= $#lines ) { |
1194 |
|
my $line = $lines[$lnum]; |
1195 |
|
$lnum++; |
1196 |
|
|
1197 |
|
#warn "## $line\n"; |
1198 |
|
if ($line && $line =~ m/^\Q$border\E(:END)*$/) { |
1199 |
|
$isend = $1; |
1200 |
|
last; |
1201 |
|
} |
1202 |
|
|
1203 |
|
if ($line =~ /\t/) { |
1204 |
|
my ($k,$v) = split(/\t/, $line, 2); |
1205 |
|
$hints->{$k} = $v; |
1206 |
|
} |
1207 |
|
} |
1208 |
|
|
1209 |
|
my $snum = $lnum; |
1210 |
|
|
1211 |
|
while( ! $isend && $lnum <= $#lines ) { |
1212 |
|
my $line = $lines[$lnum]; |
1213 |
|
#warn "# $lnum: $line\n"; |
1214 |
|
$lnum++; |
1215 |
|
|
1216 |
|
if ($line && $line =~ m/^\Q$border\E/) { |
1217 |
|
if ($lnum > $snum) { |
1218 |
|
my $rdattrs; |
1219 |
|
my $rdvector; |
1220 |
|
my $rdsnippet; |
1221 |
|
|
1222 |
|
my $rlnum = $snum; |
1223 |
|
while ($rlnum < $lnum - 1 ) { |
1224 |
|
#my $rdline = $self->_s($lines[$rlnum]); |
1225 |
|
my $rdline = $lines[$rlnum]; |
1226 |
|
$rlnum++; |
1227 |
|
last unless ($rdline); |
1228 |
|
if ($rdline =~ /^%/) { |
1229 |
|
$rdvector = $1 if ($rdline =~ /^%VECTOR\t(.+)$/); |
1230 |
|
} elsif($rdline =~ /=/) { |
1231 |
|
$rdattrs->{$1} = $2 if ($rdline =~ /^(.+)=(.+)$/); |
1232 |
|
} else { |
1233 |
|
confess "invalid format of response"; |
1234 |
|
} |
1235 |
|
} |
1236 |
|
while($rlnum < $lnum - 1) { |
1237 |
|
my $rdline = $lines[$rlnum]; |
1238 |
|
$rlnum++; |
1239 |
|
$rdsnippet .= "$rdline\n"; |
1240 |
|
} |
1241 |
|
#warn Dumper($rdvector, $rdattrs, $rdsnippet); |
1242 |
|
if (my $rduri = $rdattrs->{'@uri'}) { |
1243 |
|
push @docs, new Search::Estraier::ResultDocument( |
1244 |
|
uri => $rduri, |
1245 |
|
attrs => $rdattrs, |
1246 |
|
snippet => $rdsnippet, |
1247 |
|
keywords => $rdvector, |
1248 |
|
); |
1249 |
|
} |
1250 |
|
} |
1251 |
|
$snum = $lnum; |
1252 |
|
#warn "### $line\n"; |
1253 |
|
$isend = 1 if ($line =~ /:END$/); |
1254 |
|
} |
1255 |
|
|
1256 |
|
} |
1257 |
|
|
1258 |
|
if (! $isend) { |
1259 |
|
warn "received result doesn't have :END\n$resbody"; |
1260 |
|
return; |
1261 |
|
} |
1262 |
|
|
1263 |
|
#warn Dumper(\@docs, $hints); |
1264 |
|
|
1265 |
|
return new Search::Estraier::NodeResult( docs => \@docs, hints => $hints ); |
1266 |
|
} |
1267 |
|
|
1268 |
|
|
1269 |
|
=head2 cond_to_query |
1270 |
|
|
1271 |
|
Return URI encoded string generated from Search::Estraier::Condition |
1272 |
|
|
1273 |
|
my $args = $node->cond_to_query( $cond ); |
1274 |
|
|
1275 |
|
=cut |
1276 |
|
|
1277 |
|
sub cond_to_query { |
1278 |
|
my $self = shift; |
1279 |
|
|
1280 |
|
my $cond = shift || return; |
1281 |
|
croak "condition must be Search::Estraier::Condition, not '$cond->isa'" unless ($cond->isa('Search::Estraier::Condition')); |
1282 |
|
|
1283 |
|
my @args; |
1284 |
|
|
1285 |
|
if (my $phrase = $cond->phrase) { |
1286 |
|
push @args, 'phrase=' . uri_escape($phrase); |
1287 |
|
} |
1288 |
|
|
1289 |
|
if (my @attrs = $cond->attrs) { |
1290 |
|
for my $i ( 0 .. $#attrs ) { |
1291 |
|
push @args,'attr' . ($i+1) . '=' . uri_escape( $attrs[$i] ); |
1292 |
|
} |
1293 |
|
} |
1294 |
|
|
1295 |
|
if (my $order = $cond->order) { |
1296 |
|
push @args, 'order=' . uri_escape($order); |
1297 |
|
} |
1298 |
|
|
1299 |
|
if (my $max = $cond->max) { |
1300 |
|
push @args, 'max=' . $max; |
1301 |
|
} else { |
1302 |
|
push @args, 'max=' . (1 << 30); |
1303 |
|
} |
1304 |
|
|
1305 |
|
if (my $options = $cond->options) { |
1306 |
|
push @args, 'options=' . $options; |
1307 |
|
} |
1308 |
|
|
1309 |
|
push @args, 'depth=' . $self->{depth} if ($self->{depth}); |
1310 |
|
push @args, 'wwidth=' . $self->{wwidth}; |
1311 |
|
push @args, 'hwidth=' . $self->{hwidth}; |
1312 |
|
push @args, 'awidth=' . $self->{awidth}; |
1313 |
|
|
1314 |
|
return join('&', @args); |
1315 |
|
} |
1316 |
|
|
1317 |
|
|
1318 |
=head2 shuttle_url |
=head2 shuttle_url |
1319 |
|
|
1320 |
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 |
1321 |
master. |
master. |
1322 |
|
|
1323 |
my $rv = shuttle_url( $url, $content_type, \$req_body, \$resbody ); |
my $rv = shuttle_url( $url, $content_type, $req_body, \$resbody ); |
1324 |
|
|
1325 |
C<$resheads> and C<$resbody> booleans controll if response headers and/or response |
C<$resheads> and C<$resbody> booleans controll if response headers and/or response |
1326 |
body will be saved within object. |
body will be saved within object. |
1427 |
} |
} |
1428 |
|
|
1429 |
|
|
1430 |
=head2 set_info |
=head2 set_snippet_width |
1431 |
|
|
1432 |
|
Set width of snippets in results |
1433 |
|
|
1434 |
|
$node->set_snippet_width( $wwidth, $hwidth, $awidth ); |
1435 |
|
|
1436 |
|
C<$wwidth> specifies whole width of snippet. It's C<480> by default. If it's C<0> snippet |
1437 |
|
is not sent with results. If it is negative, whole document text is sent instead of snippet. |
1438 |
|
|
1439 |
|
C<$hwidth> specified width of strings from beginning of string. Default |
1440 |
|
value is C<96>. Negative or zero value keep previous value. |
1441 |
|
|
1442 |
|
C<$awidth> specifies width of strings around each highlighted word. It's C<96> by default. |
1443 |
|
If negative of zero value is provided previous value is kept unchanged. |
1444 |
|
|
1445 |
|
=cut |
1446 |
|
|
1447 |
|
sub set_snippet_width { |
1448 |
|
my $self = shift; |
1449 |
|
|
1450 |
|
my ($wwidth, $hwidth, $awidth) = @_; |
1451 |
|
$self->{wwidth} = $wwidth; |
1452 |
|
$self->{hwidth} = $hwidth if ($hwidth >= 0); |
1453 |
|
$self->{awidth} = $awidth if ($awidth >= 0); |
1454 |
|
} |
1455 |
|
|
1456 |
|
|
1457 |
|
|
1458 |
|
=head1 PRIVATE METHODS |
1459 |
|
|
1460 |
|
You could call those directly, but you don't have to. I hope. |
1461 |
|
|
1462 |
|
=head2 _set_info |
1463 |
|
|
1464 |
Set information for node |
Set information for node |
1465 |
|
|
1466 |
$node->set_info; |
$node->_set_info; |
1467 |
|
|
1468 |
=cut |
=cut |
1469 |
|
|
1470 |
sub set_info { |
sub _set_info { |
1471 |
my $self = shift; |
my $self = shift; |
1472 |
|
|
1473 |
$self->{status} = -1; |
$self->{status} = -1; |