4 |
use strict; |
use strict; |
5 |
use warnings; |
use warnings; |
6 |
|
|
7 |
our $VERSION = '0.00'; |
our $VERSION = '0.01'; |
8 |
|
|
9 |
=head1 NAME |
=head1 NAME |
10 |
|
|
205 |
|
|
206 |
sub attr_names { |
sub attr_names { |
207 |
my $self = shift; |
my $self = shift; |
208 |
croak "attr_names return array, not scalar" if (! wantarray); |
return unless ($self->{attrs}); |
209 |
|
#croak "attr_names return array, not scalar" if (! wantarray); |
210 |
return sort keys %{ $self->{attrs} }; |
return sort keys %{ $self->{attrs} }; |
211 |
} |
} |
212 |
|
|
222 |
sub attr { |
sub attr { |
223 |
my $self = shift; |
my $self = shift; |
224 |
my $name = shift; |
my $name = shift; |
225 |
|
return unless (defined($name) && $self->{attrs}); |
226 |
return $self->{'attrs'}->{ $name }; |
return $self->{attrs}->{ $name }; |
227 |
} |
} |
228 |
|
|
229 |
|
|
237 |
|
|
238 |
sub texts { |
sub texts { |
239 |
my $self = shift; |
my $self = shift; |
240 |
confess "texts return array, not scalar" if (! wantarray); |
#confess "texts return array, not scalar" if (! wantarray); |
241 |
return @{ $self->{dtexts} }; |
return @{ $self->{dtexts} } if ($self->{dtexts}); |
242 |
} |
} |
243 |
|
|
244 |
|
|
252 |
|
|
253 |
sub cat_texts { |
sub cat_texts { |
254 |
my $self = shift; |
my $self = shift; |
255 |
return join(' ',@{ $self->{dtexts} }); |
return join(' ',@{ $self->{dtexts} }) if ($self->{dtexts}); |
256 |
} |
} |
257 |
|
|
258 |
|
|
461 |
sub attrs { |
sub attrs { |
462 |
my $self = shift; |
my $self = shift; |
463 |
#croak "attrs return array, not scalar" if (! wantarray); |
#croak "attrs return array, not scalar" if (! wantarray); |
464 |
return @{ $self->{attrs} }; |
return @{ $self->{attrs} } if ($self->{attrs}); |
465 |
} |
} |
466 |
|
|
467 |
|
|
646 |
|
|
647 |
sub doc_num { |
sub doc_num { |
648 |
my $self = shift; |
my $self = shift; |
649 |
return $#{$self->{docs}}; |
return $#{$self->{docs}} + 1; |
650 |
} |
} |
651 |
|
|
652 |
|
|
718 |
}; |
}; |
719 |
bless($self, $class); |
bless($self, $class); |
720 |
|
|
721 |
if (@_) { |
my $args = {@_}; |
722 |
$self->{debug} = shift; |
|
723 |
warn "## Node debug on\n"; |
$self->{debug} = $args->{debug}; |
724 |
} |
warn "## Node debug on\n" if ($self->{debug}); |
725 |
|
|
726 |
$self ? return $self : return undef; |
$self ? return $self : return undef; |
727 |
} |
} |
868 |
return unless ($self->{url}); |
return unless ($self->{url}); |
869 |
$self->shuttle_url( $self->{url} . '/out_doc', |
$self->shuttle_url( $self->{url} . '/out_doc', |
870 |
'application/x-www-form-urlencoded', |
'application/x-www-form-urlencoded', |
871 |
"uri=$uri", |
"uri=" . uri_escape($uri), |
872 |
undef |
undef |
873 |
) == 200; |
) == 200; |
874 |
} |
} |
1051 |
croak "id must be numberm not '$a->{id}'" unless ($a->{id} =~ m/^\d+$/); |
croak "id must be numberm not '$a->{id}'" unless ($a->{id} =~ m/^\d+$/); |
1052 |
$arg = 'id=' . $a->{id}; |
$arg = 'id=' . $a->{id}; |
1053 |
} elsif ($a->{uri}) { |
} elsif ($a->{uri}) { |
1054 |
$arg = 'uri=' . $a->{uri}; |
$arg = 'uri=' . uri_escape($a->{uri}); |
1055 |
} else { |
} else { |
1056 |
confess "unhandled argument. Need id or uri."; |
confess "unhandled argument. Need id or uri."; |
1057 |
} |
} |
1096 |
|
|
1097 |
sub name { |
sub name { |
1098 |
my $self = shift; |
my $self = shift; |
1099 |
$self->set_info unless ($self->{name}); |
$self->_set_info unless ($self->{name}); |
1100 |
return $self->{name}; |
return $self->{name}; |
1101 |
} |
} |
1102 |
|
|
1109 |
|
|
1110 |
sub label { |
sub label { |
1111 |
my $self = shift; |
my $self = shift; |
1112 |
$self->set_info unless ($self->{label}); |
$self->_set_info unless ($self->{label}); |
1113 |
return $self->{label}; |
return $self->{label}; |
1114 |
} |
} |
1115 |
|
|
1122 |
|
|
1123 |
sub doc_num { |
sub doc_num { |
1124 |
my $self = shift; |
my $self = shift; |
1125 |
$self->set_info if ($self->{dnum} < 0); |
$self->_set_info if ($self->{dnum} < 0); |
1126 |
return $self->{dnum}; |
return $self->{dnum}; |
1127 |
} |
} |
1128 |
|
|
1135 |
|
|
1136 |
sub word_num { |
sub word_num { |
1137 |
my $self = shift; |
my $self = shift; |
1138 |
$self->set_info if ($self->{wnum} < 0); |
$self->_set_info if ($self->{wnum} < 0); |
1139 |
return $self->{wnum}; |
return $self->{wnum}; |
1140 |
} |
} |
1141 |
|
|
1148 |
|
|
1149 |
sub size { |
sub size { |
1150 |
my $self = shift; |
my $self = shift; |
1151 |
$self->set_info if ($self->{size} < 0); |
$self->_set_info if ($self->{size} < 0); |
1152 |
return $self->{size}; |
return $self->{size}; |
1153 |
} |
} |
1154 |
|
|
1155 |
|
|
1156 |
|
=head2 search |
1157 |
|
|
1158 |
|
Search documents which match condition |
1159 |
|
|
1160 |
|
my $nres = $node->search( $cond, $depth ); |
1161 |
|
|
1162 |
|
C<$cond> is C<Search::Estraier::Condition> object, while <$depth> specifies |
1163 |
|
depth for meta search. |
1164 |
|
|
1165 |
|
Function results C<Search::Estraier::NodeResult> object. |
1166 |
|
|
1167 |
|
=cut |
1168 |
|
|
1169 |
|
sub search { |
1170 |
|
my $self = shift; |
1171 |
|
my ($cond, $depth) = @_; |
1172 |
|
return unless ($cond && defined($depth) && $self->{url}); |
1173 |
|
croak "cond mush be Search::Estraier::Condition, not '$cond->isa'" unless ($cond->isa('Search::Estraier::Condition')); |
1174 |
|
croak "depth needs number, not '$depth'" unless ($depth =~ m/^\d+$/); |
1175 |
|
|
1176 |
|
my $resbody; |
1177 |
|
|
1178 |
|
my $rv = $self->shuttle_url( $self->{url} . '/search', |
1179 |
|
'application/x-www-form-urlencoded', |
1180 |
|
$self->cond_to_query( $cond ), |
1181 |
|
\$resbody, |
1182 |
|
); |
1183 |
|
return if ($rv != 200); |
1184 |
|
|
1185 |
|
my (@docs, $hints); |
1186 |
|
|
1187 |
|
my @lines = split(/\n/, $resbody); |
1188 |
|
return unless (@lines); |
1189 |
|
|
1190 |
|
my $border = $lines[0]; |
1191 |
|
my $isend = 0; |
1192 |
|
my $lnum = 1; |
1193 |
|
|
1194 |
|
while ( $lnum <= $#lines ) { |
1195 |
|
my $line = $lines[$lnum]; |
1196 |
|
$lnum++; |
1197 |
|
|
1198 |
|
#warn "## $line\n"; |
1199 |
|
if ($line && $line =~ m/^\Q$border\E(:END)*$/) { |
1200 |
|
$isend = $1; |
1201 |
|
last; |
1202 |
|
} |
1203 |
|
|
1204 |
|
if ($line =~ /\t/) { |
1205 |
|
my ($k,$v) = split(/\t/, $line, 2); |
1206 |
|
$hints->{$k} = $v; |
1207 |
|
} |
1208 |
|
} |
1209 |
|
|
1210 |
|
my $snum = $lnum; |
1211 |
|
|
1212 |
|
while( ! $isend && $lnum <= $#lines ) { |
1213 |
|
my $line = $lines[$lnum]; |
1214 |
|
#warn "# $lnum: $line\n"; |
1215 |
|
$lnum++; |
1216 |
|
|
1217 |
|
if ($line && $line =~ m/^\Q$border\E/) { |
1218 |
|
if ($lnum > $snum) { |
1219 |
|
my $rdattrs; |
1220 |
|
my $rdvector; |
1221 |
|
my $rdsnippet; |
1222 |
|
|
1223 |
|
my $rlnum = $snum; |
1224 |
|
while ($rlnum < $lnum - 1 ) { |
1225 |
|
#my $rdline = $self->_s($lines[$rlnum]); |
1226 |
|
my $rdline = $lines[$rlnum]; |
1227 |
|
$rlnum++; |
1228 |
|
last unless ($rdline); |
1229 |
|
if ($rdline =~ /^%/) { |
1230 |
|
$rdvector = $1 if ($rdline =~ /^%VECTOR\t(.+)$/); |
1231 |
|
} elsif($rdline =~ /=/) { |
1232 |
|
$rdattrs->{$1} = $2 if ($rdline =~ /^(.+)=(.+)$/); |
1233 |
|
} else { |
1234 |
|
confess "invalid format of response"; |
1235 |
|
} |
1236 |
|
} |
1237 |
|
while($rlnum < $lnum - 1) { |
1238 |
|
my $rdline = $lines[$rlnum]; |
1239 |
|
$rlnum++; |
1240 |
|
$rdsnippet .= "$rdline\n"; |
1241 |
|
} |
1242 |
|
#warn Dumper($rdvector, $rdattrs, $rdsnippet); |
1243 |
|
if (my $rduri = $rdattrs->{'@uri'}) { |
1244 |
|
push @docs, new Search::Estraier::ResultDocument( |
1245 |
|
uri => $rduri, |
1246 |
|
attrs => $rdattrs, |
1247 |
|
snippet => $rdsnippet, |
1248 |
|
keywords => $rdvector, |
1249 |
|
); |
1250 |
|
} |
1251 |
|
} |
1252 |
|
$snum = $lnum; |
1253 |
|
#warn "### $line\n"; |
1254 |
|
$isend = 1 if ($line =~ /:END$/); |
1255 |
|
} |
1256 |
|
|
1257 |
|
} |
1258 |
|
|
1259 |
|
if (! $isend) { |
1260 |
|
warn "received result doesn't have :END\n$resbody"; |
1261 |
|
return; |
1262 |
|
} |
1263 |
|
|
1264 |
|
#warn Dumper(\@docs, $hints); |
1265 |
|
|
1266 |
|
return new Search::Estraier::NodeResult( docs => \@docs, hints => $hints ); |
1267 |
|
} |
1268 |
|
|
1269 |
|
|
1270 |
|
=head2 cond_to_query |
1271 |
|
|
1272 |
|
Return URI encoded string generated from Search::Estraier::Condition |
1273 |
|
|
1274 |
|
my $args = $node->cond_to_query( $cond ); |
1275 |
|
|
1276 |
|
=cut |
1277 |
|
|
1278 |
|
sub cond_to_query { |
1279 |
|
my $self = shift; |
1280 |
|
|
1281 |
|
my $cond = shift || return; |
1282 |
|
croak "condition must be Search::Estraier::Condition, not '$cond->isa'" unless ($cond->isa('Search::Estraier::Condition')); |
1283 |
|
|
1284 |
|
my @args; |
1285 |
|
|
1286 |
|
if (my $phrase = $cond->phrase) { |
1287 |
|
push @args, 'phrase=' . uri_escape($phrase); |
1288 |
|
} |
1289 |
|
|
1290 |
|
if (my @attrs = $cond->attrs) { |
1291 |
|
for my $i ( 0 .. $#attrs ) { |
1292 |
|
push @args,'attr' . ($i+1) . '=' . uri_escape( $attrs[$i] ); |
1293 |
|
} |
1294 |
|
} |
1295 |
|
|
1296 |
|
if (my $order = $cond->order) { |
1297 |
|
push @args, 'order=' . uri_escape($order); |
1298 |
|
} |
1299 |
|
|
1300 |
|
if (my $max = $cond->max) { |
1301 |
|
push @args, 'max=' . $max; |
1302 |
|
} else { |
1303 |
|
push @args, 'max=' . (1 << 30); |
1304 |
|
} |
1305 |
|
|
1306 |
|
if (my $options = $cond->options) { |
1307 |
|
push @args, 'options=' . $options; |
1308 |
|
} |
1309 |
|
|
1310 |
|
push @args, 'depth=' . $self->{depth} if ($self->{depth}); |
1311 |
|
push @args, 'wwidth=' . $self->{wwidth}; |
1312 |
|
push @args, 'hwidth=' . $self->{hwidth}; |
1313 |
|
push @args, 'awidth=' . $self->{awidth}; |
1314 |
|
|
1315 |
|
return join('&', @args); |
1316 |
|
} |
1317 |
|
|
1318 |
|
|
1319 |
=head2 shuttle_url |
=head2 shuttle_url |
1320 |
|
|
1321 |
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 |
1322 |
master. |
master. |
1323 |
|
|
1324 |
my $rv = shuttle_url( $url, $content_type, \$req_body, \$resbody ); |
my $rv = shuttle_url( $url, $content_type, $req_body, \$resbody ); |
1325 |
|
|
1326 |
C<$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 |
1327 |
body will be saved within object. |
body will be saved within object. |
1328 |
|
|
1329 |
=cut |
=cut |
1330 |
|
|
1331 |
|
use LWP::UserAgent; |
1332 |
|
|
1333 |
sub shuttle_url { |
sub shuttle_url { |
1334 |
my $self = shift; |
my $self = shift; |
1335 |
|
|
1348 |
return -1; |
return -1; |
1349 |
} |
} |
1350 |
|
|
1351 |
my ($host,$port,$query) = ($url->host, $url->port, $url->path); |
my $ua = LWP::UserAgent->new; |
1352 |
|
$ua->agent( "Search-Estraier/$Search::Estraier::VERSION" ); |
|
if ($self->{pxhost}) { |
|
|
($host,$port) = ($self->{pxhost}, $self->{pxport}); |
|
|
$query = "http://$host:$port/$query"; |
|
|
} |
|
|
|
|
|
$query .= '?' . $url->query if ($url->query && ! $reqbody); |
|
|
|
|
|
my $headers; |
|
1353 |
|
|
1354 |
|
my $req; |
1355 |
if ($reqbody) { |
if ($reqbody) { |
1356 |
$headers .= "POST $query HTTP/1.0\r\n"; |
$req = HTTP::Request->new(POST => $url); |
1357 |
} else { |
} else { |
1358 |
$headers .= "GET $query HTTP/1.0\r\n"; |
$req = HTTP::Request->new(GET => $url); |
1359 |
} |
} |
1360 |
|
|
1361 |
$headers .= "Host: " . $url->host . ":" . $url->port . "\r\n"; |
$req->headers->header( 'Host' => $url->host . ":" . $url->port ); |
1362 |
$headers .= "Connection: close\r\n"; |
$req->headers->header( 'Connection', 'close' ); |
1363 |
$headers .= "User-Agent: Search-Estraier/$Search::Estraier::VERSION\r\n"; |
$req->headers->header( 'Authorization', 'Basic ' . $self->{auth} ); |
1364 |
$headers .= "Content-Type: $content_type\r\n"; |
$req->content_type( $content_type ); |
|
$headers .= "Authorization: Basic $self->{auth}\r\n"; |
|
|
my $len = 0; |
|
|
{ |
|
|
use bytes; |
|
|
$len = length($reqbody) if ($reqbody); |
|
|
} |
|
|
$headers .= "Content-Length: $len\r\n"; |
|
|
$headers .= "\r\n"; |
|
|
|
|
|
my $sock = IO::Socket::INET->new( |
|
|
PeerAddr => $host, |
|
|
PeerPort => $port, |
|
|
Proto => 'tcp', |
|
|
Timeout => $self->{timeout} || 90, |
|
|
); |
|
1365 |
|
|
1366 |
if (! $sock) { |
warn $req->headers->as_string,"\n" if ($self->{debug}); |
|
carp "can't open socket to $host:$port"; |
|
|
return -1; |
|
|
} |
|
|
|
|
|
warn $headers if ($self->{debug}); |
|
|
|
|
|
print $sock $headers or |
|
|
carp "can't send headers to network:\n$headers\n" and return -1; |
|
1367 |
|
|
1368 |
if ($reqbody) { |
if ($reqbody) { |
1369 |
warn "$reqbody\n" if ($self->{debug}); |
warn "$reqbody\n" if ($self->{debug}); |
1370 |
print $sock $reqbody or |
$req->content( $reqbody ); |
|
carp "can't send request body to network:\n$$reqbody\n" and return -1; |
|
1371 |
} |
} |
1372 |
|
|
1373 |
my $line = <$sock>; |
my $res = $ua->request($req) || croak "can't make request to $url: $!"; |
1374 |
chomp($line); |
|
1375 |
my ($schema, $res_status, undef) = split(/ */, $line, 3); |
warn "## response status: ",$res->status_line,"\n" if ($self->{debug}); |
|
return if ($schema !~ /^HTTP/ || ! $res_status); |
|
|
|
|
|
$self->{status} = $res_status; |
|
|
warn "## response status: $res_status\n" if ($self->{debug}); |
|
|
|
|
|
# skip rest of headers |
|
|
$line = <$sock>; |
|
|
while ($line) { |
|
|
$line = <$sock>; |
|
|
$line =~ s/[\r\n]+$//; |
|
|
warn "## ", $line || 'NULL', " ##\n" if ($self->{debug}); |
|
|
}; |
|
1376 |
|
|
1377 |
# read body |
return -1 if (! $res->is_success); |
1378 |
$len = 0; |
|
1379 |
do { |
($self->{status}, $self->{status_message}) = split(/\s+/, $res->status_line, 2); |
1380 |
$len = read($sock, my $buf, 8192); |
|
1381 |
$$resbody .= $buf if ($resbody); |
$$resbody .= $res->content; |
|
} while ($len); |
|
1382 |
|
|
1383 |
warn "## response body:\n$$resbody\n" if ($resbody && $self->{debug}); |
warn "## response body:\n$$resbody\n" if ($resbody && $self->{debug}); |
1384 |
|
|
1386 |
} |
} |
1387 |
|
|
1388 |
|
|
1389 |
=head2 set_info |
=head2 set_snippet_width |
1390 |
|
|
1391 |
|
Set width of snippets in results |
1392 |
|
|
1393 |
|
$node->set_snippet_width( $wwidth, $hwidth, $awidth ); |
1394 |
|
|
1395 |
|
C<$wwidth> specifies whole width of snippet. It's C<480> by default. If it's C<0> snippet |
1396 |
|
is not sent with results. If it is negative, whole document text is sent instead of snippet. |
1397 |
|
|
1398 |
|
C<$hwidth> specified width of strings from beginning of string. Default |
1399 |
|
value is C<96>. Negative or zero value keep previous value. |
1400 |
|
|
1401 |
|
C<$awidth> specifies width of strings around each highlighted word. It's C<96> by default. |
1402 |
|
If negative of zero value is provided previous value is kept unchanged. |
1403 |
|
|
1404 |
|
=cut |
1405 |
|
|
1406 |
|
sub set_snippet_width { |
1407 |
|
my $self = shift; |
1408 |
|
|
1409 |
|
my ($wwidth, $hwidth, $awidth) = @_; |
1410 |
|
$self->{wwidth} = $wwidth; |
1411 |
|
$self->{hwidth} = $hwidth if ($hwidth >= 0); |
1412 |
|
$self->{awidth} = $awidth if ($awidth >= 0); |
1413 |
|
} |
1414 |
|
|
1415 |
|
|
1416 |
|
=head2 set_user |
1417 |
|
|
1418 |
|
Manage users of node |
1419 |
|
|
1420 |
|
$node->set_user( 'name', $mode ); |
1421 |
|
|
1422 |
|
C<$mode> can be one of: |
1423 |
|
|
1424 |
|
=over 4 |
1425 |
|
|
1426 |
|
=item 0 |
1427 |
|
|
1428 |
|
delete account |
1429 |
|
|
1430 |
|
=item 1 |
1431 |
|
|
1432 |
|
set administrative right for user |
1433 |
|
|
1434 |
|
=item 2 |
1435 |
|
|
1436 |
|
set user account as guest |
1437 |
|
|
1438 |
|
=back |
1439 |
|
|
1440 |
|
Return true on success, otherwise false. |
1441 |
|
|
1442 |
|
=cut |
1443 |
|
|
1444 |
|
sub set_user { |
1445 |
|
my $self = shift; |
1446 |
|
my ($name, $mode) = @_; |
1447 |
|
|
1448 |
|
return unless ($self->{url}); |
1449 |
|
croak "mode must be number, not '$mode'" unless ($mode =~ m/^\d+$/); |
1450 |
|
|
1451 |
|
$self->shuttle_url( $self->{url} . '/_set_user', |
1452 |
|
'text/plain', |
1453 |
|
'name=' . uri_escape($name) . '&mode=' . $mode, |
1454 |
|
undef |
1455 |
|
) == 200; |
1456 |
|
} |
1457 |
|
|
1458 |
|
|
1459 |
|
=head2 set_link |
1460 |
|
|
1461 |
|
Manage node links |
1462 |
|
|
1463 |
|
$node->set_link('http://localhost:1978/node/another', 'another node label', $credit); |
1464 |
|
|
1465 |
|
If C<$credit> is negative, link is removed. |
1466 |
|
|
1467 |
|
=cut |
1468 |
|
|
1469 |
|
sub set_link { |
1470 |
|
my $self = shift; |
1471 |
|
my ($url, $label, $credit) = @_; |
1472 |
|
|
1473 |
|
return unless ($self->{url}); |
1474 |
|
croak "mode credit be number, not '$credit'" unless ($credit =~ m/^\d+$/); |
1475 |
|
|
1476 |
|
my $reqbody = 'url=' . uri_escape($url) . '&label=' . uri_escape($label); |
1477 |
|
$reqbody .= '&credit=' . $credit if ($credit > 0); |
1478 |
|
|
1479 |
|
$self->shuttle_url( $self->{url} . '/_set_link', |
1480 |
|
'text/plain', |
1481 |
|
$reqbody, |
1482 |
|
undef |
1483 |
|
) == 200; |
1484 |
|
} |
1485 |
|
|
1486 |
|
|
1487 |
|
=head1 PRIVATE METHODS |
1488 |
|
|
1489 |
|
You could call those directly, but you don't have to. I hope. |
1490 |
|
|
1491 |
|
=head2 _set_info |
1492 |
|
|
1493 |
Set information for node |
Set information for node |
1494 |
|
|
1495 |
$node->set_info; |
$node->_set_info; |
1496 |
|
|
1497 |
=cut |
=cut |
1498 |
|
|
1499 |
sub set_info { |
sub _set_info { |
1500 |
my $self = shift; |
my $self = shift; |
1501 |
|
|
1502 |
$self->{status} = -1; |
$self->{status} = -1; |
1511 |
|
|
1512 |
return if ($rv != 200 || !$resbody); |
return if ($rv != 200 || !$resbody); |
1513 |
|
|
1514 |
chomp($resbody); |
# it seems that response can have multiple line endings |
1515 |
|
$resbody =~ s/[\r\n]+$//; |
1516 |
|
|
1517 |
( $self->{name}, $self->{label}, $self->{dnum}, $self->{wnum}, $self->{size} ) = |
( $self->{name}, $self->{label}, $self->{dnum}, $self->{wnum}, $self->{size} ) = |
1518 |
split(/\t/, $resbody, 5); |
split(/\t/, $resbody, 5); |