4 |
use strict; |
use strict; |
5 |
use warnings; |
use warnings; |
6 |
|
|
7 |
our $VERSION = '0.00'; |
our $VERSION = '0.04_1'; |
8 |
|
|
9 |
=head1 NAME |
=head1 NAME |
10 |
|
|
12 |
|
|
13 |
=head1 SYNOPSIS |
=head1 SYNOPSIS |
14 |
|
|
15 |
use Search::Estraier; |
=head2 Simple indexer |
16 |
my $est = new Search::Estraier(); |
|
17 |
|
use Search::Estraier; |
18 |
|
|
19 |
|
# create and configure node |
20 |
|
my $node = new Search::Estraier::Node; |
21 |
|
$node->set_url("http://localhost:1978/node/test"); |
22 |
|
$node->set_auth("admin","admin"); |
23 |
|
|
24 |
|
# create document |
25 |
|
my $doc = new Search::Estraier::Document; |
26 |
|
|
27 |
|
# add attributes |
28 |
|
$doc->add_attr('@uri', "http://estraier.gov/example.txt"); |
29 |
|
$doc->add_attr('@title', "Over the Rainbow"); |
30 |
|
|
31 |
|
# add body text to document |
32 |
|
$doc->add_text("Somewhere over the rainbow. Way up high."); |
33 |
|
$doc->add_text("There's a land that I heard of once in a lullaby."); |
34 |
|
|
35 |
|
die "error: ", $node->status,"\n" unless ($node->put_doc($doc)); |
36 |
|
|
37 |
|
=head2 Simple searcher |
38 |
|
|
39 |
|
use Search::Estraier; |
40 |
|
|
41 |
|
# create and configure node |
42 |
|
my $node = new Search::Estraier::Node; |
43 |
|
$node->set_url("http://localhost:1978/node/test"); |
44 |
|
$node->set_auth("admin","admin"); |
45 |
|
|
46 |
|
# create condition |
47 |
|
my $cond = new Search::Estraier::Condition; |
48 |
|
|
49 |
|
# set search phrase |
50 |
|
$cond->set_phrase("rainbow AND lullaby"); |
51 |
|
|
52 |
|
my $nres = $node->search($cond, 0); |
53 |
|
if (defined($nres)) { |
54 |
|
# for each document in results |
55 |
|
for my $i ( 0 ... $nres->doc_num - 1 ) { |
56 |
|
# get result document |
57 |
|
my $rdoc = $nres->get_doc($i); |
58 |
|
# display attribte |
59 |
|
print "URI: ", $rdoc->attr('@uri'),"\n"; |
60 |
|
print "Title: ", $rdoc->attr('@title'),"\n"; |
61 |
|
print $rdoc->snippet,"\n"; |
62 |
|
} |
63 |
|
} else { |
64 |
|
die "error: ", $node->status,"\n"; |
65 |
|
} |
66 |
|
|
67 |
=head1 DESCRIPTION |
=head1 DESCRIPTION |
68 |
|
|
74 |
It is implemented as multiple packages which closly resamble Ruby |
It is implemented as multiple packages which closly resamble Ruby |
75 |
implementation. It also includes methods to manage nodes. |
implementation. It also includes methods to manage nodes. |
76 |
|
|
77 |
|
There are few examples in C<scripts> directory of this distribution. |
78 |
|
|
79 |
=cut |
=cut |
80 |
|
|
81 |
=head1 Inheritable common methods |
=head1 Inheritable common methods |
157 |
} elsif ($line =~ m/^$/) { |
} elsif ($line =~ m/^$/) { |
158 |
$in_text = 1; |
$in_text = 1; |
159 |
next; |
next; |
160 |
} elsif ($line =~ m/^(.+)=(.+)$/) { |
} elsif ($line =~ m/^(.+)=(.*)$/) { |
161 |
$self->{attrs}->{ $1 } = $2; |
$self->{attrs}->{ $1 } = $2; |
162 |
next; |
next; |
163 |
} |
} |
164 |
|
|
165 |
warn "draft ignored: $line\n"; |
warn "draft ignored: '$line'\n"; |
166 |
} |
} |
167 |
} |
} |
168 |
|
|
256 |
|
|
257 |
sub attr_names { |
sub attr_names { |
258 |
my $self = shift; |
my $self = shift; |
259 |
croak "attr_names return array, not scalar" if (! wantarray); |
return unless ($self->{attrs}); |
260 |
|
#croak "attr_names return array, not scalar" if (! wantarray); |
261 |
return sort keys %{ $self->{attrs} }; |
return sort keys %{ $self->{attrs} }; |
262 |
} |
} |
263 |
|
|
273 |
sub attr { |
sub attr { |
274 |
my $self = shift; |
my $self = shift; |
275 |
my $name = shift; |
my $name = shift; |
276 |
|
return unless (defined($name) && $self->{attrs}); |
277 |
return $self->{'attrs'}->{ $name }; |
return $self->{attrs}->{ $name }; |
278 |
} |
} |
279 |
|
|
280 |
|
|
288 |
|
|
289 |
sub texts { |
sub texts { |
290 |
my $self = shift; |
my $self = shift; |
291 |
confess "texts return array, not scalar" if (! wantarray); |
#confess "texts return array, not scalar" if (! wantarray); |
292 |
return @{ $self->{dtexts} }; |
return @{ $self->{dtexts} } if ($self->{dtexts}); |
293 |
} |
} |
294 |
|
|
295 |
|
|
303 |
|
|
304 |
sub cat_texts { |
sub cat_texts { |
305 |
my $self = shift; |
my $self = shift; |
306 |
return join(' ',@{ $self->{dtexts} }); |
return join(' ',@{ $self->{dtexts} }) if ($self->{dtexts}); |
307 |
} |
} |
308 |
|
|
309 |
|
|
512 |
sub attrs { |
sub attrs { |
513 |
my $self = shift; |
my $self = shift; |
514 |
#croak "attrs return array, not scalar" if (! wantarray); |
#croak "attrs return array, not scalar" if (! wantarray); |
515 |
return @{ $self->{attrs} }; |
return @{ $self->{attrs} } if ($self->{attrs}); |
516 |
} |
} |
517 |
|
|
518 |
|
|
576 |
my $self = {@_}; |
my $self = {@_}; |
577 |
bless($self, $class); |
bless($self, $class); |
578 |
|
|
579 |
foreach my $f (qw/uri attrs snippet keywords/) { |
croak "missing uri for ResultDocument" unless defined($self->{uri}); |
|
croak "missing $f for ResultDocument" unless defined($self->{$f}); |
|
|
} |
|
580 |
|
|
581 |
$self ? return $self : return undef; |
$self ? return $self : return undef; |
582 |
} |
} |
750 |
|
|
751 |
my $node = new Search::HyperEstraier::Node; |
my $node = new Search::HyperEstraier::Node; |
752 |
|
|
753 |
|
or optionally with C<url> as parametar |
754 |
|
|
755 |
|
my $node = new Search::HyperEstraier::Node( 'http://localhost:1978/node/test' ); |
756 |
|
|
757 |
|
or in more verbose form |
758 |
|
|
759 |
|
my $node = new Search::HyperEstraier::Node( |
760 |
|
url => 'http://localhost:1978/node/test', |
761 |
|
debug => 1, |
762 |
|
croak_on_error => 1 |
763 |
|
); |
764 |
|
|
765 |
|
with following arguments: |
766 |
|
|
767 |
|
=over 4 |
768 |
|
|
769 |
|
=item url |
770 |
|
|
771 |
|
URL to node |
772 |
|
|
773 |
|
=item debug |
774 |
|
|
775 |
|
dumps a B<lot> of debugging output |
776 |
|
|
777 |
|
=item croak_on_error |
778 |
|
|
779 |
|
very helpful during development. It will croak on all errors instead of |
780 |
|
silently returning C<-1> (which is convention of Hyper Estraier API in other |
781 |
|
languages). |
782 |
|
|
783 |
|
=back |
784 |
|
|
785 |
=cut |
=cut |
786 |
|
|
787 |
sub new { |
sub new { |
799 |
}; |
}; |
800 |
bless($self, $class); |
bless($self, $class); |
801 |
|
|
802 |
if (@_) { |
if ($#_ == 0) { |
803 |
$self->{debug} = shift; |
$self->{url} = shift; |
804 |
warn "## Node debug on\n"; |
} else { |
805 |
|
my $args = {@_}; |
806 |
|
|
807 |
|
%$self = ( %$self, @_ ); |
808 |
|
|
809 |
|
warn "## Node debug on\n" if ($self->{debug}); |
810 |
} |
} |
811 |
|
|
812 |
$self ? return $self : return undef; |
$self ? return $self : return undef; |
1263 |
|
|
1264 |
my $rv = $self->shuttle_url( $self->{url} . '/search', |
my $rv = $self->shuttle_url( $self->{url} . '/search', |
1265 |
'application/x-www-form-urlencoded', |
'application/x-www-form-urlencoded', |
1266 |
$self->cond_to_query( $cond ), |
$self->cond_to_query( $cond, $depth ), |
1267 |
\$resbody, |
\$resbody, |
1268 |
); |
); |
1269 |
return if ($rv != 200); |
return if ($rv != 200); |
1357 |
|
|
1358 |
Return URI encoded string generated from Search::Estraier::Condition |
Return URI encoded string generated from Search::Estraier::Condition |
1359 |
|
|
1360 |
my $args = $node->cond_to_query( $cond ); |
my $args = $node->cond_to_query( $cond, $depth ); |
1361 |
|
|
1362 |
=cut |
=cut |
1363 |
|
|
1366 |
|
|
1367 |
my $cond = shift || return; |
my $cond = shift || return; |
1368 |
croak "condition must be Search::Estraier::Condition, not '$cond->isa'" unless ($cond->isa('Search::Estraier::Condition')); |
croak "condition must be Search::Estraier::Condition, not '$cond->isa'" unless ($cond->isa('Search::Estraier::Condition')); |
1369 |
|
my $depth = shift; |
1370 |
|
|
1371 |
my @args; |
my @args; |
1372 |
|
|
1376 |
|
|
1377 |
if (my @attrs = $cond->attrs) { |
if (my @attrs = $cond->attrs) { |
1378 |
for my $i ( 0 .. $#attrs ) { |
for my $i ( 0 .. $#attrs ) { |
1379 |
push @args,'attr' . ($i+1) . '=' . uri_escape( $attrs[$i] ); |
push @args,'attr' . ($i+1) . '=' . uri_escape( $attrs[$i] ) if ($attrs[$i]); |
1380 |
} |
} |
1381 |
} |
} |
1382 |
|
|
1394 |
push @args, 'options=' . $options; |
push @args, 'options=' . $options; |
1395 |
} |
} |
1396 |
|
|
1397 |
push @args, 'depth=' . $self->{depth} if ($self->{depth}); |
push @args, 'depth=' . $depth if ($depth); |
1398 |
push @args, 'wwidth=' . $self->{wwidth}; |
push @args, 'wwidth=' . $self->{wwidth}; |
1399 |
push @args, 'hwidth=' . $self->{hwidth}; |
push @args, 'hwidth=' . $self->{hwidth}; |
1400 |
push @args, 'awidth=' . $self->{awidth}; |
push @args, 'awidth=' . $self->{awidth}; |
1405 |
|
|
1406 |
=head2 shuttle_url |
=head2 shuttle_url |
1407 |
|
|
1408 |
This is method which uses C<IO::Socket::INET> to communicate with Hyper Estraier node |
This is method which uses C<LWP::UserAgent> to communicate with Hyper Estraier node |
1409 |
master. |
master. |
1410 |
|
|
1411 |
my $rv = shuttle_url( $url, $content_type, $req_body, \$resbody ); |
my $rv = shuttle_url( $url, $content_type, $req_body, \$resbody ); |
1415 |
|
|
1416 |
=cut |
=cut |
1417 |
|
|
1418 |
|
use LWP::UserAgent; |
1419 |
|
|
1420 |
sub shuttle_url { |
sub shuttle_url { |
1421 |
my $self = shift; |
my $self = shift; |
1422 |
|
|
1435 |
return -1; |
return -1; |
1436 |
} |
} |
1437 |
|
|
1438 |
my ($host,$port,$query) = ($url->host, $url->port, $url->path); |
my $ua = LWP::UserAgent->new; |
1439 |
|
$ua->agent( "Search-Estraier/$Search::Estraier::VERSION" ); |
1440 |
|
|
1441 |
if ($self->{pxhost}) { |
my $req; |
1442 |
($host,$port) = ($self->{pxhost}, $self->{pxport}); |
if ($reqbody) { |
1443 |
$query = "http://$host:$port/$query"; |
$req = HTTP::Request->new(POST => $url); |
1444 |
|
} else { |
1445 |
|
$req = HTTP::Request->new(GET => $url); |
1446 |
} |
} |
1447 |
|
|
1448 |
$query .= '?' . $url->query if ($url->query && ! $reqbody); |
$req->headers->header( 'Host' => $url->host . ":" . $url->port ); |
1449 |
|
$req->headers->header( 'Connection', 'close' ); |
1450 |
|
$req->headers->header( 'Authorization', 'Basic ' . $self->{auth} ) if ($self->{auth}); |
1451 |
|
$req->content_type( $content_type ); |
1452 |
|
|
1453 |
my $headers; |
warn $req->headers->as_string,"\n" if ($self->{debug}); |
1454 |
|
|
1455 |
if ($reqbody) { |
if ($reqbody) { |
1456 |
$headers .= "POST $query HTTP/1.0\r\n"; |
warn "$reqbody\n" if ($self->{debug}); |
1457 |
} else { |
$req->content( $reqbody ); |
|
$headers .= "GET $query HTTP/1.0\r\n"; |
|
1458 |
} |
} |
1459 |
|
|
1460 |
$headers .= "Host: " . $url->host . ":" . $url->port . "\r\n"; |
my $res = $ua->request($req) || croak "can't make request to $url: $!"; |
|
$headers .= "Connection: close\r\n"; |
|
|
$headers .= "User-Agent: Search-Estraier/$Search::Estraier::VERSION\r\n"; |
|
|
$headers .= "Content-Type: $content_type\r\n"; |
|
|
$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, |
|
|
); |
|
1461 |
|
|
1462 |
if (! $sock) { |
warn "## response status: ",$res->status_line,"\n" if ($self->{debug}); |
|
carp "can't open socket to $host:$port"; |
|
|
return -1; |
|
|
} |
|
1463 |
|
|
1464 |
warn $headers if ($self->{debug}); |
($self->{status}, $self->{status_message}) = split(/\s+/, $res->status_line, 2); |
1465 |
|
|
1466 |
print $sock $headers or |
if (! $res->is_success) { |
1467 |
carp "can't send headers to network:\n$headers\n" and return -1; |
if ($self->{croak_on_error}) { |
1468 |
|
croak("can't get $url: ",$res->status_line); |
1469 |
if ($reqbody) { |
} else { |
1470 |
warn "$reqbody\n" if ($self->{debug}); |
return -1; |
1471 |
print $sock $reqbody or |
} |
|
carp "can't send request body to network:\n$$reqbody\n" and return -1; |
|
1472 |
} |
} |
1473 |
|
|
1474 |
my $line = <$sock>; |
$$resbody .= $res->content; |
|
chomp($line); |
|
|
my ($schema, $res_status, undef) = split(/ */, $line, 3); |
|
|
return if ($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}); |
|
|
}; |
|
|
|
|
|
# read body |
|
|
$len = 0; |
|
|
do { |
|
|
$len = read($sock, my $buf, 8192); |
|
|
$$resbody .= $buf if ($resbody); |
|
|
} while ($len); |
|
1475 |
|
|
1476 |
warn "## response body:\n$$resbody\n" if ($resbody && $self->{debug}); |
warn "## response body:\n$$resbody\n" if ($resbody && $self->{debug}); |
1477 |
|
|
1549 |
} |
} |
1550 |
|
|
1551 |
|
|
1552 |
|
=head2 set_link |
1553 |
|
|
1554 |
|
Manage node links |
1555 |
|
|
1556 |
|
$node->set_link('http://localhost:1978/node/another', 'another node label', $credit); |
1557 |
|
|
1558 |
|
If C<$credit> is negative, link is removed. |
1559 |
|
|
1560 |
|
=cut |
1561 |
|
|
1562 |
|
sub set_link { |
1563 |
|
my $self = shift; |
1564 |
|
my ($url, $label, $credit) = @_; |
1565 |
|
|
1566 |
|
return unless ($self->{url}); |
1567 |
|
croak "mode credit be number, not '$credit'" unless ($credit =~ m/^\d+$/); |
1568 |
|
|
1569 |
|
my $reqbody = 'url=' . uri_escape($url) . '&label=' . uri_escape($label); |
1570 |
|
$reqbody .= '&credit=' . $credit if ($credit > 0); |
1571 |
|
|
1572 |
|
$self->shuttle_url( $self->{url} . '/_set_link', |
1573 |
|
'application/x-www-form-urlencoded', |
1574 |
|
$reqbody, |
1575 |
|
undef |
1576 |
|
) == 200; |
1577 |
|
} |
1578 |
|
|
1579 |
|
|
1580 |
=head1 PRIVATE METHODS |
=head1 PRIVATE METHODS |
1581 |
|
|
1582 |
You could call those directly, but you don't have to. I hope. |
You could call those directly, but you don't have to. I hope. |
1604 |
|
|
1605 |
return if ($rv != 200 || !$resbody); |
return if ($rv != 200 || !$resbody); |
1606 |
|
|
1607 |
chomp($resbody); |
# it seems that response can have multiple line endings |
1608 |
|
$resbody =~ s/[\r\n]+$//; |
1609 |
|
|
1610 |
( $self->{name}, $self->{label}, $self->{dnum}, $self->{wnum}, $self->{size} ) = |
( $self->{name}, $self->{label}, $self->{dnum}, $self->{wnum}, $self->{size} ) = |
1611 |
split(/\t/, $resbody, 5); |
split(/\t/, $resbody, 5); |