| 1 |
dpavlin |
1 |
package CouchDB::Estraier; |
| 2 |
|
|
|
| 3 |
|
|
use strict; |
| 4 |
|
|
use warnings; |
| 5 |
|
|
|
| 6 |
|
|
use Search::Estraier; |
| 7 |
|
|
use Data::Dump qw/dump/; |
| 8 |
|
|
|
| 9 |
|
|
=head1 NAME |
| 10 |
|
|
|
| 11 |
|
|
CouchDB::Estraier - Hyper Estraier full-text search for CouchDB |
| 12 |
|
|
|
| 13 |
|
|
=head1 METHODS |
| 14 |
|
|
|
| 15 |
|
|
=cut |
| 16 |
|
|
|
| 17 |
|
|
our $c = { |
| 18 |
|
|
node_url => 'http://localhost:1978/node/', |
| 19 |
|
|
debug => 0, |
| 20 |
|
|
estuser => 'admin', |
| 21 |
|
|
estpasswd => 'admin', |
| 22 |
|
|
quiet => 0, |
| 23 |
|
|
FullTextSearchQueryServer => 0, |
| 24 |
|
|
DbUpdateNotificationProcess => 0, |
| 25 |
|
|
}; |
| 26 |
|
|
|
| 27 |
|
|
|
| 28 |
|
|
=head2 run |
| 29 |
|
|
|
| 30 |
|
|
Process command line options and start helper |
| 31 |
|
|
|
| 32 |
|
|
CouchDB::Estraier->run; |
| 33 |
|
|
|
| 34 |
|
|
=cut |
| 35 |
|
|
|
| 36 |
|
|
sub run { |
| 37 |
|
|
my $self = shift; |
| 38 |
|
|
GetOptions($c, qw/node_url=s debug+ quiet+ estuser=s estpasswd=s dbuser=s dbpasswd=s/) or die $!; |
| 39 |
|
|
warn "# c: ", dump($c) if ($c->{debug}); |
| 40 |
|
|
if ( $c->{FullTextSearchQueryServer} ) { |
| 41 |
|
|
while ( 1 ) { |
| 42 |
|
|
my $database = <>; |
| 43 |
|
|
my $query_string = <>; |
| 44 |
|
|
chomp $database; |
| 45 |
|
|
chomp $query_string; |
| 46 |
|
|
$self->search( $database, $query_string ); |
| 47 |
|
|
} |
| 48 |
|
|
} else { |
| 49 |
|
|
while ( 1 ) { |
| 50 |
|
|
my $database = <>; |
| 51 |
|
|
my $query_string = <>; |
| 52 |
|
|
chomp $database; |
| 53 |
|
|
chomp $query_string; |
| 54 |
|
|
$self->add( $database, $query_string ); |
| 55 |
|
|
} |
| 56 |
|
|
} |
| 57 |
|
|
} |
| 58 |
|
|
|
| 59 |
|
|
|
| 60 |
|
|
=head2 add |
| 61 |
|
|
|
| 62 |
|
|
CouchDB::Estraier->add( $database, $data ); |
| 63 |
|
|
|
| 64 |
|
|
=cut |
| 65 |
|
|
|
| 66 |
|
|
sub add { |
| 67 |
|
|
my ( $self, $database, $data ) = @_; |
| 68 |
|
|
|
| 69 |
|
|
# create and configure node |
| 70 |
|
|
my $node = new Search::Estraier::Node( |
| 71 |
|
|
url => $c->{node_url} . $database, |
| 72 |
|
|
user => $c->{estuser}, |
| 73 |
|
|
passwd => $c->{estpasswd}, |
| 74 |
|
|
croak_on_error => 1, |
| 75 |
|
|
create => 1, |
| 76 |
|
|
debug => $c->{debug} >= 4 ? 1 : 0, |
| 77 |
|
|
); |
| 78 |
|
|
|
| 79 |
|
|
# create document |
| 80 |
|
|
my $doc = new Search::Estraier::Document; |
| 81 |
|
|
|
| 82 |
|
|
if (my $id = $data->{_id}) { |
| 83 |
|
|
$doc->add_attr('@uri', $id); |
| 84 |
|
|
} else { |
| 85 |
|
|
die "can't find pk_col column _id in results\n"; |
| 86 |
|
|
} |
| 87 |
|
|
|
| 88 |
|
|
while (my ($col,$val) = each %{$data}) { |
| 89 |
|
|
|
| 90 |
|
|
if ($val) { |
| 91 |
|
|
# add attributes (make column usable from attribute search) |
| 92 |
|
|
$doc->add_attr($col, $val); |
| 93 |
|
|
|
| 94 |
|
|
# add body text to document (make it searchable using full-text index) |
| 95 |
|
|
$doc->add_text($val); |
| 96 |
|
|
} |
| 97 |
|
|
|
| 98 |
|
|
} |
| 99 |
|
|
|
| 100 |
|
|
warn "# doc draft: ",$doc->dump_draft, "\n" if ($c->{debug} >= 2); |
| 101 |
|
|
|
| 102 |
|
|
die "error: ", $node->status,"\n" unless (eval { $node->put_doc($doc) }); |
| 103 |
|
|
|
| 104 |
|
|
} |
| 105 |
|
|
|
| 106 |
|
|
=head2 search |
| 107 |
|
|
|
| 108 |
|
|
Implementation specification: L<http://wiki.apache.org/couchdb/FullTextSearch> |
| 109 |
|
|
|
| 110 |
|
|
CouchDB::Estraier->search( $database, $query ); |
| 111 |
|
|
|
| 112 |
|
|
=cut |
| 113 |
|
|
|
| 114 |
|
|
sub search { |
| 115 |
|
|
my ( $self, $database, $query ) = @_; |
| 116 |
|
|
|
| 117 |
|
|
# create and configure node |
| 118 |
|
|
my $node = new Search::Estraier::Node( |
| 119 |
|
|
url => $c->{node_url} . $database, |
| 120 |
|
|
user => $c->{estuser}, |
| 121 |
|
|
passwd => $c->{estpasswd}, |
| 122 |
|
|
croak_on_error => 1, |
| 123 |
|
|
create => 1, |
| 124 |
|
|
debug => $c->{debug} >= 4 ? 1 : 0, |
| 125 |
|
|
); |
| 126 |
|
|
|
| 127 |
|
|
my $cond = new Search::Estraier::Condition; |
| 128 |
|
|
$cond->set_phrase( $query ); |
| 129 |
|
|
my $nres = $node->search($cond, 0); |
| 130 |
|
|
|
| 131 |
|
|
if ( defined($nres) ) { |
| 132 |
|
|
|
| 133 |
|
|
print "ok\n"; |
| 134 |
|
|
for my $i ( 0 ... $nres->doc_num - 1 ) { |
| 135 |
|
|
my $rdoc = $nres->get_doc($i); |
| 136 |
|
|
print $rdoc->attr('@uri'),"\n",$i,"\n"; |
| 137 |
|
|
} |
| 138 |
|
|
print "\n\n"; |
| 139 |
|
|
|
| 140 |
|
|
} else { |
| 141 |
|
|
print "error\n", $node->status, "\n"; |
| 142 |
|
|
} |
| 143 |
|
|
|
| 144 |
|
|
|
| 145 |
|
|
} |
| 146 |
|
|
|
| 147 |
|
|
|
| 148 |
|
|
1; |
Parent Directory
| 
Revision Log